Help with discrete opamp

Thread Starter

liquidair

Joined Oct 1, 2009
192
Hi all, I have a recent project (which I'm 0 for 2 with) where I needed an opamp to take a negative DC voltage and output 2x the input voltage. This would be an easy task if it not for the fact that I need to do this with high negative voltages (VOut max = 110V or so) and I need this opamp to draw ~5-8mA total max.

Admittedly, solid-state is not my strong point, but I was able to whip up a simple design in LTSpice that seemed to work well. On first power on, I noticed the supply was overheating, and I traced this down to about 33mA flowing in the circuit's output stage (T11, R74, R75, T12). LTSpice shows the whole circuit consuming 6mA or so.

Any clue what could be going on, or how I should be approaching this?
Opamp - Copy.jpg
 

sghioto

Joined Dec 31, 2017
5,380
I think the bias on T11 and T12 is too high. Remove D5 or D6 and replace with a resistor. Try starting with a 1K and see if that helps.
 

Audioguru again

Joined Oct 21, 2019
6,674
The base-emitter of the transistors and the diodes have a range of forward voltage. Some will produce a current too high and some will produce a current too low. You need a trimpot to set the bias current for the output transistors.
 

Thread Starter

liquidair

Joined Oct 1, 2009
192
I think the bias on T11 and T12 is too high. Remove D5 or D6 and replace with a resistor. Try starting with a 1K and see if that helps.
Thank you! I popped this into the simulator. While swapping one of the diodes for the 1k resistor, the current in the output stage shot up to 60mA vs the 1.3mA I had with the 2 diodes. While that appears to be a step in the wrong direction, it's the first time I was able to duplicate what I was seeing in real life so you're onto something I think!
 

Thread Starter

liquidair

Joined Oct 1, 2009
192
The base-emitter of the transistors and the diodes have a range of forward voltage. Some will produce a current too high and some will produce a current too low. You need a trimpot to set the bias current for the output transistors.
AHHH, so this would explain why I get a low current in Spice but the real world differs...LTSpice is using perfect components. Is there a way to get around the trimpot and have something that always works? It would help with space and error potential.
 

michael8

Joined Jan 11, 2015
410
I don't understand your input bias. Normally the inputs of an opamp are biased between the +/- supplies. All you have is gnd and -120v with the bases of T7 and T8 returned to gnd (which is V+!)
 

Thread Starter

liquidair

Joined Oct 1, 2009
192
I don't understand your input bias. Normally the inputs of an opamp are biased between the +/- supplies. All you have is gnd and -120v with the bases of T7 and T8 returned to gnd (which is V+!)
T7 will always be getting a negative DC voltage as it's input, somewhere between -20 to -55V ish.

But are you commenting on the ground connection on R69? If so, that's a good catch. Initially I had this circuit powered from +30V on the positive rail. When I first saw the -120V supply overheating, I surmised the problem could be that since I'm testing the device for the first time, the fact that the 30V supply is not powered on yet might be causing problems. But I realized this circuit doesn't really need 30V since we'll never want a positive output voltage, so I grounded the positive rail instead, and left R69 the way it was originally.

So does that make grounding R69 inappropriate? If so, what is a better option? Like I said, I'm out of my element on this one...tubes came naturally, integrated opamps make sense to me, but solid-state never clicked except for switching.
 
Last edited:

Thread Starter

liquidair

Joined Oct 1, 2009
192
Try a 100 ohm resistor.
100 ohm knocked it down to like 70nA or something in the sim, but I'm seeing what you are doing and how it's affecting the output current. It looks like I can at least alleviate the overheating problem by replacing D5 or D6 with a low ohm resistor and work up until I can get a trimmer in there. Thank you!
 

Ian0

Joined Aug 7, 2020
9,671
If it's only a DC amplifier, and there are not requirements for good crossover distortion performance, remove D5 an D6 altogether and connect both output transistor bases together. If you ask it to amplify a sinewave, it will be rubbish, but for DC it will do the job and bias won't be a problem.
What load is connected to the output? You may not need to provide output drive in both directions.
 

Thread Starter

liquidair

Joined Oct 1, 2009
192
If it's only a DC amplifier, and there are not requirements for good crossover distortion performance, remove D5 an D6 altogether and connect both output transistor bases together. If you ask it to amplify a sinewave, it will be rubbish, but for DC it will do the job and bias won't be a problem.
What load is connected to the output? You may not need to provide output drive in both directions.
Ya, It's just to provide DC bias. It may need to absorb any AC signals from the signal path, but those signals should be minor. I can always sim it since sghioto and Audioguru got me straightened out that its a bias issue causing the high dissipation, likely not something wrong.
 

Ian0

Joined Aug 7, 2020
9,671
Do you really need push-pull operation, or is it just a resistor to ground? If so you can leave out D5, D6, T11, T12, R74 and R75.
 
Top