GroupDIY Bridge Compressor.

[Nishmaster]

Dave,

The issue isn't so much either inadequate smoothing or bridge DC from the sidechain per se.

As you change the level of the AC component of the waveform, because there is DC present on the output of the input triode, the DC level changes also. The resistive divider created by the bridge also divides this DC component, remember. This change, when coupled with the time constant of the attack/release network, becomes an AC component of the signal. If your corner frequency for the coupling caps is 20Hz, it only takes 1/20th second attack, or 50ms, to show up as a 20Hz thump on the output. That's a pretty slow attack. A faster attack rides even higher in the audible band, and likely your coupling cap corner frequency is much lower than that for best performance.

In other words, the only time thump is not present is if the coupling cap corner frequency is higher than the attack network time constant.

To combat this, the bridge was originally designed in a balanced configuration. In this configuration, the DC gain shift is (ideally) balanced between the two gain reduction tubes, thus cancelling out any control voltage feedthrough.

Does that make more sense?

-Matt

 

[Jean Clochet]

Post removed.

 

[DaveP]

Nishmaster,

Thanks for that explanation, I had neglected the effect of the DC rising and falling on the bridge, which of course makes AC!  I will go back to the drawing board on this.  I did originally get it working ok with push pull tubes but it was trickier to get the tubes balanced and it consequently lost some of its compression range.

Jean,

Thanks for your contribution, I will make sure the time constant on the caps is properly checked to avoid blocking.

best
DaveP

 

[Jean Clochet]

Post removed.

 

[DaveP]

Jean,

The dif amp has an input Z of 1~2Meg depending on the values of grid resistor chosen, so the loading is very slight.  The caps are of the order of 10~22nF.

I'll come back to this later cos I'm rewiring the circuit now.

best

DaveP

 

[DaveP]

Jonte,

Thanks for bringing up a potential problem, although it took someone else to convince me!  Not sure how I would make out in Finnish!

I am now re-doing the circuit basically as the original but will be using two dif amps instead of a transformer, it will all be balanced push-pull.

Thanks to Matt and Jean for their advice

best
DaveP

 

[PRR]

> the attachment feature?  I can never get anything in preview

File attach is not committed until POST. Attach then Preview, the file is just dropped.

Put in some text, preview, _then_ Attach and Post. Now go back and do any Editing needed.

(You may see that many of my posts have edit times 1-10 minutes after post time.)

 

[DaveP]

Thanks PRR, I'll try that next time.

A few corrections:-

The cathode bias on the long-tailed pairs should read 270 ohms not 1k, this puts the 12AT7's in a linear operating area.

The 6AQ5's are only rated for 250V in triode mode so are not a direct sub for 6V6's, I will use the B9A 6CM6's instead which are 300V.  I'm basing the power supply on 320V so I don't want to waste the power on excessive resistor drops.

best
DaveP

 

[DaveP]

I have to decide where the threshold for compression is going to start.  At the moment it turns over at about +18dB output, whereas the 26W is +25.5dB.

The 26W has a tapped primary which I calculated is about halfway.  Because my transformer uses all the primary the compression is starting about 6dB sooner.

Can I have some suggestions for the output level and the input sensitivity please?  I have a feeling that the threshold may have to be fixed.

The compression chart I posted did not include the input transformer so the gain is quite high as it stands.  The transformer was put in just to give a balanced input.

best
DaveP

 

[emrr]

I have modified a lot of tube limiters to have variable threshold/ratio, and it's useful.  Requires enough input sensitivity/padding, enough output headroom, and suitable gain adjustment to explore the range.  There's a learning curve there, but makes it much less of a one trick pony. 

Your current compression curve is high ratio, so that's covered. 

 

[ruairioflaherty]

I have modified a lot of tube limiters to have variable threshold/ratio, and it's useful.  Requires enough input sensitivity/padding, enough output headroom, and suitable gain adjustment to explore the range.  There's a learning curve there, but makes it much less of a one trick pony. 

Your current compression curve is high ratio, so that's covered.

I'm in agreement with Doug here from a users perspective.  For my uses (70% mastering/30% mixing) a fixed threshold is less flexible especially if I need to run the unit hot to hit it.

It may not be something you want to do though.

 

[plumsolly]

If the threshold does end up being fixed, I think it makes sense to error on the side lower rather than higher. I usually need to set the input on my 26C at 75% or more to get the compression going, even after I modified mine for a slightly lower threshold. This is just conjecture, but I would guess these broadcast limiters were designed to leave the signal alone until it needed to clamp down to keep the transmitter from over-modulating - hence a high threshold. In the studio, however, if we are running through the box, we probably want it to act, so a lower threshold makes sense.

Best,

Ben

 

[emrr]

Confusing the issue here is that threshold and ratio are the same thing.  One moves, so does the other.  Input attenuator takes care of pure threshold issues already.  I question whether anyone really wants a slow attack fast release fixed high ratio unit.  Once you start playing with slower attack times seems you want ratio control too.

 

[plumsolly]

Confusing the issue here is that threshold and ratio are the same thing.  One moves, so does the other.

There is a some explanation of this interaction in the Sta-Level manual.

-Ben

 

[DaveP]

Driving home from work I think I've worked out how it can be done.

There is a -30V bias on the CV signal and this has to be overcome by the positive rectified audio from the OPT.  The bias needs to get to ~-7V before the tube turns on and reduces the signal, so if I put a pot on the -30V supply and reduce it then the tube will turn on sooner and the threshold will be lower.  If I make the bias higher then the tube will turn on later, in other words, controlling the bias controls the threshold.

If I make the control vary the threshold from the stock figure down to about +4dB, will that keep the customers satisfied?

best
DaveP

 

[emrr]

I would run a bench test and see where standard ratios fall.  Use 10 dB stepped sine for tests.  If you can get 2:1 sounding good, sort amp requirements on that basis.

 

[plumsolly]

Here is the relevant Sta-Level info.

Best,

Ben

 

[DaveP]

Thanks for the feedback everyone.
just trying out the attachment gismo with Log version of compression.


Thanks plumsolly, good explanation there.
I will try various bias levels and plot the results next week when I can get back to the bench.

Is it worth making the prototype a pucca job for rack mounting or a stand alone unit, I guess everyone will make it slightly differently anyways.  Not sure that Indian word is in the US vocabulary, it means a proper job.
best
DaveP

 

[emrr]

Pucca that.  I'm always in for a rack job.  Tall and skinny I like'em; old style tubes off the back for ventilation, in an open frame rack.  Hard to find in new builds. 

Stalevel bit is among the best manual descriptions out there. 

 

[DaveP]

Doug,

I'm happy to go with the flow here but we are talking 6U, can people spare that amount of rack space?  It would make life easier for me though, I must admit.

best
DaveP