Compressor, substituting VTL5C10 for NSL 32. what to modify?

[Morocotopo]

Hi people, searching for a nice compressor (I´ve already built the Ross and the DOD ones, nice but not what I wanted) I prototoyped the one found in a link in the first post of this thread:

http://www.diystompboxes.com/smfforum/index.php?topic=75884.msg667636

But I don´t have the specified opto, the VTL 5C10.

Datasheet:

http://www.datasheetarchive.com/VTL5C10-datasheet.html

So I used a NSL32.

Datasheet:

http://pdf1.alldatasheet.com/datasheet-pdf/view/205826/SILONEX/NSL-32.html

The thing works, and has the compression characteristics that I like, but there´s one problem: It doesn´t get, with the comp pot at max, a whole lot of compression. I´d like it to have more compression at the top of the pot, now with the pot at max it gets a subtle compression, wich is nice but I´d like the option for having more if I need it.

Sooo, the question is, does anyone knows what part of the schem to modify to get more compression? I´ve tried some mods, but not knowing exactly how the circuit works, It´s a blind search really, I don´t think that´s a very good way of working...

If anyone can help me I´d really appreciate it.

Thanks!

[Jarno]

The VTL5C10 is a bit special, so I don't think an NSL32 is totally the same. Especially the Roff of te C10 is quite low, I am currently using two parallel LDR's and a blue LED in my compulator.
It still has some issues though, so I went ahead and ordered the C10 part, probably need to do some other work as well, there is a bit of distortion (trimpot needs adjustment probably).

[Mark Hammer]

I can't see the schematic here at work, but the problem may not be the optoisolator itself.  Rather, it may simply be the amplitude of the signal you are feeding the circuit.

The usual strategy is to either increase the gain of the circuit segment which is driving the LED half of the optoisolator, or alter the manner in which the LDR half affects the gain of the stage it is part of, or both.

The other thing to consider is that the Vactrol appears to have a faster onset (drop to min resistance) and longer decay time (return to max resistance) than the Silonex unit does.

[Morocotopo]

Jarno, Mark, thanks for responding.

Jarno: yes, from what I´ve read in several places the VTL5C10 does have some special characteristics that other optocouplers don´t have, I might eventually order one of those, but I´m trying to make what I have work, although I know that most certainly it won´t work exactly as the original design.

Mark: Yes, that´s my line of thinking, that modifying some part of the circuit I can make it work to my satisfaction with a different opto. The only problem is that I don´t know where to mod... but your suggestins are good, very logical. I´ll review my opamp configurations theory to see if I can make sense of it.

Thanks!

[Eb7+9]

replace the highlighted gain resistor by a 100k trimmer - adjust to taste
this will allow to balance the circuit response against cell sensitivity and won't alter the amplitude limiter action before it

[gigimarga]

I've just finished one using VTL5C10 and I have the same question...so thank you!

On the other hand, is it possible to wire an attack pot?

[Eb7+9]

hmm .. not really since the cell diode is being hit by AC

[Morocotopo]

Eb, thanks a lot, extremely helpful. Right now I´m off to work but I´ll try it as soon as possible, and report how it worked. Then I´ll try to figure how the mod works, ´cause it´s nice to learn the why together with the how.

[gigimarga]

hmm .. not really since the cell diode is being hit by AC

What means AC?

[Eb7+9]

sorry ... what I meant was the LED is being hit by un-rectified AC (alternating current) signal

if you want to slow the response down you can always stick a largish capacitor across the LED
but with a 619 ohm resistor between the output of the op-amp and the LED it will take
a very large cap before you notice any effect - I'd try 100uF ... don't forget to tie the (-) of the cap to Vref

by doing this you'll also slow down the release ... can't avoid that here

if that doesn't slow things enough try increasing the 619ohm to 2k2 or something to help bring the RC time constant
up to human time frame ...

[petemoore]

  Is the NSL-32 achieving the same extreme R value as the original Vactrol ?
  Push signal into the input, and read the R value of the 32's LDR side.
  See if someone with the 'right' Vactrol will do the same so you'll know the target Value for what makes the compressor 'squish' to normal level.
  Compare data sheets to see how they compare as far as what it takes to 'fully' light [without toasting] the LED side.
  More light I would guess = a drop in gain.
  I don't want to get you to blow your NSL set, but maybe wonder if the R under lighted LED conditions is getting to a low enough value.

[gigimarga]

sorry ... what I meant was the LED is being hit by un-rectified AC (alternating current) signal

if you want to slow the response down you can always stick a largish capacitor across the LED
but with a 619 ohm resistor between the output of the op-amp and the LED it will take
a very large cap before you notice any effect - I'd try 100uF ... don't forget to tie the (-) of the cap to Vref

by doing this you'll also slow down the release ... can't avoid that here

if that doesn't slow things enough try increasing the 619ohm to 2k2 or something to help bring the RC time constant
up to human time frame ...

Thx a lot Eb7+9!
I will try what you said next days.

[Eb7+9]

In thinking about this a little further I realize that the 619 ohm resistor at the output of the op-amp will set the time constant only before the LED turns on ... once the LED turns on its incremental resistance might drop lower than the 619, or whatever resistor you have at the output of the op-amp, and discharge the cap faster ... I'm not sure what it is for an LED but for a small-signal silicon diodes 150 ohms is customarily chosen in hand analysis as a guestimate/starting value  ... before the LED turns on the incremental resistance of the LED will very high and the 619 ohm path will dominate (and set the time constant)

that means the cap will discharge more quickly across the LED turned on than when it's being charged up by the output resistor and the LED hasn't turned on yet (ie., during the attack period) ... so, curiously - though in principle you can increase both the attack and release time by adding a large cap across the LED (might have to be super large in fact) you're likely gonna end up with a release time that's shorter than the attack time ... I'm not sure you're going to like having a shorter release but see what happens - I know from experience that this is a very limited method for adding retardation to an opto-coupler

[Morocotopo]

replace the highlighted gain resistor by a 100k trimmer - adjust to taste
this will allow to balance the circuit response against cell sensitivity and won't alter the amplitude limiter action before it

OK people, I tried this but it doesn´t allow me to increase the compression, as I move the trimmer there´s a "sweet spot" where it compresses, higher or lower it just varies the gain of the circuit and it stops compressing. I measured the trimmer´s "sweet spot" value, it´s around 7,7 K ohms, so pretty close to the original 10K R.
Now What?

Grrr...

The thing is that, when working, it´s a really really really nice sound... but I want more of it! (I never heard the original, BTW). But my guess is that this depends on a certain range of resistance from the opto, something not easily achievable with a different one. I guess it´s time to open my wallet and stop trying to make things do what they can´t (nice phrase, right?  )

[Eb7+9]

as I move the trimmer there´s a "sweet spot" where it compresses, higher or lower it just varies the gain of the circuit and it stops compressing.

sorry I mistook GND for Vref on the downside of 4k7 - that's not an amplitude limiter but a biasing circuit
with the added gain trimmer you're modulating the intentional Bias offset ... put the 10k gain resistor back

the RHS 10k resistor is drawing a DC current from the (-) input node by way of the double-diode drop and so setting the DC of the op-amp through the 10k gain resistor ... try the 100k trimmer instead on the RHS 10k, you're looking to establish an op-amp offset of around 1.5 to 1.6 volts above Vref to bias the NSL-32 on the edge of turn-on ... careful when you do this - power up the circuit with the trimmer set to max resistance and slowly turn it while measuring the resistance on the cell side

at this point the signal gain should suddenly start to drop as you lower the trimmer resistance - if I'm right this trimmer will also act as a rough Threshold control ... set the Comp control off before you do this just in case there's another offset voltage present at the input of the Comp pot

after Biasing the cell on the edge, as you sweep the Comp control follow the DC of this op-amp with a meter to see if another offset component is present at the input of the Comp control - it shouldn't move

[Morocotopo]

Eb, first of all, many thanks for your assistance!

Now, I tried your suggestion, it does, when moving the trimmer, at certain point, just between the point where the signal gain changes, act somewhat like a threshold control, but it also has a "sweet spot", below or above that it stops compressing. And the amount of compression doesn´t really change from what it had before, it´s about the same.
One thing I forgot to mention is that instead of the 619 ohms R, I have 560 ohm ones, but I tried it with 680 ones also, and there wasn´t much audible change.

Some sound samples, comp at max, first clean, then with the FX:

Direct into mixer via a buffer:
Comp single notes direct.mp3 - 0.55MB

Same, but with amp emulation soft in a Twin setting:
comp clean.mp3 - 1.15MB

Should I post voltages?

[Eb7+9]

a sweet spot is to be expected ...

first, it would be good to confirm at least that the cell resistance drops monotonically
past a certain point by sweeping the threshold control I mentioned above
please confirm voltage variation at the op-amp's output

just remember that LED current can climb real quick with only 500 ohms of resistance in series
the LED is easy to burn

follow the cell resistance with your meter while you do your sweep, don't go below 500 ohms
you need to know if the cell resistance can't drop past a certain "high" point for some reason

two things come to mind :

the data sheet of the VTL5C10 gives a typical LED operating voltage of 2.2 volts ...
while the turn-on voltage of the NSL-32 is around 1.6volts ... this is saying the VTL LED needs a higher bias voltage
is it possible that the stock circuit values provide an overly high bias voltage for the NSL-32 ??
and the NSL-32 LED is damaged ?

but, hand analysis says that with two 10k resistors in the circuit you should have an
idling voltage of 1.4volts (two diode drops) ...
can you confirm this voltage when two 10k resistors are present in the circuit ?!

unless my analysis is way off it would mean that both LED are under-Biased at idle ... the VTL one being more so
in principle that shouldn't burn the NSL-32 LED

gonna sim this out to make sure I'm not missing something

---

NSL-32's should typically be able to dip down to about 500 ohms on 2mA of LED current ...

10.5k to 0.5k is a 21:1 drop (or -26.5dB) ...

yeah, that should give plenty of compression

---

ok, try this out ... in the signal divider formed by the 10k resistor and grounded cell replace
the 10k resistor by a higher value pot, this will give you a lower output scaling factor even if your cell is stuck at
a certain minimum value - now you suddenly have a rough ratio control ... adjust your threshold control accordingly
expect the threshold control to be touchy