Mix Bus Compressor from scratch

[Adolph Ernst]

One thing I notice from the Grundig schematic is that the first stage has grid leak bias. Large enough positive signal peaks will charge the input cap negative after initial clipping. EF83 also has capacitor coupled input and bias to cathode is taken from a voltage divider making cathode voltage fixed, so same effect should happen at the EF83 stage too. Fast attack and release compression is happening by grid current charging input caps and very slow release gain changes happening because of the sidechain.

 

[DaveP]

EF83 also has capacitor coupled input and bias to cathode is taken from a voltage divider making cathode voltage fixed, so same effect should happen at the EF83 stage too.

This system was used in the mono versions, the stereo versions (like mine) use another circuit.
best
DaveP

 

[DaveP]

The mains transformer is now replaced and all is working normally on the HT/B+

I have made some progress on the gain reduction problems but work continues.
best
DaveP

 

[DaveP]

I have stabilised the compression at slow attack and release times and this has given the following compression curve:-



thats all for tonight
best
DaveP

 

[DaveP]

A Day in the life: My partner needed a few days away because as a farmer she has no real holidays. We went to the wonderful Brittany coast for some walks on the coastal path. Day one was fine and we did 15km, Day 2 we get a phone call saying our cat was run over, partner in tears. Day 3 we did 18km but I felt really ill afterwards with shivers and chills. Day 4 I have the flu and can't do anything. Day 5 we drive home. C'est la Vie.
best
DaveP

 

[ruffrecords]

A Day in the life: My partner needed a few days away because as a farmer she has no real holidays. We went to the wonderful Brittany coast for some walks on the coastal path. Day one was fine and we did 15km, Day 2 we get a phone call saying our cat was run over, partner in tears. Day 3 we did 18km but I felt really ill afterwards with shivers and chills. Day 4 I have the flu and can't do anything. Day 5 we drive home. C'est la Vie.
best
DaveP

I know only too well just how you feel. Get well soon.

Cheers

Ian

 

[thomasdf]

Sorry about that Dave! C’est la vie, but sometimes la vie is sad… I live with a cat I love and your story breaks my heart. Sending you good vibes !

 

[IBart]

Yes I was also sad to read this, how horrid

Hope you feel much better soon Dave !!

 

[Johnny1234]

Hi Dave,
Hope you recover quickly. and are back even stronger than before.

 

[DaveP]

Day 6: I dig a hole to bury the cat, feeling like death warmed up.

It's my partner I feel sorry for, she really needed a break and had been planning it for 18 months, her job won't allow another break until September.
Thank you all for your kind words, it helps!
best
DaveP

 

[DaveP]

Back on the case.

This is the frequency response with no compression at 0dB output.

More tests to follow.
best
DaveP

 

[DaveP]

This is the frequency response with 2dB of compression. It was done the same way as the first chart, but the threshold was reduced until it showed 2dB loss on the VU meter (which is surprisingly accurate).

The obvious difference is that the frequency response has improved at both ends of the spectrum, with just a slight boost at 40Hz.
More tests to follow
best
DaveP

 

[thomasdf]

That's interesting. Some impedance variation ? A 40hz 1dB boost can be a nice addition to a mixbus Weird the HF response becomes flat under compression !

 

[gyraf]

..or the expected compress-distortion at low frequencies, putting measurement off

 

[zamproject]

Things with side chain active is that you also measure the side chain (inverse) response.
Bass boost curve slope tend to target a +3dB around 15Hz, which seem coherent with the input network of the side chain 10nF/1MΩ ?!?
If you want a flat response maybe try a 100n here to lower the fc @1.5Hz
Or if you want a less bass pumping compressor try a 1n for a fc @ 150Hz

 

[DaveP]

Thanks Zam, now I have a bench mark I can start tweaking the details.
best
DaveP

 

[DaveP]

Time to give some figures:

The compressor has two stages/channel. The EF83 input tube and the 6DT8/E83F output stage.
Starting with the output stage connected to a 600 ohm resistance load. 330mVrms input to the E83F gives an output to the OPT of 7.84Vrms, this corresponds to an OPT output into 600 ohms of 1.22Vrms and approximately 0VU on the meter.
7.84Vrms is ~22Vp-p and the threshold is capable of reacting to this if it is dialled-in. The output stage has a lot of headroom, and the input can be increased up to 1.5V before any serious waveform bending takes place, but bending the VU meter needle is another thing altogether!

I have added these voltages to the schematic to make it easier to follow.


The EF83 needs an input of 67mV to give an output of 1.32V, which becomes the 330mV at the input of the E83F, the NFB in the output stage is responsible for this reduction. The EF83 is capable of giving an output of over 10V without visible distortion, but this would totally overload the output stage, it is nevertheless a useful overload margin. These figures were all obtained with the threshold turned to maximum (no compression), but use of the threshold control would allow higher inputs than 67mV without overloading the output stage.

Next I will try to explore the compression characteristics.
best
DaveP

 

[DaveP]

The following chart shows the effect of varying the threshold voltage. In the Grundig circuits, that this is borrowed from, 18V was the preset voltage for the mono models, 16V was the threshold for the stereo models. It was a factory preset to guard against tape overload, I have made it variable but this should be used within limits.

With the 16V setting, it kicks in earlier and so gives more control before exceeding the VU output level. This Mix bus compressor was not just designed as a compressor, it was concieved to also give a final "gloss and sheen" to audio (whatever that means!). Whether it achieves that objective or not, I will leave for others to decide. I cannot claim the attack and release works as designed, the 10uF cap does not like being adjusted, in fact it will only tolerate being fed by an output cap of one tenth of its own capacity, i.e 1uF. This was something that PRR noticed back in 2005 https://groupdiy.com/threads/ef83-6bk8-circuit-examples.9471/#post-112594
best
DaveP

 

[C.B. - Boudio]

Could you give some directions to make its output with a simple cathode follower instead of the parafeed design? (for I am some kind of a stingy teenager who cannot afford the nescessary iron... )

 

[DaveP]

Could you give some directions to make its output with a simple cathode follower instead of the parafeed design?

I have modified the circuit for you and it should work, but the original circuit was chosen for its sound, so no guarantees as to what this will sound like.
Also the original output stage had NFB and so a low output Z, however this cathode follower should still have an output Z of around 500 ohms which will be fine into a 10k load. This is a non-feedback design so you must run the EF83 at low input voltages to avoid distortion. The diode is out of phase compared to the original and may need reversing, it may make little difference, I can't be sure.

Bonne chance!
best
DaveP