GAIN11

The Art & Science of Audio & Video

John Oram

A RECIPE FOR SUCCESS

Courtesty of: prosoundweb.com
John Oram now puts his name to his own range of mixers and studio outboard, after a varied career including many years of design experience with some of the big names in British audio. PAUL WHITE enjoys a slice of sonic history.

John Oram’s life is a bit like Dr. Who’s TARDIS — it doesn’t seem big enough to hold everything he’s managed to cram into it! He’s played drums for Marianne Faithfull, helped Stan Cutler develop the first wah-wah pedal, created Reflex active guitar pickups (and still makes active electronics for electro-acoustic guitars), been a restaurant owner and chef, run a marine navigation school, owned his own recording studio and been instrumental in a number of groundbreaking audio products, including Trident’s first parametric equaliser and a number of Trident recording consoles. Now, thanks to the excesses of American PR, John’s been landed with the title ‘Father of British EQ’, which he accepts in a kind of tongue-in-cheek way, but there’s nothing frivolous about the way he approaches audio design.

John now has his own company building analogue mixers and studio outboard gear, but he started in the industry working for Vox, back in the heyday of the AC30, after gaining an HNC in Electronics. From Vox he moved to Trident, where his relationship with the company lasted 14 years. He designed their respected parametric equaliser as his first job after joining. Now John’s customer base reads like a Who’s Who of top engineers and producers, and includes the George Lucas Skywalker Ranch. Obviously, there’s something a little bit special about the Oram approach to electronics, and I was keen to find out what it is. John started by filling in a little of his history.

EARLY DAYS

“I cut my teeth with Vox, in the 1960s, where I worked on amplifers, organs and instrument equalisers, but as bands got bigger, more sophisticated equalisers found their way into PA equipment. EQ is all a matter of degree — instrument EQs might have boosts of up to 24dB, whereas musical EQ requires more like 6 to 10dB of boost. Trident saw my first exploration into meaningful EQ, where I had to ask myself what EQ really meant. Was it just another tone control, or did it have to add something? What I learnt early on is that with EQ, it’s often more important to take away rather than to add, and I developed a design brick that used subtractive EQ rather than conventional boost. For example, I found that if you were to take away the two ends of the spectrum and leave the middle, that was as quiet as at unity gain. To create EQ boost, you can take away sections of the frequency spectrum, then place them in the negative feedback loop so that you’re effectively improving headroom while maintaining low noise.”

Much is said about the way the shape of EQ curves affects the subjective sound. What’s your view on this?

“EQs designed by purely theoretical engineers tend not to make the most interesting musical sounds. A theoretical EQ design might produce the classic 6dB/octave slope, but when I started playing around with EQ for Trident, half the design was done on paper and the rest by listening. With the ratios of values I arrived at, virtually all my EQs were more like 4dB/octave in real terms, which was really a function of circuit loading, which a ‘by the book’ designer might well have tried to avoid by using buffers. However, providing the degree of loading doesn’t increase distortion, you find you have a smoother curve that can sound more musical.
“There were lots of state-variable-filter circuits around at the time, that required dual-gang pots, but I developed a frequency-sweep section that used a single-ganged pot hooked into a parallel T-filter circuit, allowing the notch to be swept over the 10:1 range demanded by the design brief. The first console I ever designed for Trident was the Fleximix, and one of the briefs was a given price point. The EQ spec was to include a high shelf, a low shelf and a sweep mid, so I wondered if I could do that without using any more op-amps. One of my design goals is also to use the minimum number of active elements. If I can get away with a piece of wire and a resistor, I’d rather use that than an op-amp.

“The real novelty of that Fleximix mid band was a circuit that satisfied the 10:1 sweep requirement with a fixed Q, but because it had no active gain around it, the slope was no steeper than 4dB/octave. This worried me at first, because I thought the customers might want to focus in on sounds more, but one of the major customers of Trident at that time was Autograph Sound, who were using the console live for West End musicals. It turned out that what they really loved about the console was the mid-band EQ, because they could use it to add colour and warmth without provoking feedback.”

What did studio users like about this EQ approach?

“If I were to criticise other designers, not mentioning any names, some have gone for every EQ bell and whistle in their mixer channel strips. There’s no technical reason why you shouldn’t do this, but musically I think simplicity plays a big part, and the number of guys who come back to me having used the Oram stuff, and the Tridents, tell me that the more channels they put into the mix, the bigger the sound gets. With other consoles, they found that the more channels they added, the thinner the sound got. Why is that? There is no such thing as a zero-phase analogue EQ, and if you’re going to put variable-Q sweep filters in there, that EQ has to be used very carefully, or when you bring the mix together, some signals will be out of phase with others at certain frequencies, which is why the sound gets thinner.

“I ensure that the phase shift in my high-shelving EQ is equal and opposite to the phase shift in my low EQ, because I know that most people tend to add some top and some bottom at the same time. When the Trident Series 80 got a reputation for good tight mixes, one of the measurements I made was with the EQ set up by engineers so that I could see what was really going on. In fact I’ve never insisted that the frequency response of the EQ in its flat position should be perfectly flat. One of the things that was criticised by several reviewers on the Hi-Def EQ was that in the flat position, there’s a little boost at the top and a fairly steep droop at the bottom. However, I think that actually sounds better when used in conjunction with other gear — it’s not that we can’t make it flat, but there are times when you have to apply a philosophy which doesn’t conform exactly to the textbook.

“With EQ, I’ve always believed that if you want a good bottom-end response, the HF has got to be smart — you have to have good transient response. Similarly, if the HF is going to sound good, the LF has to be controlled as well, which is where our OramSonics concept comes in.
“I’ve always known what sounds good and what doesn’t, and in my early days when I played drums professionally, one thing that always hit me about a good studio installation was the hiss. In those days, if you could hear hiss, it meant the system had some HF, whereas these days hiss would be a bad thing. But what always sounded good to me was a sound with lots of bottom and lots of top. And when you look at the way DJs or live sound engineers set up their graphic EQs, the middle’s pulled down and you have that classic smile curve or swan’s wings.”

NO LIMITATIONS

“When I was with Vox, I worked with a guy called Dick Denney (he designed the AC30 guitar amp), who was a brilliant and totally empirical engineer with no theoretical knowledge to speak of — as proven by the explosions that used to come out of his lab! He’d come out covered in soot and say, ‘Whatever you do, don’t connect this electrolytic capacitor [a 25 volt!] to a valve anode with 300 volts on it!’ But he taught me that there were no limitations — it didn’t matter how you did the job so long as you got the sound you wanted.

“Back then, everything was tube technology, but when transistors came along Vox had the first transistor amp in production, called the T60. What was apparent was that while valve circuitry used very high-value resistors and very low-value capacitors, transistor circuits worked at a lower impedance and so used lower value resistors and large-value capacitors. Valves are voltage-driven devices and transistors are current-driven, and it was a bit of a problem, because both Dick and I agreed that the transistor EQ sounded rather feeble compared to the valve equivalent. Valves also had that inherent odd harmonic distortion, while transistors produced even harmonic distortion, which wasn’t very pleasant from a guitarist’s point of view.

“At the time it was a mystery, because although the time constants in the EQs were exactly the same on paper, the valve version sounded noticeably better. When I moved to working as a consultant for Trident, it turned out that my EQ circuits used larger resistor values and smaller capacitor values compared to most designs, and that again had this characteristic ‘better’ sound.

“It wasn’t until 1987, when I started the Reflex company making active guitar pickups, that I had access to a computer with a decent amount of power — a 286 PC with a 20Mb hard drive and a whole megabyte of RAM — and that was brilliant, because there was a lot of useful mathematics that I had studied at college, but which I hated because it was hard to do. With this computer, I could manipulate numbers easily, and I discovered some quite useful tools that allowed me to delve into my circuits to analyse what was going on. I also had a circuit simulator package which, amongst the obvious parameters such as frequency response, calculated group delay — the inherent time a transient takes to pass through a circuit. It was intriguing to see that you could get this sub-low lump of time delay that might be as much as 20ms at 2Hz, and it really got me thinking. I put all my existing circuit designs, and those designed by Dick Denney, into the computer, and they all showed this same characteristic. Substituting more conventional values for resistors and capacitors showed that the effect was no longer there. Sometimes with electronics we take too much for granted, and it can be beneficial to go back to basics.

“The way I visualise alternating signal voltages is as DC levels moving up and down — they’re simply being adjusted from one level to another at different times and different rates.”
Does working with what are inherently high-impedance filter circuits mean that you need to use additional buffering in your designs?

“Where needed, yes, but sometimes you can use the loading effect of the circuit to advantage, in order to produce gentler shelving characteristics. Now we see to it that if we build even a line amp or a buffer stage, we purposely put in enough LF group delay to make it sound interesting, and we’ve never had a situation where we’ve had too much group delay. We get a lot of feedback from users, and when they say our EQ passes the blink test (a snare drum playback makes you blink!), or when a user says the LF is so tight and solid, I know we’re doing something right. Moneypit in Nashville, who could afford to have any mixer they want, decided that ours was punchier than their existing Trident desk, which I can only put down to the Analogue Devices input amplifiers we use and the surface-mount technology, because nothing else has really changed that much. Some people are surprised that we use TL074 op-amp chips in the signal path, but I was actively involved in the design of that particular chip, and the Trident Fleximix was the first console I know of that used op-amps at all.”

OP ART

Why is it that op-amps have such a bad name in purist audio circles?
“People push them too hard — the spec sheets show impressive figures, so designers take them at their word and try to get as much performance as possible out of a single op-amp. Both Hugh Ford and, I think, Rupert Neve, were quoted in published articles before I did the Trident design, saying that op-amps were all but unusable in professional audio. If you were to take a standard 741 op-amp (which the data sheet claims has 70dB of gain) and try building an amp with 70dB of gain, you’d find it had 4% distortion, a noise floor of -50dB, and an HF roll-off 3dB down at 9kHz — appalling. In the Trident Fleximix design, I took two 741s (which was the op-amp of the day), ran them at 35dB of gain each, and got a response flat to 20kHz with 0.05% THD and a low noise floor. Now that we use surface-mount technology for everything, all the op-amp chips are actually quad op-amps, so you have four devices per package to play with.”

One of the comments I hear from designers is that in EQ circuits, op-amps run out of gain-bandwidth at high cut and boost settings.

“The reason that happens is that the conventional design is arranged so that when you want more EQ boost, the op-amp has to do more work. We, on the other hand, employ negative feedback in the EQ, so when you ask for more boost you’re asking the op-amp to deliver less. Going back to the AC signal as a moving DC level, whatever circuit it goes through has to be able to go up and down at the same rate. You have to decide where you want your gain and how you’re going to use it. If you want the noise and distortion to go up as you increase the EQ gain, then fine, because that’s the usual way of doing it, but if you stand on your head and do it the other way around with cut in the negative feedback circuit, it’s never any worse than the unity gain condition.

“There are times when I cascade gain stages, like with the console buss mix amp. Most mixers use resistors coming out of each module that go into a virtual earth mix amplifier, and the more channels you have, the more distortion and noise you get. You also get 180 degrees of phase shift, which you have to put right elsewhere in the circuit, so my approach might be to make the first stage of the mix amp a times 10 attenuator, by making the feedback resistor 10 times smaller, which makes the circuit quieter, and the crosstalk less, because the buss impedance is lower. That’s one of the reasons we can use ribbon cable busses and still get such crosstalk figures. To get the gain back up, the mix amp looks straight into another op-amp with a gain of 10, which brings the overall gain back to unity and puts the phase of the output signal right at the same time.”

I guess it’s the same with mic amps where you use an Analogue Devices chip, albeit a special version customised to your requirements, instead of the more usual discrete transistors.
“The ultimate criterion is: can we make something a musician will get excited over? Of course, we also sell to movie companies and production companies, who’re not so interested in music but are interested in a low noise floor and low distortion, and we can satisfy them too. The ultimate priority, though, is sound — that’s what it’s all about, and I think our mic amp design performs extremely well.”

What are your future plans?

“At the moment, we’re working to finish our Series 48 digitally controlled analogue console. That’s been a massive project with a lot of investment and we have customers wanting them as soon as possible. There may well be spin-offs from the digital control circuitry used in that console, but at the moment we have to get it finished. After that, I’d also be interested in looking at analogue-to-digital converters, though I don’t believe digital audio will really be any good until we use much higher sampling frequencies.”

Advertisements

March 2, 2008 - Posted by | Pioneers in Audio | , , , , , , , , , , , , , , , , , , , ,

No comments yet.

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: