Igor Levin Talks About the Zodiac Platinum DSD DAC


 

 
What inspired the development of the Zodiac Platinum DSD DAC?
 
Well, as you know we’ve been shipping our very very successful audiophile DAC, which is the Zodiac Gold and having shipped that product for a number of years now we have learned a great deal about what it is that the customers actually want in a DAC.
 
So, Platinum, that grows out of our original work over the Gold contains two aspects to it, it’s got the evolutionary development and it’s got a revolutionary development of the Gold converter. What regards to the evolutionary stuff is it mostly comes from listening to the customers and taking into account their suggestions, for example people said they want to have DSD – we’ve said OK fine we’re gonna’ put in the DSD… They talked about how they would want to be able to connect 10M to our DAC the way that pro audio folks are currently doing – we said OK we’re gonna put in the 10M input. They talked about the fact they love our relays and the precision and volume adjustment, but the relays are a little bit noisy – you can hear the clicking, so we’ve went on and we have redesigned the relay controls choosing the relays that don’t click. And then other things, including the power supply which is very very major. People wanted to have a specialized power supply and we went on and we have improved Voltikus to now Voltikus 2, which is bundled with the product. We have also improved the cabling and the connections that are beefed-up between the power supply and the converter. And even small things like say the remote control – some people felt that the angle-range was a little bit restrictive, so we redesigned the remote control now it works from a great distance. And then comes the second aspect, something engineers get excited about is the revolutionary stuff. We like to be able to introduce something that’s unique and we’re always looking for some ways to make a big difference so when we came out with the Gold we came out with the USB supported 384 kHz and we were one of the first people to have ever done that. So, when we went ahead to do the DSD, we said well, we don’t just want to do a DSD, we want to do the best DSD. How can we improve the DSD and that’s where we came up with this idea of the DSD upsampling which to my knowledge is extremely unique. So now we can say okay we’ve got the best PCM DAC and we got the best DSD DAC out there.
 
How does the quad DAC architecture affect the perceived sound?
 
The idea is using multiple DACs working in parallel – by utilizing this parallel combination what actually happens is that the noise cancels out whereas the signal reinforces, so this is a technique to improve signal to noise ratio as well as it actually improves stereo separation because the left channel DACs and the right channel DACs are no longer in one case in one chip, they’re actually different chips, so this improves stereo imaging, it improves signal-to-noise ratio and the linearity of the DACs.
 
What are the benefits of the upsampling done by the Zodiac Platinum DAC?
 
The whole idea of upsampling occurred to me when I was looking to improve the DAC conversion quality, which I started by doing this Quad technique, that I mentioned earlier. In the process of actually paralleling the DACs, I realized
there is another limitation which is the quality of the filters inside the DACs themselves.
 
Many people don’t realize it but inside a D to A converter chip the first thing that the chip does is it takes the audio signal and upsamples it, so that all modern DACs work on that basis. They do not convert at the audio rate, they actually convert at a rate that’s typically say, 64 times faster than the audio rate and so the first thing that the chip does is it goes through an upsampling filter.
 
Then I started thinking about this upsampling filter – how much does that filter affect the sound quality. Is that filter as good as it can possibly be?
 
One thing you realize, that when you’re a chip manufacturer and trying to make a chip like a DAC, there’s only a limited amount of space you have on the converter, plus you need to be concerned with the issues of heat and power dissipation. So when you’re making a chip it’s an art of compromise, you’re not necessarily going to be able to put the best digital circuit, the best upsamplers out there because when you use more bits, which is something that improves the sound quality you’re gonna need to dissipate more heat, you’re gonna need more space on the chip. It’s something that the chip manufacturers cannot afford. So I realized really have a great chip – and I’m very very happy about the T.I. chips that we have been using top-of-the-line TI DACs, and this chip has one of the best analog parts of the circuit possible. I really like the sound quality. That’s why I stick with that and not use another converter manufacturer which is currently a bit of a fad. But the issue is the upsampler, so it goes back to the upsampler and so what we did is that’s where I started thinking about the upsampling and how important is the upsampling and what can you do with the upsampling. We have chosen an algorithm that we believe is the best sounding, the most audio-like algorithm in reconstructing the wave and the precision that we’re carrying out is much much higher than is done on the chip. For example the chip interpolators are typically computed with twenty-something bits 22, 23, 24, 28. We’re actually computing this thing with 64 bits. It’s a lot more accurate computation and that’s what makes the difference when you’re computing the missing points and with you being able to fit the curve afterwords, which is done by the analog part of the D/A converter.
 
Why isn’t the chip doing the same thing that we’re doing?
 
Well, as I mentioned the issue is space, the heat and the cost-effectiveness. We’re using a whole big FPGA to do these computations whereas on the DAC you just have a few square millimeters of space to fit this and it’s simply not possible.
 
How did you arrive at the idea of using DSD upsampling on the Zodiac Platinum?
 
When I started working with DSD, I realized that there’s the whole aspect of DSD that makes upsampling even more necessary, even more relevant to DSD than it ever was to a PCM. Well, why’s this? Well, the DSD is a one-bit signal. It does its trick by means of upping the sample rate at which one bit conversion is performed. In the standard DSD, the conversion is performed at 64 times the sampling rate and it’s got only one bit to play with and because it tries to construct all this dynamic range that would have come from having multiple bits, you really need this very very high frequency at which this one bit is manipulated, so the thing about the DSD, this technique of constructing multiple bits from just one bit – the higher the sample rate the better it works. So that if we could have recorded DSD instead of the 64 times using a 64 time DSD rate or single DSD rate, some people do like a dual DSD rate which runs at 128 times but the higher you run it the better it gets, so that’s just the way – that just comes out of mathematics of sigma-delta modulators. Once I realized that, then I said well OK even in cases where the original material was not recorded at higher rates we can still use the same mathematics that we use in the upsampling with the PCM in order to produce higher sample rate DSD and that’s going to make a big difference when the DSD is reconstructed. So, in a way I would go so far as to say, well we’ve built the best DSD that’s ever been done by actually improving it through this upsampling process.