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 said:
Hi Pierre,

That was quick!! Let me guess what you did. Your signal comes is as normal via the mic or line input via the push buttons to the mic transformer. You took out the link the connects the transformer secondary to the input of the first amp and connected the transformer secondary to the 10K pot. That could be your first problem. The mic transformer really needs to be loaded with no more than 150K. The 10K pot is reflected to the primary as 100 ohms  so it is not surprising you need a lot of signal to get anything out. I cannot think of an easy fix for this. One possibility is to remove the 10K pot entirely and rely on the mic pre gain switch to set the gain. Then the worst case load is the 47K resistor in series with the opto. If you made this 100K it would not change the threshold but it would increase the available gain reduction and perhaps alter the ratio a little. The load on the transformer would be 100K worst case which does not affect the reflected primary impedance anywhere near as much as the 10K.

You then connect OUT1 to IN2 via the 10K threshold pot and connect the opto LEDs to OUT2 via a 2K resistor. The audio output comes from OUT1.  You probably already have about 20dB gain in the second amp if it was set up to be gain make up for the Pultec so that should be fine.

To test it, turn the threshold fully off and feed 0dBu into the line input. Adjust the gain switch position for approximately 0dBu out. You should then be able to turn up the threshold pot and see the output level reduce. If you turn it up so it is 6dB lower you have approximately 6dB of gin reduction. You can then adjust the gain switch to get the level back to about 0dBu.

Cheers

Ian

Hi Ian,
    Thats more or less how i connected things except i respected the schematics and inverted the amp 1 and 2 .
So i got rid of the 10k gain pot and replaced it with a 100k resistor.
I set everything so that i get something that still sounds good  to my ears  at the maximum compression setting wich equals to around 12 db...
More than that and it squishes too much.
I guess we would need a third amp to makeup the gain lost with the compression. 

Sound pretty good!

Cheers!

Pierre



 
anjing said:
I guess we would need a third amp to makeup the gain lost with the compression. 

Sound pretty good!

Cheers!

Pierre

Glad you like it.  The original design assumed a 10K:10K transformer for which the 10K gain pot is fine. Using it with a mic transformer means you need to lose the 10K pot. You now have no gain make up control!! A solution to this is to use the negative feedback gain  resistor as the make up gain control. If you wired the mic transformer to the normal mic amp in the Eurochannel you could then use the normal mic gain switch as a make up gain control. Use IN2 and OUT2 for the side chain. The only problem with this is the audio comes out of OUT1. So if you have the amps swapped you could replace the preset gain pot  of IN2/OUT2 with a pot.

Cheers

Ian
 
This looks like something I might try in the near future (nearness depending largely on cost of build for two units). I'm happy with the Edcor output transformers in my Classic Preamps and would likely use them again here. Your PCB accepts Cinemag, Sowter and Jensens at the input. Could I conceivably use an Edcor there, too?

http://www.edcorusa.com/xsm_series

Ralph
 
rmaier said:
This looks like something I might try in the near future (nearness depending largely on cost of build for two units). I'm happy with the Edcor output transformers in my Classic Preamps and would likely use them again here. Your PCB accepts Cinemag, Sowter and Jensens at the input. Could I conceivably use an Edcor there, too?

http://www.edcorusa.com/xsm_series

Ralph

With Cinemag/Sowter/Jensen you pay fro quality of build, core size and a mu metal screen.  Edcor input transformer usually have a smaller core size and lower incuctance and no screening. That is why they are cheaper. For this application it is possible that none of thse limitations matters. Provided you feed from a low source impedance, the inducrance does not matter too much. As long as you don't want to feed it is signals up to +20dBu then the smaller core does not matter. As it is a line level signal going through it you can quite probably  get away with no screening.

So I think it is definitely worth a try especially as the XSM series has a nice big core.

Edit: you could prbably use the smaller WSM series. I have never tried these so I would be interested in the results.

Cheers

Ian
 
Interesting. In addition to being  smaller and cheaper, the specs look pretty comparable. Do you think they are worth trying on the output as well? And no, I don't see myself hitting these with anything remotely approaching +20dB.

Thanks!

Ralph
 
WSM range worked good enough even as outputs for mic amps, iirc it was recommended over XSM for MILA by designer (better response or something). I've used them on some other projects and WSM are nice at line levels when one wants sound from iron core like Carnhills and a few Sowters too. Never had problems with inductance even when it was driven by typical cap coupled triode into line level out.
Like Ian says, should be ok here. The only reason against it i can think of is having some nickel in the core for sound, if that is what you want.
 
rmaier said:
Interesting. In addition to being  smaller and cheaper, the specs look pretty comparable. Do you think they are worth trying on the output as well? And no, I don't see myself hitting these with anything remotely approaching +20dB.

Thanks!

Ralph

According to Edcor the WSM maximum signal voltage is 10V rms which is +22dBu. If used as an output transformer, the 2:1 ratio required means the masimum output level will be +16dBu. probably OK for many applications but definitely lacking a bit in headroom. The 10K:10K type will probably be OK for line inputs.

Cheers

Ian
 
Thanks Ian and My3gger, lots to chew on. It probably makes the most sense to go with WSMs on the inputs and XSMs on the outputs - I need to crunch some numbers and mull it over or the Poor Man could pretty quickly become another Poshman! BTW, FEDEX just brought my VTL5C3s. I wonder if I managed to get the last 4 on the planet? If so, I'm feeling a pretty keen sense of obligation to completing this build  ;D.

Ralph 
 
Wow, 18 months and I'm just getting ready to start ordering parts for a build. Time flies...
I have a few questions before I begin placing orders:

1) A number of the earlier compressor drafts had a provision for a G.R. meter. Would it be terribly difficult to include one here (and how, please)?

2) Is it possible to stereo link a pair of them?

Thanks!
Ralph
 
rmaier said:
Wow, 18 months and I'm just getting ready to start ordering parts for a build. Time flies...
I have a few questions before I begin placing orders:

1) A number of the earlier compressor drafts had a provision for a G.R. meter. Would it be terribly difficult to include one here (and how, please)?

2) Is it possible to stereo link a pair of them?

Thanks!
Ralph

Hi Ralph, good to hear from you again. This thread took on a new life here:

https://groupdiy.com/index.php?topic=52847.msg674382#msg674382

Which included ideas for including a GR meter and stereo linking. However, I have not had time to t4est any of these so you would be pioneering.

Cheers

Ian
 
Pioneering! How cool its that! Thanks, Ian, I'll check it out and post further questions in the other thread.

Ralph
 
For a long time I was intrigued by the question whether it is possible to be without a complex detector in a tube compressor.
I assembled this my circuit on the layout and was even surprised that everything worked well.
The control scheme GR has not been tested so far, only theoretically :
5vdjly.jpg


Vladimir P.
 
There is a problem with connecting the diode bridge directly across the output transformer secondary - it is a non-linear load which will cause distortion in the output signal, just like a regular passive VU meter does.

Also there is no gain after the threshold pot so there will be no compression until the output level reaches the sum of the diode drop plus the LED drop - just over 2V in total or about +5dBu.

Cheers

Ian
 
> possible to be without a complex detector

Certainly. I used to limit power amplifiers with a pot and bridge off the speaker side, to an LED on an R and LDR on the input side. Rude, not musical, but avoided clipping into speakers.
 
PRR said:
> possible to be without a complex detector

Certainly. I used to limit power amplifiers with a pot and bridge off the speaker side, to an LED on an R and LDR on the input side. Rude, not musical, but avoided clipping into speakers.

This works because power amps have very low output impedances - a few ohms at most and the non-linearity of the bridge and LED has a much smaller effect. I would not recommend using it on a 600 ohm transformer output.

Cheers

Ian
 
ruffrecords said:
There is a problem with connecting the diode bridge directly across the output transformer secondary - it is a non-linear load which will cause distortion in the output signal, just like a regular passive VU meter does.
Cheers
Ian
Hi  Ian!  Your predictions of the problems were not confirmed by the results of the measurements. To control the LED of opthopair  we do not need a current of 10-40 mA,  for a compression of 10-20 dB, enough 10-40 microamperes,
a thousand times less.  This can be seen in the table in the figure.
   In my scheme, there are limiting resistors with a nominal value of 5 kilooms, even in limiter mode (ratio 1:20),
  in the compressor mode, the resistance of the RATIO controller is added 10 kΩ (1:10) - 30 kΩ (1: 5).
  Therefore, the load of a 600 Ohm transformer is tens of kilo,
  I assembled a model for testing optocouplers, took three VTL5C # 1, # 2, # 3 and one my handmade (12 years ago).
  The table also shows that the VTL5C is quite different in characteristics.
6zy3rp.jpg
 
vari-mu said:
Hi  Ian!  Your predictions of the problems were not confirmed by the results of the measurements. To control the LED of opthopair  we do not need a current of 10-40 mA,  for a compression of 10-20 dB, enough 10-40 microamperes,
a thousand times less. 
I made no mention of the amount of LED current. What I said was the diode bridge plus LED across the 600 ohm utput is a non-linear load and will cause distortion. I also sai the threshold will be at an output of about +5dBu. Both statements are still true.
  I assembled a model for testing optocouplers, took three VTL5C # 1, # 2, # 3 and one my handmade (12 years ago).
  The table also shows that the VTL5C is quite different in characteristics.

I am concerned about these results. For 23dB attenuation, the resistance of the VTL5C3 needs to be about 3K. Looking at the VTL5C3 datasheet, for a resistance o 3K, the LED current needs to be 6mA not the 50uA you measured. Either the datasheet is completely wrong or your measurements are or the VTL5C3 you tested is not a VTL5C3.

I have plenty of VTL5C3 so I will repeat your experiment and publish the results.

Cheers

Ian
 
Back
Top