AudioRail® Technologies:   It simply works.

Note: This web page is now deprecated, and is available only for historical reference. See http://www.audiorail.com

    - Conservative engineering
    - Patented, elegant simplicity of design
    - Complete accuracy and robustness
    - Profound cost effectiveness

Click to look inside the box

Click on pictures to enlarge:

ADAT rx32tx32 User Manual (PDF)

If this webpage is too technical,
Pricing details here:

$500 per
64-channel
point of
connection
In stock! U.S. domestic and international orders welcome.
European customers: Click here for WEEE and RoHS compliance information.
AudioRail customers looking for ADT1616 interface cards for the DDX3216? We have a few available. Click here.
Compatibility issue with the PreSonus DigiMax FS and JetPLL: Click here.

This website is committed to delivering straightforward information with no marketing hype. All correspondence should be directed to info@audiorail.com or telephone inquiries to (978) 461-0177.

Table of contents:

Info: Use Cat5 to fiber converters to extend AudioRail's reach.

Info: Behringer ADA8000 analysis

Introduction: The technology

AudioRail® is a simple and elegant, patented method for transporting many channels of digital audio over a daisy chain of standard Category 5 UTP (Unshielded Twisted Pair) cable or other LAN interconnect media using off the shelf Ethernet transceiver technology, forming a digital audio snake.

Ethernet physical layer transceivers contain the technology to transport 100 Mbps or more of full duplex data over inexpensive Category 5 or other conventional Ethernet cable. In the case of Fast Ethernet, 100 Mbps of bandwidth can easily accommodate the equivalent of 32 channels of analog audio in each direction simultaneously (64 channels aggregate) at a sample rate of 48 KHz (or half of that number of channels for 96 KHz sample rate) at 24 bit digital audio resolution. Instead of using conventional packet/frame store and forward methods, AudioRail streams time division multiplexed data directly in and out of the phy transceiver's MII, GMII, Serdes, or comparable off the shelf phy interface port, with only a few microseconds latency, end to end. more on latency

No software, firmware, DSP, or any kind of microprocessor intelligence is employed in the entire network audio mechanism.

AudioRail Technologies, a new startup venture, is developing a family of products based on this concept. more

The implementation is accomplished in modular devices that are strung together in a daisy chain of Category 5 cable connections. Each unit has optical or coaxial S/PDIF or other digital or analog audio connections and two RJ45 sockets. Adjacent units can be separated by up to 100 meters, essentially forming a digital audio snake. This distance can be increased using Cat5 to fiber converters

The digital audio transport latency is on the order of a few microseconds, end to end, which makes it essentially negligible, and ideally applicable for live sound applications. more on latency The system is simple and straightforward, requiring no software or computer control to transport the many digital audio channels from end to end. The system comes up and becomes completely functional within about two seconds after power is applied.

The system configuration is versatile, as any number of modular units can be daisy chained together. Each individual incoming audio port in the configuration can be routed in either direction and each outgoing audio port can be switched to tap off of the TDM stream from either direction. There can also be more than one tap to retrieve data from the same digital audio channel in the stream. Click here for a more detailed explanation.


Click here to see how this is useful.

Each digital audio stream is independent of the next, and does not require a common clock. Click here for an explanation from our manual of how to configure word clocks with AudioRail.

Each modular unit has two RJ45 connectors and a number of input and/or output channels. Units receiving audio inputs have switches to direct where to route the signal. Units generating audio outputs have switches to select where to get the signals from.

Category 5 cable costs only 4 to 6 cents per foot in bulk. For example: (no affiliation) Home Depot, CableMAX, Cablewholesale, Computer Gate, CablesDirect, QualityCables

Category 5 cable can be obtained from Radio Shack (Catalog #278-1583) for 15 cents a foot, by the foot. Radio Shack's crimp tool (Catalog #279-405) is a professional quality tool that we have found to be as good or better than some other industrial quality brands, and costs $31.99. Radio Shack's 8-pin modular plugs (Catalog #279-406) come in a package of 5 for $3.89. Ready-made 100 foot cable can be obtained from CompUSA, for example, for $30.

Our product user manual gives instructions on how to crimp cable ends.

The following pictorial illustrates the difference between the AudioRail networked audio transport mechanism and a conventional Ethernet LAN approach to transporting digital audio. AudioRail employs a continuous TDM (time division multiplexed) stream, instead of the traditional packet store and forward approach, making it orders of magnitude faster, more efficient, and less expensive. (Note: The diagram on the left is a generic block diagram and should not be construed as being an exact representation of any company's particular product on the market.)

Another way that this technology can be implemented to accomplish the same or similar result is to apply the technology so that other professional audio products have the essential design embedded within their products, interfacing to the outside world using the RJ45 or other "Ethernet" media connectors.

Our vision is for a future where digital audio components don't just continue to replace and displace their analog predecessors, but are also able to be inexpensively interconnected with the same sort of cabling and electromechanical interconnect that is now so widely available for local area networking.

We intend to reach that segment of the professional audio industry that would be eager to step up from conventional point to point analog audio snakes, analog building wiring, patch panels and cabling to a means of simply and economically transporting many channels of digital audio over a digital snake using standard off the shelf Category 5 cabling schemes. Both off the shelf Category 5 cable and installed Category 5 cable wiring can be used. Multiple devices can be placed in various locations as needed to pick up and deliver audio from one place to another over, each unit being separated from the next one by as much as 100 meters (or more, using fiber converters), a distance far exceeding the practical length that conventional analog audio cable can be employed, all the while retaining the advantage that noise-free CD-quality digital audio transmission provides.

As a functional by-product of the design, there can be multiple devices selected to get audio from the same originating sources. The technology is able to transport industry standard IEC 958 (S/PDIF), as well as Tascam TDIF, Alesis ADAT light-pipe, and other audio related signals, such as SMPTE time code, word clock and MIDI interfaces. Units with these digital audio connections can be directly attached to industry standard digital audio ports on industry products capable of digital audio connections, or can be front ended with any digital to analog and analog to digital converter products of the user's choice to accomplish conventional analog connections. Off-the-shelf digital to analog and analog to digital converters, such as the $230 Behringer ADA8000 (see our analysis) that provides 8 mic/line inputs and 8 line outputs, bring the cost of conversion to/from analog to $14.38 per port.

AudioRail products are very user friendly and trivial to set up and operate. The important digital audio sample rates (in the first implementation, 44.1 KHz, 48 KHz, and 96 KHz) are autosensed. The only essential user controls are front panel switches to choose which direction and/or which group of TDM channels to route the individual digital audio streams. There is no software required and nothing else to configure.

State of progress: Optical S/PDIF (AES3/IEC958) transport

The logic design for conventional digital audio (AES3/IEC958) transport over 100Base-T Ethernet transceivers was completed, synthesized, and fits in a mating pair of 100K gate FPGAs (Altera EPF20K100ETC144-2, -3). Hardware simulation was completed which simulated the operation of eight adjacent devices routing digital audio signals through with the simulation checking the integrity of the data end to end. Prototype hardware has been designed, built, debugged, and is completely functional. Extensive signal integrity testing with a high-end, 2 GHz bandwidth oscilloscope, logic analyzer, and end-to-end data integrity checks have been performed. Data is transported without error over multiple hops of Category 5e cable of 100 meters or more. Toslink to Toslink plastic optical fiber (POF) links function with margin up to 72 feet (well over 2x the 10 meter maximum spec for POF) at 48K and up to 42 feet at 96K, using the cheapest low end plastic fiber optic cable generally available on the market.

The most extensive testing was done with a single 96K digital audio feed that was routed through a group of four adjacent units, each separated by 100 meters or more of cable, then looped back around through the system 16 times using cheap 18 ft. fiber optic cables. In doing so, the digital signal traveled repeatedly through the system, thereby traversing over 10,500 feet of Category 5e cable and over 300 feet of fiber optic cables, before it finally reached its destination, completely error free.

State of progress: Alesis ADAT Lightpipe transport

This product is now shipping.

A significant feature of Alesis ADAT Lightpipe is that each fiber connection already time division multiplexes eight channels of audio to begin with. A single AudioRail device with four optical Toslink receivers and four optical Toslink drivers can therefore send the equivalent of 32 channels of audio in each direction simultaneously.

Because of this format provides the most cost effective implementation, this product has been chosen to be the flagship product.

The logic design for transporting four transmit-receive pairs of Alesis ADAT light-pipe connections was coded, synthesized, and fits in a single 100K gate FPGA (Altera EPF20K100ETC144-2). Simulation was completed which successfully transports eight unidirectional streams (four bidirectional pairs) with end to end checking. Prototype hardware was designed, built, debugged and is completely functional. 44.1K and 48K data is transported without error over multiple hops of Category 5e cable of 100 meters or more. Those familiar with the ADAT protocol will appreciate the fact that input streams with up to 60 ns p-p jitter have been successfully tracked entirely in the digital domain without employing any PLLs. Given that the unit interval (UI) time is 81 ns, that corresponds to an eye pattern with only about 25% eye opening.

The most extensive testing was done with a single Alesis ADAT 48K digital audio feed that was routed through a daisy chain of six units, the end units being Alesis ADAT units, with four other units in between, spanning three sections of 100M Category 5e cable (over 1000 ft total) and two shorter cables. The Alesis ADAT Lightpipe signal was looped back around using optical cables eight times using cheap 12 ft fiber optic cables. In doing so, the digital signal traveled repeatedly through the system, thereby traversing over 8000 feet of Category 5e cable and 108 feet of fiber optic cables, before it finally reached its destination, completely error free.

Prototypes have been deployed successfully and without incident regularly in live sound reinforcement (configuration).

We completed regulatory approvals and are now starting volume production builds. (See list/explanation of regulatory standards and related FAQ item)

Other work in progress:

Analog front-end interfaces are also in design, as well as interfaces for coaxial S/PDIF, Tascam TDIF, word clock, MIDI, and SMPTE time code interfaces.

Latencies

Specific latencies measured were:

44.1K S/PDIF optical Toslink-Toslink = 6.5 us
48K S/PDIF optical Toslink-Toslink = 6 us
96K S/PDIF optical Toslink-Toslink = 4 us

44.1K Alesis ADAT optical Toslink-Toslink = 4.4 us
48K Alesis ADAT optical Toslink-Toslink = 4 us

These were between two adjacent units with a very short Category 5 cable.

Each additional hop from device to adjacent device incurs only about 300 ns (0.3 us) delay, and additional Category 5 cable delay is about 5 ns (0.005 us) per meter of cable. (Note: The several microsecond "Toslink-Toslink" delay is not compounded from hop to hop. That delay is incurred only once, and is associated with the encoding and decoding of the digital audio stream at each end.)

In non-technical terms, think of the speed of sound as being on the order of 1000 feet per second (roughly). A "microsecond" ("us") is one millionth of a second. If the AudioRail delay were 10 us, then this would correspond to sound traveling roughly one hundredth of a foot, which would be roughly a tenth of an inch. In live sound production, to compensate for this delay, one would only need to move the speakers one tenth of an inch forward from where they would have been if AudioRail had not been employed and straight digital audio cable connections were substituted instead. This is what we mean by "negligible."

Engineering documentation and testing

Circuit schematics, parts list, PCB layout, photographs

The first prototypes consisted of a board that accepts eight channels of Toslink optical S/PDIF digital audio, a board that delivers eight channels of Toslink optical S/PDIF digital audio, and a board that has four pairs of Alesis ADAT Lightpipe optical connections. The following are the circuit schematics and screen shots of the PCB artwork, which was done on a 4-layer board (top and bottom signal routing layers, with power and ground inner layers). In the artwork screen shots, the white dots are spaced at a 1 inch grid, to provide perspective of the size.

Both the circuit schematics and PCB layout screen shots show the simplicity of the AudioRail design at the board level. All of the AudioRail know-how is really contained in a single chip for each board, which in this case is an Altera FPGA.

Schematics of board with 4 ADAT Lightpipe interface pairs.
PDF of PCB artwork of board with 4 ADAT Lightpipe interface pairs.
High resolution photograph showing inside of assembly.
User Manual

Schematics of board with 8 S/PDIF Toslink receivers.
Schematics of board with 8 S/PDIF Toslink transmitters.
Screen shot of PCB artwork of board with 8 S/PDIF Toslink receivers.
Screen shot of PCB artwork of board with 8 S/PDIF Toslink transmitters.
Photograph of S/PDIF board in unpainted enclosure.
Photograph of S/PDIF logic components and RJ45 interface.

Note 1: The Toslink receiver outputs in the S/PDIF version are now each buffered by a 74LVC1G04, not yet reflected in the schematic or PCB artwork screen shots, for improvement of signal integrity. These are reflected in the ADAT Lightpipe version.

Note 2: The S/PDIF versions shown above will probably not be the initial S/PDIF offering. The initial S/PDIF product will likely be a single, exact electrical/mechanical replica of the Alesis ADAT Lightpipe version, coded for S/PDIF. Consequently, it would have 4 Toslink inputs and 4 Toslink outputs. This decision would be for the purpose of keeping development costs down.

The following is the bill of materials showing all the parts required to build a mating pair of S/PDIF assembled units (one with 8 Toslink receivers and one with 8 Toslink transmitters), and the bill of materials for the ADAT Lightpipe version, which is one unit with 4 pairs of ADAT lightpipe connections. It is important to point out that this hardware was designed very conservatively, using older, state-of-the-shelf parts, to minimize sourcing risks and ensure a trouble-free path from proof of concept to volume production. There are areas where the cost and number of components can be reduced. Intel offers a fairly new dual 10/100 Ethernet transceiver chip (the LXT973). RJ45 connectors are available with integrated magnetics. Using Altera's more recently introduced Cyclone FPGA family would significantly reduce the cost of the FPGA chips, as compared to the older APEX 20KE family now being employed. Reduction in the number of discrete passive components is also possible.

Parts list for ADAT Lightpipe version: PDF
Parts list for ADAT Lightpipe version: XLS
Parts list for ADAT Lightpipe version (RoHS): PDF
Parts list for ADAT Lightpipe version (RoHS): XLS

Parts list for mating pair of S/PDIF interface units (transmit + receive): PDF
Parts list for mating pair of S/PDIF interface units (transmit + receive): DOC

The following is the 3-dimensional mechanical design for the ADAT Lightpipe version enclosure.

3-dimensional mechanical model for ADAT Lightpipe version: DXF
front and rear panel silkscreens for ADAT Lightpipe version: PDF

The flagship product is the ADAT Lightpipe version, as it provides the most cost effective solution for digital audio transport.

Regulatory approvals

AudioRail products are tested at accredited test sites in accordance with the following regulatory standards (see related FAQ item):

The following are the test reports from accredited regulatory test sites for the ADAT Lightpipe version:

Source and destination Fast Fourier Transforms

Posted here are some of the highlights of the extensive testing done to ensure that data is transferred over AudioRail error free. Unless otherwise noted, the optical S/PDIF test configuration was four adjacent AudioRail devices separated by 1000 feet of Category 5e cable that was cut into three sections, with a 96K 24-bit digital audio signal looped around the unit 16 times with cheap 18 ft plastic optical fiber cables. The Alesis ADAT Lightpipe test configuration added an additional pair of end nodes to the above configuration, with the optical signal looped back around the units 8 times with cheap 12 ft plastic optical fiber cables.

Audio was generated from a WAV file editor, then recorded at its destination. An FFT was done on each. Clicking on any of the following links will download zip files containing the full resolution Windows Bitmap (BMP) images.

For optical S/PDIF:

27.5 Hz sine wave at 96K for 10 seconds: Source
27.5 Hz sine wave at 96K for 10 seconds: Destination

440 Hz sine wave at 96K for 10 seconds: Source
440 Hz sine wave at 96K for 10 seconds: Destination

7040 Hz sine wave at 96K for 10 seconds: Source
7040 Hz sine wave at 96K for 10 seconds: Destination

White noise at 96K for 10 seconds: Source
White noise at 96K for 10 seconds: Destination

White noise at 96K for 10 seconds (linear scale): Source
White noise at 96K for 10 seconds (linear scale): Destination

Pink noise at 96K for 10 seconds: Source
Pink noise at 96K for 10 seconds: Destination

Pink noise at 96K for 10 seconds (linear scale): Source
Pink noise at 96K for 10 seconds (linear scale): Destination

For Alesis ADAT Lightpipe:

27.5 Hz sine wave at 48K for 10 seconds: Source
27.5 Hz sine wave at 48K for 10 seconds: Simple cable loopback
27.5 Hz sine wave at 48K for 10 seconds: Through AudioRail

440 Hz sine wave at 48K for 10 seconds: Source
440 Hz sine wave at 48K for 10 seconds: Simple cable loopback
440 Hz sine wave at 48K for 10 seconds: Through AudioRail

7040 Hz sine wave at 48K for 10 seconds: Source
7040 Hz sine wave at 48K for 10 seconds: Simple cable loopback
7040 Hz sine wave at 48K for 10 seconds: Through AudioRail

White noise at 48K for 10 seconds (linear scale): Source
White noise at 48K for 10 seconds (linear scale): Simple cable loopback
White noise at 48K for 10 seconds (linear scale): Through AudioRail

Pink noise at 48K for 10 seconds: Source
Pink noise at 48K for 10 seconds: Simple cable loopback
Pink noise at 48K for 10 seconds: Through AudioRail

Pink noise at 48K for 10 seconds (linear scale): Source
Pink noise at 48K for 10 seconds (linear scale): Simple cable loopback
Pink noise at 48K for 10 seconds (linear scale): Through AudioRail

Note: FFT comparisons show AudioRail to perform as well or better than just a simple fiber optic cable in its place. For this test, an inexpensive (MSRP $199) PC sound card, the Event EZ8, was used. This sound card uses the standard Alesis Semiconductor AL1401A and AL1402 integrated circuits for encoding and decoding the ADAT Lightpipe protocol.

Live sound configuration, tested with analog paths, using PC sound cards for 8 analog in/out:

Channels 1/2 out (110/220/440 Hz mix): Through AudioRail
Channels 1/2 out (110/220/440 Hz mix): Eliminate AudioRail

Channels 3/4 out (110/220/440 Hz mix): Through AudioRail
Channels 3/4 out (110/220/440 Hz mix): Eliminate AudioRail

Channels 5/6 out (55/110/220/440/880/1760/3520/7040 Hz mix): Through AudioRail
Channels 5/6 out (55/110/220/440/880/1760/3520/7040 Hz mix): Eliminate AudioRail

Channels 7/8 out (55/110/220/440/880/1760/3520/7040 Hz mix): Through AudioRail
Channels 7/8 out (55/110/220/440/880/1760/3520/7040 Hz mix): Eliminate AudioRail

Note: FFT comparisons show the system with AudioRail to perform essentially the same as when AudioRail is removed from the configuration (i.e. ADAT lightpipe cables connecting A/D and D/A conversion boxes directly to digital mixing console). Keep in mind that the noise floor of approximately -80 dB is due to multiple analog components being in the path, which included 1/4" unbalanced "instrument cable" connections. Also compare channels 5/6 with 7/8, which both represent exactly the same mix, in the "Eliminate AudioRail" scenarios, to see the lack of ability to perfectly correlate two different paths/scans that should theoretically be the same. Color graded FFT time vs. frequency scans (not shown) were also taken to search for noise events. None were seen above the existing noise floor. This live sound configuration is currently being employed on a regular basis for a 3 hour gig per week, without incident. (By "without incident" we mean that we have yet to experience a single glitch).

Jitter measurements and susceptibility

Of particular concern in equipment employing digital audio is cumulative and cycle to cycle jitter in the digital audio bitstream. Although each bit may be clocked in correctly with digital accuracy, the point of time when each data bit is clocked in may cause variations in the exact point that each digital sample is reconstructed at its destination. These variations in time can theoretically translate into sonic aberrations. And if jitter is of sufficient magnitude, samples can be dropped due to the inability of the receiving circuitry to track the signal.

AudioRail's receiving circuitry re-clocks the incoming signal from the optical receivers, so as to completely remove short term jitter, and has additional mechanisms in place to deal with long term jitter.

The FFT plots in the preceding section would show sonic degradation if there was any significant long-term cumulative jitter introduced by AudioRail into the audio spectrum.

On a cycle by cycle basis, short term jitter is limited for the IEC958 professional audio format to 0.25 UI ("Unit Interval"), where "Unit Interval" is the quantum in the raw digital audio bitstream. For 96K digital audio 0.25 UI is 20.3 ns, and for 48K digital audio it is 40.7 ns. (Alesis does not specify jitter characteristics for compliance to their ADAT Lightpipe specification.)

The optical S/PDIF and Alesis ADAT Lightpipe versions of AudioRail use the fastest Toslink components available from Toshiba, the TORX141 and TOTX141 (see Toshiba product table and product brochure). Components of this speed were only recently introduced by Toshiba and Sharp Microelectronics to meet market demand for high (88.2K/96K) transfer rates associated with DVD and other products. Most professional audio products on the market today still use the slower devices.

For optical S/PDIF, the devices consume all of the 0.25 UI margin (at the specified maximum 10 meter POF length), leaving none for the designs which employ them. Consequently, no product on the market today can claim conformance to the specification, on paper. Fortunately, testing has shown that the Toslink devices in practice operate far better than specified. This, combined with good high-speed digital design techniques that were employed in the AudioRail products have yielded plenty of margin.

Although Alesis has not specified a jitter value corresponding to their ADAT Lightpipe specification, they have cited the older Toshiba and Sharp optical components, which are rated at half the speed as those being employed in AudioRail.

The following screen shots were taken with a 2 GHz bandwidth Tektronix TDS794D digitizing oscilloscope of a mating Toslink pair of devices with various lengths and quality of optical cable between them. The stimulus was a 25 MHz clock source, rather than a digital audio stream. The jitter screen shots show the transitions one cycle after the transition (rising OR falling) that triggered the sweep. The persistence is infinite, color graded over approximately 10 million sweeps for 1 minute. Horizontal scale is 1 ns per division. Vertical scale is 0.5V per division. 0V reference is one division from the bottom (arrow on left side). Trigger voltage is 1.5 V (arrow on right side). Signals are LVTTL (low threshold 0.8V, high threshold 2.0V).

Signals driving Tx Toslink. (not 25 MHz source)
Signal driven from Rx Toslink (at trigger).
Rx Toslink jitter with ILS200 Monster Cable (1 meter).
Rx Toslink jitter with ILS100 Monster Cable (1 meter).
Rx Toslink jitter with un-named 2 meter cable from Circuit City.
Rx Toslink jitter with Steren 260-012 (4 meter, $6).
Rx Toslink jitter with NXG brand NX-706 (6 meter, $14).
Rx Toslink jitter with two 4m cables plus Radio Shack #M5-1583 coupler.

Note: All oscilloscope screen shots that follow (dated 1/22/2003 and later) are with a 1 GHz bandwidth Tektronix TDS784D digitizing oscilloscope.

The following screen shot shows AudioRail receiving looped back 96K digital audio data that has traversed 42 feet of the cheap fiber optic cable (33 feet is the Toslink design limit), including 3 cheap Radio Shack (#M5-1583) mechanical Toslink couplers, with one cable not inserted into its coupler all the way. The cursors show the 20.3 ns jitter spec. that has been exceeded (dark blue dots outside of that zone). It also traversed the grand journey 16 times through AudioRail with maximum length (100 m) Cat 5 cables. The source was a 48KHz square wave for 1 minute. The result at the other end was recorded as a text file, so that the entire waveform could be searched for any bad samples. There were none.

Rx Toslink jitter at 96K with 42 feet of cable.
Resulting "WAV" textfile with no errors. Caution: ZIP file is 192KB, but size of textfile within is 80MB in size

The following screen shot shows AudioRail receiving looped back 48K digital audio data that has traversed 72 feet of the cheap fiber optic cable (33 feet is the Toslink design limit), including 3 cheap Radio Shack (#M5-1583) mechanical Toslink couplers, with one cable not inserted into its coupler all the way. The cursors show the 40.7 ns jitter spec. that has been exceeded (dark blue dots outside of that zone). It also traversed the grand journey 16 times through AudioRail with maximum length (100 m) Cat 5e cables. The source was a 24KHz square wave for 1 minute. The result at the other end was recorded as a text file, so that the entire waveform could be searched for any bad samples. There were none.

Rx Toslink jitter at 48K with 72 feet of cable.
Resulting "WAV" textfile with no errors. Caution: ZIP file is 96KB, but size of textfile within is 40MB in size

The following screen shot shows AudioRail receiving 48K Alesis ADAT Lightpipe data that has traversed 48 feet of cheap fiber optic cable (33 feet is the Toslink design limit), including 3 cheap Radio Shack (#15-1583) mechanical Toslink couplers. This puts it just at the threshold of error. The cursors show 61 ns of peak to peak jitter, out of a unit interval time of only 81.38 ns, which yields an eye closure value of 75% (eye opening of only 20 ns). It also traversed the grand journey 8 times through six AudioRail nodes with a total of over a 1000 feet of Cat 5e cables. The sources were 4 stereo channels of different varying square wave patterns, all running for 1 minute, that were optimized by trial and error to yield the maximum jitter that they could. The result at the other end was recorded as a group of 4 text files, so that the entire waveform could be searched for any bad samples. There was one.

Note: The threshold was intentionally pushed to the point of error for illustrative purposes. It is not a statistical function of cable length. If the fiber optic cable length and couplers are reduced slightly, the error rate goes completely to zero.

Rx Toslink jitter, 48K square waves with 48 feet of cable.
Resulting "WAV" textfiles with 1 error total. Caution: ZIP file is 2.3MB, but size of textfiles within total over 200 MB in size

The following screen shots show AudioRail receiving 48K Alesis ADAT Lightpipe data consisting of four stereo pairs of sine waves, at frequencies 27.5 Hz, 55 Hz, 110 Hz, and 220 Hz, for a total of one minute. The same AudioRail test configuration is used as in the previous example. Since the source material consists of low frequency sine waves, FFT scans, rather than "WAV textfiles" are performed to validate data. Any bad bits show up as single sample anomalies that are easily spotted in an FFT scan that covers frequencies on a linear vertical scale up to the Nyquist limit of 24 KHz. In this example, the jitter is 60 ns, and there is one error occurring in each channel. The error is shown to illustrate the relative sensitivity of the accompanying FFT scans. The FFT scans were done using Blackmann-Harris windowing of 8192 sample buckets, with color graded intensity on the vertical scale (frequency) of linear values plotted 5000% of full scale.

Note: The threshold was intentionally pushed to the point of error for illustrative purposes. It is not a statistical function of cable length. If the fiber optic cable length and couplers are reduced slightly, the error rate goes completely to zero.

Rx Toslink jitter, 48K sine waves with 48 feet of cable.
Channels 1/2: 27.5 Hz original
Channels 1/2: 27.5 Hz AudioRail
Channels 1/2: 27.5 Hz error shown (too small to see)
Channels 3/4: 55 Hz original
Channels 3/4: 55 Hz AudioRail
Channels 3/4: 55 Hz error shown
Channels 5/6: 110 Hz original
Channels 5/6: 110 Hz AudioRail
Channels 5/6: 110 Hz error shown
Channels 7/8: 220 Hz original
Channels 7/8: 220 Hz AudioRail
Channels 7/8: 220 Hz error shown

Similarly, the following screen shots show AudioRail receiving 44.1K Alesis ADAT Lightpipe data. The jitter is 65 ns p-p, and there are no errors.

Rx Toslink jitter, 44.1K sine waves with 48 feet of cable.
Channels 1/2: 27.5 Hz original
Channels 1/2: 27.5 Hz AudioRail
Channels 3/4: 55 Hz original
Channels 3/4: 55 Hz AudioRail
Channels 5/6: 110 Hz original
Channels 5/6: 110 Hz AudioRail
Channels 7/8: 220 Hz original
Channels 7/8: 220 Hz AudioRail

Note: We do NOT recommend using long fiber optic cable lengths with Alesis ADAT Lightpipe connections. (Neither does Alesis.) The jitter in products using ADAT Lightpipe is many times the magnitude of conventional optical S/PDIF connections, and long cables aggravate it. This is not fundamentally the fault of either Alesis or other vendors employing the ADAT Lightpipe interface. It is due to the greater variability in data patterns in the protocol, as compared to that using optical S/PDIF.

We recommend keeping cable lengths for Alesis ADAT Lightpipe connections short, and using good quality cable.

The higher jitter in Alesis ADAT Lightpipe should also not be considered a threat to audio fidelity for several reasons. First, the jitter is due to duty cycle distortion that contributes frequency components that are not less than about 1 MHz. Also, the mechanisms employed to recover the signals use PLLs (or the digital equivalent of PLLs) that have loop time constants that are orders of magnitude less, effectively filtering that jitter out. In the AudioRail implementation, synchronization to the signal being decoded occurs only once per frame (i.e. at the sample rate). And the data re-encoded at the far end knows nothing even of this jitter component, because it is absorbed in elasticity buffering along the way.

The issue, then, is whether the data can be accurately tracked without mis-clocking a bit in (i.e. altering a bit). Artifacts of this sort, when they do appear, surface as harsh static and crackling noise, rather than subtle sonic variations, and are very easy to spot. We have observed varying results with different vendors' ADAT Lightpipe equipped devices with respect to jitter tolerance.

AudioRail has been tested at 96K, 48K, and 44.1K rates with a host of consumer and professional audio products on hand, including sound cards, digital audio converters, and discrete standalone digital audio components.

Power margins

The DC (3.3 Volt) current draw of the receive S/PDIF version is 0.46 A, the transmit S/PDIF version is 0.64 A, and the ADAT version is 0.63 A. The AC/DC switching power supply inside the S/PDIF unit is capable of delivering 10.7 A maximum, and 8.0 A continuous current. With this margin, things are barely getting warm inside the box. A smaller power supply was chosen for the ADAT Lightpipe version, bringing it down to a rating of 5.0 A maximum continuous current. It is still barely getting warm.

Timing margins and signal integrity

Successful compilations of the Altera FPGA for each of the units that contains the AudioRail logic design have been yielding maximum clock rates of around 65-70 MHz for for the FPGA used in the 50 MHz receive design and 108-113 MHz for the FPGA used in the 100 MHz S/PDIF transmit design and Alesis ADAT design.

Signal integrity checks with a 2 GHz bandwidth oscilloscope showed about 0.3V 0.3V overshoot/undershoot on the FPGA to Toslink connection (see the Signals driving Tx Toslink. screen shot above), which is not high enough to activate the rail clamping diodes inside the semiconductor devices, and all other signals are better than that, most of which have no overshoot or undershoot at all.

All clock signals have monotonic rising and falling edges, and there are no setup/hold violations on paper or observed in the lab.

50 MHz clock screen shot
100 MHz clock screen shot.

Note from the schematics and PCB etch screen shot that all signals are series terminated at the source.

Logic design method

The logic design consists entirely of Mealy/Moore sequential state machines and logic with no additional combinatorial logic, and was written in VHDL. All asynchronous and isochronous inputs are front-ended with synchronizers.

"Unexpected events" testing and consequential results

A substantial amount of time was spent pulling out and plugging back in fiber and Category 5 cables, flipping switches, and power cycling units live to make sure that the AudioRail components do not lock up. In all cases, normal operation is restored in less than 2 seconds from the point that the connections or power are restored.

If the fiber Toslink connections are pulled out or inserted slowly enough (deliberately much slower than they ever would in actual practice), there is a point at which the digital audio stream will waver between losing and regaining synchronization so as to cause serious and sustained noise in the analog realm. Other than that, the system's audible response to live insertion and removal of connections and power cycling events is similar to what one would expect in traditional analog sound systems when they are power cycled live or have cables inserted or pulled live: A single audible click, pop, or crackle noise.

AudioRail business and engineering FAQs

The questions most asked:

The "AudioRail" scheme

Competing schemes

Costs and warranty

Business

Are you selling them, what is the track record, and are your customers happy?

We have been selling the AudioRail ADAT rx32tx32 both domestically and internationally since late February of 2004. We have employed them ourselves in live sound production regularly since June of 2003. Our customers are happy with this product and the feedback we have been getting is very good.

This website documents more technical information about our design and testing than any other company that we know of in this field, including schematics, parts lists, PCB layout, and other information. Our design methods are conservative and straightforward in every way. We highlight the simplicity, rather than complexity or sophistication, of our approach to transporting digital audio, which means that there are fewer things that can go wrong. Of particular interest to those in professional audio is that our entire design is contained on one PC board (apart from the power supply), has few components, and has no cabling harnesses besides that connecting to the power supply. The power supply, a frequent point of failure in many industry products, is in our case a switching power supply that is eight times over-rated for the amount of current it is supplying, and so barely gets warm inside the product enclosure.

Click here for a high-resolution photograph showing what is inside the box.

Results of regulatory testing, in particularly the area of EMC immunity, document our product being subjected to thousands of volts of a.c. line surges, ESD hits, and induced Cat5 cable hits, such that we are qualified to offer our product to customer applications classified as "Heavy industrial environment and environments close to broadcast transmitters" (EN55103-2, category E5) in terms of immunity (as well as category E1 "Residential environment" in terms of the most stringent product RF emission requirements).

The product is built into a steel frame with aluminum front and back panels that is bolted all around. One unit was pounded and dropped repeatedly during required regulatory safety testing (associated with requirements that a unit not break open and expose a user to hazardous voltages, but without any regard as to whether the unit would functionally survive the test). We tightened a few loose screws and have continued to use this unit in routine lab testing since December 2003, and have deployed it in critical live musical theater performances with paying public audiences where failure would have been a show stopper.

These test results from A2LA accredited regulatory test facilities are posted right here on our website (also see related FAQ item).

All of our units are carefully inspected personally by us after final assembly, and are tested against a standard of having an error rate of zero, which is what it is designed for.

Given all of the above, we are confident in the reliability of our product, and we have no hesitation in offering a 5 year transferable warranty.

Also see the related FAQ Why should we trust you or your products?

The bottom line is that we feel we can offer potential customers the assurance that AudioRail is a "safe" investment, so that they can focus on determining whether AudioRail is the "right" investment for their particular sound application.

Is the AudioRail scheme compatible with Ethernet?

Electrically, yes. In terms of protocol and interoperability, not at all.

What happens if you plug an AudioRail component into a local area network, or vice versa?

First of all, the electrical signals and pin to pin connections are the same. You will do no damage, electrically or physically, by doing so. It is the protocols, that is, the format of the data and the way it is packaged and transferred, that are mutually incompatible.

Plugging an AudioRail component into a local area computer network will cause the local area network to stop working for as long as the AudioRail component is plugged into it. AudioRail will essentially hog all the bandwidth of the local area network with its unending TDM stream, preventing any of the local area network components from successfully transferring any information. The local area network will resume normal operation as soon as the AudioRail component is removed.

Plugging a local area network or any LAN component into AudioRail will corrupt the AudioRail stream for as long as it is plugged in. The LAN components will repeatedly try to transmit packets into the AudioRail system, corrupting it. It is possible that this will cause a horrible noise on the audio end of things, if and whenever the garbled LAN packets are interpreted by AudioRail to be legitimate. The AudioRail system will resume normal operation as soon as the LAN components are removed.

Why did you not make AudioRail compatible with Ethernet?

The elegant simplicity and low cost of AudioRail is obtained largely by throwing out all the unnecessary overhead associated with Ethernet LAN transport mechanisms. This was the intent of the original concept, not a compromise or oversight. The simple, straightforward architecture of AudioRail does not leave any room for LAN compatibility.

There are other, higher-end products on the market that are compatible with Ethernet LAN. Peak Audio's CobraNet is one example. Gibson's MaGIC is another. Digigram's Ethersound is yet another. These are good solutions for those who need LAN compatibility and require the advanced features that they provide. They are also much more expensive and all but Digigram's Ethersound products incur latency that is substantially greater.

Does AudioRail convert between digital audio formats?

AudioRail does not presently convert between digital audio formats. For example, you cannot put Alesis ADAT Lightpipe in one end and get S/PDIF out the other end. It also does not do sample rate conversion. For example, you cannot put 44.1K audio in one end and get 48K audio out the other end.

Can different AudioRail products interoperate on the same cable?

Most definitely yes. The AudioRail mechanism recognizes the difference between different formats (SPDIF/AES3, ADAT optical, TDIF, MIDI, SMPTE, etc.) and transports them all on the same CAT5 cable. One does not interfere with the other. Each format simply takes its share of the available bandwidth of the cable. As different AudioRail family products are introduced, the intention is that they all interoperate with each other on the same wire.

Do your products require CAT5, or CAT5E cable?

The current products are based on 100 Mbps Ethernet technology, which only requires CAT5 cable. This is a significant advantage for users employing already existing CAT5 rated cabling, and moot for new cable purchases, since CAT5 cable is not being made anymore (CAT5E has replaced it). 100 Mbps Ethernet employs a 125 MHz baseband signal. CAT5 typically does not accommodate over 250 MHz, whereas CAT5E is typically rated to 350 MHz. The difference is only in the stricter tolerances and twist requirements used in manufacturing the twisted-pair cable. CAT5 was marginal for Gigabit Ethernet, which employs 250 MHz baseband duplex signaling with DSP driven recovery techniques. CAT5E is required for Gigabit Ethernet, and will be required for Gigabit AudioRail, when it is eventually introduced.

What about CAT3 cable? Will it work?

No. CAT3 cable is rated to only 16 MHz and is good for 10 Mbps Ethernet at best. It will not work with AudioRail, which employs technology corresponding to 100 Mbps Ethernet.

Can AudioRail CAT5 cable hops exceed 100 meters (328 feet)?

We are conservatively specifying 100 meters (328 feet) as the maximum length per hop. The actual maximum length could be more, we have tested to more, and other companies may claim more in their CAT5-based products. The practical limit above 100 meters depends on the quality of cable used, as well as other factors such as how well building wall outlets and patch panels are constructed.

We recommend that fiber converters be used to extend the range of AudioRail beyond 100 meters per hop, which then allows vast distances to be achieved. Click here for more information.

Note that two fiber converters connected with a short, 1 meter fiber cable can also function as a simple CAT5 cable repeater. Since the distance from each AudioRail unit to each fiber converter can be up to 100 meters, the effective reach is immediately doubled to 200 meters in this way.

How susceptible is the CAT5 cable connection to picking up noise?

The CAT5 cable connection is far superior to a conventional analog connection. It is hard to quantify the difference, because there are so many factors involved. To give a qualitative feel for the difference, consider the following points: The above is, admittedly, quite a bit of hand-waving. Again, it is hard to quantify the difference, other than to say that it is far more robust than conventional analog. See related FAQ describing accredited regulatory immunity testing, description of regulatory standards that AudioRail complies with, and further information in the user manual.

How durable is a CAT5 cable, and how susceptible is it to electrical degradation?

It is anticipated that most long CAT5 cable runs will consist of cable bought in bulk that is intended for installed wiring. It consists of solid, not stranded internal wires, so is not quite as flexible as analog cabling. Still, we have found that it is very manageable and flexible enough for repeated use.

CAT5 cable isn't as durable as off the shelf microphone, instrument, and speaker cables that are designed for professional audio applications. It doesn't have the thick rubberized protective outer insulation and flexibility. For this reason, a bit more care must be taken in portable sound venues to protect it from damage. However, because the cable is a single thin one, and can be run for longer lengths, it lends itself better to be run out of the way and tucked out of the way. It can be covered by a carpet or cable protection products much more easily than an analog snake. Obviously, we would caution against just throwing it down through the center of an auditorium to be trampled on or tripped over by the masses.

Standard cable wrapping and unwrapping practices as commonly employed by live sound personnel in the industry should be used, so that it does not get profoundly twisted and kinked. However, the susceptibility to electrical signal degradation should not be of undue concern. Internally, each wire pair has four twists per inch, so will not easily be separated. All "CAT5" cable purchased today is actually Category 5e, typically rated for 350 MHz and capable of transporting Gigabit Ethernet signaling, which is far faster than that employed by the present technology that we are using (corresponding to 100 Mb Ethernet.) It is also a question of what other parameters are being pushed to the limit, such as cable length, additional CAT5 interconnect hardware, and so on. In the majority of applications, it should not pose a problem.

Because the CAT5 cable is so inexpensive, it can simply be replaced at more frequent intervals than analog cables, if subject to excessive abuse on a regular basis. At around 5 cents a foot, a typical 150 foot cable can cost less than 10 dollars to replace.

There are CAT5 cables that are designed to be more rugged. Here are examples of rugged Cat5 cables:

Whirlwind EC-SPEC (formerly the "EB-100")

Gepco CT504HD raw cable

How durable and reliable is a CAT5 modular plug and socket?

We often hear that people are hesitant to use a CAT5 connection scheme for live sound. From a purely emotional perspective, a modular plug and socket does look and feel flimsy. And the little plastic locking tab does tend to snag and break off easily. But in practice, the mated connection stands up well when compared against conventional analog connectors.

Modular plugs and sockets as presently designed have been used for many decades, dating back to their first widespread use by the telephone industry in the 1970s. They aren't immune to failure, but neither are conventional analog connectors. Most people should recollect plenty of instances where they pulled a telephone off of a table and it fell halfway to the floor, dangling by the modular phone jack, and yet suffered no damage. Those of us with decades of experience in the computer and datacomms industry are accustomed to working in engineering labs and equipment rooms full of Category 5 network cables running all over the place, to an extent comparable and completely analogous to professional audio work with its inevitable mess of cables running here and there. In the computer lab environment, you might have a fifty pound machine on a roll around cart that you start pushing, only to find out that it had a network cable on it that you forgot to disconnect. There are network cables running everywhere that you yank, stretch, trip over, re-route, and so on, day after day.

Do these network cables ever fail at the connector? Yes, of course they do. Once in a while. But they are amazingly resilient, and certainly are not inferior to most professional audio cable connectors, in terms of reliability and durability. An XLR cable connector may seem to be more robust, again from an emotional perspective, but if the cable is pulled, will the strain relief hold? If it has been a while since it was made, the cable insulation has been compressed and thinned out, and the two little screws are not holding it as tightly against the strain relief. Yank on it and the internal connections will break. Keep tightening the two little screws, and the wires will eventually break within the strain relief as they are bent back and forth against the ever tightening grip.

Audio 1/4" TRS cables don't always stay in place, and are easily pulled out by accident. Slight differences in jack vs. plug geometries cause occasional, but potentially maddening reliability problems. Similar problems occur with RCA plugs and jacks.

Banana plugs for speakers pull out. Wires that are screwed down fray, short, and/or break. Perhaps the Neutrik Speakon design could be considered sufficiently robust.

In conclusion, although modular plugs and jacks are by no means bullet proof, we do not feel that modular plugs and jacks are any less durable or reliable than most other plug+jack connection schemes currently being used for professional audio work.

Why didn't you use Neutrik Ethercon connectors on your product?

We did in fact consider using them, and perhaps in retrospect we should have.

Neutrik has improved upon the modular RJ45 design by adding the familiar metal "XLR" shell, while still allowing their RJ45 socket to accept cables with the standard plastic modular plug, as well as their more rugged version.

Click on the following photographs to see high resolution pictures of their product:

Neutrik Ethercon plug
Neutrik Ethercon jack

AudioRail can still be used with a Neutrik Ethercon cable approach. Just not to the rear of the AudioRail product itself. Neutrik provides panel mount connectors that allow a user fashion his own "back of the rack" panel connection scheme. Click on the following photographs to see how this can be done:

Neutrik Ethercon panel mount with punch-down terminals in back
Neutrik Ethercon panel mount with regular RJ45 jack in back
Neutrik Ethercon coupler

Note that with these panel mount connectors with Ethercon RJ45 in the front and regular RJ45 or punch-down in the back, one can use regular RJ45 from the back of AudioRail to the rack panel, then Neutrik Ethercon out from the back of the rack.

Do you recommend using STP (shielded twisted pair) CAT5 cable?

For the vast majority of cases, this is not necessary. In what the standards refer to as a "heavy industrial environment and environments close to broadcast transmitters", we call for the use of STP wiring. We discuss this in the user manual.

Do you support fiber, as an alternative to CAT5, for greater distances?

Certain kinds of Category 5 to fiber media converters work with AudioRail. The cost of these start in the vicinity of $100 each. Not all fiber converters work with AudioRail, but only those that convert the electrical data to optical data without looking at the data or combining it in a CSMA/CD scheme. Click here for more information.

What about word clock synchronization?

Each digital audio stream in AudioRail follows the timing of its source, as it appears at its input connector. They are completely independent, offering the flexibility of running different end to end connections at different speeds and clock synchronization schemes, even though they are time division multiplexed over a single CAT5 cable. AudioRail does not require a common clock source, and it does not require an additional parallel word clock cable to run alongside it. Users simply select which digital audio stream to extract the word clock from.

Users do need to be careful to follow manufacturer's instructions to properly synchronize digital audio components connected together in any digital audio system, regardless of whether it is extended by AudioRail or simply interconnected using only straight digital audio cables. This is typically simply a matter of selecting one as the master clock source and the rest as slaves set to follow the timing of that source. Equipment options vary, however, so equipment user manuals and specifications need to be consulted on an individual basis.

Click here for a picture with accompanying explanation.

What about the inevitable digital-analog conversion delays?

AudioRail accomplishes its transport of digital audio with negligable delay (more on actual latency numbers). Latencies in professional digital audio and DSP-driven products vary widely. Our goal was to develop a product that did not add any significant delays over using a straight wire or fiber connection in its place, and we have succeeded in doing that.

Minimally, a mating pair of good quality 24 bit ADC and DAC integrated circuit chips running at 96K might add as little as 70 microseconds, increasing to several hundred at 48K. The multitude of DSP-driven audio processing devices that have proliferated the market, especially those that have analog input and output interfaces, and are cascaded in serial, will add more. It is important for users today to be aware of these latencies, because they can add up, and often you have to dig into manufacturers' specifications to find them. Obviously, advances in semiconductor technology will work to bring the numbers down, while equipment developers in some cases will take advantage of this opportunity to employ more complex DSP functionality, driving the delays back up, much the way things operate in the PC hardware/software realm (i.e. it tends to take as much time to start and run applications today as it did when computers operated at small fractions of the speeds that they now do).

What is the meaning behind the name "AudioRail®"?

It works like a train, which runs along a track, stopping at each stop to pick up and drop off passengers. It is low-tech, like a train, employing fairly simple technology. We are trying to emphasize this. Keeping it simple and trouble-free. A train is simple in design, and does not require much skill to pilot. Think of conventional analog as being like cars and pedestrians on conventional roads, with individuals going this way and that, point to point, as needed. Think of the higher-end solutions like air travel. The higher end solutions are more complex in design, more expensive, and incur much more overhead in getting on and off the ground, as well as in air traffic control, but once in the air, they do really fly, and there are applications where they are legitimately appropriate.

We are introducing the digital audio "railway" as a means to get from here to there simply, efficiently, reliably, and without a lot of overhead. That was the thinking behind the name.

Are you competing with Peak Audio's CobraNet or Gibson's MaGIC?

We prefer to say no. We are creating a different price-performance niche in the market that is clearly between conventional analog audio and their elaborate LAN transport mechanisms in terms of price and advanced features. We encourage users to explore all the market options in terms of what they really need to solve their particular problems, rather than making exaggerated market claims and hype that suggest that AudioRail is the answer to any and every need.

Users should not buy AudioRail products if a $300, 100 foot, 16+4 channel audio snake would suffice. Users should not buy AudioRail products if they really do need the sophistication and functionality of CobraNet or MaGIC. See http://www.peakaudio.com/ for more information on CobraNet. See http://magic.gibson.com/ for more information about Gibson's MaGIC initiative.

We are often asked, how much does something like CobraNet cost, compared with AudioRail? Beyond saying that the difference is very large, it is a hard question to answer, because CobraNet is implemented in many different ways. A case in point that may be a fair comparison is to take as an example the Rane NM84, paired up with the Rane NM48. This would give you 8 analog channels one way plus 4 the other way, for a total of 12 analog channels of audio. One particular distributer sells the NM84 for $3.3K and the NM48 for $2.4K, which would be $5.7K total for 12 analog channels end to end. In comparison, two AudioRail units, at $500 each, paired up with two Behringer ADA8000 units, at $230 each, would total $1.5K for 16 channels (8 in each direction). So in this example, AudioRail would be about 25% of the cost.

Scaling up in channels with AudioRail in the above example would be much more significant. 64 channels of AudioRail would scale up to only $2.8K by adding six more ADA8000 conversion units, whereas 60 channels of NM84/NM48 combinations would scale up to $28.5K. So in this example, AudioRail would be about 10% of the cost.

In another example, the Yamaha MY16-AT interface card for their line of digital mixers has a typical street price of $430 for a 16-channel ADAT optical interface, whereas we found their new 16-channel MY16-C CobraNet interface card priced at $1300 (AValive, ProAudioKit).

Note: The above are example street prices from distributer sales literature at this time of writing, based on comparing only one product set. There are many other licensees and products employing CobraNet. Users are advised to carefully do their own research on products and pricing before making decisions.

CobraNet provides complete Ethernet LAN compatibility, remote configuration using standard PC applications, and other benefits. AudioRail provides no Ethernet LAN compatibility and no remote configuration. It just gets the audio from here to there with minimal latency, and that is it.

CobraNet latencies are selectable at 1.33, 2.67, or 5.33 milliseconds. This is something significant to take into account.

Ethernet LAN compatibility, remote configuration, latency, and price would be the main factors to consider in making a decision between CobraNet and AudioRail.

We actually hope that both Peak Audio and Gibson are successful in their ventures. They have some promising technology and features that should be taken seriously by the professional audio market.

What about Aviom's A-Net "digital snake" products?

Aviom started out by developing specialized monitor mix products, but has since developed more generic digital snake products in their ever growing product set. They transfer 16 channels of 24-bit audio in one direction on a Cat5 cable, with the additional ability to aggregate four of these streams into a single Cat5 cable run, providing up to 64 channels in a one direction with the purchase of an additional concentrator product.

We found the following approximate street prices for their competing products to be as follows, at the time of this writing (Sweetwater)

Model Description Cost
AN-16/i 16 channels line in $800
AN-16/i-M 16 channels mic/line in $1800
AN-16/o 16 channels line out $800
16/o-Y1 16 channels Yamaha output $500
AN-16SB 64 channel aggregation bridge $225

Based on the above numbers, we constructed a comparison table. See http://www.aviominc.com/ for more information. Since we are operating a no-hype website that provides an engineering, rather than marketing point of view, it is easy for us to compare what we feel are fair trade-offs between the two implementations. Each has advantages and disadvantages. Here is how we see them: Note: Check the Aviom website and distributor pricing to validate this information. They will always have more up to date information on their products than we have on their products!

Compare Aviom AudioRail Notes
Min. cost (analog end to end solution) $1,600 for 16 channels one way (line in) $1,460 for 16 channels (8 each way) Aviom: line in (no mic. preamps!)
AudioRail: Qty=2 ADAT version, Qty=2 Behringer ADA8000 (with mic. preamps!)
Max. cost (analog end to end solution) $10,850 for 64 channels one way $2,840 for 64 channels (32 each way) Aviom: mic/line in
AudioRail: Qty=2 ADAT version, Qty=8 Behringer ADA8000 (with mic. preamps!)
Cost for digital end to analog end solution No product yet available $1,920 for 64 channels (32 each way) AudioRail: Qty=2 ADAT version, Qty=4 Behringer ADA8000 (with mic. preamps!)
Assumes one end has digitial audio connections
Cost for digital end to digital end solution No product yet available $1,000 for 64 channels (32 each way) AudioRail: Qty=2 ADAT version,
Assumes both ends have digitial audio connections
Min cost for Yamaha digital mixer interface $500 for 16 channels output $930 for 32 channels (16 each way) Aviom: Single Yamaha interface card
AudioRail: Qty=1 ADAT version + Yamaha MY16-AT interface card
Max cost for Yamaha digital mixer interface $1000 for 32 channels output $1360 for 64 channels (32 each way) Aviom: Two Yamaha interface cards
AudioRail: Qty=1 ADAT version + Two Yamaha MY16-AT interface cards
Rack space required for 8 analog channels audio delivery 1U 2U Aviom: AN-16/i or /o product
AudioRail: 1U, plus 1U for analog/digital conversion unit
Rack space required for 64 analog channels audio delivery 4U or 12U in, 4U out 5U Aviom: Mic/line in is 3U each unit
AudioRail: 1U, plus Qty=4 Behringer ADA8000
Rack space required for 64 digital channels audio delivery n/a 1U AudioRail: Qty=1 ADAT version
Ethernet compatibility No No
Can use Hubs/Switches No No
Ability to fan out audio network to multiple parallel daisy chain branches (tree structure) Yes No
Can share with PCs on LAN No No
Can configure with remote PC No No
Number of Cat 5 cable runs needed for bidirectional sound 2 1
Word length/speed 24 bit, n/a 24 bit, 44.1K/48K/96K Note: 96K halves the number of available AudioRail channels, doubles cost/channel for ADAT
Transport latency order of 1 ms (incl. converters) Negligible
(few microseconds, no converters)
Unconfirmed Aviom number
Alesis ADAT Lightpipe, S/PDIF connectivity tbd Yes Important for direct digital audio connections (example)
EN55103 Level E5 immunity "Heavy industrial environment..." ? Yes See user manual for more information
Product availability Now Now
Company with established reputation and market presence Yes No
Risk if company goes out of business Low Low Products are self-sufficient, no ongoing upgrades necessary, will continue to work

There are a couple of other strategic things to consider in the competing two product approaches: In Aviom's favor, an investment in their digital snake product makes the user's purchase compatible with their personal monitor mix suite of products. In AudioRail's favor, AudioRail is a more generic digital audio building block that allows the user to change and upgrade A/D and D/A converters to suit their changing applications, while keeping the digital audio snake infrastructure constant. This also applies to easier migration with digital audio interfaces on digital mixing boards and recorders.

What about BSS Soundweb, Aviom's other A-Net products, and Hear Technologies?

These are specialized systems. BSS Soundweb networks signal processing devices together, configured in a ring network, transporting eight channels of 48K audio in each direction. See http://www.bss.co.uk/ for information on BSS Soundweb. Aviom's monitor mix products are even more specialized, focused on providing an elegant product set that delivers very flexible monitor mixes to live performers. See http://www.aviominc.com/ for information on Aviom's products. Hear Technologies, a division of Quantum Technologies, has introduced a smaller scale system for monitor mixes. They transport 8 channels of audio and power through a Cat 5E cable. It is worth noting that they also offer a single, inexpensive ADAT Lightpipe to Cat 5E converter for MSRP $69 (a pair is required) that will simply extend the reach of a single ADAT Lightpipe connection to 500 feet. See http://www.heartechnologies.com/ for more information on their products.

Note that these technologies, like AudioRail, depart from standard Ethernet LAN compatibility. This strategy is not unique to AudioRail.

Although AudioRail products are more general in application and not intended to compete with these products, it is worthwhile mentioning again that AudioRail provides a digital audio transport latency of only a few microseconds. Almost all other technologies of this nature incur latencies that are substantially greater.

What about Digigram's Ethersound products?

Digigram Ethersound definitely competes with AudioRail in concept and implementation, especially since it likewise minimizes latency. See http://www.digigram.com/ for more information. Since we are operating a no-hype website that provides an engineering, rather than marketing point of view, it is easy for us to compare what we feel are fair trade-offs between the two implementations. Each has advantages and disadvantages. Here is how we see them: Note: Check the Digigram website and distributer pricing to validate this information. They will always have more up to date information on their products than we have on their products!

Compare Ethersound AudioRail Notes
Min. cost (est. street price, for analog end to end solution) $2,600 for 8 channels one way $1,460 for 16 channels (8 each way) Ethersound: ES8in (no mic. preamps!) + ES8out

AudioRail: Qty=2 ADAT version, Qty=2 Behringer ADA8000 (with mic. preamps!)
Max. cost (est. street price, for analog end to end solution) $21,000 for 64 channels one way $2,840 for 64 channels (32 each way) Ethersound: 8x ES8in (no mic. preamps!) + 8x ES8out

AudioRail: Qty=2 ADAT version, Qty=8 Behringer ADA8000 (with mic. preamps!)
Cost for digital end to analog end solution No product yet available $1,920 for 64 channels (32 each way) AudioRail: Qty=2 ADAT version, Qty=4 Behringer ADA8000 (with mic. preamps!)
Assumes one end has digitial audio connections
Cost for digital end to digital end solution No product yet available $1,000 for 64 channels (32 each way) AudioRail: Qty=2 ADAT version,
Assumes both ends have digitial audio connections
Rack space required for 8 analog channels audio delivery 1U 2U Ethersound: ES8-class product

AudioRail: 1U, plus 1U for analog/digital conversion unit
Rack space required for 64 analog channels audio delivery 8U 5U Ethersound: Qty=8 ES8-class products

AudioRail: 1U, plus Qty=4 Behringer ADA8000
Rack space required for 64 digital channels audio delivery n/a 1U AudioRail: Qty=1 ADAT version
Ethernet compatibility Yes No
Can use Hubs/Switches Yes No Be careful!
Ability to fan out audio network to multiple parallel daisy chain branches (tree structure) Yes No
Can share with PCs on LAN Yes No Note: Potential risk during live performance*
Can configure with remote PC Yes No
Number of Cat 5 cable runs needed for bidirectional sound 2 1
Word length/speed 24 bit, 44.1K/48K 24 bit, 44.1K/48K/96K Note: 96K halves the number of available AudioRail channels, doubles cost/channel for ADAT
Transport latency Negligible
(microseconds)
Negligible
(microseconds)
Alesis ADAT Lightpipe, S/PDIF connectivity tbd Yes Important for direct digital audio connections (example)
EN55103 Level E5 immunity "Heavy industrial environment..." ? Yes See user manual for more information
Product availability Now Now
Company with established reputation and market presence Yes No
Risk if company goes out of business Low Low Products are self-sufficient, no ongoing upgrades necessary, will continue to work

* Note: Our concern about sharing audio with PCs on a computer LAN during live performance is that there is always a certain amount of uncertainty about whether a PC running Microsoft Windows with any number of network and user applications is going to function (or malfunction) predictably. In a studio environment, an operating system, network, or application failure of either the system being used or another rogue PC or device on the network would only mean a re-take or "undo" operation. In a live performance or live recording environment, the implications are potentially much more grave. Although AudioRail does not offer any method of interoperability with Ethernet LAN components, let alone PC applications, we feel that there is something also to be said for the peace of mind that something as simple as AudioRail is, by design, going to have far fewer possible failure modes. It is, again, a tradeoff that a sound professional must make in choosing equipment that best suits the particular target application or installation.

What about the Helios digital snake products?

Helios has introduced a proprietary digital audio snake product that can communicate over CAT5, coax, and fiber connections. It is not Ethernet compatible, but does provide remote control capabilities. A 24+6=30 channel analog end to analog end solution starts at about $10K. The product set is very modular, and scales up to 120 channels. Their website is at http://www.heliosdigital.com.

What about Otari Lightwinder?

The Otari Lightwinder Colloseum LW-50 is a system based on standard 62.5um Multi-mode fiber optic connections. It transfers up to 64 channels of audio in each direction using up to 16 stations separated by fiber connections of up to 3000 meters. The latency is 1.6 ms. The sample rates are 44.1K/48K with sample rate conversion also provided. Each of up to 16 stations can have up to 8 modules of 1U rackspace each, and each module handles 8 channels of audio. Analog audio modules with mic. preamps and 20 bit A/D or D/A conversion are available, or AES/EBU (pro S/PDIF) input or output modules. The system can be configured and controlled with a remote PC connection and software application. A connection matrix can assign any input to any output or outputs.

This is a very capable system. It is also very large and expensive. We do not know the actual list or street prices, but it must clearly be in the 5-digit range, like other high-end systems in this class. A 32-channel two station unit previously used at a trade show was recently posted for sale on eBay at a starting bid of $25.5K and a "buy it now" price of $28K (eBay item #3706664688).

What about FiberPlex LightViper?

Fiberplex, Inc. has developed a fiber optic based 32 input 8 output digital snake with integral 24 bit 96 KHz analog/conversion and XLR/TRS connectors, called LightViper. A March 2004 press release cites a price of $7.5K for a VIS-1832 base system with 300 feet of fiber optic cable. The maximum distance cited is 6600 feet. We could not find a latency specification. This looks like a very nicely packaged product, with the familiar look and feel of an audio snake (click for photo).

What about Optocore?

Optocore is a family of very high end fiber optic products forming a distribution system that transfers audio, video, and several other interfaces. We found the DD32 listed at one European distributor (Thomann) for EUR 5190 (about $6.6K USD). Latency is cited as 41 microseconds.

Are you competing with MADI?

Perhaps, although not intentionally. MADI is a 1991 AES (Audio Engineering Society) standard for transmitting up to 56 channels of up to 24 bit digital audio at 32 KHz to 48 KHz sample rate, point to point, in one direction, over a 75 Ohm coaxial cable connection. MADI is also known as AES10.

In contrast, we are transmitting up to 32 channels (at 48K) in both directions simultaneously, in a daisy chain network, using commodity parts (the Ethernet transceivers) and less expensive Cat 5 cable.

MADI was a good idea that was far ahead of its time in 1991. It would have been very difficult for them 12 years ago to foresee the rapid proliferation and market acceptance of Ethernet technology that would take place over the years to follow. In 1991, it was not even clear that Ethernet would prevail over IBM Token Ring networks. In fact, at that time it looked to many of us that Token Ring would prevail over Ethernet, Digital Equipment Corporation's FDDI, and all other competing networking technologies.

Now, a few dollars buys you an Ethernet transceiver chip that contains some very sophisticated signal generation and recovery techniques. That, with a few more dollars of passive components, including the RJ45 connector, completes the interface. Category 5 cable can be bought in bulk for five or six cents a foot, and ready made cables for only a few dollars. We are capitalizing on this technology, which has become a very low cost commodity on account of its massive proliferation.

We feel therefore, that we are in a good position to displace many MADI applications with our technology, because our solution is less expensive and more versatile.

What about the RME ADI-648?

The RME ADI-648 is a MADI coax/fiber based product that uses the Alesis ADAT Lightpipe optical interface and is a fair competitor to the AudioRail ADAT rx32tx32 product. It has twice as many channels (64 + 64) for roughly four times the cost of AudioRail ($2K/unit vs. $500/unit). It requires two 75 Ohm coaxial cables to transfer data both ways, compared to AudioRail, which requires one Category 5 cable. Or it can transfer data over an integral multi-mode fiber connection, which can extend its reach to up to 2 km (1.2 miles). (AudioRail can do this with outboard media converters. Click here for more information.)

Note that stacking two independent AudioRail systems on top of each other will also provide 64 + 64 channels of audio transport, also using two cables (Cat5, in the case of AudioRail), at roughly half the price of a single ADI-648 solution.

Note that any number of ADI-648 units can be configured in a daisy chain that can be looped back all the way to the source. But the daisy chain then becomes a 64-channel unidirectional path.

For the additional cost, the ADI-648 provides a computer controllable crossbar matrix which allows you to assign any ADAT input connection to appear on any ADAT output connection. In contrast, AudioRail simply hard-maps ADAT input connectors to corresponding output connectors at the other end.

For the additional cost, the ADI-648 also has both 75-Ohm coax and fiber connectors integral to the unit. AudioRail currently only has Cat5 RJ45 connections. Media converters must be purchased to convert the Cat5 to fiber, if fiber is needed for long runs. (Click here for more information.)

There is another difference between AudioRail and the RME that can work either to advantage or disadvantage, depending on the application. The RME locks its whole system to a word clock source of either internal 44.1K, internal 48K, MADI, or "ADAT", and requires that all ADAT connections be in sync to work. In AudioRail, each ADAT output simply follows the timing of its corresponding source, with no common clock and no synchronization constraints. On one hand, AudioRail is more flexible, even allowing different clock domains to run at the same time and at different speeds, but if you are running only one clock domain you have to make sure they are all individually sync'ed up, which may be a nuisance to those who only have one clock domain.

See http://www.rme-audio.com/english/adi/adi648.htm for more information on the RME ADI-648.

Have you heard about Sony SuperMAC?

Sony's UK division has recently submitted a proposal to the AES, designated AES-X140, to standardize its SuperMAC ("Super Multi-Channel Audio Connection) interconnect, formerly known as MAC-DSD. This is a point to point interconnect that transfers up to 48 channels (at 48K sample rate), or alternatively other formats, most notably their DSD ("Direct Stream Digital") format, plus a synchronization signal in each direction, on a Category 5 cable. The synchronization signal provides both synchronization of the data link and an embedded word clock. This extra signal is routed on unused pairs of the Category 5 cable, in addition to the standard 100Base-T Ethernet transceiver interface.

To fan out the point to point connection, a specialized router product is used.

Their URL is http://www.sonyoxford.co.uk/pub/supermac/index.html.

The AES proposal cites this as being "designed for use in a studio environment" and as an "up to date alternative to MADI."

What other things are going on in this area of the industry?

Pioneer and Gibson have a proposal into the IEEE entitled "Synchronous Ethernet". An overview brochure and specification draft for discussion purposes are available.

Nine Tiles Networks Ltd in the UK has a product that transfers audio using ATM (Asynchronous Transfer Mode) using Ethernet hardware, conforming to AES47-2002, a digital audio over ATM standard.

There is also a technical committee in the AES devoted to discussing networked audio systems.

What are the costs for analog converters and other interface gear?

AudioRail products will interface directly with professional audio products that have digital audio interfaces. This includes digital mixers, digital recorders, digital audio workstations, standalone A/D and D/A converters, and other devices. The cost of AudioRail is currently $500 per 64-channel point of connection, excluding CAT5 cable and digital audio cable costs, which are minimal.

The following table lists standalone A/D and D/A converters that support the Alesis ADAT Lightpipe interface. The Alesis ADAT Lightpipe interface is what is implemented on our flagship product. It can be seen that a simple digital audio snake with 16 channels of analog connections at each end can be implemented for about $1400, and one with 64 channels of analog at each end can be implemented for $2600. (This assumes cost of AudioRail at $500 per unit). A Behringer DDX3216 digital mixer with ADAT interface option cards has a street price of about $1300, so a simple 32 input, 16 output digital mixing configuration with converters at the front of house could cost as little as $3100.

Product Description price
Behringer ADA8000 - analysis) 8 line/mic in + 8 line out to lightpipe in/out $230 street
Alesis AI3 8 line in + 8 line out to lightpipe in/out $400 street
PreSonus DigiMAX FS 8 mic/line in to lightpipe $600 street
PreSonus DigiMAX LT 8 mic/line in to lightpipe $800 street
PreSonus DigiMAX 96K 8 mic/line in to lightpipe $1.3K street
Marian ADCON 2 mic/line + 6 line in, 8 out to lightpipe in/out $600 street
Frontier Design Group Tango24 8 line in + 8 line out to lightpipe in/out $700 street
CreamWare A16 Ultra 16 line in + 16 line out to 2x lightpipe in/out $850 street
Focusrite OctoPre LE 8 line in to lightpipe out $600 street
Focusrite OctoPre 8 line in to lightpipe out $800 street
Graham-Patten ADAT-1 8 line in to lightpipe out $1.5K
RME ADI-8 PRO 8 line in + 8 line out to lightpipe in/out $2K list
RME ADI-8 DS 8 line in + 8 line out to lightpipe in/out $2.25K list
Lucid ADA88192 8 line in + 8 line out to lightpipe in/out $2.6K list
Apogee AD16 16 line in to 2x lightpipe $3K list
Apogee AD-8000 8 line in to lightpipe $5K list
Apogee DA16 2x lightpipe to 16 line out $3K list
Apogee Mini-DAC lightpipe to 2 line out $1K list
Apogee PSX-100 2 line in to lightpipe $3K list
Apogee RosettaAD 2 line in to lightpipe $1.4K list
Apogee Trak2 2 mic to lightpipe out $4K list

Category 5 cable costs only 4 to 6 cents per foot in bulk. For example: (no affiliation) Home Depot, CableMAX, Cablewholesale, Computer Gate, CablesDirect, QualityCables

Category 5 cable can also be obtained from Radio Shack (Catalog #278-1583) for 15 cents a foot, by the foot. Radio Shack's crimp tool (Catalog #279-405) is a professional quality tool that we have found to be as good or better than some other industrial quality brands, and costs $31.99. Radio Shack's 8-pin modular plugs (Catalog #279-406) come in a package of 5 for $3.89. A ready-made 100 foot cable from CompUSA, for example, costs $30.

Our product user manual gives instructions on how to crimp cable ends.

Toslink fiber optic cables can be bought for as little as $2.59 for a 3 foot cable.

What about warranty, return policy and customer support?

The products currently carry a 5-year transferable warranty, as explained in the user manual. We consider this is a very robust product, due to its simplicity and conservative design.

We offer a 30-day return policy that has both generous and restrictive aspects to it. If the product does not work in an application that we say it ought to, due to some incompatibility or our not advertising or documenting the product properly, the product can be returned for full refund of purchase price, including any U.S. sales tax incurred, and including U.S. domestic ground shipping charges. We stand by our products with honesty and integrity, and will not turn down a legitimate complaint or ignore a legitimate problem.

We regret that a small business like ours does not have the financial means to risk the kind of "no questions asked" full refund policies of major distributers and retailers. We wish we could, but we cannot. In a scenario where a customer desires a "no questions asked" return, changes his mind, has misunderstood the product's application where clearly documented in the website or user manual (for example, if a customer does not understand the difference between optical S/PDIF and ADAT Lightpipe, or perhaps did not realize that his existing building wiring was CAT3 or telephone grade cable and cannot be changed), or any other scenario that involves misjudgment on the part of the customer, then we will still be happy to take the return, charging a restocking fee to simply compensate us for what we anticipate it will cost us to sell it again. We have no intention on making a profit from a dissatisfied customer.

Again, we are a new startup doing our own direct sales and cannot afford to risk generous return offers. Our product functionality is so simple and straightforward that there should be no question as to the suitability of the product in any given application, compared with kinds of professional audio products that may legitimately require a trial period for a customer to decide whether a product's functionality is suitable or desirable for his application.

As eager as we are to sell our products, we believe that customers in this industry should carefully research potential product investments, technical documentation, and competitive alternatives with diligence and sobermindedness, rather than simply purchase a product on impulse because "it's cool".

Please do not hesitate to contact us at info@audiorail.com, or telephone us at (978) 461-0177 to discuss your product application or ask any questions. Your successful product application is of utmost concern to us, and we will not pressure anyone to invest in our products simply to make a sale, or if they are not right for the application.

What is the timeframe for deployment of your products in the open market?

We are in volume production and are shipping the ADAT rx32tx32 product domestically and internationally. We are currently operating as a small business with direct sales only. Long term business plans are still open for discussion and we are open to the potential for business relationships that will make a more formidable market impact in this industry.

What regulatory approvals do you have to market your products?

EMC (ElectroMagnetic Compatibility) testing was conducted at Compliance Worldwide, which is an A2LA accredited test site (Certificate Number 1673-01) with specific accreditation for testing to standards governing consumer and professional audio products. Safety testing was conducted at TUV's American corporate headquarters. The product has passed radiated and conducted RF emissions testing for FCC Part 15 Class B, EN55103-1, and EN55013 (CISPR13). It passed the first time, with more margin than they have ever seen in their site's testing history. It passed immunity testing to EN55103-2, as well. The product was tested to comply with all EN55103 categories (E1 through E5), covering a range of environments from residential (E1) to heavy industrial (E5). Safety testing was conducted, demonstrating compliance to UL6500 and EN60065. The product is therefore completely qualified for sale both in the United States and internationally.

See related information describing standards above

What technical qualifications do you have? Or are you just an electronics hobbyist?

The principle engineer who started AudioRail Technologies has a B.S. in Electrical Engineering from Northwestern University, 1979, plus over 30 years of experience in industry in a large corporate environment, almost all of which was spent actively doing computer hardware and digital systems design/development. A substantial amount of that time was spent in the computer datacomms segment, designing WAN/LAN computer networking hardware. This engineer has also spent the last 22 years doing a substantial amount of professional audio work on the side as a charity, consisting of live sound reinforcement for live musical theater productions and contemporary church settings, as well as studio recording, with some clients' released albums to his credit. He has at various times also performed regularly as a non-professional musician and capitalizes on a solid musical background in facilitating studio recording work.

There are other technical contributors with engineering background that add substantially to the capabilities of AudioRail Technologies. There are no "hobbyists" involved in the technical know-how that we have.

What is the specific nature of the patent claims?

First of all, as a disclaimer, the language of the actual issued patent is the only legal, authoritative explanation. But to roughly (and unofficially) summarize in a nutshell, the innovation has to do with using Ethernet transceiver technology to transport digital audio and audio-related signals using a time division multiplexed scheme, along with the method used to efficiently sample and reconstruct the signals while maintaining their timing and synchronization end to end. It should be pointed out that this "innovation" is not one of technological prowess, but of reduction of complexity, making the system simpler in design and implementation than all other existing approaches.

Ethernet transceivers were not designed to implement a time division multiplexed scheme. They were and are designed to implement Carrier Sense Multiple Access with Collision Detect (CSMA/CD) in a packet store and forward scheme, to support the local area networking of computers. Putting together digital audio and audio related signals with a TDM stream and an Ethernet transceiver was an unlikely and non-obvious marriage of technology. But it works elegantly.

What direction are you headed, in terms of technology, improvements, and growth?

We currently have prototypes with optical S/PDIF interfaces. The ADAT Lightpipe version, which is the most cost effective implementation, is in volume production.

We are developing a similar product with 75-Ohm coaxial S/PDIF interfaces, which are more prevalent in the professional audio industry.

We were developing products with integral analog interfaces to provide an economical analog end-to-end solution, but for $230, Behringer's new ADA8000 (8 mic/line inputs + 8 line outputs) makes it questionable whether it is worthwhile for us to continue this effort (see our test results) .

We are developing products that will transport the Tascam TDIF interface.

We are developing interfaces to transport SMPTE time code, word clock and MIDI.

We are working on a Gigabit Ethernet transceiver solution, and are trying to decide when the time will be right to introduce it, based on the price of Gigabit Ethernet components. This will allow for hundreds of channels of audio. There is nothing precluding us taking that step in the market today. It is technologically the same concept that we employ now. But we want to start with a lower priced product offering. Compatibility of future Gigabit offerings with the current 100 Mb solution will likely involve a simple aggregation bridge product that will time division multiplex up to 10 streams of 100 Mb products into a Gigabit "fat pipe".

Nothing precludes us from developing a product with integral fiber transceivers to replace the Category 5 cable connections. We are weighing the cost of doing that vs. continuing to recommend separate off the shelf fiber media converters to accomplish the same thing.

What does it take to get your present designs to production?

The flagship product is the ADAT lightpipe version. Pre-production prototypes have been operational and employed in live sound for some time. The regulatory approvals process has been completed. We have started volume production builds and are shipping. Volume component purchase and PCB assembly costs approximately $28K for each build of 100 units (see BOM). To get each unit out the door, including packaging and credit card processing fees, costs us about $300.

Longer term, we want to do some smart consolidation of different variations into one or a very few PCB designs (example: optical/coaxial/AES/receive/transmit/IEC958/ADAT), re-pinout using lower cost Altera Cyclone FPGAs, employ RJ45 connectors with integrated magnetics, dual Ethernet transceiver (Broadcom BCM5222, Intel LXT973, or equiv.), while continuing to pursue our technology map (Tascam TDIF, integrated line/mic ADC/DAC, MIDI, Gigabit AudioRail, etc.

What are your target product costs?

We are pricing our first units at $500 per unit. The bill of materials for our first products work out to be in the vicinity of $900 in quantities of 1s, and $300 in quantities of 100s (PDF XLS). Clearly it is more expensive than running an analog snake or two. Clearly it is less expensive than higher-end products on the market. We are currently running our startup business with no debt and very little overhead. Our intention is to minimize end cost without sacrificing quality, and maintain an honest, middle-class standard of living for ourselves while attempting to recover our business investment.

Are all elements of your design available?

We are happy to show many elements of our design, such as circuit schematics, parts lists, PCB layout drawings, etc. They consist of well established, industry standard design techniques with off the shelf components and engineering CAD tools. Much of this is posted right here at this web site, for the purpose of substantiating the design to those interested in reviewing it.

Isn't that FPGA expensive? Why didn't you develop an ASIC?

Yes, an FPGA is expensive compared to an ASIC. But ASIC development is costly and a small business like ours could not afford it. The FPGA is comparatively inexpensive to develop, at the expense of the parts costing more. It does provide us with very fast development cycles, and provides flexibility to quickly spin variants of the design.

What business options are you open to, in terms of buyouts, partnering, licensing, etc.

We are open to all options. A complete buyout by a larger corporate entity would be good for the technology and the market, because we are a small business with limited resources at our disposal, and limited funds to invest. Licensing the technology is definitely an option, and we would carefully weigh our desire for short term financial gain against providing incentive to make it easy for a licensee to add market presence to AudioRail. Partnering is also an option with similar factors to weigh in. Individuals and companies should not hesitate to contact us, even informally or casually, to discuss any ideals or proposals they have.

What is your current investment, capital and otherwise, and how is your business organized?

The business is officially structured as a sole proprietorship. Approximately 2.5K man-hours and $82K of NRE have been spent to date, and up to $28K per 100 exists in inventory (cost) from any sequential volume build. Completing the patent process is expected to cost $5-10K more. No resources from any other company were employed in this venture. AudioRail Technologies is completely self-funded and self-equipped. There is currently no debt, and no contractual or investment obligations or other corporate overhead to account for in what is quite a grass roots startup. The implications to the potential cost of a company acquiring or utilizing our technology should be obvious.

What are the names of the individuals involved, business location, and other specifics?

Please contact us at info@audiorail.com and we will be happy to provide this information. We don't want to publish it for search engine robots and casual or unintentional browsers.

Why should we trust you or your products?

That is a fair question indeed. Why should you trust a "grass roots" startup that has sprung up out of nowhere? This is one of the reasons we have offered so much technical information at this site (besides the fact that we think that companies should, anyway). The engineering documentation, including circuit schematics, PCB layout, and full bill of materials are all posted for review by anyone. Highlights of testing, including oscilloscope screen shots, FFT scans, complete WAV files, and testing methodology are posted. Documents from regulatory testing are posted. The user manual is posted. We are being up front about everything we have done, including even citing relevant competition and prior art. It is all there for open appraisal.

Investing in a new product offered by a new company is indeed an understandable concern. Larger, well established companies can bank on their reputation and track record. We are offering very thorough documentation in its place. We hope that by doing so we can be considered an equally "safe" investment.

This website leaves a lot to be desired, appearance-wise.

Our focus is on engineering, not marketing. If you do a "show document source," you will see that this web page is nothing more than a textfile that has been manually annotated with html tags. If you are looking for attention grabbing advertising gimmicks, slick colors, blinking links, frames, pop-ups and cute java applets with animated choo-choo trains chugging across the screen, you have come to the wrong place. The 13x170 pixel tan/beige background bitmap tile of the railroad tracks is all we have to offer you. If you are a professional looking for technical information, you have come to the right place. We hope that the information provided here is helpful to those who are interested in understanding our products and progress, including potential customers, as well as potential investors and potential partners.

Thank you for your interest in AudioRail. Please stay tuned for more information and new developments as they happen.

HOME