"Distortastudio" Colour Design Pt. 2 April 17, 2015 17:58
In last week's post we identified three distinct distortion sounds of the Portastudio: mixer-only, tape NORMAL speed, and tape HIGH speed. We asked you which sound you preferred and the consensus was that... there was no consensus and we should try to incorporate options for all three sounds.
So we're going to try our best to make a Colour that incorporates all three tones of the Portastudio. Hopefully we'll be able to fit all three modes onto the board along with jumpers to set your preference. Since all three modes include the mixer-only circuitry, we're starting our circuit R&D there.
The Portastudio Channel Strip
After confirming that it did indeed sound awesome, we began our investigation into the mixer section by opening up the 464. It's actually a pretty complex piece of gear; there's a forest of film and electrolytic capacitors, a few big noise-reduction chips, the motorized parts for the cassette deck, etc.
The mixer itself however, is fairly simple. The green board you see here is the underside of the mixer. It consists almost entirely of opamps and resistors (including potentiometers) with a few caps in the filter section. On top of that is the a sub-board for the balanced input circuitry, but since that comes before the stages we're distorting we can ignore it.
It's simple enough, in fact, to identify the major circuit blocks without doing any reverse engineering. An adjustable gain stage (Trim) is followed by a couple adjustable filters (EQ section), followed by a passive fader (channel fader) followed by a buffer, followed by a passive pan control.
Just in case there was something fancy going on, we ordered the service manual from eBay so we could look at the schematic on paper. As suspected, besides the discrete balanced input circuitry (which we're ignoring), the 464's channels strip is basically a series of basic opamp circuits. And from what we can tell from other schematics available online, every model and revision of the Portastudio uses roughly the same approach for the channel strip.
So, nothing fancy about the circuitry we're distorting here. Just some textbook opamp designs being driven really hard.
Operational amplifiers (opamps) are relatively small, cheap integrated circuits that provide different amounts of amplification depending on the circuitry you place around them. They're kind of like cheat codes for electronics: with just an opamp and a couple resistors you can get a very stable, low-noise gain stage that would take 10+ parts and a lot more design time to make with discrete components.
So how do we make an opamp distort? Opamps have a hard limit to the amplitude of signal they can pass. This limit is set by the voltage of the power supply voltages (also called "rails") and the opamp's own ability to "swing" within those rails. For example, the power supply of the Colour 500 Palette is the 500-series standard of +/-16VDC, or 32 volts total. Most opamps can swing within roughly 2 volts of each power rail for a maximum peak-to-peak voltage of 28V, or +22dBu (for a point of reference, that's 18dB above professional line level). That's our hard limit; nothing will be amplified past 28V.
Let's see what happens if we feed an opamp a signal that's 3V peak-to-peak and set the opamp to a gain of 10x. The opamp will try to drive those 3V peaks up to 30V. But it can't! Anything that should have been above 28V gets "rounded down" to 28V, while everything below 28V gets amplified normally. In other words: clipping! In other words: distortion!
Like microphones or preamps, opamps come in a bewildering variety: low-noise, low-power, high-current, wide-bandwidth, etc. And also like microphones or preamps, there's much debate over how significant the audible differences are between them. Some people (often musicians) claim to hear a "night and day" difference when they swap the opamps in an old piece of gear. Some people (often engineers) claim there's no difference at all.
I've spent a good bit of time listening to different opamps in a controlled environment and my take is that audible differences between opamps are vanishingly small when they're operated in their linear range. When you start to ask them to do stuff beyond what on their spec sheets, all bets are off.
So, swapping an opamp that's just a buffer between two stages in a compressor? Probably not gonna hear much of a difference. Swapping opamps in a circuit with a hot input signal and excessive amounts of gain? Check out the samples below.
The first chip we tried in our mixer-section prototype was the trusty TL071. At "normal" signal levels, it sounded very similar to the actual TASCAM channel strip. But at even moderate distortion levels there was a gross crackling on the transient hits (this was partially due to our using only one opamp for this prototype, but it was nonetheless more pronounced with this chip).
Then we swapped the crackly TL071 for a UA741, truly one of the crappiest, most obsolete opamps (on paper) still manufactured. Voila! The crackle was gone. But so was a lot of the high end and, wow, that hiss!
Finally, we tried the obvious: the NJM4565, the same opamp used in our origninal 464. Now our prototypes started sounding almost identical to the original channel strip. (The crackling was still there--better than the TL071 but worse than the UA741. But we figured out how to eliminate the crackling completely by using two opamps and described above.)
So next time someone tries to tell you that all opamps sound alike, tell them they're just not driving them hard enough!
The Circuit So Far
To get maximum clipping and character out of our opamp, we're both a) applying copious gain to get the signal into clipping territory and b) regulating the power supply voltages down to lower the clipping point of the opamps. Here's our circuit so far:
You'll notice that there are two opamp stages. We did this because we found that, while we could drive a single opamp into clipping, the gain required came with too much noise and other nasty, non-musical artifacts. By spreading the same amount of gain across two opamps, we get the same amount of distortion with far less noise.
The ICs you see in series with the power supply rails are fixed regulators, which simply knock the supply voltage down from +/-16V to +/-9V. This gets us in the clipping range of our 464 Portastudio, which is powered with +/-10V.
Roll the Tape
We feel we're pretty dang close with the mixer section. The current circuit with the NJM4565 opamp sounds very close to our 464 channel strip. All that's left is to nail down the exact internal gain settings so that users will have a wide range of tones with the Colour Palette's Saturation knob. Next week, we're onto the "tape" half of the circuit. We're aiming to incorporate the both dark, grainy tone of the normal-speed mode, and the brighter, but more rounded (than the mixer) sound of the high-speed mode. We just brainstorming right now, so if you have any circuit ideas we're all ears!