SSL Clone compressor ratio measurements

[Biasrocks]

Harpo

Been following along with great interest.

I've got a dbx202c in my GSSL build. I'm wondering if the resistor choices are optimal for this VCA. I tried to do some of the math myself, but I couldn't find the current draw for Iin + Iout for the 202C, so I'm at a loss. And the 1K seems like it should be 100R + 909R according to the SSL schematic. Also the 27K should be higher no, it's 68K1 in the original but I realize that the 5534 is on 15V rails as opposed to 18V in the SSL so that would drop it down to a 13V swing instead of 16V right?

Could you shed some light on this confusion for me.

Mark

 

[Harpo]

Mark,
not much help, couldn't find the limits of signal current for the 202C. The DBX202 paper showing a 50k for Rin and Rfb with unknown source level. Lowest Rin and Rfb around a +/-15V supplied circuit I could find was 33k in a MCI-500. Assuming Iin and Iout same value + a little quiescent current, a 202C might allow about 1mA for safe operation. Maybe ask/pm member gkhebert for a probably better answer to this question. The 68k was probably chosen in favour of lower THD figures at the cost of SNR.
Resistor at control port 100R+909R for what type of VCA ? The original (ssl_82e26/ssl_82e27) looks like 100R+4k7 at audio-VCA as well as sidechain-VCA for a 202C (or 1k3 for a 202X from a different original), but the surrounding circuit isn't exactly the same either, compared to the GSSL.

glbaudio,
increasing the Rx would increase noise modulation, so I'd favour the decreasing of Rfb at the TL072. YMMV. Just try both variants yourself. It's only a single resistor. Increasing the cap across this Rfb might help better sleep.

 

[Biasrocks]

Good info, thanks Harpo.  8)

Mark

 

[MeToo2]

Thanks for the info. Initially I was measuring too close to the knee and worrying that my ratios weren't working correctly. So this thread put my mind at rest.

Now got the actual GR, the GR meter, and the ratio to track within a dB (Black Can DBX 202 with 3k3 series resistor + 2K trim pot, instead of the 1K of the original gssl 202c design) Noise floor looks pretty good too.

 

[MikeClev]

I finally took the time to measure and calibrate my dual GSSL. (2 GSSLs in a fiery red 3ru case).

I read this thread thoroughly and have read the entire GSSL help thread. I think I'm ready to tackle it now!

I put a -15dbfs 1000hz sine tone (Cubase test tone oscillator) into the GSSL inputs. (-15dB was low enough to be below the compression threshold)

Took the GSSL outputs back into the audio interface and metered the input on my DAW, Cubase.

I set the ratio to 2:1

I set attack to fast

I set release to 1.2s

Make up gain to 0

Bypass disengaged

Turn threshold anticlockwise until the input meter was reading 10dB lower than before, meaning 10dB of compression was happening. 10dB compression should be far enough away from the knee to get a good measurement.

Then I increase the amplitude of the sine tone from Cubase by 10dB

Then read the level of the tone coming back out of the GSSL. 

From this we can determine how much compression is happening.

So, if I start at -25dBfs and the level increases (by 5dB) to -20dBfs (when I raise the input by 10dB), then I have 10dB:5dB = 2:1 compression. Voila!

So on to the thorough testing...

There are 2 variables I wanted to vary during this test. the 100k/127k resistor between the TL074 and Sidechain VCA, and the two 1k resistors feeding the DBX202 emulation circuit's control voltage port. I added a trimpot in series with a 100k resistor for the 100k/127k resistor, and added a 2k trimpot instead of the jumper that feeds the two 1k resistors. This allowed careful tuning of all three resistances.

I used 2180LB VCAs in the audio paths and That2150A in the sidechain. I carried out the alterations per VCAs detailed here: http://diy.fischerworks.com/gssl_vca.shtml

The results:

Sidechain Resistor: 127k
Control Voltage resistor: 1k
2:1    5.9dB  (should be 5dB)
4:1    3.1dB  (should be 2.5dB)
10:1  1.8dB  (should be 1dB)

Sidechain Resistor: 120k
Control Voltage resistor: 1k
2:1    5.7dB
4:1    2.7dB
10:1  1.3dB

Close, very close. But I wanted to see what the other variables would give me. I also wanted that 10:1 to be a bit harder.

Sidechain Resistor: 100k
Control Voltage resistor: 1k
2:1    5.4dB
4:1    1.5dB
10:1  -0.3dB  Beyond infinity compression!

Sidechain Resistor: 120k
Control Voltage resistor: 1k+120ohm trimmer in series
2:1    6dB
4:1    2.9dB
10:1  1.8dB

Sidechain Resistor: 120k
Control Voltage resistors: 1k+240 ohm trimmer in series
2:1    6.3dB
4:1    3.5dB
10:1  2.4dB

Sidechain Resistor: 120k
Control Voltage resistors: 890ohm. I lowered them both by sticking a resistor in parallel with each temporarily.
2:1    7.4dB
4:1    7dB
10:1  6.7dB Totally going in the wrong direction! Flattening all the ratios out, making them closer together.

Sidechain Resistor: 100k. Changed this to what was originally specified for the That 2150 GSSL circuit.
Control Voltage resistor: 1k plus 240 trimmer in series
2:1    5.3dB (should be 5dB)
4:1    2.3dB (should be 2.5dB)
10:1  0.9dB (should be 1 dB)

Jackpot!  Thats close enough for me!

So what have I learned... 

That some vital details are left out of the GSSL documentation, or scattered all over the forums, so its no wonder people get odd results when they build their GSSLs. If these details were collected in one place, people would be less likely to run into trouble.

Also I think this tells me that the choice of 100k/120k/127k sidechain resistor depends on what VCA you use in the sidechain. 100k is correct for the That 2150A, and 120k is apparently correct for the 218x series. (according to what other people have said, I haven't tested this!) This resistor appears to shift the amount of compression at all ratios up and down together.

The choice of resistor for the audio VCA control voltage also depends on the choice of VCA, and for the That 2180LB,  240ohm in series before the two 1k resistors worked for me. Increasing this value 'flattens' the ratios out, making them closer in value, so if you are getting negative compression / past infinity ratios, this is where to look.

I hope this information helps somebody!

-Mike C

 

[guze]

Very nice
Thanks

 

[bobschwenkler]

Just came across this thread for the first time... I'll grab mine at some point and check out how it's performing. I think though, that I may also experiment with negative ratios. For what it's worth to anyone else to know, when I used to use Waves stuff, the C1 compressor could be set with negative ratios and I used it multiple times this way to create some really cool sounds, especially with very short attack and release times. It could be very useful to me if I could get anything similar out of this box, install a switch to change "modes".

 

[tekno808]

Hi,

Finished my GSSL with Turbo using THAT 2180C , changed the resitors in the vca according to the THAT2180c and the R in the sidechain 100k to 122k, added VR 4.7k for the meter calibration and the bypass mod from SSLtech !
Ratios are ok for me with 10db increase 2:1 = 4.9db  4:1= 2.8db   10:1= 0.9db !

Now i checked the Makeup and i see that i have 16db so it's going from 0 to 16db !, so if i want 20db i have to change the 100r resistors in the output section?

Thx....
Danny.

 

[MeToo2]

Hi,

Finished my GSSL with Turbo using THAT 2180C , changed the resitors in the vca according to the THAT2180c and the R in the sidechain 100k to 122k, added VR 4.7k for the meter calibration !
Ratios are ok for me with 10db increase 2:1 = 4.9db  4:1= 2.8db   10:1= 0.9db !

Now i checked the Makeup and i see that i have 16db so it's going from 0 to 16db !, so if i want 20db i have to change the 100r resistors in the output section?

Thx....
Danny.

Nope.Leave the 100R resistors alone. You should probably try changing the 620K resistor near the TLO72. It's acting as a straight inverting amp so the gain is Rf/Rin on the DC offset provided by the make-up gain the pot. Rin is currently 620K. Rf is now 122K. You can then work out the new required value for Rin from that..... 122/620 -> 16dB. You want 20dB. The VCA already works with log gain values, so therefore Rin will scale linearly and should be approx 16/20*620 = 487K in E48 or 510K as the nearest standard E24 value.

 

[tekno808]

Nope.Leave the 100R resistors alone. You should probably try changing the 620K resistor near the TLO72. It's acting as a straight inverting amp so the gain is Rf/Rin on the DC offset provided by the make-up gain the pot. Rin is currently 620K. Rf is now 122K. You can then work out the new required value for Rin from that..... 122/620 -> 16dB. You want 20dB. The VCA already works with log gain values, so therefore Rin will scale linearly and should be approx 16/20*620 = 487K in E48 or 510K as the nearest standard E24 value.

Thx for the quick reply, i did a quick mod by soldering a 2.2M resistor in parallel to the 620k and now i have a range of 20 db ! but still have to mod the output section so that i have in bypass 6 db more so the input and output have the same level ( in bypass output is 6db less) and also with bypass off no compression and the makeup fully ccw !

 

[MeToo2]

Nope.Leave the 100R resistors alone. You should probably try changing the 620K resistor near the TLO72. It's acting as a straight inverting amp so the gain is Rf/Rin on the DC offset provided by the make-up gain the pot. Rin is currently 620K. Rf is now 122K. You can then work out the new required value for Rin from that..... 122/620 -> 16dB. You want 20dB. The VCA already works with log gain values, so therefore Rin will scale linearly and should be approx 16/20*620 = 487K in E48 or 510K as the nearest standard E24 value.

Thx for the quick reply, i did a quick mod by soldering a 2.2M resistor in parallel to the 620k and now i have a range of 20 db !

Good.

but still have to mod the output section so that i have in bypass 6 db more so the input and output have the same level ( in bypass output is 6db less) and also with bypass off no compression and the makeup fully ccw !

This I don't understand. Setting the make-up pot fully CCW should give 0V at the input to the 620K resistor. Switching in the bypass in should also give 0V at the input to the 620K resistor, and also ensure no signal gets into the side chain amp (so there'll be no compression). Maybe you should check the wiring of your bypass switch and these voltages relative to signal ground.

 

[tekno808]

This I don't understand. Setting the make-up pot fully CCW should give 0V at the input to the 620K resistor. Switching in the bypass in should also give 0V at the input to the 620K resistor, and also ensure no signal gets into the side chain amp (so there'll be no compression). Maybe you should check the wiring of your bypass switch and these voltages relative to signal ground.

I mean @ the input on the GSSL i give a test signal 1khz -10 db and i get @ the output on GSSL -16 db with the makeup fully CCW and also in bypass mode !
So normally it should be the same value , Input GSSL -10db --- Output GSSL -10db , in bypass and whit the the makeup fully CCW.
( the bypass wiring is ok , in bypass mode the input to the 620k resistor is switched to the ground ! )
So thats why i want to mod the outpustage so it give me 6db more @ the output on the GSSL.

Danny.

 

[kante1603]

I mean @ the input on the GSSL i give a test signal 1khz -10 db and i get @ the output on GSSL -16 db db with the makeup fully CCW and also in bypass mode !
So normally it should be the same value , Input GSSL -10db --- Output GSSL -10db , in bypass and whit the the makeup fully CCW.
( the bypass wiring is ok , in bypass mode the input to the 620k resistor is switched to the ground ! )
So thats why i want to mod the outpustage so it give me 6db more @ the output on the GSSL.

Danny.

Hi Danny,

just a quick thought.
Maybe(!) it´s just a balancing issue arround the NE5532s at the output (trace both o/ps from Pin 1 to pos. XLR pin 2 and from pin 7 to XLR pin 3-these "values" must be equal to each other).If one is is missing you´ll have a loss of 6dB in all modes.
You can also measure the pins 2 and 3 against the pin 1(gnd) on XLRs while feeding a signal.

Best,

Udo.

 

[MeToo2]

This I don't understand. Setting the make-up pot fully CCW should give 0V at the input to the 620K resistor. Switching in the bypass in should also give 0V at the input to the 620K resistor, and also ensure no signal gets into the side chain amp (so there'll be no compression). Maybe you should check the wiring of your bypass switch and these voltages relative to signal ground.

I mean @ the input on the GSSL i give a test signal 1khz -10 db and i get @ the output on GSSL -16 db db with the makeup fully CCW and also in bypass mode !
So normally it should be the same value , Input GSSL -10db --- Output GSSL -10db , in bypass and whit the the makeup fully CCW.
( the bypass wiring is ok , in bypass mode the input to the 620k resistor is switched to the ground ! )
So thats why i want to mod the outpustage so it give me 6db more @ the output on the GSSL.

Danny.

Ah OK. Understood. Then you need to check the relative values of the 27K on the input feeding the DBX202 (emulator) compared to the 15K feedback resistor (in parallel with the 100pF) of the NE5532, although it seems odd/unlikely that that this ratio is so far out.  Could also be that your balancing amp isn't working properly on the 2nd half of the 5532. Or your output XLR isn't wired up correctly. Also be aware that if you are using unbalanced cables on the output that you will lose 6dB on the output stage as pin 3 of the XLR will be held at ground, even though the 5532 may be driving the other side of the 100uF cap correctly.

Assuming you have a -10dBV 1KHz test tone, 0 compression, 0 make up gain.....

Just trace through with an AC voltmeter to see where the "problem" is i.e. measure

1) between pins 2 & 3 of the input XLR: this should measure the same as your input test tone i.e. 0dB gain = -10dBV.
2) Pin 6 of the 5524 to ground (should be the same as your test tone i.e. 0dB gain = -10dBV).
3) Pin 1 of the 5532 to ground (should be down 6dB relative to the test tone = -16dBV).
4) Pin 7 of the 5532 to ground (should also be down 6dB relative to the test tone but opposite phase to above = -16dBV)
5) Pin 2 of the output XLR to ground (should be 6dB down relative to the test tone = -16dBV.)
6) Pin 3 of the output XLR to ground (should be 6dB down relative to the test tone but opposite in phase to above = -16dBV)
7) Between pins 2-3 of the output XLR (should be the same as your test tone = 0dB down on the input = -10dBV)

 

[tekno808]

Thx.. MeToo2 and kante1603 !!!

My mistake , the cabels i use is a xlr to mono jack so it's connected balanced to unbalanced going to my AD/DA converter that's the reason for 6 db loss in signal !
Removing on my GSSL the xlr output connection give me the right output Between pins 2-3 of the output XLR (Input -10dBV output -10dBV).
My AD/DA conveter can handle balanced and unbalanced signal.
Tomorrow i'm gonna make new cabels, xlr to stereo jack !

Many thx for all your help !

Danny.

 

[kante1603]

Thx.. MeToo2 and kante1603 !!!

My mistake , the cabels i use is a xlr to mono jack so it's connected balanced to unbalanced going to my AD/DA converter that's the reason for 6 db loss in signal !

Congrats-nice fix!

Have fun,

Udo.

 

[Suburban Ninja]

Sidechain Resistor: 100k. Changed this to what was originally specified for the That 2150 GSSL circuit.
Control Voltage resistor: 1k plus 240 trimmer in series
2:1    5.3dB (should be 5dB)
4:1    2.3dB (should be 2.5dB)
10:1  0.9dB (should be 1 dB)

Jackpot!  Thats close enough for me!

Hi
I have a doubt about the placement of the trimmer.
i did a quick draw if someone can tell me more.
Thanks you

 

[Suburban Ninja]

 

[Suburban Ninja]

or