Threshold values in a vari-mu comp

Hi all...

I have built a vacuum tube vari-mu comp and to control the amount of signal through the side chain I use a 6pos/ 2pol switch with resistors. This is to get as accurate balance as possible. Everything is working perfect and the comp has very fine values and sounds very good.

But... I wonder how to proceed to set specific threshold values. It is not extremely important but it would be nice to know how to measure or calculate these values.

I do not have a complete schematic to post at the moment because I have it on another laptop and it has been changed quite a lot and is therefore not "updated" But I guess that the question could be answered in a more common way.

The comp has a gain of 26dB but since it has to handle signals from a previous preamp I guess that the signal from the preamp sets the threshold. I guess (again) that I could use a preamp measure the output signal of... lets say 50dB and then connect the comp and elaborate with a pair of resistors until the comp show gain reduction and so on? It will take som serious time to do it of course

Sorry for a maybe stupid question.


Best regards

/John

What is "threshold"?

Let's ponder a very different limiter. You hire/draft people for an army. They come in all different heights, midgets to giants. OK, except anybody over 5'9" (172cm) won't fit in the cheap Jeep your army bought. So you build a machine to trim the tall people down to size.

You could just use a blade at 5'9". However this does something bad to the tall guys.

What you would like to do is apply a shrink-ray, but only to the top of the tall ones. If you set the ray to shrink 50% from 5'9" on up, the 6'9" guy comes out 6'3", still too tall. Start lower, say 5'0". At 50%, 2:1 crunch, people up to 6'6" will come out <= 5'9", as needed. But that 6'9" guy is still too tall.

The usual point of a limiter is so the next stage of the chain (transmitter, tape, PA amp) will NOT "hit its head". But peaks can come way over-size. You could one day get a recruit over 7 feet (Robert Wadlow was 8'11"). So the "threshold" is really totally about knowing your program material and how much head-room dent can be tolerated for occasional Wadlows.

maybe you could temporarily replace the switch with a dual pot, set it to where you want the first threshold to be, take it out of circuit and measure its resistance?  repeat for each of the six positions you want on your switch.

Matt C said:

maybe you could temporarily replace the switch with a dual pot, set it to where you want the first threshold to be, take it out of circuit and measure its resistance?  repeat for each of the six positions you want on your switch.

Click to expand...

Thank you sir

I have thought of that but since it is hard to find a double pot that is accurate over both gangs I believe it could cause imbalance and maybe just stir things up in the process? Would you say its the best way?

Best regards

/John

PRR said:

What is "threshold"?

Let's ponder a very different limiter. You hire/draft people for an army. They come in all different heights, midgets to giants. OK, except anybody over 5'9" (172cm) won't fit in the cheap Jeep your army bought. So you build a machine to trim the tall people down to size.

You could just use a blade at 5'9". However this does something bad to the tall guys.

What you would like to do is apply a shrink-ray, but only to the top of the tall ones. If you set the ray to shrink 50% from 5'9" on up, the 6'9" guy comes out 6'3", still too tall. Start lower, say 5'0". At 50%, 2:1 crunch, people up to 6'6" will come out <= 5'9", as needed. But that 6'9" guy is still too tall.

The usual point of a limiter is so the next stage of the chain (transmitter, tape, PA amp) will NOT "hit its head". But peaks can come way over-size. You could one day get a recruit over 7 feet (Robert Wadlow was 8'11"). So the "threshold" is really totally about knowing your program material and how much head-room dent can be tolerated for occasional Wadlows.

Click to expand...

Hmm... fair enough.

Threshold has no universal definition; each designer/manufacturer has his own.
However it seems the consensus is "the threshold is reached when some amount of GR is noted"; obviously it depends on how much is "some". Usually something like 0.5 to 3 dB.
The subject is made more complex by the fact that compression curves are somewhat continuously variable, as is the case with feedback detection compressors. Typically, there is a quite large range of input voltage around the threshold where GR varies from imperceptible to a fraction of dB.

In most compressors there is a rectifier device in the side chain. Below a certain voltage this rectifier typically outputs nothing. Above this voltage it begins to conduct and an generates a gain reducing voltage. It is this device that in most compressors defines the threshold. As Abbey has pointed out, the actual value of the threshold is a bit cloudy.

In most compressors the side chain is driven from the output of the compressor so the threshold signal level is related directly to the output level. If, for example the threshold is 0dBu then any signal that produces an output lower that 0dBu will not be compressed. If you have 26dB of gain in the compressor from input to output then the input signal threshold would be -26dBu. If you add gain between the output and the rectifier you can reduce the threshold further. In summary, the input threshold is the rectifier threshold less  the total gain in the loop from the input to the rectifier. Changing the nominal gain anywhere in this loop changes the threshold ( and nothing else).

Cheers

Ian

abbey road d enfer said:

Threshold has no universal definition; each designer/manufacturer has his own.
However it seems the consensus is "the threshold is reached when some amount of GR is noted"; obviously it depends on how much is "some". Usually something like 0.5 to 3 dB.
The subject is made more complex by the fact that compression curves are somewhat continuously variable, as is the case with feedback detection compressors. Typically, there is a quite large range of input voltage around the threshold where GR varies from imperceptible to a fraction of dB.

Click to expand...

Thank you sir

I have thought about this back and forth and I realize that there could be various definitions where this threshold would be. I guess I can settle with the idea that it does not have to be determined that precise and therefore just leave it with, in my case, six different settings. I just had an idea that it maybe could be measured or calculated. I have a couple of FMR RNC 1773 compressors that have the threshold noted -40, -20, -10, 0, +10 and +20... I cannot really tell what it all means 

Best regards

/John

ruffrecords said:

In most compressors there is a rectifier device in the side chain. Below a certain voltage this rectifier typically outputs nothing. Above this voltage it begins to conduct and an generates a gain reducing voltage. It is this device that in most compressors defines the threshold. As Abbey has pointed out, the actual value of the threshold is a bit cloudy.

In most compressors the side chain is driven from the output of the compressor so the threshold signal level is related directly to the output level. If, for example the threshold is 0dBu then any signal that produces an output lower that 0dBu will not be compressed. If you have 26dB of gain in the compressor from input to output then the input signal threshold would be -26dBu. If you add gain between the output and the rectifier you can reduce the threshold further. In summary, the input threshold is the rectifier threshold less  the total gain in the loop from the input to the rectifier. Changing the nominal gain anywhere in this loop changes the threshold ( and nothing else).

Cheers

Ian

Click to expand...

Thank you sir

Yes, this is understood to some extent... varying the gain affects the GR signal as well. That was the reason for my initial question if it all could be measured and determined... but as I understand now it would be very tricky to do so.

I had a glance at the Manley vari-mu comp and they do not mark any values on their frontpanel so I guess that I could do without it as well?

Best regards

/John

johnheath said:

I had a glance at the Manley vari-mu comp and they do not mark any values on their frontpanel so I guess that I could do without it as well?

Click to expand...

I like it better when gear knobs are marked.
For example less/more, left/right, stinks/rules, sucks/rocks  ;D

abbey road d enfer said:

I like it better when gear knobs are marked.
For example less/more, left/right, stinks/rules, sucks/rocks  ;D

Click to expand...

Haha - yes... some marking is probably needed for the personal well-being

Do you have input and output attenuators or is the threshold the only way of adjusting the amount of compression?

ruairioflaherty said:

Do you have input and output attenuators or is the threshold the only way of adjusting the amount of compression?

Click to expand...

Yes sir, attenuators both at input and output

Best regards

/John

johnheath said:

Yes sir, attenuators both at input and output

Best regards

/John

Click to expand...

Great!  I use a custom Vari-Mu daily in my mastering that has both input and output attenuators as well as threshold.

My whole chain is pretty predictable level wise but with the Input and Output attenuators I really only vary the threshold over a small range.  It's best to determine what works for you by ear and match your workflow.  If you like to hit the A/D at a certain level etc.

Looking forward to seeing what you've made.

johnheath said:

Yes sir, attenuators both at input and output

Best regards

/John

Click to expand...

I many compressors (LA2A, Neve 2254) the output gain control is outside the loop so it does not affect the threshold.

Cheers

Ian

ruffrecords said:

I many compressors (LA2A, Neve 2254) the output gain control is outside the loop so it does not affect the threshold.

Cheers

Ian

Click to expand...

Indeed.  I asked because in modern recording/mixing/mastering we are often using a Vari Mu compressor for it's inherent distortion and tone.  Without both input and output attenuators as well as threshold you often end up hunting for a sweet spot where you are compressing the desired amount and driving the tubes/transformers into distortion in the desired amount and achieving a suitable output level.

Two hypothetical extremes in my daily workflow
1) Attenuate input, open output for a  cleaner sound and compressing quite a lot by bringing the compressor threshold down (dynamic jazz record)

2) Indie rock record that already compressed too much where I set the input attenuator wide open, attenuate the output and set the threshold so that it's barely compressing.  This gives plenty of tone and distortion from hitting the transformers and tubes hard and on occasion can be just the thing (although not nearly as often as Gearslutz would have you believe).

For mastering I'd probably want a 12 way switch for threshold but in recording & mixing I could see that a 6 way would be enough.


 

ruairioflaherty said:

Great!  I use a pretty incredible Vari-Mu daily in my mastering - The Magic Death Eye - http://magicdeatheye.com/stereo-compressor/

My whole chain is pretty predictable level wise but with the Input and Output attenuators I really only vary the threshold over a small range.  It's best to determine what works for you by ear and match your workflow.  If you like to hit the A/D at a certain level etc.

Looking forward to seeing what you've made.

Click to expand...

Thank you sir

That machine looks awesome! It is surely a beast... mine is a lot more simple I guess (probably looks more like an Altec 436 or an UA 175).

ruairioflaherty said:

Indeed.  I asked because in modern recording/mixing/mastering we are often using a Vari Mu compressor for it's inherent distortion and tone.  Without both input and output attenuators as well as threshold you often end up hunting for a sweet spot where you are compressing the desired amount and driving the tubes/transformers into distortion in the desired amount and achieving a suitable output level.

Two hypothetical extremes in my daily workflow
1) Attenuate input, open output for a  cleaner sound and compressing quite a lot by bringing the compressor threshold down (dynamic jazz record)

2) Indie rock record that already compressed too much where I set the input attenuator wide open, attenuate the output and set the threshold so that it's barely compressing.  This gives plenty of tone and distortion from hitting the transformers and tubes hard and on occasion can be just the thing (although not nearly as often as Gearslutz would have you believe).

For mastering I'd probably want a 12 way switch for threshold but in recording & mixing I could see that a 6 way would be enough.


 

Click to expand...

That seems to be a nice tip. I could easily change the threshold (6-pos) switch to a 12-pos one and divide the resistors to get 12 smaller and still useful steps?

Best regards

/John

johnheath said:

That machine looks awesome! It is surely a beast... mine is a lot more simple I guess (probably looks more like an Altec 436 or an UA 175).

Click to expand...

Yes, it's a beast that can also tip toe ballet style   I can honestly say that if you offered me a Fairchild 670 in exchange for it I would refuse, it's better in every way.

As for your simple and focused is good, we have all the (often needless) complexity we could ask for in plug ins these days.

That seems to be a nice tip. I could easily change the threshold (6-pos) switch to a 12-pos one and divide the resistors to get 12 smaller and still useful steps?

Click to expand...

I hate feature creep and adding stuff for the sake of it but in this case I think it might be a good idea, it will give access to a broader range of in/out gain options and thus a broader pallet.

ruairioflaherty said:

Yes, it's a beast that can also tip toe ballet style   I can honestly say that if you offered me a Fairchild 670 in exchange for it I would refuse, it's better in every way.

As for your simple and focused is good, we have all the (often needless) complexity we could ask for in plug ins these days.

I hate feature creep and adding stuff for the sake of it but in this case I think it might be a good idea, it will give access to a broader range of in/out gain options and thus a broader pallet.

Click to expand...

Yes, simple is the word but I do in fact have attack and release controls and a few other features on mine... Altec436 is little too simple for me (built a pair and did not really like them... but that is my personal opinion)

And I agree that changing just for the cause of changing is bogus to me but as mentioned a 12-pos switch really wouldn't change much except increasing the number of settings

Best segrads

/John

The tradition with feedback limiters is to provide an input gain control and an output level control. The threshold is fixed inside of the device, but by using these two controls, you can place the gain reduction wherever you want relative to your signal levels. Many classic ones use switched attenuators, so you can set and reset easily, or match two channels.

You could fold all of that up into one switch, but it becomes very complex and counter-intuitive for those who have used classic feedback limiters.

If you want to change the actual threshold level of the limiter, you can change the DC value sent to the rectifiers that sample the output plate voltage. However, all you're doing is changing the gain structure inside of the limiter, which will usually just make the limiter noisier. Most vari-mu stages are prone to hum from imbalance and power supply ripple, since one can't put feedback across them and still get variable gain. So, using anything lower than the 'standard' threshold voltage will probably just add hum.

I say there's no problem to fix - you just need to get used to non-VCA type processors. The whole idea of 'threshold' is from feedforward processors that have a dB-linear VCA and a real level detector, something that didn't exist until the 70s.