Need help with Engineer's Thumb compressor build

[Rob Strand]

While fiddling with things I have witnessed some distortion on very small signals, and shorting the LED fixed it. Possibly it was because, when not compressing, little or no current flows in the LED and therefore its dynamic resistance was high, meaning the Vref supply was higher-impedance than normal, which would introduce some unwanted feedback between the various Vref connections. However, I was also doing other things at the same time, so it may have been a red herring. I only mention it here to simplify rankot's troublshooting situation.

Good to know.  So short the LED ASAP!

[rankot]

Shorting LED didn't change a thing. I will now try to drive BC327 base, and finally LM13700 adjust pin directly, to see if it changes anything.

EDIT: Driving BC327 base works the same. I unshorted LED, and what I can also see is that LED is glowing brightly when distortion occurs.

EDIT 2: Driving adjust bias pins directly (through 220 Ohm resistors) makes distortion on only one side of the pot, but I can measure voltage in the range from 1.5 to 3.6 V on the pot - so maybe my LM13700 is faulty, or it simply can't work in parallel? I will try to disconnect one trans opamp.

EDIT 3: Distorting even with the one trans op amp disconnected...

EDIT 4: Now, I connected everything again and those are my voltages, with all the pots at maximum and no input signal, compared to ltspice simulation values (separated by dash):

NE5532 (A&D)
1: 6.02 / 5.86 (probably different Vf on LED)
2: 4.33 / 4.34
3: 4.32 / 4.34
4: 0
5: 3.69 / 3.78
6: 4.11 / 3.78 <--
7: 4.17 / 4.09
8: 8.69

TL072 (B&C)
1: 4.25 / 3.78 <--
2: 4.10 / 3.78 <--
3: 3.69 / 3.78
4: 0
5: 3.85 / 3.78
6: 4.07 / 3.78
7: 3.48 / 3.21
8: 8.69

[merlinb]

I will now try to drive BC327 base, and finally LM13700 adjust pin directly, to see if it changes anything.

Why? We already know that part works fine.

NE5532 (A&D)

It helps if you use the right opamp. Use a TL072.

[Rob Strand]

From you tests there's two zone of distortion small signals (Vcontrol = 4.3V) and large signals (Vcontrol < 0.78).    They maybe be from different causes.

Can I suggest a few things to try:

1) Try replacing the NE5532 with a TL072.     (As Merlin already suggested.)
     The NE5532's have input diodes which can do funny things when the opamp clips.

2)  Increase R12  to say 1.2k, or better for the experiment use 1.5k.
      The current going down to the OTA is pretty high maybe that's enough to shave some current off.
      At 0.78V on opamp C,  (4.5 - 0.78) / 1k = 3.72mA,  spits into two two each OTA, so IABC = 3.72/2 = 1.86mA each.
      If you got to 1.5k then that drops to about 1.86/1.5 = 1.2mA.   The datasheet only shows upto 1mA.

Through 220 Ohm resistors) makes distortion on only one side of the pot, but I can measure voltage in the range from 1.5 to 3.6 V on the pot - so maybe my LM13700 is faulty, or it simply can't work in parallel? I will try to disconnect one trans opamp.

The way to check the OTAs are sharing current it is set the voltage on opamp C to say 2V then measure the voltage across each of the 220R resistors they should be pretty much equal voltages.

I'm not sure what's going on with your spice sim.  The NE5532 has a high input bias current and that can shift the DC voltages, especially with 1M input resistors.

[Eb7+9]

... or it simply can't work in parallel? I will try to disconnect one trans opamp.

of course OTAs can run in parallel, it's just not a good idea to do that here as it cuts your chances of success significantly ... you guys still not looking in the right place here

other main question, what size input signal are you using ??

there are two things that can screw up the performance of this circuit ... as a result, some get lucky and some won't ... you want to eliminate the luck part of the equation so DIY'ers can build with 100% success ...

it's simple, as I said in the main thread go look at the AGC application diagram in the NE570/571 datasheet ... basically the same topology

[rankot]

1) Try replacing the NE5532 with a TL072.     (As Merlin already suggested.)
     The NE5532's have input diodes which can do funny things when the opamp clips.

I will, if I manage to desolder it, since this is a dual layer board with riveted holes and it is very hard to do it.

2)  Increase R12  to say 1.2k, or better for the experiment use 1.5k.
      The current going down to the OTA is pretty high maybe that's enough to shave some current off.
      At 0.78V on opamp C,  (4.5 - 0.78) / 1k = 3.72mA,  spits into two two each OTA, so IABC = 3.72/2 = 1.86mA each.
      If you got to 1.5k then that drops to about 1.86/1.5 = 1.2mA.   The datasheet only shows upto 1mA.

I have tried this, used 2k resistor for R12, thus limiting OTA current to less than 1mA, but it didn't improve a thing...

EDIT: No way I can desolder this NE5532... I will try to take it to the guy who has a SMD desoldering station at Monday.

[merlinb]

EDIT: No way I can desolder this NE5532... I will try to take it to the guy who has a SMD desoldering station at Monday.

Generally you snip through the legs to remove the body, then remove the remains of the pins with tweezers and a solder sucker.
It's a big mistake to solder things in permanently when making unknown mods to a circuit, you can't just swap ICs willy nilly and expect it to work. Next time use IC sockets or just tack things in temporarily with scraps of wire

[merlinb]

From you tests there's two zone of distortion small signals (Vcontrol = 4.3V) and large signals (Vcontrol < 0.78).    They maybe be from different causes.

No that's how the circuit is supposed to work. Those voltages are beyond normal operation, stop obsessing over them.

2)  Increase R12  to say 1.2k, or better for the experiment use 1.5k.

The circuit works with the component values in the schematic, changing them isn't going to fix anything, it's more likely to create new problems.
I believe this is simply the result of using an inappropriate opamp; the NE5532.

[rankot]

From you tests there's two zone of distortion small signals (Vcontrol = 4.3V) and large signals (Vcontrol < 0.78).    They maybe be from different causes.

No that's how the circuit is supposed to work. Those voltages are beyond normal operation, stop obsessing over them.

2)  Increase R12  to say 1.2k, or better for the experiment use 1.5k.

The circuit works with the component values in the schematic, changing them isn't going to fix anything, it's more likely to create new problems.
I believe this is simply the result of using an inappropriate opamp; the NE5532.

I believe you're right. I wanted to further reduce noise, although not completely understanding differences between NE5532 and TL072. ETC which I made first, and the only one working as expected, is using TL074. The second one used NE5534, and this one is using NE5532, both of them experiencing distortion and almost not compressing at all. I will try to remove NE5534 in the second one, it has a home made single layer PCB and it's quite simple to desolder.

[Rob Strand]

From you tests there's two zone of distortion small signals (Vcontrol = 4.3V) and large signals (Vcontrol < 0.78).    They maybe be from different causes.

No that's how the circuit is supposed to work. Those voltages are beyond normal operation, stop obsessing over them.

Come on, even you stated you have seen problems at low control currents  .    That was the idea behind shorting the LED.

    2)  Increase R12  to say 1.2k, or better for the experiment use 1.5k.

The circuit works with the component values in the schematic, changing them isn't going to fix anything, it's more likely to create new problems.

When you go past what's written in the datasheet you can only say the circuit *usually* works.   Its only *after* rankot did the test we *know* it's not that.

I believe this is simply the result of using an inappropriate opamp; the NE5532.

BS aside, it's highly likely - and even more so *after* rankot's tests.

[rankot]

OK, I have just tried with TL071 instead of NE5534 in my second build and it sounds the same. Then I tried to remove the JFET input buffer, proposed by Jonny Reckless in some of the posts of the main thread, because that was the last thing not the same comparing builds one and two. And it seems that skipping the buffer did the trick - it works as expected! I just notice some faint rattling on the very end of the note decay when Threshold pot is at minimum - pin 3 of B is at 4.23V. When turned to max, pin 3 gets at 3.83V and no rattling at all. Is that normal, or I have to look for something else? Maybe cause I used 100n instead of 10n for C3?

I will try to put NE5534 back and see what happens now. Maybe the input buffer was the only problem!

One more question for Merlin: if I use 2u2 for C4 and C5, shall it make this sound better with bass, or not?

[rankot]

Definitely NE5534 behaves different - it does the compression, but it seems that it's not sustaining here??? Maybe I'm just too tired of different tries, but seems to be like that, so I will leave TL071. In case of my third build, where I used dual op amp in this position (one half is maintaining V_ref like in ET issue 4), I will try to desolder NE5532 at the service, and then to put a socket, so I can try LM4562 there, for example.

[PRR]

....remove the JFET input buffer....

I do not remember you mentioning a JFET buffer in this thread??

In detective stories, it is very rude to introduce new suspects on the last page.

[rankot]

I thought I have already posted schematic, but maybe it was in another thread? I'm little bit confused those days. 

[Rob Strand]

I do not remember you mentioning a JFET buffer in this thread??

In detective stories, it is very rude to introduce new suspects on the last page.

I thought I have already posted schematic, but maybe it was in another thread? I'm little bit confused those days. 

In reply #709 you posted Issue 4 (no JFET) , which is what I thought we were working with so I tacked it onto this thread in reply #33,

https://www.diystompboxes.com/smfforum/index.php?topic=97123.700

[Eb7+9]

Definitely NE5534 behaves different - it does the compression, but it seems that it's not sustaining here???

if you get a chance, with the 5534 op-amp in there, measure the voltage difference between the output of the op-amp and the ~4.5v mid-rail reference (pin 4 or 13 on the ota's) ... how many mV's do you see there, and does it vary much while varying the RATIO (1M) resistor ?!

[rankot]

Definitely NE5534 behaves different - it does the compression, but it seems that it's not sustaining here???

if you get a chance, with the 5534 op-amp in there, measure the voltage difference between the output of the op-amp and the ~4.5v mid-rail reference (pin 4 or 13 on the ota's) ... how many mV's do you see there, and does it vary much while varying the RATIO (1M) resistor ?!

Measurements (please note that I don't have that 4k7 resistor on wiper), plus probe is at op amp output pin 6 and minus probe at 4.5V Vref:

With NE5534: -434mV with Ratio at minimum (0 Ohm) and -195mV with Ratio at max (1M).
With TL071: 0mV with Ratio at minimum (0 Ohm) and -6mV with Ratio at max (1M).

Quite a difference.

[rankot]

So, I have finally fixed my second build, and settled it with TL071 and without the input buffer. I put it back into the enclosure, but it suddenly lost a lots of it's volume, so I have to debug it again and I really have no nerve for that, already loosing two days on it, so it will sit on a shelf until I get more time.

Since I have a third build too, but it is made on a professional PCB and I can't desolder NE5532 from it, I tried to do what I can and see what happens - and I simply removed a buffer, cause it was easy to desolder and - voila - everything started to work! No distortion, compression works fine etc, even with NE5532. Even compression indicator diode works well. I have also measured voltage difference, with 4k7 resistor on wiper - plus probe at op amp output pin 6 and minus probe at 4.5V Vref:

With NE5532: -300mV with Ratio at minimum (0 Ohm) and -195mV with Ratio at max (1M).

So it is similar to NE5534 (of course), but it is working?

Furthermore, I replaced R18 (220 Ohm, ET issue 4 original schematic for reference) with a 500 ohm pot in series with 47 Ohm resistor (that is the limit before it starts distorting) and I have a nice Gain adjustment - handy indeed.

[merlinb]

Its only *after* rankot did the test we *know* it's not that.

Yes. So why were you still talking about it after we know it's not that?

Until Ranknot tells us what circuit he thinks he has built, this thread is just the blind leading the blind.

[rankot]

Until Ranknot tells us what circuit he thinks he has built, this thread is just the blind leading the blind.

Most of my posts are about issue 3 - the second ETC I've ever built.

But this post is about issue 4 (the third one I built):

So, I have finally fixed my second build, and settled it with TL071 and without the input buffer. I put it back into the enclosure, but it suddenly lost a lots of it's volume, so I have to debug it again and I really have no nerve for that, already loosing two days on it, so it will sit on a shelf until I get more time.

Since I have a third build too, but it is made on a professional PCB and I can't desolder NE5532 from it, I tried to do what I can and see what happens - and I simply removed a buffer, cause it was easy to desolder and - voila - everything started to work! No distortion, compression works fine etc, even with NE5532. Even compression indicator diode works well. I have also measured voltage difference, with 4k7 resistor on wiper - plus probe at op amp output pin 6 and minus probe at 4.5V Vref:

With NE5532: -300mV with Ratio at minimum (0 Ohm) and -195mV with Ratio at max (1M).

So it is similar to NE5534 (of course), but it is working?

Furthermore, I replaced R18 (220 Ohm, ET issue 4 original schematic for reference) with a 500 ohm pot in series with 47 Ohm resistor (that is the limit before it starts distorting) and I have a nice Gain adjustment - handy indeed.

Sorry for making such a confusion, but I've spent most of the past two days trying to fix those pedals.