Interview with Paul McGowan of PS Audio

August 31, 2010


Paul McGowan of PS Audio has agreed to an exclusive interview with the Ultra High-End Audio and Home Theater Forum to discuss his cutting edge PerfectWave Transport and PerfectWave DAC, both of which have been receiving enthusiastic reviews since their recent introduction.

Thank you for agreeing to be interviewed for the Ultra High-End Audio and Home Theater Forum again. Your recent introduction of the PerfectWave DAC and the PerfectWave Transport has a lot of people talking, and I am sure our members are anxious to learn move about them. Let me begin by asking you to describe generally the PerfectWave DAC.

Thank you. I appreciate the opportunity to share with you and your readers. The DAC’s one of the most exciting products we’ve ever made. In fact, the PerfectWave Series is really the most exciting audio product I’ve ever been involved with over the last 35 years.

Let me give you a little history. We started this project a couple of years ago and the impetus for it was based on a comment one of our dealers, Listen Up in Denver, made. One of the owners, Walt Stinson, called me and asked if I had any clue why the Sonos music server they were selling didn’t sound as good as a CD player plugged into our DAC – playing the same music in the same format and resolution. Walt had experimented with hard drive based music systems and knew that hard drive recorded music tended to sound better than what comes off the CD – and he was a bit baffled as to why it seemed to get worse when sent over the network through the Sonos.

This then started our quest which has ended in the PerfectWave series. We knew that technically there’s no reason why digital audio should sound different – regardless of how it is stored or transmitted. This means that it shouldn’t matter how you store it (hard drive or optical) or transmit it (optical, coax, network, internet). If done properly, then it should all sound the same.

After initially examining what was happening in the field we came to a conclusion: most of the music servers out there then (and today) were designed by computer companies and not audio companies. The computer people don’t seem to pay a lot of attention to issues we in the high-end audio community are familiar with and we reasoned this was one of the main stumbling blocks for music servers and network connected DACS.

We set out to build a system that stayed true to our goals: playing back digital audio with the highest level of audio performance possible, regardless of how that music was stored or delivered. This meant we had to first design an optical player that proved this (the PWT memory player) and secondly, we had to design a high-end DAC that could resolve these differences. The result is the PerfectWave DAC.

Many of our members really like to get into the inner workings of their audio components. Would you tell us what DAC chip you selected – Burr-Brown, Wolfson, AKM – and why you chose it, focusing in particular on its sonic advantages?

We chose the Wolfson 8742 because of its sound and because of its advanced filter set. The Wolfson family is at the cutting edge of technology while the current crop of Burr Brown’s is at the end of its life. We’ve been using Burr Browns forever but once the new Wolfson set came out it was a no-brainer. They just kicked butt!

What’s interesting is that the actual DAC chip is not as important as you might think. In a DAC the specs are getting near ridiculous with signal to noise ratios well beyond the capacity of humans to perceive the small differences. What’s really important are the filters and – most important, the analog stage at the output of the DAC. This is where the magic happens in a high-end DAC. Between the power supply and the audio stage, the biggest levels of differences can be heard.

In the PWD we went to a great deal of trouble to build the finest power supply and analog output stage we’ve every built in over 35 years of building amplification stages. Jung regulators, all discrete through hole parts in the signal path such as PRP resistors, film capacitors, low ESR devices etc. No surface mount parts are in the signal path whatsoever and the topology itself is fully class A and balanced with extremely low feedback. The circuit is direct coupled from input to output and uses an invisible sounding servo circuit that allows this DAC to go well below 1Hz. Bass performance is simply stunning.

I understand that the PerfectWave DAC can be connected directly to a power amplifier, thereby eliminating the need for a preamplifier. Would you tell us how you implemented the volume control? Is it analog, digital or a combination of the two?

It is digital. This tends to make people nervous when they hear this so please allow me to explain. The way this control is implemented there is zero loss of resolution in the loudest 50% of the control, thus for any reasonable listening level there’s no loss of any kind. For lower levels you do have a one or two bit lost but who cares since the level is low enough?

I think it’s important to realize that a preamp adds a significant amount of coloration to the signal. You can certainly set our control to 100% if you want to go through your preamp (and some do) but in my opinion, there’s no preamp like no preamp.

The PerfectWave DAC accepts up to 24-bit/96kHz data streams over USB. Is the USB interface itself unable to handle higher resolution digital signals? Did you consider using a Firewire interface in addition to or in lieu of the USB interface?

The USB input is unable to handle higher but that limitation is not in the DAC, it is the interface itself. USB is not a great medium to start with but we did want to make sure that if you use USB you didn’t have to downsample like you do on any number of other DACS. Firewire’s an option but we did not include it because with the introduction of the upcoming optional network Bridge, there’s no sense to it.

Once you implement the Network Bridge, all bets are off. The Bridge will be able to accept 192kHz 24 bit data and has a Digital Lens built into it. This means that any data coming over the network will sound as good as if it were played on the PerfectWave memory transport. There will be no loss, so this is the best solution.

The PerfectWave DAC incorporates a Digital Lens. I take it that this mechanism has its origins in the Genesis Digital Lens from the mid-1990s? What improvement have the PS Audio engineers brought to bear on this venerable technology?

The DAC does not, but the transport as well as the Bridge does. Indeed, this is based on the work we did on the original Genesis Digital Lens. Each unit has one built in (the transport and the Bridge). It is an intelligent buffer, in essence. We use this giant buffer to store the raw digital data, then rebuild that same data on the output of the Lens with a low jitter asynchronous clocking system and then deliver this data via either I2S or S/PDIF.

The beauty of the Lens is that once we get the raw data cleaned up, stripped of all its timing and jitter-prone components, it’s placed into the memory. Once it’s in the memory we can easily say it doesn’t matter how it got there: via taking off an optical disc, or a hard drive, sending it over the network or the internet – it really doesn’t matter. Whatever gets into the Lens, comes out the same. Perfect.

You can hear the results in the PWT. Even Red Book CD’s played on the PWT are jaw dropping. Most people who hear the PWT for the first time are, frankly, stunned that it could sound so good. They never realized how CD’s could sound.

You implement apodizing filters in the PerfectWave DAC. Would you outline their operation and advantages? Do you offer an option to select from among different filters?

Yikes. That’s a whole interview in itself and we should probably reserve space for this discussion. Basically, this type of filter is designed to eliminate a very unnatural sounding artifact of filtering: pre-echo. As you probably know, digital filters (linear phase filters) have a strange attribute where they do everything right – linear phase and group delay – but at the expense of pre-echo. Pre-echo is like what you read about in modern physics where something appears in two places at the same time – one event happening slight ahead of the next. So, instead of ringing (after echo), pre-echo actual is a type of ringing that actually happens before the actual sonic event. This never happens in nature and so the ear finds this quite unnatural. The apodising filters eliminate or reduce this effect quite nicely.

Indeed, you can choose auto filter or select any of the 5 choices you wish.

You have some interesting views on upsampling, stating that in some cases upsampling digital signals can make them sound worse. Can you elaborate on your ideas and experiences in this area?

Upsampling is digital manipulation. It does not generally sound good. One of the worst things it does is change the timing on digital data (and the group delay and phase response). Upsampling is used by 99% of all DAC manufacturers (including our older products) as a means of getting rid of jitter. Upsamplers do a nice job of eliminating jitter, but they do so at the expense of the performance.

This is easier to hear than explain. If any of you have a chance to audition a PWD, try it in Native Mode (no upsampling) and compare to any upsampled frequency on the DAC. The vast majority of people that try this always go for native mode when there’s no jitter to start with.

That last sentence is the key “no jitter to start with”. So, if you’re listening to the PWT – then there’s no jitter to start with and Native will sound better. If you’re listening to the Bridge, same thing applies. If you’re listening to one of the other inputs, then one of the upsampled choices might be preferable – although it is indeed a compromise.

Tell us about the Network Bridge addition to the PerfectWave DAC. When will it be available? In particular, tell us about the user interface. Many of us use iTunes or Media Monkey and are curious how we would access our files.

The Bridge should be available at the end of the year and initially we plan on having an iPhone app that you can download to access your library. The iPhone (or iPod Touch) is a killer way to go because it’s cheap ($199 for the Touch) and brilliant in the way it works. It will be able to access your library if it is on a NAS or the computer. It is a fully DLNA UPnP compatible system so any of these aforementioned servers (Twonky) among others will work just fine.

More info as it is unveiled, but needless to say all you’ll need is a NAS and a network connection and you’re good to go.

You can also access the library from the PWD’s front panel touch screen.

Moving on to the PerfectWave Transport, would you describe its unique features compared to a standard transports?

Sure. There are so many, where do start? J First, let me describe 99% of all CD transports/players so we’re all on the same page.

In a transport DAC setup, the transport acts as the master to the system. This means that everything is clocked from the transport. The clocks come from the CD or DVD mechanism’s electronics which are all built in using a self-contained mechanism.

Pretty much every single player/transport ever sold uses and off-the-shelf CD or DVD mechanism made by any number of companies, but all pretty much based on a couple of reference designs from Sony and Phillips. Regardless, whoever’s brand they use the mechanism is an electro-optical device that is self contained. This means we give it certain commands and it handles the rest and spits out data and clocks that drive the system to us. All the error correction, speed control, servo, laser, clocks etc. are all handled and controlled by this self-contained mechanism and other than mechanical issues such as damping and mounting, there’s not a lot engineers can do to make it better or worse.

So the very first difference is here: in the mechanism which is the heart of any CD transport. We decided early on that building the PWT could never happen if we used an off-the-shelf CD or DVD mechanism our own error correction and our own data collection without regard for any timing information.

So, we chose instead to use a DVD ROM drive as a start. Using this device means that we control every aspect ourselves. This decision allows us to use a form of EAC for error correction (where we go back and make multiple comparisons of the data until we’re convinced it’s correct) and, of course, place everything from the drive into our famous Digital Lens and then out through a completely asynchronous clock.

What this means is that we have a true memory player. Anything you play on the PWT is completely stored in the memory and you never listen directly off of the optical disc. You can even eject the disc on the PWT and the music will continue to play for several minutes as an example.

Lastly, as a nod towards the wave of the future – we spent nearly 4 months of engineering time programming our device to be able to read WAV files up to 192kHz 24 bits right off a DVD. This was really a huge deal and we have the first transport/player to ever accomplish that. One other company has followed suit after our launch but their player is many times more expensive than ours.

One format unavailable through the PerfectWave Transport is SACD. What factors lead to your decision not to incorporate it?

A couple. To read SACD we’d need a drive mechanism that could read it. There still are no optical ROM readers other than a Blu-ray drive that can do this. Secondly, there are so many DRM issues involved that we felt it would be too compromised to really be what we want. I wouldn’t discount it for the future – I see that Oppo has taken an off-the-shelf Blu-ray player and made their digital out available on HDMI. We’ll watch and see if they get in a tangle with Sony or not. In any case, using an off-the-shelf drive mechanism will never work for us.

From the internal pictures of the PerfectWave Transport you appear to have incorporated a standard computer DVD-ROM drive. Have you made any special modifications?

Nope. Fortunately there’s no need to. The drive is only a slave and there’s zero processing going on. We take the raw IDE data out of the drive and process it ourselves in our FPGA (Field Programmable Gate Array) which is the heart of the PWT. We can substitute any number of drive mechanisms in here and they all sound the same. The beauty of this system, among the many, is that if there’s ever a breakdown of the ROM drive it’s easy an inexpensive to replace.

How do you achieve bit-perfect transfers of data from optical discs without using error correction?

We use something similar to Exact Audio Copy (EAC). In this system there’s no interpolation or guessing. It works by comparison. So the ROM drive reads a section of the disc and places that data into a temporary buffer. Then it is told to go back and read it again. If the data comes back a perfect match, then the original data is transferred over to the internal Digital Lens and accepted as valid. If it does not get a match, then it goes back and reads it again and again until we have that perfect match. We can look up to 80 times before the data is determined to be corrupt for that word and we move on. This is a very accurate way of retrieving data – and it’s one of the primary reasons we had to build a memory transport as we never know how long it will take to get accurate data.

Do CD transports sound different and, if so, why?

Indeed they do and up until the introduction of the PerfectWave transport you could easily suggest they all sound different: mostly not great. The biggest reason they don’t sound great is the fact that they read right off the disc, go through a short buffer and then the data and clock signals are fed into the DAC. This becomes the master for the entire digital audio system. The quality of the clock and the data make a huge difference sonically. By using the off-the-shelf mechanisms – even the great ones like the top loading Phillips unit – you are stuck with accepting the crappy master, bit and word clocks spewed out by the mechanisms. The fact is, mechanics plays a little role in the performance, but only in keeping the clock problems to a minimum.

By using an internal Digital Lens for the data, the quality of the drive and its electronics is completely ignored. It doesn’t matter about clocks or timing or anything at that point. In fact, as I mentioned in the first part of this interview, the Digital Lens is the key to making this all work. The Lens makes it possible to not care how the data is put into the Lens: from an optical disc, from the internet, from USB, from any means possible, once it’s in the Lens memory it’s all the same. It is the quality of the fixed asynchronous clock that makes all this wonderful sound possible.

The fixed asynchronous clock on the PWT is the one thing an ordinary transport or player can’t really take advantage of like we can. Why? Because the only way to keep jitter low is to have the output clock fixed and not related in any way to the incoming data (which is what asynchronous means – not synchronized). The problem with just throwing an asynchronous clock onto a standard drive mechanism is there’s no Lens in between the two. You can’t know what speed the data is coming off of the optical disc and you can’t control it. The only way this works is to have an intelligent memory setup that grows and shrinks as needed to feed the output clock perfectly.

Jitter is always a hot, and hotly debated, topic among audiophiles. Tell us how you have addressed the issue in both the PerfectWave Transport and the PerfectWave DAC. At what point do you believe jitter becomes audible?

Jitter is very audible and I think I’ve covered it pretty well above. If you use the PerfectWave System (and it is a system – although each piece is just fine independent of the others) you won’t have jitter to deal with.

What comes out of the Transport is jitter free and we can use our HDMI connection to deliver I2S data to the DAC. I2S data is what is the inherent internal operating system within every CD player and contains a separate bit clock, word clock and master clock along with the separate musical data. Once you scramble all this together in the S/PDIF format for sending over a single cable, you’re going to introduce jitter. In the PerfectWave system that’s not necessary and so we can send jitter free audio right into the DAC and it sounds glorious – no oversampling necessary thank you!

The same will be true with the Bridge. Whatever data we get from a NAS or computer or even the internet will go through the same Lens as is in the PWT. The clocks will be stripped out and the data stored until it’s needed. It’s really a gorgeous system.

Is there anything else you would like to tell us about the PerfectWave DAC and Transport?

Get one or at least get in a position to try it. Listening to red book CD’s on the PWT is a treat. You will simply not believe just how great this 20 year old medium is until you hear it on a PWT and PWD combo. These are truly ground breaking products and we’re very proud of them and very excited.

Thank you again for joining us today.