Ahead of the Game: John Mingo of Baetis Audio shares his knowledge

Anyone who wants to begin to have a better understanding of computer audio would do well to heed the words of designer John Mingo of Baetis Audio. John's insights have far-reaching impact and bode well for the continued progress and excellence of his music computers at Baetis Audio. Here is vital detail that John shared with me from a recent interview.

Baetis Audio™ – Our Evolving Philosophy About the Design of the Very Best Audiophile Computers.

Since its inception in early 2012, Baetis has emphasized building audiophile computers with the very best in 2-channel audio via SPDIF or AES. Our philosophy has evolved rather considerably since then, but sometimes the audiophile industry is slow to change its own understanding of what an OEM is trying to achieve. For example, in a recent issue of The Absolute Sound, the description of our Reference model (as a TAS Editor’s Choice) still emphasizes our superlative AES/EBU output -- but we have evolved to be a lot more than just AES.

In late 2017, our view is that a media server (computer) must be designed to provide the best digital audio output possible for the most commonly used input on a customer’s DAC or pre/pro (including the inputs of DACs or pre/pros that the customer is thinking seriously about adding to his equipment rack). And, we do spend a huge amount of our time and money auditioning DACs because we do not learn very much from many of the major audio magazines’ DAC reviews. First, these reviews do not often provide comparison across the various digital inputs of the DAC; many reviews ONLY use the DAC’s USB input because of convenience, or an outdated bias the reviewer may have. Second, the reviews hardly ever compare any built-in volume control of the DAC to the pre-amp being used within the reviewer’s own system. Sometimes the reviewer appears to think that he should ALWAYS use a separate pre-amp, no question, especially when the DAC has a pre-amp in the digital domain. Berkeley, or Robert Harley of The Absolute Sound, would likely laugh at this notion. But these reviews keep on coming.

Indeed, the industry calls a “purpose-built” computer a server, not a computer. Even though the server uses a CPU, a Motherboard, and RAM – and “plays” computer files. In short, the degree of misleading information or lack of sufficiently in-depth testing done by the industry is appalling when the subject is digital audio. At Baetis, we have spent more than 6 years of day-after-day testing and comparisons.

Here is what we have learned in these years of listening to servers and DACs and MCh DACs (or pre/pros) in our own rooms with equipment costing way more than the average audiophile’s room:

1. Any USB output from a factory computer suffers from immense common-mode noise. Even when the computer builder uses a much better USB output (e.g., an SOtM™ PCI-bus-based audio card), the PCI-bus process results in common-mode noise that is unacceptably high. The “blacks” are not black-enough for our taste, and some detail is lost.

1. Many DACs do have an AES input that is its best type of input, when using a Baetis computer to generate the AES output. Generally, a regular computer only has an AES output if the DIY owner installs a PCI-bus-based AES card. These PCI-based AES output cards, costing up to about $2k, might be better than a factory computer’s USB output, but they also have unacceptably high common-mode noise.

1. You have to be extremely selective when choosing a motherboard. A Baetis computer, any model, any price, always uses a motherboard that contains its own SPDIF output header. Only a few brands of MB, and only a few models of such brands, have this SPDIF header. The MBs that we choose not only have such an SPDIF header but also have one of only two HD Audio Codecs (the audio chip on the MB) that sound truly good for 2-channel audio. No DIY advisor or “expert” has written about the differences in Audio Codecs, nor do most DIYers have a good enough audio system to be able to perceive the differences.

1. Once Baetis started using these special Motherboards we learned over the next 4 years the following:

1. The speed of the CPU and the RAM was an important determinant in the quality of audio. Essentially, most all of the “purpose built computers” (including Aurender™, our main competitor) use CPUs whose speed is 2GHz or slower. Our models always use at least 3.0Ghz CPUs.

2. Same thing with RAM – the brand and speed of RAM can, in small increments, improve audio. For MCh audio through a really good pre/pro, faster CPU and faster RAM can be heard to improve audio quite clearly (with a decent pre/pro – not the MCh DACs in inexpensive receiver models that abound in many audio systems).

3. There is a general way to improve the SPDIF signal coming from the MB of these few brands/models of MB:

1. The inherently lower common mode noise of the SPDIF header can be lowered even further (significantly) by using a “daughterboard” in which the world’s best galvanic isolation transformers are used.

1. The right capacitors in the right circuitry can increase the voltage output of the SPDIF signal (well above the 30-year old “rules” that are still adhered to). Some DACs respond as if on cue to this increased voltage – producing much better 2-channel audio.

1. AES output can be achieved through proper use of those isolation transformers that will do a second duty -- transforming a 75-ohm SPDIF signal into a 110-ohm AES signal. The resulting higher voltage of the AES output is what the DAC manufacturers take advantage of – if the DAC designer is doing his job optimally. Our people are not DAC designers; we don’t know WHY this is the case. We just hear the result. Higher output voltage, other things equal, makes 2-channel audio via SPDIF or AES sound better.

352. The sound coming from a Baetis daughterboard can be significantly improved by using cryo-silver internal signal cabling (standard in our high-end models). Since we are talking about an AES signal, this benefit of certain internal wiring is NOT used in our lower cost models including the Prodigy-X model so highly reviewed by Kal Rubinson of Stereophile. If you want the very best AES signal, the cable from the MB to the Daughterboard, and the cables from the DB to the AES output connector, and to the BNC-SPDIF output connector, should be made of the very best cryo-silver wire (we use Revelation Audio Labs™ cables made especially for us) with the very best output connector quality. Incidentally, Kal is mostly reviewing MCh systems, which means he is NOT using our AES outputs and is also listening to bit depths that exceed that of a 2-channel, 192/24 high-def file. An AES output, as optimal as it is for files originating from redbook CDs and hi-def downloads (and even DSD converted to PCM), cannot produce native DXD (PCM at 352.8 or 384) or native, true DSD.

1. If you wish to improve the 2-channel audio quality of ANY Baetis computer, we have also learned that VERY good linear PSUs manufactured by HD-Plex™ can achieve even more improvement. But the amount of the improvement depends on the construction of the DC cabling coming from the external PSU to the Baetis, and within the Baetis, the quality of the cabling connecting the Neutrik™ Powercon™ DC connector on the chassis of the computer to the ATX board (DC/DC distributor) on the MB of the computer. Again, it costs more, but you can immediately hear the impact of using cryo-silver DC cabling everywhere in the DC chain.

4. All of this we learned by spending time and money up through about 2015. What we learned in 2016 and 2017, however, has made us what we now That is, we asked ourselves – what about the other major inputs into a DAC or pre/pro (or outputs from an audiophile computer) besides AES? – i.e., what about the way in which USB, HDMI, and Ethernet processes can be improved? Here is what we learned:

2016. USB’s major shortcoming is that common mode noise from the MB via USB or common-mode noise via a USB port on a PCI-bus-based card is simply too high. Some OEMs were experimenting with ways of having such signals subjected to galvanic isolation. The best of these methods became available in 2016. SOtM™, the maker of the best USB cards using a PCI-bus, figured out how to reduce common mode noise by using galvanic isolation techniques on a USB board that was NOT PCI-bus-based. The Board, and its two USB outputs, are put into a PCI physical slot, but connected instead directly to a USB header on the MB (rather than to a PCI-bus). Also, this card can be connected to an SOtM clock that improves the clocking of the USB signal from that of the computer’s motherboard. Two birds killed quite elegantly within a single solution. So, now 4 of our 6 models, large enough to fit the SOtM boards, can be fitted with this expensive but wonderful USB option. It is this SOtM™USBhubIN option that Kal Rubinson is, as of this writing, using for his Reference System via the Baetis Prodigy X model (for MCh audio into DACs or groups of DACs with acceptable USB inputs).

1. In 2016 we also learned that the special cryo-silver cable we used for our internal AES board wiring could also be used to power the SOtM USB card (that was non-PCI-bus-based). Kal Rubinson, who introduced us to the SOtM USBhubIN process, is one of the recipients of this discovery, which uses a Revelation Audio Labs 9v DC cable from one of the 4 DC outputs of the HD-Plex 400W linear PSU in order to power the SOtM USBhubIN cards in a Baetis computer. Similarly, Andy Quint of The Absolute Sound uses this SOtM USBhubIN process with a T+A™ DAC 8 DSD for true DSD music from our Baetis Reference 2 model.

1. What about HDMI? Would any audiophile actually use HDMI for hi-def audio? We think the industry has, again, missed the boat on this subset of digital audio issues. HDMI, besides its ability to send video, can indeed send a high quality PCM signal to a DAC. In fact, as always, the quality of the HDMI signal depends on a) the quality of the computer generating the HDMI signal; b) the quality of digital-to-analog conversion process in the DAC or pre/pro; and c) the quality of the HDMI cable, which is not as crucial for sending video as it is for the best audio. We’ll have more to say about this in the future, but we have already learned that some pre/pros (NOT the $20k ones, all of which have certain features designed for professional sound recording – production features that serve to hobble their ability to produce the best audio reproduction) can make a PCM signal sound wonderful in either 2-channel or up to 13 channels, when the signal is sent from one of our 7^th generation CPU-equipped Baetis models. The exact sound is highly dependent on the computer and pre/pro or DAC, and also highly dependent on the brand and model of HDMI cable. But what Baetis has learned is that some mid-level priced pre/pros can play the PCM signal via an HDMI cable in a way that sounds better than some higher priced DACs using AES!! This finding has implications for audiophiles on a budget, and implications for anyone focusing on multi-channel audio.

1. Finally, what about sending a computer file to the DAC or pre/pro via Ethernet cable? There are only a few DACs, and more pre/pros, that have an Ethernet input, and all of these will literally take only a true bit-streamed signal via Ethernet. That is, the Ethernet cable cannot deliver an AES or SPDIF signal. Rather, the DAC must receive the entire computer file, unchanged by any DSP in the computer, sent via Ethernet (when the computer file is either stored inside the computer or anywhere else on the home network). There are many issues that must be considered here, including whether the user wants to employ some DSP in the computer (such as, say, Digital Room Correction). But there are some DACs/Mch DACs that are truly good when used in this fashion – the most widely known one being NADAC™ (which Kal has reviewed very highly in its MCh incarnation). So, you guessed it. One of the improvements that Baetis will hopefully announce in the future will concern the best way(s) to improve the quality of the Ethernet signal carrying the bit-streamed computer file to the DAC or pre/pro via its Ethernet input. Our objective, as always, will be to make these Ethernet-based DACs sound their very best.

We have not touched on MQA™ in this brief discussion, but Baetis computers are all MQA-ready and Roon™-ready. The DAC must be MQA-capable, while the MQA Core-decoding can be done either in the computer (depending on software) or in the DAC. Not all purpose-built servers can handle all of these software choices. Still another reason to start out with a real computer when you are trying to get to the best 2-channel or MCh music reproduction.