generic VariMu sidechain buffer

Hi. I am trying to come up with a simple DC amplifier for buffering variable mu Sidechain before it reaches the compression tube grids.

This is very useful for a series of reasons: faster attack times, simple ratio control, lower load on output tubes, etc...

But the circuit should stand up to -50v output for driving remote cutoff tubes to its maximum compression.  Also, it should be able to provide DC gain (for ratio control), and have low output impedance.

This could be made into a small board and used to upgrade existing vari mu comps.

I was thinking about a pair of transistors with a feedbak loop, but I wasn't able to come up with a functional circuit, as I have no experience with DC amplifiers.

So, in the end, I am using high power opamps, like the LM3886 (overture series / gainclones) for this purpose with a single negative supply, connected like in the picture attached. It shows a simple DC amp with a gain of 10. I used an audio amplifier because that is what I know.

I though about posting and asking for opinions. Maybe someone knows there is a much better chip for this job, or maybe someone could help me come up with a simple discrete solution, like a transistor pair, that could be simpler and cheaper. A FET hiZ input and  powerfull Darlington output  sounds to me like a good idea. But I wasn't able to come up with a working discrete solution...

If we're using transistors in the sidechain why not make the control voltage positive and bring it to remote cutoff cathodes. Then we wouldn't even necessarily need input transformer for the vari gain stage.

Here's one example.
http://ghr.fi/varimu/varimu.pdf
Here the max control voltage is about 20V but the same idea could be used for as high control voltage as you want.

I am not sure an LM3886 will do a right thing with its input essentially AT its positive rail. At least you might need a +5V rail.

> a pair of transistors with a feedbak loop

Three transistors makes it much more precise. 100V transistors are readily available. If you do not have a +55V rail, increase the upper resistors in proportion. For+250V try 2.5Meg and 250K.

Thank YouTube, PRR. In fact I have tried Thiago solucionar, but I was using a five transistor discrete opamp. This one that you posted makes It much more viable.

Would It be possible to use just one rail for this, grounding the positive rail? I though so, as it doesn't need to output any positive voltages...

Hmmm... Maybe I wrong...

As it is a mic input compressor, I could use the Phantom PSU, making somewhat beafier...

Here's one example.

Click to expand...

Looks very nice, innovative.

I don't see a rectifier on that sidechain. Care to explain???

Also, I see RV10 as threshold, RV8/9 as ratio and RV5 as attach.

But what does RV4, RV6 and RV7 varies as pots?

I don't see a rectifier on that sidechain. Care to explain???

Click to expand...

Output of Q4 and Q5 is unidirectional pulses.

Also, I see RV10 as threshold, RV8/9 as ratio and RV5 as attach.

But what does RV4, RV6 and RV7 varies as pots?

Click to expand...

RV10 DC threshold
RV8/9 AC threshold
RV7 ratio
RV6 release
RV5 attack
RV4 is U2, U3 bias adjustment. It also affects threshold when Q3 starts conducting.

The sidechain designs is based on PRR's designs from 12 years ago https://groupdiy.com/index.php?topic=11469.msg135692#msg135692

> it doesn't need to output any positive voltages...

It gets tougher to swing TO the rail. How close does it have to get? Since input range and output range may include zero, we may work transistors at zero C-E. Less than roughly 0.1V a lot of parameters fall-off. Less than 0.050V may be impossible unless we hand-carve the transistors (as IC designers can).

But it would be nice to have a module that could be used with single or dual supply without any mods. It turns out  that the best I could come up with is using that simple UREI opamp schem:

with single supply, like you see in the picture attached, would get me up to about -8mV.  That is not entirely bad in a vari mu compressor sidechain, as I could get rid of the cathode bias resistor in the input tubes and use that small voltage as the bias in the zero compression point (although using a common cathode bias resistor makes CMRR and thumping canceling better).

We could also use the 6v or 12v heater supply to as a low volts sidechain buffer positive rail, so we can get confortably to zero volts on the output.

I couldn't figure out how to save one transistor by eliminating the VAS in that opamp schem without messing with the output bias. I like the PP output. That means more current drive and lower attack times possible for hard limiting.

Professor PRR, if you have the time, please, show us your implementation of the special single supply version that can come very close to zero volts. 0.1 or 0.05 would be negligible. If you could make it with a high power/current output, would be even better...

I aso made a higher current version of your first proposed schematic, just making all resistors 1/10th of the value. I really want very fast attack.

P.S.: what is that diode doing in the output??? 

> PP output. That means more current drive and lower attack times possible for hard limiting.

Then I do not understand what you are doing. Surely it is cheaper to do you timing at low level, then use a moderately fast DC amplifier to boost that timed signal up.

And even for high-level timing, for most audio, attack is much faster than release. Your PULL must be large but the push can be small, probably passive.

> save one transistor by eliminating the VAS in that opamp schem without messing with the output bias

I can not get that scheme to sim properly. For -0.2V to -5V input it does alright, but below 0.18V it SLAMS to the -50V rail. Q1 has no Vce left. Some fiddling could reduce this on sim, but in real life it is liable to cut-off the quietest passages. (Call it "noise gate"?)

The fix for the near-zero case can be found on the LM380/386. Quasi-Darlington the inputs. There's a small error near zero, which may be fiddleable, but I don't think it will SLAM until the input goes 0.5V positive.

The diode on the plans with +50V was to catch the output so it could NOT drive the vary-gain grids positive, which might be distressing.

It seems to work just fine in my LTspice sim... Maybe my transistor models are playing tricks on me???

Only drawback is that is starts at -730mV when feeding 0v to it. But one could get rid of the cathode resistor and use this as the tube bias. It is very close to what I am using anyway.

Look:

BUT... there is a sweet spot where both 0v points meet, by using a positive 0.74v supply.

Hmmm... How much trouble would it be to make a 0.74v regulator? 

I was REALLY inclined to use your simpler 3-transistors version. It has only 120mv difference at the beginning of the curve at 0v, and I could easily compensate for that by making the cathode resistor a little smaller.

BUT, it shows an anomaly in the curve after the output reaches 29v. I don't have a clue WHY!!! But it does - at least it does in LTSpice with standard models.

On the other hand, my circuit has a "lag" in the starting point of the curve, so it would probably frustrate my desire for fast attack times anyway, making it all worthless.  :-[

OK, Tried the Urei opamp with dual supply and added a 560k resistor on inverting input to +v to compensate for the zero difference. Curve looks almost perfect. I think I am ready to breadboard that thing now.

Hi did you get any further with this
very interested to see how you would implement this with  a vary mu circuit