Threshold control on single ended tube limiter

I've been tinkering with this single ended tube limiter I picked up recently.  Luckily the schematic was pasted inside the chassis.  I've redrawn it and attached here.  Looks to me like this is from the late 30's or early 40's
It was modified (with the side chain removed and a 6SF5 in V1) before I got it but I was able to bring it back to  stock.  The caps are mostly oil filled and I've read these are pretty reliable so I replaced C12 & C13 with 8uF electrolytics and fired it up. 

The tertiary winding on the output transformer feeds the sidechain for compression.
Turns out the output transformer was no good, very low output - but luckily I had an output transformer with a similar tertiary winding and getting good output now.  The main problem I'm faced with now is that the threshold of limiting is all out of whack.  P1 is a stepped attenuator and the amp needs a very hot output to trigger the compression, but it also means the output is severely distorted by that point.  I assume P2 is a threshold control?  I have a couple vari mu compressors but I have a pretty rudimentary understanding of how the diode tubes actually work.  I was hoping for some help in modifying the threshold on this to be in a useful range

This is obviously designed for speech (measured very little bass response and a sizable presence boost around 7K). I'm not expecting fidelity with this but its a beautifully made piece of equipment and I'd love to get it into a little bit more useable state with minimal modification. thanks!

No voltages measured?

Short R24, see if you get anything better. 

Tubes all good?  Double checked resistance values? 

I'd immediately suspect the original tertiary was a step-up, and it's not getting enough with the new one.  One example that's tough to replicate is the Dept. of Commerce limiters, where the tertiary is a step-up to 450ish K from 30-60K.  One solution would be adding another tube stage for side chain gain.

Some more info would help, who made it, pictures, etc.  You never know who actually has documents or first hand knowledge. 

In fact this was made for the CAA.  It's a similar circuit and tube line up to the Signal Corps AM-447. I took the replacement output transformer from one of those.  Original OT was 6k:600&6k, the AM447 OT is 5k:600&5k. I'll upload some photos when I'm back at the bench

Earlier I did try shorting R24 and it helped get the threshold closer to a usable range but output is still clipping or near clipping by the time the threshold starts to engage.  DMM across R21/R22 (they are in parallel) measured ~19K which is low. I'll verify other resistance readings and get some voltage readings next chance I get.

hereforever said:

DMM across R21/R22 (they are in parallel) measured ~19K which is low.

Click to expand...

With R24 shorted and P2 at ground that shouldn't matter one bit. 

If P2 wiper is at ground it should compress all the time, with a lower ratio, assuming there's signal present to be rectified. 

Interesting difference in the AM-447, that tap is used for headphone and side chain comes off a voltage divider from the plate.  You could try that if you suspect anything about the transformer.    You can raise the divider point to increase signal and see, maybe try it with a 100K pot. 

Maybe use a separate step up transformer to boost level from the tertiary winding?

Back with a guts photo and some voltages...

V1
----
3/5: 235 mV
4: -0.59 V
6: 15.3 V
8: 108.2 V

V2
----
3: 209 V
4: 17.9 V
5/8: 0.6 V
9: 0 V (-7mV measured)

V3
----
3: 270 V
4: 272 V
5: 0 V (4mV measured)
8: 30.8 V

V4
----
5: -209mV
8: 11.5 V with P2 fully ccw, 26.7 V P2 fully cw

Another issue is that the 1ma gain reduction meter isn't reading correctly.  With no gain reduction happening it reads more like .8ma

I'll leave this cropped GIF here; easier than opening a huge PDF.

At these voltages, the 6SK7 needs around -3V on G1 to give substantial (~~20dB) limiting. If that is truly a 10k winding to the diode, this would be near 0.6V on the 600r winding. If the primary is up near 25k then the 6F6 should be nowhere near clipping. (It's a 280V 8mA stage and should deliver +30dBm clipped, loafing at -2dBm.) So I dunno.

I suspect the plate/screen feed of V2 has a filter cap on it.

PRR said:

I suspect the plate/screen feed of V2 has a filter cap on it.

Click to expand...

Correct- .5uF cap there, error when I transcribed the schematic
PRR thank you for converting the schematic to be more legible and adding voltages

EmRR said:

Interesting difference in the AM-447, that tap is used for headphone and side chain comes off a voltage divider from the plate.  You could try that if you suspect anything about the transformer.    You can raise the divider point to increase signal and see, maybe try it with a 100K pot. 

Click to expand...

Since I'm not sure about the OT I think I might give this a try