Universal Audio 176 compressor - threshold points

Hi everybody

I have recently built a 176 vari mu compressor clone using Sowter 1285 output transformer. They offer this model specifically for the 176 compressor, like the original it has primary taps for the different ratio settings.

The compressor seems to be operating as intended and it sounds very nice, however I am concerned about the threshold points at the different ratio settings.

Basically the 12:1 and 8:1 settings have a much higher threshold point than the 2:1 and 4:1 settings.
12:1 in particular is not very useful, because the threshold point is way above the point where the amplifier starts clipping.

I have managed to alter the threshold points by changing the resistors on the ratio switch - R39 for 12:1 and R45 for 8:1
Smaller resistors lower the threshold voltage, so the different ratio settings operate at similar threshold points.

I don't understand why the stock resistor values don't really work in my build. Maybe the Sowter 1285 is not exactly a drop in replacement and it actually requires some tweaks to the original circuit ?

Forgot to mention that some of the voltages marked on the UA schematic are wrong. This has been pointed out in previous discussions on the forum.

The schematic shows 45V on the 6AL5 tube when the ratio is set at 12:1
Actually this should be about 38V

My problem is, at 38V the threshold point is still too high, and I had to lower R39 value in order to get the 12:1 ratio setting working at a useful threshold point.

Hi Beatnik!
I am sorry I don't have a reply for you since I am not that expert + I don't do DIY if not on a much simpler scale of difficulty.

Nevertheless... I am putting together some literature, notes, schematics etc. for an educational project I am "building" regarding compression and compressors.

Correct me everybody if I am wrong: is the UA176 the first Vari-MU compressor (the first compressor ever actually, since it was all Vari-MU back then) to sport a RATIO control on the front panel?
Again, correct me if I am wrong: isn't it true that what UA called Ratio is actually a Threshold parameter, where different resistors alter the DC going to the 6AL5?

If I am not wrong... what is the difference between RATIO and THRESHOLD in the UA176?
Maybe there was no difference back then? In mathematical words, I can understand it: by altering the threshold you changed the ratio between input and output signal.
But nowadays, in a world of VCA compression? How does it work? What is actually doing THTRESHOLD parameter and what is RATIO doing?
I am also thinking that the 1176LN works the same way, doesn't it?

Thanks in advance everybody.

Peace, Gabe

> What is actually doing THTRESHOLD parameter and what is RATIO doing?

Threshold is a level. Ratio is a slope. Read 00A.

Design Note 00A - Basic Compressor/Limiter Design

When you do not have a real sidechain computer, the two concepts get entangled.

Hi Gabe

a world of VCA compression is a very sad world

I don't know if the 176 is historically  the first comp to have a ratio control

From what I have managed to understand of this comp, the ratio switch controls BOTH the threshold and the ratio

The different ratios are achieved by taps on the output transformer primary winding

The threshold points are determined by a voltage divider that alters the dc voltage on the 6al5 tube

This combination makes the threshold being ratio dependant - lower ratios have a lower threshold point

This makes sense because usually the higher ratios are best to limit higher peaks, whereas broader amount of compression sounds more natural at lower ratios

The 1176 has the same behaviour, although it's an entirely different compressor

You can doublecheck you have the OPT hooked up correctly and it is doing it's job correctly.
http://www.sowter.co.uk/schematics/UA176-177-OutputXfmr.jpg

That sheet gives the winding specs. Put a 1v sine wave into the primary of the OPT and measure the voltage on each output winding (with the varimu OFF).
You should get
blk: 12:1 = 1v
sl: 8:1 = .63v
vio: 4:1 =  .32v
brn: 2:1 = .13v
{edit} transformer output (9:1) should be .11v

If you get those measurements, then the transformer is doing it's job.  Possibly you have the windings flipped around or the sowter has the color coding wrong?
The transformer should be increasing the voltage being sent to the sidechain as you go to higher ratios.
At lower ratios, the voltage going to the sidechain is lower.

In conjunction with this, the threshold voltage is being changed by the switch as well.  There is a voltage divider formed by R38 and the switched in resistor. V_thr = 100v*R/(82k+R)
So the threshold voltages (measured at the R36/37/38 node) should be:
12:1 = 36.43v 
8:1 = 19.6v
4:1 = 7.66v
2:1 = 1.2v

If your build passes those tests then it is operating correctly according to the build documents.  Varimus have a compression curve set by the characteristics of the tube, which is not of a single slope - varimus have a curving compression characteristic, so they are not as easily dialed in by the sidechain as FET or other solid state compressors. The threshold has to change to get to a different part of the slope.

Some tube varimus NEED high internal signal levels to drive the sidechain - that's why they have attenuators on the output.

Notice also that if you had a signal of 36v in the amplifier going into the OPT, your sidechain levels would be:

blk: 12:1 = 36v  (vs thr=36.43v)
sl: 8:1 = 22v        (vs thr=19.6v)
vio: 4:1 =  11v    (vs thr=7.66v)
brn: 2:1 = 4.7v    (vs thr=1.2v)

So you should be pretty close to the knee of compression as you flip through the ratios, but you should definitely be seeing compression at lower ratios, while as you get to 12:1 you are just reaching the threshold.

Of course, to have a reasonable level on the output, you would need some attenuation after the OPT. 

{edit} the OPT has a 9:1 drop for the output winding, so you should have 4v output and need maybe 12 dB attenuation at least.

PRR said:

> What is actually doing THTRESHOLD parameter and what is RATIO doing?

Threshold is a level. Ratio is a slope. Read 00A.

Design Note 00A - Basic Compressor/Limiter Design

When you do not have a real sidechain computer, the two concepts get entangled.

Click to expand...

A bit unfair to extrapolate words like that. Having used compressors for more than 20 years now, I know very well the difference between Threshold and Ratio. These days I am getting deeper, analyzing old circuits and getting to know compression from the "electrical" side.

beatnik said:

[...]

From what I have managed to understand of this comp, the ratio switch controls BOTH the threshold and the ratio

The different ratios are achieved by taps on the output transformer primary winding

The threshold points are determined by a voltage divider that alters the dc voltage on the 6al5 tube

This combination makes the threshold being ratio dependant - lower ratios have a lower threshold point

This makes sense because usually the higher ratios are best to limit higher peaks, whereas broader amount of compression sounds more natural at lower ratios

The 1176 has the same behaviour, although it's an entirely different compressor

Click to expand...

Yes, I was missing the important part explained later by dmp: the AC voltage  changes on the various taps. That's an actual change of ratio between Input and Output Level. I wouldn't say though that threshold is ratio dependant, rather they work together. As you understand, that's exactly what I meant when referring to the 1176. A 20:1 ratio is actually compressing less, because of a major variation of the threshold. It IS an entirely different beast!

dmp said:

[...]

That sheet gives the winding specs. Put a 1v sine wave into the primary of the OPT and measure the voltage on each output winding (with the varimu OFF).
You should get
blk: 12:1 = 1v
sl: 8:1 = .63v
vio: 4:1 =  .32v
brn: 2:1 = .13v
{edit} transformer output (9:1) should be .11v

If you get those measurements, then the transformer is doing it's job.  Possibly you have the windings flipped around or the sowter has the color coding wrong?
The transformer should be increasing the voltage being sent to the sidechain as you go to higher ratios.
At lower ratios, the voltage going to the sidechain is lower.

In conjunction with this, the threshold voltage is being changed by the switch as well.  There is a voltage divider formed by R38 and the switched in resistor. V_thr = 100v*R/(82k+R)
So the threshold voltages (measured at the R36/37/38 node) should be:
12:1 = 36.43v 
8:1 = 19.6v
4:1 = 7.66v
2:1 = 1.2v

[...]

Click to expand...

Thank you smp!!! Upon replying to beatnik, you made it perfectly clear to me.  All I wanted is in your reply!

GabrieleP said:

Yes, I was missing the important part explained later by dmp: the AC voltage  changes on the various taps. That's an actual change of ratio between Input and Output Level. I wouldn't say though that threshold is ratio dependant, rather they work together. As you understand, that's exactly what I meant when referring to the 1176. A 20:1 ratio is actually compressing less, because of a major variation of the threshold. It IS an entirely different beast!

Click to expand...

The 1176 changes the threshold along with the ratio, but it is very different, as it is a FET type (the universal audio 1176 manual does a nice job explaining this).

There is a ratio of the tertiary windings and a ratio of compression - but these are separate "ratios" - they are not directly related.
If you look at the example figures I put above and carry it a step further: suppose the amplifier voltage doubles to 72v. Then the voltages for each tertiary winding are:
blk: 12:1 = 72v  (vs thr=36.43v)
sl: 8:1 = 45v        (vs thr=19.6v)
vio: 4:1 =  23v    (vs thr=7.66v)
brn: 2:1 = 9.4v    (vs thr=1.2v)

Notice all of these voltages are double what they were before as well. The different ratios of the tertiary windings did not change the "ratio" of voltage sent to the sidechain when the amp voltage changed. What this does though is allow the threshold to be changed without requiring the signal level in the amplifier to be significantly adjusted.
It might be helpful to read over the sta level manual as it is another vari mu compressor and the manual explains how changing the sidechain threshold changes the compression ratio (page 7 and drawing A-11734)
http://www.realoldiesradio.com/docs/StaLevel/Sta-Level_Manual.pdf
 
"The reason the compression ratio is changed is because the RMS voltage is increased logarithmically with the increase in level in dB. The rectified bias is still a function of the RMS voltage. Thus it does not require such a  change in dB to obtain the required bias level with high signal levels as with low signal levels"

Note that in the sta level they discourage having a threshold/ratio control since the amp settings (i.e. level) would have to be re set so drastically, since it does not have the separate windings for SC voltage to automatically scale it for you.

dmp said:

[...]

In conjunction with this, the threshold voltage is being changed by the switch as well.  There is a voltage divider formed by R38 and the switched in resistor. V_thr = 100v*R/(82k+R)
So the threshold voltages (measured at the R36/37/38 node) should be:
12:1 = 36.43v 
8:1 = 19.6v
4:1 = 7.66v
2:1 = 1.2v

[...]

Click to expand...

Allow me smp... How did you get these threshold values? Following voltages and applying math to the schematic?
In other words: are these  DC BIAS voltages or rather AC RMS values?  (following terminology of the Gates STA-Level manual which I read after  your precious suggestion!).

If they are DC BIAS voltages, AC signals of the amp would have to be 0,707 times those for compression to begin, correct?
So, for the 36,43v of the 12:1 ratio, AC signal from the amp would have to be 25,75v... 2,86v after the 9:1 conversion of the OPT as you suggest...  11,34 dBu/7,34 dBVU.

GabrieleP said:

How did you get these threshold values? Following voltages and applying math to the schematic?

Click to expand...

Those are DC voltages calculated with a voltage divider between 100v and 0v. The signal voltage (AC) will trigger compression when it exceeds the threshold.

OK! Also reading pag. 8 from the Fairchild 670 Instructions makes a lot more sense now.

Referring to the UA 176, I understand that setting the RATIO equates to changing simultaneously DC and AC thresholds.

The only reason for this would be to maintain a constant Gain Reduction dB value for different levels at the input (in old words the Program Material).

Correct?

The explanation on the Gates STA-Level manual is revelatory: "The compression ratio may be increased up to 6:1 instead of 3.3:1 by raising the bias on the cathode of V6. [...] The output and input levels of the amplifier must be raised to start the compression action and the distortion will increase along with these levels. The reason the compression ratio is changed is because the RMS voltage is increased logarithmically with the increase in level in dB. The rectified bias is still a function of the RMS voltage. Thus, it does not require such a change in db to obtain the required bias level with high signal levels as with low signal levels".

To put in other words: the only way to change Ratio is to turn up INPUT level (as in every other Vari-MU limiting amplifier).
Being logarithmic math, a slight level variation of signals already high at the input corresponds to a big variation of the DC bias. The higher you raise the DC bias, the lesser you need to raise the INPUT level in order to maintain constant Gain Reduction. Although I'd rather think in terms of RMS voltage (Ratio) and re-adjust DC bias (Threshold) after...

It is clear now to me that on the UA 176 all this is taken care by the RATIO parameter, without the need to reach the INPUT knob. Smart!

Thanks again smp, great inputs!

dmp said:

Put a 1v sine wave into the primary of the OPT and measure the voltage on each output winding (with the varimu OFF).
You should get
blk: 12:1 = 1v
sl: 8:1 = .63v
vio: 4:1 =  .32v
brn: 2:1 = .13v
{edit} transformer output (9:1) should be .11v

Click to expand...

Thanks for the very useful information.

I have made sure that the transformer is wired correctly, and the Sowter wiring colour code matches the original

I have done the measurements you suggested and my results are very close to your calculations

blk: 12:1 = 1V
sl: 8:1 = .58V
vio: 4:1 =  .29V
brn: 2:1 = .12V

The threshold voltages determined by the resistor network seem to be correct as well (keep in mind I have more than 100V coming off the regulator tube)

12:1 = 39.8V
8:1 = 21.4V
4:1 = 8.4V
2:1 = 1.3V

I guess my build is performing correctly according to the original schematic, however I find that the 12:1 ratio setting has too high threshold point, and you can get it compressing only when the amplifier is already distorting quite heavily, but maybe this is just how these units were designed ?

beatnik said:

I guess my build is performing correctly according to the original schematic, however I find that the 12:1 ratio setting has too high threshold point, and you can get it compressing only when the amplifier is already distorting quite heavily, but maybe this is just how these units were designed ?

Click to expand...

Hey beatnik,
Looks like the sidechain is correct and you could adjust the resistances to move the thresholds so it works better for you.
But it sounds strange to me that the amplifier is distorting heavily.
Vari mus are meant to have high internal amplifier levels in order to drive the sidechain.
Can you measure the voltages on the anode and cathodes of the tubes?
What did you use for interstage transformer and is it connected correctly? I believe it should be 80k:15k, so stepping down the signal by a little over half.

Also part of threshold, just like an 1176:  the input level control.  Change ratio, also change input level. 

If the attack time is very slow and the release very fast, the timing cap has little chance of charging to provide compression, and with maximum ratio by nature being near amplifier clipping point, peaks will get by and clip.  This is true in every vari-mu I've worked on/with/modified.