Andrew Berliner – Comments On The Console

THE CRYSTALAB LOGO #3 – ANDREW’S FAVORITE

I never really found out who the incredible artist was who did the Crystalab artwork and comics. If you’re out there I’d love to know!

There are MANY more stories associated with the building of the console(s) and the rooms they were in — and that’s just for the 5+ years or so that I was there.

Here’s an example. Studio B was built by Bugs Pemberton, who, as far as I know, was a drummer doing a session, when he lent a hand putting up some wood shelves. Andrew noticed that he had a special talent for woodwork, (one of the understatements of the century) and asked him to build the studio. So he did an incredible job – notice in the picture (which also opens in a larger view to another page) the brown velvet-looking flat sections framed by the quarter-rounds: the velvet sections are wrapped fiberglas sections and the curved parts are cut sonotube pieces, with veneer glued to them. Every screw head has been flush plugged with a contrasting plug. The burl veneer affixed to the machined aluminum top and back of the console has many layers of a clear acrylic, all hand rubbed, and of course ALL the markings line up perfectly. And the amazing picture on the back wall, was placed there in honor of Stevie Wonder. The scene is composed of thousands of inlaid veneer pieces, all flush cut with an exacto knife — it’s about 7 feet wide. The different parts of the inlaid flowers are scented with essential oils, in honor of Stevie’s album, Secret Life of Plants.

Also notice the EMT-250, one of Stevie Wonder’s. When there were 2 units in the room, we would patch them in a sort of quad feedback loop, adjusting the digital attenuators 1/8 dB at a time, until myriad feedback loops would be produced, in an infinity of 3- dimensional psychedelic electronic music involvement! The bongos as a flower pot were a special touch of the era.

The monitor speaker cabinet was also a series of experiments; the sections are separate and we did some tests of what you might call a “mechanical time and phase alignment system” — the tweeter could be moved in and out of the cabinet a little bit while white noise was playing, and the imaging would snap to focus. This was one of the earlier tests of my white noise alignment system, a more recent version of which is presented in my “White Paper”, HERE.

The mixing board described here was the first to use 5534 opamps (and 530’s as well) in an audio product. We received special foundry samples color coded with dots and hand serial numbered of various test engineering samples most likely from Signetics. They sometimes came with a secret sheet of paper with matching colored dots which had pencilled-in explanations of the various characteristics of the sample IC’s. Almost all the IC’s were in ceramic DIP packages with gold mil leads. Similarly, the fets used in the attenuator were some special samples obtained by Carl Todd, one of the other developmental engineers, and a true genius in his own right. I can safely say that in many instances, it was the summation of all our energy that got us through the fine points of the intense R&D associated with this project.

For example, this was the first audio board to have both analog and digital clocking circuitry running around inside. We used to do extreme tests to determine (and make sure) that the digital switching current noise did not get into the analog portion. It was discovered that when the segmented LED’s used in the numerical display of the dB attenuation would switch, they would introduce noise. So we removed that section of circuit board, and Andrew and I developed a current mirror circuit and a slightly differing bypass scheme; once that was retrofit, you could literally open the gain up full on ALL the channels — including the submasters and main faders — and then hear absolutely no noise as the displays toggled.

One day Andrew had an idea to try a new bypassing scheme. We hopped in his car and went off to one of the magic surplus stores. Some of these stores knew us quite well, and we’d literally take a shopping cart around the back isles filling up small brown paper bags with all sorts of electronic part goodies. Try doing that today! So we returned with a bag of 10 mF tantalums. In those days, as well as today, sometimes the manufacturer marks the + side and sometimes they mark the – side. Andrew spent half the night laying on his back on the floor soldering in well over 100 caps. I came in and then he said “Let’s try it!” So he turned on the power and one by one, just like a sci-fi effect in a movie, every cap exploded — they were all soldered in backwards ! That was a very welcome moment in what had been a technically very tense couple of months!

According to my experiments and exhaustive tests, one of the reasons the mic preamps were so incredibly quiet, and the crosstalk so incredibly low, is a methodology I developed which to my knowledge no one else has bothered to do. The trick essentially is the mic preamps are inverting. Positive pressure on the diaphragm of a mic gives a + voltage on pin 2. The mic preamp has up to 70dB of gain. It is inverting. Then ALL the subsequent circuits are noninverting. That essentially means that any instantaneous current draw into the mic preamp section from the rails is matched and opposite as you go through the rest of the chain. Therefore the current modulation noise on the rails is nullified. However at the “end” you come out with the correct absolute polarity, since all the internal circuitry is unbalanced. It’s a unique trick. That’s one reason why the noise floor of the entire console was about -90 dBv and the clipping level at +32: that’s a 122 dB dynamic range, with ALL the channels set at unity gain, far superior to anything else. It’s also worth mentioning that the 0 reference level internally was -6 therefore between “0 VU” and clipping the headroom was 38dB. I am reminded that during one session in Studio B, Roy Thomas Baker was extremely annoyed he couldn’t get the equalizers to clip, even when he patched two channels in series! So much for one of his favorite British console tricks!

Nowadays the digital attenuator could be built as a substrate, an ASIC, as a LSI, etc. In those days it was on a rather large (11 inches!) PC board, all laid out visually so it could initially be experimented upon. Fortunately most of the breadboarding was “finished” and therefore not too many REV’s of the final PC board were necessary. Each section of the resistive ladder described above has a fet around each resistor. There is also an extremely clever part to the circuit which Carl Todd suggested, then we all put our two cents in… when the signal goes through each section of resistive attenuation, that means the FET is off, so the audio is going through the resistor part. When the fet “shorts” the resistor, the audio is going through the fet, therefore there “might” be some distortion. So Carl and Andrew developed a secondary opamp section with an identical FET (in fact the two FETs are on one substrate for thermal tracking) and the second FET is in the feedback loop such that its distortion cancels out any possible distortion caused by the first switching fet. On paper this seems subtle. On the layout board it was a lot of work, but with 12 of these sections in series, every distortion-lowering effort was worthwhile. When we measured circuitry like this, Andrew loathed THD measurements as much as I did. THD measurements are essentially useless, since they do not tell you whether the distortion is even order or odd order harmonic distortion, and they sound completely different. So we used the wonderful HP 3580 and later the 3582 spectrum analyzers for the “cleanest” measurements possible in those days; the only good thing we could do with the THD measuring devices was to use them as a plain ol’ AC voltmeter !