AudioRail® Technologies:   It simply works.

Summary:

Some diligent testing by customers in France (http://www.genesis.fr) has uncovered an incompatibility between the PreSonus DigiMax FS and AudioRail. This is probably related to this product using the TC Applied Technologies "JetPLL" method of clock recovery, implying that several other products on the market employing "JetPLL" may also have this problem.

This does not apply to other, older products by PreSonus which do not employ this method.

Genesis also reported similar problems with the Swissonic DA24.

Detailed description:

We have opened up and tested the PreSonus DigiMax FS to confirm the behavior that Genesis reported. We measured the PLL loop bandwidth at about 10 Hz, which means that it will refuse to track a digital audio signal which drifts too far in phase during even as slow as a 0.1 second time period. In its effort to make the clock "jitter free" it becomes too rigid and cannot consistently track an incoming signal in digital audio systems (such as AudioRail) where the signal is repeatedly re-clocked.

The DigiMax FS will not be a problem where it is only connected to a single original source, with nothing in between (such as directly to a computer sound card or a digital mixer), or if it is only used as a clock master with only the A/D conversion audio sent to a system like AudioRail (but not the D/A conversion returned via such a system).

Unfortunately, when the phase of the incoming digital signal drifts too far from the internal "JetPLL" clock signal, the DigiMax FS clocks samples incorrectly or drops out of sync entirely, because the "JetPLL" clock signal tracks it too slowly and will not reliably stay phase locked to the input signal. When it mis-samples the data, the effect is the same as if there were multiple "masters" (i.e. the slave does not track) with crackling noise evident on the D/A outputs of the DigiMax FS.

The following is a 32-bit mono recording of the DigiMax FS output through a chain of 4 AudioRail units, delivering a 55 Hz sine wave:

http://www.audiorail.com/FS_4-AudioRail.wav (size = 1.8 MB)

By contrast, the Alesis patent (US patent #5,297,181) for ADAT optical, which is the basis of their ADAT optical interface, recommends in its "preferred embodiment" a much more responsive PLL design. The loop filter for the 3-stage logic/analog PLL design specified by Alesis is at least 3 orders of magnitude faster and more responsive, being roughly equivalent to the frequency of the sample rate (e.g. 48 KHz), whereas the 2-stage JetPLL design uses a digital PLL with digital clock synthesizer in series with an analog PLL with analog VCO to create a 10 Hz loop filter bandwidth. (The digital PLL creates the very low loop filter bandwidth, and then the analog PLL removes the undesirable "intrinsic" jitter artifacts from the digital PLL.)

The particular Genesis configuration had a daisy chain of four AudioRail units, with various sources used to generate the audio, and the DigiMax FS at the end. To verify that this problem is specific to the Presonus Digimax FS, we substituted a Behringer ADA8000 in its place, using the same configuration, and also did an FFT scan of the result (using the ADA8000), both with and without the four AudioRail units. The following is a screen capture of that result, to show that there is no audible artifact from a chain of four AudioRail units:

http://www.audiorail.com/Compare_0_vs_4_AudioRail_48K_1.gif

In this picture, the blue line with shaded blue area is with 4 AudioRail units in the chain, and the thin green line is with the 4 AudioRail units removed and Toslink cable connected directly. The input signal is a 55 Hz sine wave at -6 dB, shown in the wave editor underneath the FFT plot, scanned for about 8 seconds out of 10 seconds total (see dark green highlighting), using Kaiser windowing on the FFT. 24 bit audio was used, giving 144 dB of digital resolution.

As will be evident from this, there are no audible artifacts from the addition of 4 AudioRail units, with the most significant artifacts being from the analog D/A and A/D conversions (less than -99 dB), which is where the green and blue match.

The Behringer ADA8000 recovers the ADAT signal (and clock) using the Alesis Semiconductor (now Wavefront Semiconductor) AL1402 ADAT optical decoder.

The conclusion from all of this is that the Presonus Digimax FS is not compatible with AudioRail, and probably is not compatible with other types of digital audio systems that similarly operate as a chain.

A few other relevant products use the "JetPLL" method, such as from M-Audio. A web search for the term "JetPLL" or a word search in any current or future product brochure will likely reveal if the product is using this method.

Genesis also tested with the Swissonic DA24, obtaining similar results. The Swissonic DA24 could not possibly use "JetPLL" because it is a much older design. But it cites in its feature list "Proprietary low-jitter clock recovery system" so it may be attempting the same goal and have the same problem. We don't have knowledge of any customers using the Swissonic DA24.

This has been a disappointment to us, particularly because the Digimax FS product seemed like a good product for customers who wanted to step up from the Behringer ADA8000 (or not use any Behringer product at all, as some do as a matter of choice). Now customers will have to continue to choose between the Behringer ADA8000 and much more expensive converter products.

We have updated our standalone converter list at the following URL:

http://www.audiorail.com/ar_converters.htm

For reference, our posted analysis of the low end Behringer ADA8000 converter product is at the following URL:

http://www.audiorail.com/ADA8000.htm