illogical transistor behaviour at voltage controlled amplifier and clipping

[Fluffyboii]

I have build two identical versions of the circuit above on the same board. First VCA works and I can adjust it for both logarithmic and linear CV response to have no output at 0V CV. The second is working at linear response yet it can't shut down the input signal at log response. What I mean by it can't shut down the signal is: Because this is an voltage controlled amplifier it needs to have 0V at output when correctly biased with 0CV input. The second VCA does work at Log response but no matter the bias pot it always lets input pass through a bit.

Logically I checked the voltage differences between these two circuits to find out what is wrong with the second one. Turns out all of the check marks are the same when Bias is turned all the way down. Only difference is at the "" mark. The voltage at the collector of PNP transistor doesn't response like the first one. I used BC548 and BC558 for these. When I found that BC558 of the second VCA is acting weird I changed it with another one and another one. I tested like 5-6 different ones. I also tried changing BC548 with no avail. Because at the 2nd VCA the voltage at 1st pin of the LM13700 is not getting as low as the fist one the signal keeps passing through. I tried changing the ICs, I tried changing R18 with 475 instead of 222. I basically checked everything yet find a problem that would cause the issue.
View attachment 138497And there is the thing above. Output is always the scaled version of the input but it is CLIPPED from the top for NO reason. I tried decreasing input signal volume with a voltage divider and nothing changed. At both VCAs and all of the CV response settings the signal has a positive voltage offset for no reason and I think as a result it clips from the top only. Since this issue can be only happening from the LM13700 I obviously checked all around it. I just can not find a reason. I don't know why voltage controlled amplifiers are this problematic.

 

[Fluffyboii]

I fixed the clipping by adding two resistors I forgot. First VCA works fine now.

I switched the inputs of the LM13700 around to see if the problem of the second VCA was from it ot the OP AMP side. It unexpectedly turns out that it is not about the LM13700 and the problematic side switched when I changed the inputs.

Checking all voltages on op amp and transistors shows that only different value is the PNP transistor collector voltage. Which is odd since base and emitter voltages are same. The problematic VCA lm13700 input pin voltage doesn't change much with the ofset pots movement while other one changes accordingly which is the reason why it doesn't work properly. Yet there is no difference between those two when I check all of the transistor connections.

 

[atferrari]

Is Q2 connected properly?

 

[eTech]

I believe there should be a positive current feed to bias pin 1 of IC2-A.
Q2 looks like E is connected wrong. E should connect to V+, C to R12 ,then to IC2-A pin 1.

Actually, now that I think about it, Q2 may not be needed.

 

[rjenkinsgb]

I believe there should be a positive current feed to bias pin 1 of IC2-A.
Q2 looks like E is connected wrong. E should connect to V+, C to R12 ,then to IC2-A pin 1.

Actually, now that I think about it, Q2 may not be needed.

The bias current is supposed to be sourced from 0V; the IC internal bias current circuit operates at about two diode drops from negative supply.

The weird Q2 configuration is a logarithmic converter, common (in some form) in synths.

Checking all voltages on op amp and transistors shows that only different value is the PNP transistor collector voltage. Which is odd since base and emitter voltages are same.

Double check all the resistor values??

The Q2 and OTA bias input circuits are current control, not voltage, so there should not be much voltage change at the amp control input pin over the control input range.

If it reaches even 1.5V above the negative supply, the IC may already have been damaged, or be faulty.

See the amp bias input circuit, bottom left in the diagram; two base-emitter junctions connect it to negative power. Anything more than 2mA in to that pin can damage or destroy it.

 

[Fluffyboii]

The bias current is supposed to be sourced from 0V; the IC internal bias current circuit operates at about two diode drops from negative supply.

The weird Q2 configuration is a logarithmic converter, common (in some form) in synths.



Double check all the resistor values??

The Q2 and OTA bias input circuits are current control, not voltage, so there should not be much voltage change at the amp control input pin over the control input range.

If it reaches even 1.5V above the negative supply, the IC may already have been damaged, or be faulty.

See the amp bias input circuit, bottom left in the diagram; two base-emitter junctions connect it to negative power. Anything more than 2mA in to that pin can damage or destroy it.

View attachment 138498

I know the LM13700 is not damaged since there is 20K resistor between Q2 and its amp bias pin and its voltage is always around -11 volts. I can change the connections of pin 1 and pin 16 to see that first VCA also works on other side and faulty one moves to first one. Also the linear mode works just fine on both of them. There is something wrong with Q2 or IC-1C connections. Since I checked Q2 like 20 times I can't say what is the problem with it. Will check it again.

I checked the connenctions again and they are all correct. Is is so weird. I looked everything about Log circutry but I can't find a fault. Maybe I should build one on breadboard and connect it to pin 16 instead of the one I made and see if it works. Then maybe that would make me see something obvious that I am not catching on.

 

[Fluffyboii]

rjenkinsgb I did fix the circuit. First I got the problem VCA op amp output on breadboard and connected it to the new transistor pair I placed. I bent the 1st pin of the LM13700 so that it was isolated from the original circuit. Breadboard test worked after I connected that to the LM13700. I removed the original transistors and connected the new ones on a different area on the perf board because the original area pads was destroyed from constant re-soldering. It is working fine now.

I did use BC548 and BC558 instead of 2n3004 and 2n3006 at first while making it. I bought the 2n3006 and 2n3004 today and installed those 2 instead on the problem VCA. I think BC548 and BC558 worked fine on the fist VCA by luck and there was some issue about it which got resolved when I used 2n3004 and 2n3006 instead. Do you think that was the reason of my suffering. I thought transistor model didn't matter much on these types of circuits. Or there was a persisting wrong connection on the PNP side that I didn't notice but got fixed on the new location. Thats unlikely because I used the same resistors for new BJTs and checked it like 30 times with multimeter.

 

[Tony Stewart]

http://musicfromouterspace.com/analogsynth_new/DUALVCA/DLLVCA001.html

 

[Fluffyboii]

http://musicfromouterspace.com/analogsynth_new/DUALVCA/DLLVCA001.html

I see so that one has an extra pot between inverting and non inverting pins of the lm13700 and a 4.7k resistor connected to negative supply at the output. I wonder if I need them.

 

[Dick Cappels]

atferrari already mentioned the suspicious connection of Q2. That's not the best way to make your transfer function logarithmic.

Q2 will act like a 5 volt Zener diode when the voltage on its base reaches...5 volts.

From the Fairchild 2N3906 datasheet:

Is the clipping level near +5VDC?

 

[Fluffyboii]

atferrari already mentioned the suspicious connection of Q2. That's not the best way to make your transfer function logarithmic.

Q2 will act like a 5 volt Zener diode when the voltage on its base reaches...5 volts.

From the Fairchild 2N3906 datasheet:

View attachment 138509

Is the clipping level near +5VDC?

Clipping was from missing R16 and fixed. My input is 10V peak to peak from VCO.

 

[Fluffyboii]

atferrari already mentioned the suspicious connection of Q2. That's not the best way to make your transfer function logarithmic.

Q2 will act like a 5 volt Zener diode when the voltage on its base reaches...5 volts.

From the Fairchild 2N3906 datasheet:

View attachment 138509

Is the clipping level near +5VDC?

That log converter is just too unreliable. I fixed the problem by brute forcing different transistors and yet even though now VCA works the one that uses bc548 and bc558 works better (it got higher maximum output volume) than the one I fixed with 2n3904 and 2n3906. But I can't afford to fiddle more with it since it will probably stop working if I do.
Another circuit, same transistor thing. And it is not working properly as usual. When I drive the Amp bias with another circuit thing works. But the original thing is just random mess. I catch 2n3904 overheating at some point. 100ohm pot doesn't do anything as much as I understand. The transistor connections are correct since I compared with another working circuit.

 

[Fluffyboii]

There is another log converter of Moritz Klein's VCO. This one always worked reliably in my three VCOs even though I did not match a single transistor. Why do I struggle with Ray's design?

 

[Fluffyboii]

The bias current is supposed to be sourced from 0V; the IC internal bias current circuit operates at about two diode drops from negative supply.

The weird Q2 configuration is a logarithmic converter, common (in some form) in synths.



Double check all the resistor values??

The Q2 and OTA bias input circuits are current control, not voltage, so there should not be much voltage change at the amp control input pin over the control input range.

If it reaches even 1.5V above the negative supply, the IC may already have been damaged, or be faulty.

See the amp bias input circuit, bottom left in the diagram; two base-emitter junctions connect it to negative power. Anything more than 2mA in to that pin can damage or destroy it.

View attachment 138498

If I decrease the resistance of the 20K resistor thats connected to Amp bias and let more current flow through it. Will it increase the maximum output voltage. Because this thing operates with current instead of voltage like Op Amps I am having trouble understanding it. The transistor pairs are not making it open enough because random silicon differences I think.

 

[rjenkinsgb]

Are the transistors you are using bought from a proper supplier, or some unknown source devices?

It should not be possible for the 2N3904 in that circuit to overheat. I can only guess it's a faulty / reject one or remarked device with different pinouts..

Do you have a transistor tester? They are invaluable for checking pinouts and that components are OK.
Look for something like this on ebay:

Transistor Tester Diode 328 ESR/PNP/NPN L/C/R Graphics MOS Transistor Tester* | eBay

Find many great new & used options and get the best deals for Transistor Tester Diode 328 ESR/PNP/NPN L/C/R Graphics MOS Transistor Tester* at the best online prices at eBay! Free delivery for many products.

www.ebay.co.uk

 

[Nigel Goodwin]

Are the transistors you are using bought from a proper supplier, or some unknown source devices?

It should not be possible for the 2N3904 in that circuit to overheat. I can only guess it's a faulty / reject one or remarked device with different pinouts..

Do you have a transistor tester? They are invaluable for checking pinouts and that components are OK.
Look for something like this on ebay:

Transistor Tester Diode 328 ESR/PNP/NPN L/C/R Graphics MOS Transistor Tester* | eBay

Find many great new & used options and get the best deals for Transistor Tester Diode 328 ESR/PNP/NPN L/C/R Graphics MOS Transistor Tester* at the best online prices at eBay! Free delivery for many products.

www.ebay.co.uk

Assuming he means Q1?, then the transistor goes directly from 12V (at the collector) to ground (at the emitter) via two paralleled Vbe junctions - so if the base goes positive very far you could certainly overheat Q1 as there's nothing to limit current.

 

[Fluffyboii]

Are the transistors you are using bought from a proper supplier, or some unknown source devices?

It should not be possible for the 2N3904 in that circuit to overheat. I can only guess it's a faulty / reject one or remarked device with different pinouts..

Do you have a transistor tester? They are invaluable for checking pinouts and that components are OK.
Look for something like this on ebay:

Transistor Tester Diode 328 ESR/PNP/NPN L/C/R Graphics MOS Transistor Tester* | eBay

Find many great new & used options and get the best deals for Transistor Tester Diode 328 ESR/PNP/NPN L/C/R Graphics MOS Transistor Tester* at the best online prices at eBay! Free delivery for many products.

www.ebay.co.uk

I have that exact same device. I tested them before putting them on the circuit. Transistors are bought from legit local shops. They only cause problems when I try to build that specific circuit.

 

[rjenkinsgb]

Assuming he means Q1?, then the transistor goes directly from 12V (at the collector) to ground (at the emitter) via two paralleled Vbe junctions

The lower ones are PNP, drawn upside down (just to confuse things), so the only base current should be from the 120K resistor to -12V??

Also the opamp output should never be more than a few hundred millivolts as it's "gain" is around 0.02

 

[Fluffyboii]

Also the opamp output should never be more than a few hundred millivolts as it's "gain" is around 0.02

Thats odd because I saw it swing down from -8V to positive 8V with pots movement and while doing that transistor started heating up. Sometimes CV input and CV off set pot doesn't effect its output, sometime it does. It is a total mystery.

 

[rjenkinsgb]

Are you sure R3 is the correct value?? With 2K feedback (R3) and 100K input, the output should vary just 1/50th as much as any input...

The positive input should be at approximately 1/50th of the positive supply also.