A different type of comp. (possible future project)

There are lots of opto, fet, VCA and varimu discussions here but there is a kind of comp that was just mentioned maybe two or three times.

There was posted once a schematic of the whole thing and now there is the complete manual at Jakob's page:
http://www.gyraf.dk/ look at "resources" at compressor section for the 1960 (I intentionally not mention the name of the product, to make our discussion confortable)
For a long time I read at different forums about the 1960, 1969 and 1968 type comps. There is some consensus about the fact that although the 1960 is a good comp it is not flawless so later they moded it and finally they throw away a lot of things, moded others and it looks like the 1968 is an extremelly good comp. In the new 1968 they use Burr Brown opamp for a more transparent sound, not the crapy TL07X opamps.
Manual for the 1968 here:
http://lib.store.yahoo.net/lib/mercenary-audio/1968-opman.pdf

From all the info I could gather, it looks like they throw away a lot of electronics from the audio pass to make it more transparent, and it looks like they maintain the jfet "compressione cell", and they use a different, tube based makeup gain block after compression.
The big advantage of this type of comp is that the JFET is not exposed to high levels so it is relativelly linear compared to other JFET compressors.

To better understand the whole comp circuit I redrawn it (only the comp not the whole unit) and later did some variations to discuss how we can mod, and make a new (not a clone), flexible comp based on this gain reduction idea.

Here is only the comp section from the 1960:
****REMOVED****

As you can see the "compression cell" is an instrumentation amp topology with a fet as a variable resistor. There are a lot of opamps and a tube before the gain cell. There is the balanced receiver then a tube (probably for some euphonics, look at the THD trim pot), another opamp, +4 to -10 attenuation for sidechain (did not redraw it exactly like in the original, there is option for +4 and -10 insert) two opamps for balancing the signal for the "compression cell" and many more after it (not redrawn here).

My idea is to go directly, balanced into the gain cell and use the shortest possible path for audio (just a simple gain block after the "compression cell").
Question is, what is the best compromise to make this thing work properly. Is it really necesary an aditional balanced buffer before the "compression cell"?
http://img.photobucket.com/albums/v292/chrissugar/1960/1960-comp-section-MOD1.jpg

The next thing i would ask, is there is any hidden reason to use those triodes inside the instrumentation amp or they are only for some euphonics. If they are only for adding harmonics I would leave them out. In that case the buffer opamps after the tubes can goo too:
http://img.photobucket.com/albums/v292/chrissugar/1960/1960-comp-section-MOD2.jpg

The side chain looks a little overcomplicated, maybe something like Fred's sidechain for the opto would be an interesting alternative. Also I think Fred's sidechain has the advantage that the attack and release circuits are isolated by a buffer, so they don't interact (like many other sidechains):
http://img.photobucket.com/albums/v292/chrissugar/1960/1960-comp-section-MOD3.jpg

So my first question is, what is the simplest balanced receiver that would work correctly with this comp, the second , how would compare technically the simple 3 opamp "compression cell" to the original tube based one, and third, is there any technicall reason to not use a sidechain like Fred's one.
I see so many possible variations and options for this type of comp.
What do you think?

chrissugar

P.S. I have higher res pics but Photobucket can't upload them so they have a lower resolution. Hope it is readable.

The main purpose of the tubes in there, is to fool people into thinking it's a toob compressor and ultimately command a higher price tag. A good mod for you to do is make it tube-less.

analag

I agree.

On the original, the vca circuit was a black-box, encapsulated in epoxy - said to be to protect the idea - but I always expected it to be smokescreen - because even the "support" components around the blackbox suggested a design close to that of the drawmer compressor or noisegate..

Jakob E.

Thanks Rowan and Jakob,

I'm glad we agree that probably the best thing (cleanest amplification) is to leave out the tubes and the aditional buffers. So in this case we have a classic three opamp instrumentation amplifier with a JFET as gain control.

I'm curious what the math guys (PRR, Brad Carso) think about the ballanced input. In my opinion it should work, probably all the resistors should be paired for best CMRR. There is a resistive divider at each input so they should be also equal.
If we accept as a starting point that maximum gain reduction should be 20dB then we can can calculate the input divider to be suitable from an input impedance stand point.
For a higher than 10k input Z it would be OK R19, R23 to be 10K and R20, R24 to be 1K (that would be 20dB attenuation). Then we have to calculate the feedback resistors (R35,R36,R27) for gains between 1 and 10 (0 to 20dB)
For the instrumentation amp I supose the OPA604 (or similar) would be a good candidate.

chrissugar

I'm curious what the math guys think about the balanced input.

Click to expand...

I don't see any reason why it shouldn't work. Ratio-matching of R19/R20 and R23/R24 is sufficient for good CMRR (apart from the obvious four candidates around the differential amp).

I'd make the gain of the diff-amp unity.

Samuel

I read some docs related to instrumentation amp circuits, and the articles written by Barry Porter (who favors instrumentation amp for line input circuits) and couldn't find a reason to not use the inputs of the "compression cell" as the balanced inputs of the whole compressor.

I did some calculus with the original values of the 1960.
R19, R20 and R23, R24 attenuate the signal aprox 37dB and the instrumentation amp at maximum gain (if we consider the FET a short, I know it is not ) is aprox 40dB.
So in non compressing mode the signal is attenuated 37dB then amplified aprox the same value (when FET has the lowest resistance) and when the compression starts, amplification lowers to aprox G> 1 at the higest resistance of the FET.
In my opinion attenuating the signal 40 dB then amplifiying it back is not the greatest idea. First you lose S/N ratio and also the opamps start to sound worse at 40dB gain than at a lower gain (who need more than 20dB gain reduction).
I think my initial thought to attenuate 20dB and make the max gain of the instrumentation amp 20dB is a better technical and musical solution.
Any thoughts?

chrissugar

Hmmm.

First thing or thereabouts that strikes me is that unless the signal drive is truly symmetrical the gate drive for the FET will be inappropriate. So one needs to contrive a quasi-floating gate drive that follows the common-mode signal, which ought not be too difficult. Although the original design does look needlessly elaborate, other than for intentional coloration, note the careful preparation of the signal drive so that it is symmetrical.

At least the modified circuit is all a.c. coupled, so the d.c. levels on the inputs aren't going to pass through and really screw things up.

I'm going to think about it some more, but it may be a while as I'm swamped right now.

So what about a step-down transformer instead of the voltage divider? Gives "perfect" CMRR and a tad lower noise as well.

Samuel

[quote author="chrissugar"]The big advantage of this type of comp is that the JFET is not exposed to high levels so it is relativelly linear compared to other JFET compressors.[/quote]

You still have the same problem with the FET that other FET limiters (1176) have, which is that the dynamic range of the FET is a lot smaller than the dynamic range of the input signal. Since the signal voltage at the - pin (of each op amp) is the same as that at the + pin, the signal has to be attenuated beforehand. In that regard it's really no different than the other stuff out there, which also solve the problem by pre-attenuating the signal.

But I love the way you stripped away all the bullshit from the original circuit. That's a good approach and a good skill to have. This looks like a fun idea to work with. Maybe you can find a way to screw it around and make it better.

I'm trying to think of a way to make it work by making the first two amplifiers into inverting amps. That would be one good way to keep the signal at the - pin (and thus at the FET) from getting too large. But then the gain control arrangement would have to change, defeating the purpose. Still, maybe it could be made to work.

Another fun note is that you could use an LDR instead of the FET, and have a new variation on the opto too.

[quote author="ulysses"]Another fun note is that you could use an LDR instead of the FET, and have a new variation on the opto too.[/quote]

Or how about the LDR between pins 2 and 3 in a H-pad configuration? I think the Millennia box does this. Ok, goes offtopic..

The LDR substitution for the FET has the advantage of floating relative to the control, so it solves that problem. But given that the circuit has the disadvantage of having an inverse law to the typical one, with full gain/no compression with it or the FET on, the normal gain will not be very stable. Padding it out with the series fixed R helps.

And you are locked in to the attack and (especially) decay characteristics of the LDR.

Thank you all for the answers.

Meanwhile I checked and rechecked my drawings and discovered a small error (the point from where the gain reduction signal is collected and the coresponding diode) so I exported them at higher resolution and converted them to JPG to avoid Photobucket to resize them. MOD1,2 and 3 have now much higher resolution but the original 1960 is still recompressed (I don't understand why).
So if you downloaded the schematics you can now erase the old files and use the new ones.


[quote author="bcarso"]Although the original design does look needlessly elaborate, other than for intentional coloration, note the careful preparation of the signal drive so that it is symmetrical.
[/quote]
Thank you Brad for your thoughts about the need of symetrical drive signal.

[quote author="bcarso"]At least the modified circuit is all a.c. coupled, so the d.c. levels on the inputs aren't going to pass through and really screw things up.
[/quote]
Yes it should remain AC coupled to avoid any unnecessary overcomplications. Even beeing AC coupled the moded schem has much, much more potential to be transparent compared to the original. I didn't even redraw the output stage of the original 1960 (that is five more opamps :shock: )

[quote author="Samuel Groner"]So what about a step-down transformer instead of the voltage divider? Gives "perfect" CMRR and a tad lower noise as well.
[/quote]
The transformer is the obvious candidate for this function but if it is not necessary I would avoid it. My intention is to make this comp as transparent as possible. If colour is desired there are enough options, input/output transformer and tube/transistor/jfet/opamp makeup gain.

[quote author="ulysses"]Since the signal voltage at the - pin (of each op amp) is the same as that at the + pin, the signal has to be attenuated beforehand. In that regard it's really no different than the other stuff out there, which also solve the problem by pre-attenuating the signal.
[/quote]
Yes you are right, i have no idea why I had this thought because it is obvious :? .
After clarifying this, comes the logical question: can this be the reason why the original attenuates the signal 37dB to avoid the JFET to be exposed to high levels? 20dB attenuaton would be enough? Or better, what would be the optimal input attenuation to satisfy both conditions:
1- high enough S/N level and
2- low enough signal for the JFET to behave linear?

chris

I was kindly asked by Drawmer to remove the schematic form my wepbage, as the unit turns out to be in current production. If anyone needs the schemos, simply drop me an email.

Jakob E.

OK.
Do you all think, that it would be a fair thing, to remove my redraw of the original 1960 comp section and leave only the three MODs?
I think it would be the right way to show some respect to the manufacturer (like Jakob, I didn't know the 1960 is still in production).
The idea is not about cloning but about studying the concept and develope something new and interesting, so the three MODs will serve very well.

chrissugar

I'm not sure you have to remove your redraw. The original schematic on Jakob's site is covered by their copyright, and your own drawing is not (though if you copied it from theirs, even by eye, I guess it would be...)

On the other hand, it sounds like they were pretty cool about it with Jakob, so respecting their wishes is in order. I would probably leave it up unless they say something to you directly (or to the moderators of this forum).

As a manufacturer who participates in the DIY community, I think the discussion is good and healthy for both the manufacturers and the home-builders. Your mods are a great example of how a manufacturer can have a good idea, and it can get refined through group discussion. That doesn't always happen within a company. That said, if you find ways to improve their circuit, I don't think you can complain if they (or somebody else) take some of your ideas, refine them, and incorporate them into their own redesign or new products. This back-and-forth has really, I think, advanced the art of audio electronics tremendously over the past few years, whether it be the interplay between guys like Hafler, Pass, Borbely, etc in magazines in the 80s and 90s, or guys like Jakob and PRR here on this forum. It's great, and the consumers win whether they're DIY or purchasers.

It would be nice if Drawmer posted their own schematic for the benefit of their customers who need to maintain, or even modify, their products. But now that I'm in their shoes, I can understand. I didn't make a schematic of the Folcrom available (but it's so simple that if you own one you can quickly make sense of it by looking at it). But when the topic was discussed here, different drawings were passed back and forth until NYD came up with something that is almost exactly our circuit. So now it's out there, and that's fine. I didn't invent any new circuits there - if I did, I would have patented it. We're all just passing the same ideas back and forth like tinkertoys that get reassembled in new and hopefully interesting ways. Drawmer may have a different attitude about your drawing of their circuit, but that's between them and you.

[quote author="ulysses"]On the other hand, it sounds like they were pretty cool about it with Jakob, so respecting their wishes is in order. I would probably leave it up unless they say something to you directly (or to the moderators of this forum)[/quote]
I'm with Justin
... as they were polite about it and if it is a current unit and not a reissue just cos DIY'ers were interested, then I think also being polite and in their favour is a fine idea.
Jakob has said he can share via the email system so ...
it's all good