SSL Clone compressor ratio measurements

[digiteck]

Just to be sure is it the first and the third resistors, those that are not on the picture just on the side of the VCA ? So it should be like your picture ?

Yes

Fine !

 

[digiteck]

With the part size of a trimmer between VCAs I doubt you can put your VCAs back in like before. Fitting some Milmax sockets or alike for this pin diameter to rise the VCA above the pcb might do the trick. Sockets are not needed, if you place the trimmers from traceside, obviously with the drawback of easy accessibility.

Ok, thanks for the tips, i will try this.

Seems so. If you change a parts value that is part of a R/C circuit, this will change frequency response. Check if it matters. You do the math. f=1/(2+PI()*R*C). The 22uF caps value at the output of the debalancing stage should still fit. The change from 27K to 13K in front of the audio-VCA with 47K in parallel at the sidechain summing node rises this HPF from 0.42Hz to 0.71Hz. I wouldn't care, but YMMV.

Thanks for this lesson... What is the fonction of this HPF ? Maybe something like a DC filter removal ? Maybe not because normally a DC filter start at 1 Hz if i'm not wrong ?

You seem to have read its value backwards. Looks like a 820R, not 10K.

Oh yes i miss this one ! One error it's not too bad for a start

 

[Harpo]

... What is the fonction of this HPF ? Maybe something like a DC filter removal ? Maybe not because normally a DC filter start at 1 Hz if i'm not wrong ?

Blocking DC from previous stage.
For an ideal linear frequency path you want a range between DC and infinity without loss or artefacts. Unfortunately there are no perfect active or passive components to archieve this. As compromise to come closer with real world parts, a filters -3dB rolloff point is usually set at a point outside the audible range, so that introduced phase angle by this hpf or lpf is less than +5° or -5°. For an assumed human audible range of 20Hz..20kHz, YMMV, these -3dB hpf or lpf corners for a 1st.order filter with slope falling -6dB/oct or -20dB/dec at 5° max. are set below 1.75Hz or above 228kHz. Phase angle for hpf=-ARCTAN(hpf_Hz/testfrequency_Hz)*180/PI() with amplitude response at testfrequency in dB=LOG10(-ARCTAN(hpf_Hz/testfrequency_Hz))*20 ; phase angle for lpf=ARCTAN(testfrequency_Hz/lpf_Hz)*180/PI() with amplitude response at testfrequency in dB=LOG10(ARCTAN(testfrequency_Hz/lpf_Hz))*20. This goal still might be limited by FI parts size or parts bandwidth limits.

 

[digiteck]

So here we go... I've find some components (not everything) so i decided to proceed step by step to really understand what changes make to the sound and to the circuit.
(Finally it wasn't so much difficult to solder some components on a PCB !!! It was the first time for me...)

When i removed the PCB i can see that there was a second resistor in parrallel with the 820R (in place of the 1K around the VCA) so 820R & 470R in parrallel = 298 Ohms. The value is really low comparing to 1K ! If i'm correct my ratio problem was here.
So at the moment i have only replace the 1K (mine was 298R, so replace it with the 1K) and the VCA input resistor 27K (mine was 30K, so i replace it with 13K according to what Harpo said...)

Here is the result :

- when i put the compressor in my chain, the volume increased by +5.5 db (in bypass or without comp and gain at 0)
- The ratios are much better :
ratio 2 (increased input level +10db) = output +5,3
ratio 4 (increased input level +10db) = output +3,1
ratio 10 (increased input level +10db) = output +1,8

- I have the same volume in bypass or comp but the global volume has increased by 5,5db and bring some distortion when i'm near from the max input level if i'm lower the input, the distortion disappear or is much lower...

- I have found a small noise when i turn the threshold (mostly between 0 and -15) maybe i have this before, it's not important for me now but if you have an idea it's great ! (the pots are Alpha brand or type)

- All the test have been done with a 1KHz sine wave.

So for the Threshold and ratio, it seem that i'm in a good way, i will try to find a 200K trimpot and tweak this and the 1K to find the best value. Am i correct ?

But now my main problem is the distortion and the increased volume. I understand i didn't put all the components like you said so it's probably for this reason but like i said i want to understand the circuit so i procced step by step to see what it does...

My question is : is it normal i have the increased volume on my unit because i only change the input resistors and not the output, so willl it be ok when i'll change the 15K (by 6K5) and capacitors 100pF (by 150pF) ?

What kind of AC or DC voltage should i measure at the output of the compressor and between which PIN ?

Concerning the distortion it seem to be due to the incresaed level because when i lower the input the distortion decrease and dissapear... What changes will make the 50K dist trimmer, just let us calibrate the distortion regarding to the specifications we want and the VCA installed ?

Thanks,

Lionel

 

[Harpo]

..So at the moment i have only replace the 1K (mine was 298R, so replace it with the 1K) and the VCA input resistor 27K (mine was 30K, so i replace it with 13K according to what Harpo said...)

Here is the result :

- when i put the compressor in my chain, the volume increased by +5.5 db (in bypass or without comp and gain at 0)

Lionel, the suggested 13K in front of the VCA from the THAT app note and half this value =6K5 as feedback resistor from the VCA following inverting opamp for current to voltage conversion will have a gain of -0.5 or -6dB with VCA in its bypass condition =control voltage 0V. (short version -AV=Rfb/Rin, for more acurate figures, factor in the losses in filters for your response range of interest, described in previous post). The 2nd stage of this NE5532 is an inverting stage with gain of -1, so measuring across these outputs will double the level =gain of 2 or +6dB.
With balanced output and measuring between XLR-pin2/3, there sould be overall unity gain (debalancing stage +/-0dB + VCA+following opamp -6dB + inverting 2nd.output +6dB = overall 0dB).
With unbalanced output (measure between XLR-pin2/1, leave XLR-pin3 floating, IE don't tie pin3 to gnd), level would be down by -6dB in this config, so for unbalanced operation you make the prementioned feedback resistor the same value =13K as the resistor in front of the VCA. You could make it switchable bal/unbal by paralleling a 2nd 13K.

- The ratios are much better :
ratio 2 (increased input level +10db) = output +5,3
ratio 4 (increased input level +10db) = output +3,1
ratio 10 (increased input level +10db) = output +1,8
..
So for the Threshold and ratio, it seem that i'm in a good way, i will try to find a 200K trimpot and tweak this and the 1K to find the best value. Am i correct ?

yes. Check the trimmers initial settings. You don't want to shunt the feeding stage to (virtual) ground.

But now my main problem is the distortion and the increased volume. I understand i didn't put all the components like you said so it's probably for this reason but like i said i want to understand the circuit so i procced step by step to see what it does...

My question is : is it normal i have the increased volume on my unit because i only change the input resistors and not the output, so willl it be ok when i'll change the 15K (by 6K5) and capacitors 100pF (by 150pF) ?

this should fix it (see above).

What kind of AC or DC voltage should i measure at the output of the compressor and between which PIN ?

For balanced out between XLR pin2/3, same level as input. IE 1V in = 1V out with no compression or GSSL in bypass mode.
For unbalanced out between XLR pin2/1, same level as input. (see above)

Concerning the distortion it seem to be due to the incresaed level because when i lower the input the distortion decrease and dissapear... What changes will make the 50K dist trimmer, just let us calibrate the distortion regarding to the specifications we want and the VCA installed ?

I think so.

 

[digiteck]

I made a mistake the last time i made the measurements, the increased level is +7.3 dB (not +5.5) and with the volt-meter the input 1.6 V and output 3.8 V.

Lionel, the suggested 13K in front of the VCA from the THAT app note and half this value =6K5 as feedback resistor from the VCA following inverting opamp for current to voltage conversion will have a gain of -0.5 or -6dB with VCA in its bypass condition =control voltage 0V. (short version -AV=Rfb/Rin, for more acurate figures, factor in the losses in filters for your response range of interest, described in previous post). The 2nd stage of this NE5532 is an inverting stage with gain of -1, so measuring across these outputs will double the level =gain of 2 or +6dB.
With balanced output and measuring between XLR-pin2/3, there sould be overall unity gain (debalancing stage +/-0dB + VCA+following opamp -6dB + inverting 2nd.output +6dB = overall 0dB).
With unbalanced output (measure between XLR-pin2/1, leave XLR-pin3 floating, IE don't tie pin3 to gnd), level would be down by -6dB in this config, so for unbalanced operation you make the prementioned feedback resistor the same value =13K as the resistor in front of the VCA. You could make it switchable bal/unbal by paralleling a 2nd 13K.

What i don't understand at the moment is :

-how do you translate the That DN127 doc to find the 13K value and other changes ? I can read the Rin = 13K and Rout = 13K (not 6K5) for the 2001 but if i read the 202C (those use in the SSL clone schematic) they said Rin = 50K and Rout = 50K, so the value shoud be Rin=50K instead of 27K and Rout=50K (or maybe 25K, see the next question below) instead of 15K. What is the deal ?
- the feedback resistor value should be half the value of the input resistor, is it part of the original design or just for my VCA specifications ? So regarding the original design the 27K should be 30K (and half this value for the feedback resistor = 15K like in the PCB and schematics).
- My previous input 30K resistor with 15K at the feedback resistor should also work (half the input resistor value)? What is the difference with 13K and 6K5 ? Maybe that fit better the spec of my DBX2001 ? But what it really change ?

I understand that if i divided a resistor by 2 so i've got an -6dB and if i double a resistor i wil have +6dB...

I understand that if i make some feedback resistor changes i will have to do some slight modifications like capcitors or other regarding the RC circuit.

One more time, thanks for your time !

Lionel

 

[Harpo]

I made a mistake the last time i made the measurements, the increased level is +7.3 dB (not +5.5) and with the volt-meter the input 1.6 V and output 3.8 V.

I was already wondering what other resistors might have been placed from traceside to get your +5.5dB readout.

What i don't understand at the moment is :

-how do you translate the That DN127 doc to find the 13K value and other changes ? I can read the Rin = 13K and Rout = 13K (not 6K5) for the 2001 but if i read the 202C (those use in the SSL clone schematic) they said Rin = 50K and Rout = 50K, so the value shoud be Rin=50K instead of 27K and Rout=50K (or maybe 25K, see the next question below) instead of 15K. What is the deal ?

Same value Rin and Rout are for VCA/opamp unity gain behaviour with control voltage 0V. With this balanced out stage you don't want unity gain but -6dB response for control voltage 0V.
Rin is needed for voltage to current conversion. For best noise performance you want this value as small as possible for expected range of signal voltage within the VCAs limits. Differences in the Vbe multipliers inside the VCAs or internally paralleled VCAs call for different values in this respect. A DBX2001 seems to have a typical signal current Iin+Iout of 1.15mA@15V supply for the recommended 13K Rin. This 13K leaves enough headroom for safe operation, as the driving NE5534 isn't a rail to rail opamp, IE your Rin probably won't see more than about 13Vpk. There are other DBX papers, showing a little more data for the DBX2001 than THAT DN127.

-the feedback resistor value should be half the value of the input resistor, is it part of the original design or just for my VCA specifications ? So regarding the original design the 27K should be 30K (and half this value for the feedback resistor = 15K like in the PCB and schematics).

Rfb half value Rin is part of Jakobs design, based on ssl_82e26/ssl_82e27, with added inverting -not cross-coupled- output stage for balanced connection. The original ssl_82e26 as part of a console had 68K Rin and Rout for DBX202C VCAs in these spots.

- My previous input 30K resistor with 15K at the feedback resistor should also work (half the input resistor value)? What is the difference with 13K and 6K5 ? Maybe that fit better the spec of my DBX2001 ? But what it really change ?

Allowing more signal current for the transistors/Vbe multiplier, decreasing resistive noise, maybe increasing THD.

I understand that if i divided a resistor by 2 so i've got an -6dB and if i double a resistor i wil have +6dB...

yes for an inverting opamp gainstage and within opamps limits, where 'a resistor' is the feedback resistor. If 'a resistor' is the input resistor, its just the other way round. Calculating a non-inverting gainstage will be a little different.

I understand that if i make some feedback resistor changes i will have to do some slight modifications like capcitors or other regarding the RC circuit.

yes, maybe. Just do the math to check if its required or will affect the circuits response to a noticable degree.

 

[digiteck]

Thanks Harpo, now it's much clearer for me ! I really learn some interestings things and understand the design better now with your help.

the feedback resistor value should be half the value of the input resistor, is it part of the original design or just for my VCA specifications ? So regarding the original design the 27K should be 30K (and half this value for the feedback resistor = 15K like in the PCB and schematics).

Rfb half value Rin is part of Jakobs design, based on ssl_82e26/ssl_82e27, with added inverting -not cross-coupled- output stage for balanced connection. The original ssl_82e26 as part of a console had 68K Rin and Rout for DBX202C VCAs in these spots.

But there is a last question that you didn't answer, and it's the most disturbing for me. In the schematic or the PCB (so for the 202C VCA) the input resistor is 27K and feedback resistor is 15K, if the feedback resistor should be half the input resistor, so it should be 30K for input and 15K for output or 27K for input and 13,5K for output ? Why it's not exactly half the value on the PCB and schematic ?

Lionel

 

[digiteck]

I made a mistake the last time i made the measurements, the increased level is +7.3 dB (not +5.5) and with the volt-meter the input 1.6 V and output 3.8 V.

I was already wondering what other resistors might have been placed from traceside to get your +5.5dB readout.

I think the resistor was me !!!  ;D

 

[digiteck]

the feedback resistor value should be half the value of the input resistor, is it part of the original design or just for my VCA specifications ? So regarding the original design the 27K should be 30K (and half this value for the feedback resistor = 15K like in the PCB and schematics).

Rfb half value Rin is part of Jakobs design, based on ssl_82e26/ssl_82e27, with added inverting -not cross-coupled- output stage for balanced connection. The original ssl_82e26 as part of a console had 68K Rin and Rout for DBX202C VCAs in these spots.

But there is a last question that you didn't answer, and it's the most disturbing for me. In the schematic or the PCB (so for the 202C VCA) the input resistor is 27K and feedback resistor is 15K, if the feedback resistor should be half the input resistor, so it should be 30K for input and 15K for output or 27K for input and 13,5K for output ? Why it's not exactly half the value on the PCB and schematic ?

Lionel

Someone can answer to the question above ? I really need to understand. It would mean that the 202C have an internal resistor of 3K but according to the THAT dn127 it's not the case, there is the 202X and 202XT that have an internal resistor of 1.25K but not 3K... :'(

Lionel

 

[glbaudio]

You need a distortion analyzer... measures THD. I don't have one... wish I did.

Do u know is there an amazing FREE software audio analyzer WITH a THD meter?
It is called Room EQ Wizard, it run on Win, Linux and OsX!!!

here my 50 cents...

btw just finished a Gssl with 2 DBX 202 and a 2151, and ALL the ratios are over...i've putted a 50k trimmer in series to the 100K resistor, but still not enough...it's time to increase the two 1k resistors...

 

[Harpo]

btw just finished a Gssl with 2 DBX 202 and a 2151, and ALL the ratios are over...i've putted a 50k trimmer in series to the 100K resistor, but still not enough...it's time to increase the two 1k resistors...

BS. Just compare the control voltage law from your DBX202 (-6dB/mV) with a DBX202C (-50mV/dB) and you'll know why your ratios are way off. You better adjust Rin and Rfb as well for these VCAs.

 

[glbaudio]

btw just finished a Gssl with 2 DBX 202 and a 2151, and ALL the ratios are over...i've putted a 50k trimmer in series to the 100K resistor, but still not enough...it's time to increase the two 1k resistors...

BS. Just compare the control voltage law from your DBX202 (-6dB/mV) with a DBX202C (-50mV/dB) and you'll know why your ratios are way off. You better adjust Rin and Rfb as well for these VCAs.

Amazing! This is the problem! You mean the 27k on VCA input and the 15k between -In and the Out of the 5532...any suggestion 4 new values?

 

[abbey road d enfer]

You need a distortion analyzer... measures THD. I don't have one... wish I did.

Do u know is there an amazing FREE software audio analyzer WITH a THD meter?
It is called Room EQ Wizard, it run on Win, Linux and OsX!!!

If I'm not mistaken, there's an FFT analyser, not a real THD measurement. Should not make any difference considering the particular goal, but takes some post-processing for providing a percentile figure.

 

[Harpo]

For the -50mV/dB to -6mV/dB conversion for your DBX-202 black can VCA (the GSSL schematic is for a DBX-202C)you'd need to scale the control voltage down by factor 0.12. Could be done by the substitution circuit on pcb below your audio VCAs as shown in Jakobs schematic for the THAT218x, if you can make it fit, by additional voltage divider 1k1/100R with maybe decreased performance or by decreasing the 100k Rfb between TL072 pins6/7 to 12k. This should fix your way off ratio issues.

For your DBX-202, THAT paper DN127 shows a 100k for Rin. Rfb in GSSL-balanced-out config would be half this value=50k or 100k for unbalanced-out config. Scale compensation caps down for same response.
(The debalancing NE5534 in front doesn't swing rail to rail but typ.+/-2V off its supply rails. Your DBX-202 has a 500uA Iin+Iout limit= max.250uA Iin in bypass condition, so min.value for Rin would be 52K).

 

[glbaudio]

For the -50mV/dB to -6mV/dB conversion for your DBX-202 black can VCA (the GSSL schematic is for a DBX-202C)you'd need to scale the control voltage down by factor 0.12. Could be done by the substitution circuit on pcb below your audio VCAs as shown in Jakobs schematic for the THAT218x, if you can make it fit, by additional voltage divider 1k1/100R with maybe decreased performance or by decreasing the 100k Rfb between TL072 pins6/7 to 12k. This should fix your way off ratio issues.

For your DBX-202, THAT paper DN127 shows a 100k for Rin. Rfb in GSSL-balanced-out config would be half this value=50k or 100k for unbalanced-out config. Scale compensation caps down for same response.
(The debalancing NE5534 in front doesn't swing rail to rail but typ.+/-2V off its supply rails. Your DBX-202 has a 500uA Iin+Iout limit= max.250uA Iin in bypass condition, so min.value for Rin would be 52K).

Yes tnx Harpo, i've just changed the two 1k resistors from the TL072 to the CV of the two channels VCAs with two 3K3 resistors (calculated by Ohm Law considering the difference about internal resistance of each version of VCA) to obtain a right voltage from the CV circuit...
Changed also the Rin and Rfb to 100k and the relative compensations caps (100pF to 10pF) on both channels...
now it seems to work correctly, BUT it seems i have a worst THD on both channels...i have to do a listening, probably is more than good anyway ;-)

 

[Harpo]

i've just changed the two 1k resistors from the TL072 to the CV of the two channels VCAs with two 3K3 resistors (calculated by Ohm Law considering the difference about internal resistance of each version of VCA) to obtain a right voltage from the CV circuit...

Lots of ways to shave a cat, but you seem to have a funny calculator, giving you a 3k3 substitution value. Might fit for a VCA with a -18mV/dB control law, whoever builds such thing. Sure this still will distort for your VCA. My calc for your approach comes up with a (maybe too noisy) 10k5 instead. I'd probably only decrease the 100k Rfb at TL072 to 12k, but YMMV.

 

[glbaudio]

Lots of ways to shave a cat, but you seem to have a funny calculator, giving you a 3k3 substitution value. Might fit for a VCA with a -18mV/dB control law, whoever builds such thing. Sure this still will distort for your VCA. My calc for your approach comes up with a (maybe too noisy) 10k5 instead. I'd probably only decrease the 100k Rfb at TL072 to 12k, but YMMV.

how u have calculated that value?
i've done my calculation based on the Igor indication...and seems to be correct!
http://www.groupdiy.com/index.php?topic=39459.msg501251#msg501251

 

[Harpo]

Lots of ways to shave a cat, but you seem to have a funny calculator, giving you a 3k3 substitution value. Might fit for a VCA with a -18mV/dB control law, whoever builds such thing. Sure this still will distort for your VCA. My calc for your approach comes up with a (maybe too noisy) 10k5 instead. I'd probably only decrease the 100k Rfb at TL072 to 12k, but YMMV.

how u have calculated that value?
i've done my calculation based on the Igor indication...and seems to be correct!
http://www.groupdiy.com/index.php?topic=39459.msg501251#msg501251

Dunno if my math is any better than Igors, but your quote 'BUT it seems i have a worst THD on both channels' implies, Igors 3k3 value for Rx might not be the correct substitution value, coming from a DBX-202C, at least my 10k5 is wrong as well.
Splitting hairs, Igors 197R inside the black can VCA actually is a 195R7, a voltage divider consisting of Rseries 100R and Rshunt 100R in parallel with 22R+2K2=95R7. From GSSL schematic the resistor between TL072 pin7 and DBX-202C control port is 100R+1K. For examples sake assume 1V from TL072, voltage at the 50mV/dB VCA control port with Rin 650R is at 371mV, resulting in a Log10(0.371/0.05)*20=17.4dB gain reduction. You want this same reduction for your DBX202 with a 6mV/dB control law = 6mV/50mV scale down factor 0.12 for 44.6mV at the VCAs control port.
Total resistance is 1/(0.0446/195.7)=4k391. Rx to substitute the 1k is (4k391 - 195R7 inside the VCA - 100R at opamp output) = 4k095.

 

[glbaudio]

Lots of ways to shave a cat, but you seem to have a funny calculator, giving you a 3k3 substitution value. Might fit for a VCA with a -18mV/dB control law, whoever builds such thing. Sure this still will distort for your VCA. My calc for your approach comes up with a (maybe too noisy) 10k5 instead. I'd probably only decrease the 100k Rfb at TL072 to 12k, but YMMV.

how u have calculated that value?
i've done my calculation based on the Igor indication...and seems to be correct!
http://www.groupdiy.com/index.php?topic=39459.msg501251#msg501251

Dunno if my math is any better than Igors, but your quote 'BUT it seems i have a worst THD on both channels' implies, Igors 3k3 value for Rx might not be the correct substitution value, coming from a DBX-202C, at least my 10k5 is wrong as well.
Splitting hairs, Igors 197R inside the black can VCA actually is a 195R7, a voltage divider consisting of Rseries 100R and Rshunt 100R in parallel with 22R+2K2=95R7. From GSSL schematic the resistor between TL072 pin7 and DBX-202C control port is 100R+1K. For examples sake assume 1V from TL072, voltage at the 50mV/dB VCA control port with Rin 650R is at 371mV, resulting in a Log10(0.371/0.05)*20=17.4dB gain reduction. You want this same reduction for your DBX202 with a 6mV/dB control law = 6mV/50mV scale down factor 0.12 for 44.6mV at the VCAs control port.
Total resistance is 1/(0.0446/195.7)=4k391. Rx to substitute the 1k is (4k391 - 195R7 inside the VCA - 100R at opamp output) = 4k095.

tnx a lot Harpo for your great explanation, this help me a lot to learn electronic...do you still think is better substitute the Rfb at the TL72 (instead the Rx substitution)? i think i have to change the relative capacitor too...right?