Group DIY Vari-mu

How does this idea grab the community?

I am in the process of designing a new tube vari-mu as a fundraiser for the Forum.

I have got so much out of this place and met so many great guys that I want to put something back.

The idea is this, it will be low cost but high spec i.e. fast but will be made unconventionally from little used tubes that must cost $3 or thereabouts each.

Anyone who builds it or copies the schematic is on their honour to donate the cost of a tube to the forum, say $3.  I will hand over the rights to the circuit to the Forum, anyone who contributes to the design will do the same.

This is not going to be a clone but a new take on some old ideas.  For example, the front end will not be a transformer but a tube PI, the cap charging issue will be avoided.  It will be designed to work with any mic pre that can dish out a volt into a high impedance.  The output stage will use a toroidal mains TX costing about $15 and put out 20~30Vac into 600 ohms.  It should have an attack of a few mS.

Anyway, thats the drift, over to you!

best
DaveP

Wow. Great idea, Dave. There hasn't been much in the way of reminders to donate lately, and it's the season of goodwill after all. I'm in!

Thats great DaveP! Will donate anyways

By the way, I'm not very familiar with tube topologies so tried to google what PI input is like and found this
http://satamax.free.fr/P663.jpg

Right hand side Fig. 15.43A shows a transformerless outputstage as well?

Magnetosound & Tmuikku,

Thanks for the positive comments.

Ok here goes...

This is a draft of the control amp:-



It uses 6BR6 or 6SA7 tubes with separate inputs, so the CV does not have to charge any caps from the PI. so attack stay fast.
This is a rough idea of the sidechain.



The Cathode followers will be high gm tubes (6BZ7) so output Z's will be ~150 ohms which is Fairchild territory.

This is my lot for the night.
best
DaveP

Awesome initiative Dave!

I'm in for sure. Looks like a real interesting design so far.

Has this been prototyped?

I have doubts about taking the sidechain from before the variable gain.

And cap-coupling the rectifier.

Hi PRR,

I'm glad you posted as I got the idea from one of yours on transformerless compressors.
You pointed out that the caps from a P.I. input stage would need charging, that's why I used a 6BE6.

I realised that all the "feedforward" compressors (LA-2A etc) appear to be faster than the feedback type, unless powered from a 6V6 output stage, the rest of it is from old RCA manuals where they used to reverse the CV to make Expanders.

I was testing the idea on the Drawing Board first, I'll prototype it now on The Lab and we'll see how it works out.  The caps before the rectifier may indeed be a problem and slow it down too much, but I have another sidechain in reserve to try.

I would be grateful if you could suggest a way of testing the attack speed with just a sig gen and a scope.

best
DaveP

Dave,

Fascinating circuit. What is the purpose of that 100k balancing pot? Are those the screen grids (and what are the other grids there)? Does it add something over the cathode balance circuit?

Ummm....you want to go cheap, and you use hollow state side chain? Sorry, but I don't understand why. But more important: speed+non-interstage signal path don't go very well together. So, if speed is your thing like it seems, I don't think you are on the right track now. Unless you like the combination of speed and clipping.

-Jonte

Hi Dan,

The cathode pot only balances the static current through the tubes but the advantage of using a pentode/hexode is that its possible to bring to a closer balance the dynamic amplification in use.  The older tube compressors like the BA-6A used this circuit.

Jonte Knif said:

Ummm....you want to go cheap, and you use hollow state side chain? Sorry, but I don't understand why. But more important: speed+non-interstage signal path don't go very well together. So, if speed is your thing like it seems, I don't think you are on the right track now. Unless you like the combination of speed and clipping.

-Jonte

Click to expand...

Well there's cheap and cheap.  There are a lot of guys that prefer tubes to IC's and this is for them.
I just resent paying high prices for the typical tubes when there are other lesser known Radio and TV types that are just as good.

I made a BA-6A this year without an interstage and it works fine.  Check out the schematics on the Gates SA-39, Sta-Level, Collins, 356E and 26J, you wont find an interstage.  Best part of my circuit comes straight from 1940's RCA designs so you had better take it up with them.  I haven't posted the power supply yet but that contains a mosfet voltage regulator, so I'm not totally prejudiced, only slightly!

best
DaveP

I'm sorry, but I kind of know Vari Mu designs, past and present, and what they are good for. 40's design is good for certain (many)  things, but remember, better designs did come later, for very good reasons. There is a reason Fairchild exists. Heavily limited vinyl cutting being one of the reasons. If radio transmitter overloaded it sure did sound bad, but if lathe overloaded, well... That said, going "cheap" is always one option, but knowing the limitations will not hurt anyone, quite the contrary.

OK, so you like tubes, but you have solid state rectifier there? May I suggest an alternative? Save some power, scale down the power level in your side chain, use tube rectifier (cheap) and simply put a solid state follower of any kind where you want to step up in power level. For 2 ms attack time and one pair of vari mu tubes you should be just fine with pretty simple solutions. In the 40's designers didn't have too many low impedance DC-coupled buffers or followers at their disposal. My first Vari Mu used just one jfet-follower after release caps and had 0,1 ms minimum attack time.

Any how, how much headroom have you calculated for the side chain signal in the output stage? Something like 10 dB is minimum if moderately deep fast limiting is ever needed. So, if your output stage is able to put out 30 V AC it is pretty fine for most use, 20V AC wouldn't make me feel comfortable with fast attack times and deep limiting.

It will be delicate balance between bass extension and side chain rejection with only coupling caps in the way.

It all depends what you want to use the compressor for, of course.

Another things was already expressed by PRR. Feed forward Vari Mu (side chain) is a tricky thing, needs special attention, and OTOH speed is readily available with feedback with the right solutions. Been there, done it. In general I expect feed back side chain to provide more the kind of sonic signature I think you and most builders are after.

Best,

Jonte

Dan,

I missed part of your question.

The hexode has 2 screen grids 2&4 which do just the same as in a pentode, but it also has another input grid 3 which was originally used for the CV from the AVC supply in radios.  Grid 1 has a sharp characteristic but grid 3 has a remote characteristic.  By having a separate grid for the CV there are no extra caps to charge to slow the response.  The very old 6L7 tube had the roles reversed, grid 1 remote, grid 3 sharp, that tube was used on the Western 1126 and the Gates SA-39A, its ok but it has a top electrode which is an additional complication.

best
DaveP

Jonte,

Thanks for the suggestions and other positive comments.  I have not permanently decided on the sidechain yet and I will try several ideas on the bench before making up my mind.  The vari-mu tubes only need -25v for deep limiting, they are the last types produced, it should be easy to get that level, but the trick is to get it with a low output z.  Depending on the tubes used 50Vac should be possible, but hardly needed for that circuit.

I'll try feedforward and feedback sidechains and see which works best.

Why don't you post a transistor circuit for me to try?

best
DaveP

There are many ways to skin the side chain, but this can help and is very similar to many classics. The inputs could come from output stage anodes, or, in the case of feed forward, from simple 1 stage amp (the first stage from your SC amp). I think one of the most sensible places to put ratio switch/pot is right after the vacuum diode. In this way you can put a zener after the switch/pot to protect the buffer and the buffer does not need to have super high headroom, which you would otherwise need when using low ratio and deep compression. This resembles some classics wich use output transformer taps and a rotary switch to perform essentially the same thing. Once again we see how much easier and cheaper "modern" components make our life, without no sonic penalties. Keep the resistance of the pot high enough in order to not load the output stage.

Fine component for the buffer is a high voltage j-fet or low power mosfet. Really not a critical choice.

Some hints: Since you need a negative supply for the buffer, try to arrange things so, that you can use negative heater supply too to save some components. Use series heating as much as possible so that your buffer has enough headroom with the same voltage. If not enough, use some nice voltage doubler circuit to derive the voltage from the same secondary wiring. In this case you don't even have to regulate the buffer V- since the follower wont care.

-Jonte

Jonte,

There is no point in just copying that side chain you posted as its the same as all the others, I'm trying a new approach here.

I dont think you understand my disability here, I can put together a discrete transistor circuit but I don't do ICs, I'm an old tube guy!  I won't be having any negative supplies apart from the CV.

I don't do PCB's either, I do discrete transistors on a board using the component wires joined on the underside.

best
DaveP

+1 to a new approach to vari mu.

Many other designs with 6AL5 or IC/transistor sidechains.

Don't know of any with heptode GR and not too many with tube boosted side chains.

Interesting to see your design develop.

There is no point in just copying that side chain you posted as its the same as all the others, I'm trying a new approach here.

Click to expand...

Well...close

Sometimes small features make big differences like the fact that impedance scaling will triumph and it is important to do it in the right place to get the most out of it. We have seen this before, right? People copying fairchild SC amps (OK for me as long as you don't claim it is done for sound), or doing them in SS (like EAR) but not understanding to step aside from the basic flow of power and impedances and take modern possibilities to the logical end. In otherAnd I am tired of it.

There is a certain beauty in pure tube units, of course, and PCB's are not very sexy stuff. You might have seen the thread about my "therapy pre amp" and recenly I built a pair of monoblocks featuring p2p wiring and Siemens F2a and D3a tubes and classic chassis design.

So, how exactly does your side chain work, the Fairchild way? I.e. the SC amp triodes bias points manipulated to give the threshold?

Have you analyzed or simulated the compression characteristic curve? Just wondering if you will be able to get it work without some nonlinear DC amplification element in the SC. Feedback sidechains are so freaking simple because they will self regulate the ratio, but with feed forward there is the risk that your ratio turns negative after some point. Since you don't like complexity, I guess some sort of soft saturation thingy made of resistors and zeners is the minimum you need to get feed forward SC to act in a controlled way.

I see trouble and complexity ahead.

Jonte,

I have been fixing a forum members mis-wired LA-2A clone recently but I hope to start some work on a prototype tonight.

What I would like from someone is a method for testing attack time with a scope and sig gen.  I am thinking along the lines of a square wave and checking the decay slope on the cap, anyone done anything like this before.  I do not have simulation facilities.

With a method for testing the attack I can try various designs and compare.

best
DaveP

Apart from using a cro with trace storage capability to capture the attack and release
characteristics with a slow square wave input, the only other way I have used is
a daw recording of the limiter output with a pre-prepared DAW waveform input to the limiter.

I found it difficult for a signal generator to make a sufficiently low freq square wave,
with a duty cycle able to cause useful attack/release responses from the limiter.

By creating/editing a daw input signal,  I found it easier to  customise an input signal
that would result in a measurable attack/release limiter output signal,  recorded to the daw.

Hope that is understandable

Thanks Alex,

I think you're right, getting slow enough might be a problem.

I was thinking of comparing the input square wave on one channel with the cap voltage on the other.

If all else fails I have a Hewlett Packard pulse generator at work I may get going.

Thanks for your input mate.

best
DaveP