Whilst attending a gear demonstration by Burl Audio alast week, I was able to approach the founder of the company, Rich Williams, for an interview. I had always heard glowing comments about Burl products, and now I had a chance to probe into the technical side of why their converters have been so acclaimed. Burl has been on their second tour as a company, showcasing their products at a range of studios around the world, which nicely lent itself to a discussion about their gear.
Hi Rich. My understanding is that you went to school for Computer Engineering. Was your entry into the audio industry through that education?
Yes, it was. I wanted to be involved in the music industry, but I felt like engineering was a good job skill to have in general, and if I could use it in music then it would be a bonus. I went to school for Computer Engineering and ended up working in Silicon Valley as an engineer, but I didn’t really like it all that much, so I started my own studio in Santa Cruz and became the lead singer of a psychedelic funk-rock band.
And how did you transition from engineering in Silicon Valley to engineering audio hardware?
Over time, the property that my first studio was on got sold, so I had to get a part-time job to live off until I could build another studio. I was lucky enough to get a contracting job with Universal Audio, where I was given the mission of designing an A-D/D-A converter, which ending up being the UA 2192. Whilst working on that, I had the chance to listen to a lot of electronic components and became familiar with the musicality of certain ones, versus the non-musicality of others.
Can you give examples of the ones that were musical?
When I was introduced to Class A components, which have no cross-over distortion in the signal path, versus the Class B ones that most consumer electronics use, things started to sound significantly clearer. That was the direction I felt digital recording needed to go, with an emphasis on the quality of components in the analog section. In most converters, the analog sections are very cheap and consist of too many components in the signal path, which adversely affect the tone of the signal. So I discovered that a minimalist Class A circuit sounded far better than a convoluted Class B one.
I also discovered that removing capacitors from the signal path and using transformers greatly improved the performance, due to something called “phase response”. It’s a poorly understood concept by a lot of recording engineers, most of whom think about music solely in the frequency domain – but music also has a time domain. The relative time that the high frequencies and low frequencies reach your ears is important for whether you perceive the sound as realistic or not. In general, if the high frequencies lead the low frequencies in reaching your ears, then the sound feels far more unnatural than if the two were to hit you at the same time, or even if the lows were a little ahead. When we hear someone speak, we hear the fundamental frequency from their chest in real-time, along with the harmonics coming out of their mouth and nose. But with a lot of electronics, the phase response is such that you hear the nasal part before the fundamental. No instrument works like that, be it violin, bass or guitar. You need the right phase response, not just the proper frequency distribution.
The other thing is that a minimalistic circuit path sounds better than a cluttered one because each component has its own sound that colors the signal even more. It’s very challenging to find electronic components that sound natural to our ears. Even if certain components test a certain way and look good on paper because they have low harmonic distortion and noise, they aren’t able to deliver when you place them in the circuit itself. A classic case is that you can plug in different transistors that tested the same into a circuit, and yet each one sounds different in the signal path. So finding components that not only test good, but also sound good is the challenge.
(Above: Rich Williams)
The contrast in effectiveness between capacitors, transistors and transformers seems to be a central part of Burl’s product design. Can you expound on the difference between these components a bit?
All electrical components have a sound or color, and capacitors have the most color, so eliminating them was one of our most important steps. Capacitors have more color than transistors or transformers. After that, transistors are the most colored, in my opinion. Tubes are quite transparent and clear, even though they have a reputation for sounding colored. That’s only because they have limited headroom, meaning they produce a colored tone when you drive them. A transformer, however, has very little tone to it if it’s wound correctly and is big enough.
Transformers do a few things which are beneficial to the signal path. Firstly, there are high-pass filters in most signal paths that cause the low-end to lag behind the high-end, and transformers correct the phase response issues that result from that. Secondly, transformers act similarly to tape machines, in that they turn an electric field into a magnetic field and then back again. In that process, the magnetic field acts as a shock absorber to the transients. For example, rock music can be very loud and intense, with heavy drumming and loud guitars. Tape is what smoothed out the sound of that and made it palatable. Without tape, the volume and high frequencies of rock music would have sounded much harsher. So transformers are musical in that they dampen the harshness of a recording. A violin is a good example of a single instrument that in reality can sound very harsh, but it gets smoothed out when recorded to tape. A transformer in the circuit path of an audio converter has a similar effect.
A lot of the engineers who have purchased Burl products have often said that because of the Burl Mothership, they no longer have to record to tape. Is that related to what you’re saying about transformers?
Yes, because a transformer is more effective than a tape head. A lot of recording equipment, from dynamic microphones to speakers, work in a similar way: they use magnetic fields to transfer signals. In the case of tape, the amount of magnetic field that is transferred is much less than what a transformer can do per channel. When recording with a tape machine, the thicker the tape and lower the channel count, the higher the resolution of your recording. For instance, a one-inch two-track sounds better than a half-inch two-track. The reason is that there’s a better magnetic coupling with the size increase, which leads to more linearity and better transfer of the signal, which means the sound quality is a function of size. So what we do at Burl is use really big transformers that’s the equivalent of recording to a four-inch two-track tape machine. The coupling is much greater, and that makes it superior to tape in that you get better resolution and less noise. And it’s also tailored to the modern age of hotter signal levels. In the digital domain, we have more headroom than what tape has, so recording hotter is common now. But if you tailor the transformer properly, then it saturates like tape at a higher level, which is ideal for digital recording. That’s what we have at Burl. The BX1 transformer we use is specially formulated for today’s digital recording.
Most manufacturers use what I would consider small transformers. Again, the more surface area you have, the better the coupling and the response of the lows and the highs. We use bigger transformers to get extended lows and highs, and I think other manufacturers give transformers a bad name by using under-sized ones.
You worked to design the Universal 2192 in the early 2000s, which was a unit that had a good run and is well-known. Can you talk about why it received such acclaim and how the Burl B2 Bomber has been a progression from that?
Well, when I designed the 2192, I spent a lot of time listening to different electronic components. This was when I became interested in Class A circuitry, discrete transistors and circuits without capacitors. I think the magic of the 2192 was that I figured out how to eliminate capacitors and how to have the circuit be direct-coupled. But it still didn’t sound as good as tape because of what I talked about before: phase response, transient response, saturation, etc. So the low-end sounded right, like with kick drums, but things like guitars never sounded right. The reason is that guitars have a lot of harmonic content, and tape naturally evened out the distribution of that, giving it an even spread. So what I never liked about the 2192 is that we were never able to address that aspect of getting distorted guitars to sound good through it, which is very hard to do. So when I started Burl, I was hard-pressed to top the 2192, and after a lot of testing and difficulties, I decided to try using a transformer in the signal path, which worked really well. That made the difference I needed for the B2 Bomber.
Was swapping out transistors for transformers against the prevailing engineering conventions of the time?
Absolutely. One of the most ridiculous things is the notion of “transparency”. When it comes to recording, there’s no such thing as “transparency”. The only thing your ears perceives as “transparent” are the sounds from the real world going directly into them. Outside of that, no signal path in a circuit has any “transparent” component. Everything adds tone, from wires, resistors and capacitors to converters and microphones. The “art” of recording music is to make the artist sound good coming out of any pair of speakers. You start with a microphone and end with a speaker, and you don’t just end with one speaker, but with all the possible speakers out there in people’s homes, workplaces and studios. So you have to get your recording to sound great under a million different conditions that you have no control over. That’s what tape did. It allowed engineers to make their recordings sound great under a variety of conditions because of it’s phase and transient response. So when I heard the effect of the transformer in the circuit, I was like, “That’s it! “. People are either trying to emulate the sound of tape using plugins, or they record to tape machines and dump it to digital. They’re trying every way possible to record to tape without actually recording to tape, yet the most obvious thing to aid that process is to use a transformer.
Another thing to keep in mind is that when you convert from analog to digital, you’re attenuating the signal, so you don’t need any voltage gain in the signal path. But a lot of converters have voltage gain in their signal path, which they later have to attenuate down. Burl’s A-D converters don’t have any voltage gain in the analog sections, only current gain, which adds to the quality of the sound. Even the Universal 2192 had gain in the input stage, which made it sound a little harsh, or what people call “digital”, though what they’re actually referring to is bad analog electronics.
Inside most A-D chips, almost all companies use the same components. It’s what you surround those components with and how you configure them that determine their outcome. It’s like a General Motors 350 engine. There’s a million different ways to put the carburetor and exhaust systems around it, but the better you optimize the motor itself, the more you’ll get out of it. It’s the same with converter chips. I’ve just optimized the chip.
Apart from transformers, what are other components that you’d have to pay attention to in order to preserve the density of sound going through your converters?
The transistors. You can’t just put any transistor in there. Every one has a unique sound. For our transistors, we have a JFET stage followed by a BJT stage, and you couldn’t just switch them around and expect the same sound, or even trade them out for other transistors that spec the same. Even the same transistors with the same part number, but from different manufacturers will sound different. So there’s an art to finding the right type of transistors and finding the right operating point for them and the conditions under which they perform optimally. Sometimes those conditions might be to receive a super hot signal with a lot of current. The issue with gear today is that a lot of it sounds better when it’s burning a lot of power.
I’ve heard of such things with regards to recording consoles. The original Neve desks sounded great because they used components that lent themselves to pleasant distortion, whilst the new consoles by the likes of SSL are aiming to be pristine-sounding, which has been less favorably received.
Right. When we talk about “pristine-sounding” gear, what we mean is stuff that’s been measured to sound pristine, not audibly perceived to be so. Those two things are very different. Tape is perceived to have a lot of noise and harmonic distortion, compared to digital, but we don’t hear that noise when we listen to the records. When you listen to Led Zeppelin or Jimi Hendrix, do you sit there and get annoyed by the hiss of the tape? Of course not. It has no effect on your enjoyment of the music at all. And those older machines from the 70s didn’t spec nearly as well as new machines do. Nowadays, the analog components are spec’d with crazy -120 dB THD plus noise, but that means nothing if they don’t sound good.
I’ve heard a lot of engineers say that when they switched to Burl, the sound of the recording became more “three-dimensional”. What are they trying to say, technically?
That’s the phase response – the relative speed of the highs and lows reaching your ear. When you have an analog circuit that has poor phase response, you don’t hear the density of the sound correctly, so it starts to sound two-dimensional. We don’t just hear sounds with our ears; we feel them in our bodies. When we speak to each other, you may think that what you’re hearing is just highs and mid-range, but you can also feel the low-end coming from a person’s chest. Direct coupling and using transformers gives a proper phase response where things sound three-dimensional, since you’re now able to hear the low-mids and low-end in time with the highs.
I’ve talked to a few other hardware companies about how they made their gear, and many of them say that you really have to listen to how components sound in a circuit. I can relate to how mixing engineers listen to elements of a mix to make the right value judgments, but how do you as an electrical engineer listen to the sound of a single electrical component?
We always do A-B listening tests between the components that we already know and the new one we’re testing. Then you make incremental improvements from there. With the 2192, we started with an A-B comparison between that and what became the B280C. So you tirelessly do comparisons with every instrument, from kicks, snare and hi-hats to bass and guitar. You do that for weeks, months or years until you find something better than your initial component. And also, the ultimate criteria is whatever new component you put in the circuit versus not having that component there at all. That’s the ultimate test.
Do you feel like under your current constraints, Burl has maxed out on innovating with the products you’ve released?
I don’t think we’ve maxed out, but no matter what tools you use, you have to learn how to use them in conjunction with other tools. For example, can you record to a Mothership from a mic and pre-amp without using any EQ and compression? No. Those are necessary tools. Even with tape, you’re limited by the different tape formulations, speed of the tape, etc. With 15 ips, you´ll get better low-end, and with 30 ips you’ll get better high-end. So either way you end up compensating with other gear. So I’m constantly going to refine our products, but the imperfections in the tools make them special anyway. The reason why great music of the past sounds great is the imperfections and tonality of the gear, whether it’s The Beatles or Jimi Hendrix, or even today with someone like Jack White. They work along with the flaws of the gear.
Do you know if Burl products are used on Jack White’s music?
Third Man Records has a Mothership, and I know he’s used the B1 for his guitar, though most of his own stuff is recorded to analog tape. But Vance Powell, Jack White’s engineer and producer, is a Mothership user.
How did your first “Flight of the Mothership” tour in 2011 fare compared to the current tour?
The Mothership had just been released prior to the first tour, and things have changed a bit since then. For example, only three people showed up to our Nashville stop in 2011, and now we have 30-60 people show up to the same event. But the three people from the first Nashville stop were Joe Chiccarelli, Vance Powell and Michael Wagener. From there, the Mothership exploded in Nashville. So we planted a lot of seeds during that first tour, and now we’re reaping them. When we stop in here Ibiza and twenty people show up to an event at Sonic Vista Studios, that’s miraculous. Amsterdam and Zurich also had large turnouts, even though they’re considered smaller markets. So our gear has spoken for itself. You can’t sell recording engineers on gear. They have to hear it and fall in love with it. The good ones know how to discriminate with their ears, and when they hear our stuff, they get it. In fact, the older and more experience they have, they quicker they hear the difference.
Is success for Burl mainly measured in how many units you’re able to ship, or are there other factors?
As someone who used to be a recording engineer and producer in Santa Cruz, working in the industry was frustrating because my town was a small market and your efforts can go nowhere in a place like that. Had I gone to Nashville or LA, things might have been different. So after my band, I decided to focus on Burl, and now our gear is used on a large portion of today’s music. Every year, a number of Grammy-nominated albums are created with Burl gear. An example is Chris Stapleton, who’s been working on many multi-platinum records out of Nashville. It’s all done with the Mothership, and mastered and cut to vinyl using Burl products. All of that is very satisfying for me. It’s an accomplishment for us to have survived this long, since 2006. We survived the economic downturn of that era, and now we have a company that’s well-respected and helps artists make records that people want to listen to. In the end, what matters is what people want to listen to. Back in the day, you could only listen to what was released at the time and played on radio. But I grew up in the 70s, when music was great, so it was no problem to listen to Stevie Wonder, Elton John, and The Rolling Stones. But now, new music has to compete against old music, and the vast majority of the market is now older music. When younger people get a taste of that older music, it’s only then that they realize how questionable some of today’s music can be. So Burl wants to make music that can be ranked among the legends of the past, and I think that’s what we’ve accomplished.
With pre-amps, converters, control room monitors and a mix bus among your products, is there any temptation for Burl to make a recording console?
Absolutely, but from a business standpoint, you can only bite off as much as you can chew. Every time you introduce a new product, it has to be backed financially through amassing an inventory. Sadly, doing a console probably isn’t our best business move. Using a standard form factor like we have with the Mothership is the best way to go. Wit that, we have a chassis that we can pair with our modules in a standardized way. If we design a console, which we’d have to keep in stock, it’s a massive investment. Imagine buying one console as an end user versus having to buy twenty of them as a company; you have to put a ton of money behind that. Also, if other companies like Neve and API are already doing a good job in that field, then we might want to leave it to them. I just worked on a modern API console in Amsterdam that sounds great. Obviously the old Neve Electronics stuff sounds great, and even the newer Rupert Neve stuff is awesome. The console industry isn’t going to be as big as the industry for A-D/D-A converters, because everyone needs converters. Sonic Vista Studios is the perfect example of that – they don’t have a console. They function like most modern studios by working in-the-box, so they need a quality analog path that goes in and out of the computer, and that’s where we can affect things with the Mothership business model. But if we ever get more money, I’d love to do a console.