Ask the Doctors: Gain Control
By Si Moorehead
|
Part One--The 2-1176 as Stereo Compressor
The 2-1176 has a lot of things going for it over using two individual 1176LNs and an SA. What has changed to make the 2-1176 a kick-ass stereo compressor? First off, lets mention what hasn’t changed: It is the same great circuit as the 1176LN and the same exact board used in the 6176. Now for the improvements: There is a new stereo adapter circuit, the high-quality attack and release pots are ganged on a single shaft when in stereo mode to keep things in balanced and simple, a very strict gain control FET matching process, and a completely new set of tests for stereo tracking of the audio path and sidechain. I can’t emphasize enough the amount of testing that goes in to making a matched pair.
Before I can elaborate on why a 2-1176 is better then two individual 1176LNs linked using the external SA (stereo adapter) box, I'll give you a brief description of how gain control is accomplished and how you link the 1176LNs together. Before the 2-1176 came out, the only way to get a stereo unit was to buy an external adapter--a little box with a trim pot, 2 RCA jacks, a battery, and few passive parts. This box could be connected via the stereo link RCA jacks on the back of each 1176LN. These jacks give you direct access to the gain control signal, which drives the gate of the gain control FET. Internally, the SA is basically a battery and trim pot in a box, which ties the units' sidechains together electrically. By tying the two units' gain reduction control signals together, the resulting signal is a combination of both. Each unit contributes to the gain control signal depending on its individual settings (more on this later).
A basic understanding of how the 1176LN gain control is accomplished is necessary before I can discuss stereo operation. The 1176LN uses a FET connected a DC-controlled variable resistor to accomplish gain reduction. A tap off the audio signal is taken from downstream of the gain control FET. This signal is then rectified to a level that represents the audio signal. This control voltage is then used on the gate of a carefully selected FET, connected between the audio signal path and ground, downstream from the input attenuator, and just after the input transformer. This control voltage alters the resistance of the FET causing more or less signal to be shunted to ground. You can think of the FET as a level pot, but instead of turning the pot with your hand to raise and lower the level, the resistance of the “pot” is altered with a control voltage (the DC level on the gate).
Before the 2-1176 came out, the only way to get a stereo unit was to buy an external adapter …
The more negative the FETs voltage is at its gate, the higher its resistance. So we have to carefully choose FETs that have a very specific resistance at a specific negative gate voltage. Once the FET is in the circuit, we add a small negative bias voltage that causes the FET to sit at the desired resistance. As the audio level rises, it makes the gate less negative, lowering the resistance. The bias voltage needed varies slightly from FET to FET, even after a very strict selection process, so it must be compensated for using a trim pot in the unit.
It is very much a balancing act to get everything to sit just right. The old SA is a way to get two FETs that have slightly different bias needs to work together--at least for a little while. As mentioned earlier, the old SA ties the gates of the gain control FETs together. What you end up with (unless the compression gods smile on you and give you two perfectly matched FETs) is that neither FET is now correctly biased. The adapter allows the user to add a small static positive voltage to the FET bias circuit of the unit that needed the most negative bias voltage, essentially canceling out the difference in bias needed between units. The battery provides a perfect, not-ground-referenced source that can be added to the gate that needs it. Which unit is the one that needs the voltage added to it? You just have to plug it in and try. If it doesn’t work, reverse the RCA cables and try again. Not exactly an elegant solution, but the best that could be done at the time for a reasonable price.
As you can imagine, you are going to need to trim the units as the battery dies over time. This continues until the battery no longer has the required voltage and you start over with a new battery. By going through all this, you are guaranteed that both units will start compressing at the same time on the same input signal, but not necessarily that they will track over the useable range of compression. There is no guarantee that they will both compress 10 dB for the same input if they both have the same “zero” starting point. Experience has shown that they track pretty well, but not perfectly. Once again, you are at the mercy of chance.
The new stereo adapter and testing processes fix these issues. The SA portion of the system has been greatly improved. Instead of using a battery to inject the needed voltage to get the FETs to match perfectly, the new circuit uses an isolated DC-DC converter. It takes in the already double-regulated DC voltage and creates a floating (not ground referenced) DC voltage in the proper range. Yes, even after a very strict matching process, a very small adjustment is still needed. This just highlights exactly how sensitive the system is. Unlike a battery, this voltage will not change over time to any significant degree, and will give years of adjustment-free use. The other feature in the new stereo adapter circuit is a switch that allows you to choose which unit gets the small DC bias added to it. Should you ever need to change the FETs, all you need to do is flip the switch to the side that needs bias compensation added and adjust away.
The tracking of the compression curves in the 2-1176 is accomplished with testing and matching FETs. The matched FETs are installed and calibrated, and compression curves are run using an Audio Precision test station. They must track perfectly or the process starts all over again with another “matched” pair. You may wonder what exactly a matched pair is. When matching them outside the unit, they are tested at certain points to make sure they react to a DC level in the same way. The difference between doing this static test and running actual compression curves is the number of data points. The compression curve tells you the real story, and in a 2-1176 they lay right on top of each other. This means your stereo image will remain true. If they don’t track well, your stereo image will shift from side to side, depending on which side is being compressed more or less.
The other aspect of having two sidechains tied together is that they both contribute to the single gain control signal as mentioned above. If each unit has its attack and release settings at different points, you will get the sum of those two signals controlling both units. If the attack is set faster on one side than the other, the side with the faster attack’s contribution to the gain control signal will be to reduce transients on both channels by the amount that specific unit is contributing. From a practical-use standpoint, stuff starts to get wacky really quickly, and it is almost impossible to get both units set exactly the same. Unlike input and output levels, you can’t get an easy read on attack and release settings.
Besides the technical issues, it is also a pain to have to keep jockeying back and forth. On the 2-1176, the attack and release pots are ganged together when in stereo mode. The attack and release pots on the left unit control both sidechains. When in dual mono mode, each channel uses its own pots. On the left channel, these are the front pots of the ganged pair. When stereo mode is enabled, the right unit's attack and release pots are switched out of the circuit and the rear half of the left channels ganged pots are switched in. This allows you to control both channels' attack and release with one set of pots and have them track.
The final piece of the 2-1176 design is a comprehensive stereo testing routine. We test each unit to make sure it functions as two mono units, and then do a full battery of tests to make sure that in tracks in stereo over the entire frequency range on every compression ratio.