When we reviewed the Chord Hugo it was immediately clear to us that this was not just another product, it was something very special. Subsequently, Edd Harris interviews the transportable Chord Electronics Hugo DAC/AMP designer; Robert Watts. In this interview we ask Rob questions that take us through the entire design process. At the same time we discover what led Rob to become one of the worlds finest DAC chip designers and uncover what might be coming in the future from Chord Electronics. Throughout this interview you will be able to understand just how important Hugo has been to the international HiFi and Pro Audio communities. Finally, we exclusively uncover why Rib prefers Red Book to DSD…
Considering that you are now one of the worlds leading authorities on DAC Chip implementation and designs what was it that initially started your curiosity in this complex area?
Way back in early 1989 I was dead against all things digital, it to me, sounding unmusical, hard and aggressive. But I heard the first PDM CD player using a new PDM chip from Phillips, and it blew my mind – it was smooth and musical, but had the low noise benefits of CD. So, I designed a DAC using this chip, and it went into production in 1989 – the PDM1. Now PDM is the same technically as DSD, but ran at a much faster 256 times oversampling. As time went by, better devices came from Phillips, but I noticed that the noise shaper resolution was a limiting factor. The PDM 1024 dac for example had a number of Phillips noise shaper chips, each being fed with a different dither signal, and this sounded a lot better. So I wanted to improve on the noise shaper, and also it had other issues – being very sensitive to jitter. In 1994, larger FPGA’s were starting to become available, and I realised the only way of solving the resolution and jitter problems was to design my own noise shaper. So in 1995, pulse array DAC was born, which had a 16 element discrete DAC and an FPGA noise shaper. When I think back, it was breakneck speed – going from disliking digital to designing my own DAC, and creating a new DAC architecture in six years. Considering I started as an analogue engineer, with limited digital knowledge, that is pretty fast.
Before project ‘Hugo’ begun what was it that you wanted to achieve and was it intended to for Hugo to be the epitome of ‘disruptive technology’?
I have always being pushing for better performance as ultimately I want to enjoy the music more. I didn’t expect Hugo would be revolutionary in it’s performance – I knew it would be better, but I expected it was just going to push the envelope, not too break thorough and experience a new vista!
Hugo is an immensely complex design and must have taken years to develop, what makes Hugo so special and why was it that you wanted to produce a reference quality portable DAC?
Hugo was kind of an accident. Now I had been upgrading all the digital modules that goes into a DAC over the last 6 years, to go into a reference class DAC. All of the code had been completely replaced over this period, and it was a major exercise. Also this reference DAC had a large number of improvements put into it.
Now John Franks had identified that a high-end portable DAC amp would sell well. I wanted to do such a project, simply because as a consultant to audio chip manufacturers I spend a lot of time in planes, and I want high end sound on the go. So in Autumn 2012 we went ahead with the project. At this time, Spartan-6 FPGA from Xilinx was now available in quantity. The Spartan-6 LX9 had more capacity than the FPGA I had for the prototype reference DAC, and it was much lower power. So this was the choice for Hugo, and Hugo got all the code that had been intended for the reference DAC.
Considering the technology inside, what was the main aspect of design that allowed Hugo to be produced for a fraction of the cost?
In one word – knowledge. My day job is to supply and create IP for audio chips – but audio devices are driven by cost, so the designs have capacity limits. But using an FPGA removes that capacity bottle neck – at higher cost – but simply having an FPGA is not the answer, you need a lot of knowledge and IP to make a high performance DAC using FPGA’s. But of course, the FPGA is essential too!
We know that you’re a fan of Red Book, but why not DSD considering that one of Hugo’s biggest attractions is it’s sampling capabilities?
OK, I am not a fan per se of red-book – its just that the problems of red-book are dealt with more capably than before with Hugo. I have heard very good DSD, particularly with ’70s classics that have been converted straight from analogue tape to DSD. But I have also heard awful DSD too.
My problem with DSD extends from the work I do on noise shapers, and DSD64 and DSD128 is technically severely limited. The format is stuck with an innately limited performance, but DXD standard, for example, has no format restrictions – its limited by the ADC and DAC implementation, not the format. To some extent, this is true with red-book – red-book is more limited by implementation – I am sure that red-book, properly implemented, will out perform DSD.
The reality is that 99.99% of audio in the real world is 44.1 or 48k. That’s not going to change in the foreseeable future.
Now Hugo has created a storm, what’s next and what would be your ultimate obtainable achievement?
It’s no great secret that the replacement to the QBD DAC is well under way. This has an FPGA with nearly 10 times the capacity of Hugo’s and it is proving to be very interesting.
The next big thing to me are DAC’s that can drive loudspeakers directly, and that have a single global feedback path from loudspeaker OP to digital domain. Hugo is a 1W digital amp, so upping the power is very much the goal. I have been working on digital power amps for the last ten years, and there will be some interesting developments in the coming years.
I have no idea what the ultimate is – a perfect product would be one that was completely transparent with no added audible changes. Since the perfect product doesn’t exist, we have no real idea what is possible. What treasures remain to be heard from ones old recordings? I don’t know. What I do know, is that I am still learning new things, and making sound quality improvements, and I will keep doing this until I lose my faculties, or can no longer dream up ways of making improvements!
If Hugo was an animal, what animal would it be?
Oh dear, the most difficult one. My eldest son said a panda, because he likes pandas. My youngest a duck, but quack-quack is his favourite toy. My wife said sloth, but I think she was thinking about me… To me, butterfly came to mind – the exquisite and detailed symmetrical patterns, the tremendous variation in colour, and the way it can delicately and rapidly dance around.
What would you like to say to the people who have a lot to comment on Hugo, but are yet to hear it in the flesh?
Listen to it!
Finally, in one word can you describe how Hugo sounds?
No! Because it is multi-faceted. And also puzzling. When I finally heard the initial prototype with Hugo’s initial code, I was gob-smacked. Where did that come from? Now I knew it was going to sound better than Qute, the last DAC I designed, and the one I use in my office. But listening to Hugo on my modest office system I was completely speechless, and I just was not expecting it. I had heard the modules before in isolation, but Hugo was the first when everything came together, and it was a very big surprise, as I thought it would just sound better than Qute, but it had a massive improvement in musicality, and I just was not expecting this. In musicality terms, it was like a light being turned on in a dark room – music was much more engaging, more enjoyable, and Hugo had a very addictive quality – I just wanted to hear more music. Oddly, there was no listening fatigue at all, and I could play music in the background even when solving complex problems. It is also incredibly revealing, but manages not to over analyse – it does not force detail at you. You can hear rapid changes in tempo and rhythm easily – it can sound very fast and light on its feet. One of the major changes was piano, which before sounded like a jumble of notes – but now one can hear each key starting and stopping easily, with the timbre of the piano coming across really naturally.
So the puzzle is this – Qute is very similar technically to Hugo, but Hugo is radically different musically, and I am not sure why a certain technical performance translates into crossing the Rubicon musically.