Poor Man's Tube Compressor

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

ruffrecords

Well-known member
Joined
Nov 10, 2006
Messages
16,158
Location
Norfolk - UK
Continuing  the theme of low cost construction projects of classic kit I have been thinking about a poor man's tube compressor. The classic types use vari-mu tubes in complex push pull arrangements with lots of expensive iron or they use expensive light dependent resistors and plenty of iron too.

So I have been hunting around for an alternative tube based gain control element and came across this little gem in the Radiotron Designers Handbook (see pic). It uses a pentode with 100% NFB to create a variable resistor with the value 1/gm and hence by varying the grid bias you can vary gm and hence the resistance. I have done a quick simulation using an EF86 and it looks quite promising. Distortion is below 0.5% provided the signal is no greater than 1V rms. At first I thought this might be a bit low but then I had a look at the Neve 2254 which uses a diode bridge and that operates at a nominal -31dB. Allowing 26dB headroom and assuming the nominal 0dB reference is +4dBu means its maximum operating level is only -1dBu which is less than1V rms so we are in good company. I don't have much experience with pentodes but I think I might knock this up on the bench and give it a test. As it is unbalanced we don't need push pull circuits or much in the way of iron so an inexpensive tube compressor could be a possibility.

What do you think?

Cheers

ian
 

Attachments

  • Screenshot-1.png
    Screenshot-1.png
    150.4 KB · Views: 565
He's at it again!  Looks interesting. I recall seeing another of that sort that used a 12ax7 much like an 1176 uses a fet for gain control. Maybe I can dig it up after work
 
will try simming this up later...

comp1.jpg
 
What Rob said.

Since/if the (attack) control rate is in the audio band, and the DC bias variation is MUCH more than the audio signal, it thumps.

Choice of slow attack or severe bass-cut after can make this tolerable for some uses.

> signal is no greater than 1V rms.
> Neve 2254 which uses a diode bridge and that operates at a nominal -31dB.


VCAs like this, there is a max voltage for low THD, and a minimum resistance which, with desired max loss, sets the idle impedance and thus the idle hiss.

Diodes distort before 20mV, maybe 50mV in pairs or quads. Minimum impedance runs ~~100 ohms (limited by matching and easy drive; much lower is possible). Taking -26dB or 1:20, the series resistor is 20*100 or 2K. Idle hiss is around 0.5uV.

Radiotron 16.3 does not specify the pentode or the plate/screen voltages. And I have not done a full work-up on this plan. A general pentode run rich may do 1V at grid clean, maybe 10V near cutoff. Minimum resistance near 1K (it is resistance-coupled so we can't use the show-off Gm value applicable to RF/IF duty). For 26db the series resistor should be 20*1K or 20K, for 33db or 1:45 about 45K, essentially the 50K shown. Idle hiss will be at least 3uV, and probably higher with tube-hiss.

The DC shift on a simple diode is a bit higher than the gross-THD level, say 50mV-150mV. The DC shift on the pentode is likely to be well over 100V (from cut-off to heavy limiting) and therefore may be an order-of-magnitude larger than clean signal. Thump. Diodes are cheap enuff to push-pull cancel thump (though drive and recovery schemes do cost money); tubes cost enough to resent the cost of push-pull thump cancellation.

I spent all day on a ladder so I'm not gonna think any more today.

It does have the apparent virtue of INcreasing device current for large signals, opposite of typical tube VCAs which perversely REduce current on strong signals. Such schemes must be checked for starvation on "slight reduction", -2db thru -6db, which can be where THD peaks.

Note that the control voltage is negative several volts for full-up (limiter idle), zero for full attenuation (heavy limiting). This is reverse of most tube limiters.

Note that C2 must pass full audio to do the job. 0.02u-1Meg is 8Hz. This is also 20mS, which is not an unreasonable attack time, but sometimes we'd like a little swifter action.

Note too, that in a FET limiter with a long-channel device, there is a THD minimum when half the audio swing is fed back to the gate. Some similar trick may help the tube, though the ratio probably is not quite half. The divider-action also buffers the C2 time-constant some. It may be possible to get a faster attack (with larger control voltage); it may be that with main timing network it gets a 2-pole bobble.

Accurate cheap level detector is the other big cost of a tube limiter.
 
Just did a sim and a bunch of research and that schemo I posted is a dud... the pots do nothing, and those who have built it say it just brings the level down a ton, with no actual compression happening. oh, well...
 
I have quite a few of EF86.. if you thought of a way to make them useful in a tube compressor, that would be great indeed..
 
Thanks everyone for all the comments.

Thump will indeed be an issue. If we were to aim for a 0.2mS attack time like the Fairchild it will be well within the audio band. Probably something that needs to be tested on the bench.

I am aware of the half signal across a FET to minimise distortion. I built a FET limiter using just that technique as my second year university project back in the early 70s.  Quite whether it will apply in this instance is another matter but one worth considering.

Noise is another issue and if we operate at a nominal 100mV (to allow 20dB headroom) then we need noise to be no more than 10uV for an 80dB S/N ratio. Experience has shown that tube noise is likely to be four times this level leading to a more realistic estimated S/N of 68dB.

The dc shift at the plate is very large - almost the entire supply range.

C2 is quite problematic. It needs to pass low frequencies to be fully effective as NFB yet defines the attack time. This is probably even worse than expected at first sight simply because it is connected just like a Miller capacitance and so its effect at the grid will be multiplied by whatever the stage gain happens to be.

The bottom line seems to be these issues can only be resolved by experimentation.

I had toyed with the idea of using a bridge of diodes a la Neve 2254 but I was trying to avoid semiconductors in the signal path. It does need a transformer input for balanced operation but I notice the Neve uses a couple of FET buffers on the bridge output. I was thinking it might be possible to use a long tailed pair instead and unbalance at the same time. Just had a thought - would a bridge of tube diodes work instead?

This is not going to be as speedy as the poor man's Pultec!

Cheers

Ian
 
Ian I attached a diagram, that I scanned from an old magazine that a client had about 12 years ago that I intended to do something with but probably never will.  

It was a design for a compressor for the crystal mic on a wireless transmitter.   I seem to remember they said in the accompanying text that although it's single ended, the cross coupled 6BA6's somehow cancel out the thump.  I never quite got my head round how it's doing this but being much cleverer than me you probably will !   Since you're picking pentodes to play with maybe it's possible to incorporate it into your design, maybe you could call it the "not so Poormans valve compressor" ?
 

Attachments

  • vm comp.jpg
    vm comp.jpg
    151 KB · Views: 289
gemini86 said:
Just did a sim and a bunch of research and that schemo I posted is a dud... the pots do nothing, and those who have built it say it just brings the level down a ton, with no actual compression happening. oh, well...

I bread boarded that years ago, it never worked properly then.
 
This is fascinating. I am very new to all this but this is exactly what I am interested in as I am starting. I am looking into Phantom powered Tube Mics that use a LDC Condenser capsule instead of back electret..
I will make a thread about it..
I will also bookmark this and take a look when I feel I am ready..
What kind of tubes would work in this project the best?...
Regards..
Dom..
 
Disease8 said:
What kind of tubes would work in this project the best?...
Regards..
Dom..

That's the big question. The original circuit I posted did not specify a tube type but I might try to get hold of a copy of the original reference which may give us some clues. I simulated the circuit with an EF86 and it seems to work in principle.

On the other hand, the dual pentode circuit posted by Robb Flinn seems to alter gain by varying the voltage on the suppressor grid which is a completely different method. The 6BA6 pentode used in that circuit is specified as a high slope vari mu pentode which might well make it more suitable for this sort of application.

Cheers

Ian
 
Rob Flinn said:
One good thing about the 6BA6 is that it can be got quite cheaply compared to an EF86, so it could work out cheaper using 2 x BA6 than using 1 x EF86, therefore addressing the Poor Man label better.

Yes, you are right. I just bought some NOS Siemens ones from Colomor for 6GBP each.

Cheers

Ian
 
Just in case there are some who haven't heard about it there was a poorman's varimu project in the past, the PRR Varimu.  I've started gathering parts for one, it will be my first tube build. 

Of course new projects and experiments are always fun!
 
ruairioflaherty said:
Just in case there are some who haven't heard about it there was a poorman's varimu project in the past, the PRR Varimu.  I've started gathering parts for one, it will be my first tube build. 

Of course new projects and experiments are always fun!

Yes, I have come across the PRR vari-mu but it has semiconductors in the signal path and I could never work out why he used a 12AU7 as a vari-mu tube. However, It makes me wonder if a long tailed pair could replace his output transformer which would save a lot of money.

Cheers

Ian
 
I'm working on a design thats based on the PRR and 176 that uses the 6bc8 tube as vari-mu tube. Opamp makup gain.  It doesn't have an output transformer. But its my understanding you need the input and inter-stage transformers to minimize "thump".

The original PRR does have an opamp in the output section, but also PPR outlined an option for tube output also...
 
As I remember the PRR circuit also didn't require anything fancy for transformers...I recall him even mentioning something about using the cheap little radioshack ground loop isolators. (600:600)
 

Latest posts

Back
Top