Electronics > Beginners
help, oscilating circuit
<< < (2/2)
MarkF:
I had a quick go at redrawing your circuit

   
magic:

--- Quote from: GigaJoe on March 13, 2019, 03:38:19 am ---it oscillating very stable, around 140KHz , even if I increase back-feed caps (C1, C4) to 0.3uf it oscillating on 2KHz.
--- End quote ---
Remove C4, increase C1. As it stands, U2B adds completely unnecessary delay to U2A feedback path and decreases its stability, particularly with large C4. You only need C4 if U2B oscillates at a high frequency independently from U2A, but I think it is U2A oscillating here and U2B just follows its signal.
I don't quite see the point of Q2, wouldn't it work if B-E terminals are simply shorted by a wire?
Not sure what's the point of R5 either?
Personally, I would ditch U2B, feed current sense directly into U2A and decrease R6 output voltage 10x. But that's not a killer, your circuit should work too with enough C1.
David Hess:

--- Quote from: magic on March 13, 2019, 07:37:46 am ---Personally, I would ditch U2B, feed current sense directly into U2A and decrease R6 output voltage 10x. But that's not a killer, your circuit should work too with enough C1.
--- End quote ---

I agree.  U2B and its gain is not adding anything to this circuit because it has the same precision as error amplifier U2A.

R5 should be removed also and R4 should be decreased in value or the output buffer changed to class-ab to preserve the operating point of the output buffer.  U2B could be useful here to replace R5 with a constant current sink although I would probably add a PNP transistor for a class-ab diamond buffer configuration.
GigaJoe:
Awesomnes  !!! Thanks everyone.

R2= off  C1 = 822pf   R4=470 (100mw dissipation on transistor ) -  no oscillation

so it kinda prototype,  a building block for a digital load.   I can amplify up to 5V input == 1A ( wil see how it linear are ) try to use a DAC to manage it + controller and so ...  just to see how it will work.

Q2 - a simple current amplifier, I simply unloading opamp output during transient process, discharging mosfet, and speed it up as well.
U2B - allow to make current shunt value smaller, less power dissipation, decrease thermal effect, and increase managing voltage input.

for a diamond buffer, i need a bipolar supply, I guess? does it make sense? as the  mosfet managing voltage always positive, anyway.  I'm not sure,  a next iteration would be OPA227 - current opamp ; OPA828 - compesat. opamp ...






magic:
I'm not sure if the output buffer is really needed. Particularly class A with ~10mA bias, which is less sinking current than the LM could deliver by itself. Maybe still an improvement due to the current gain, or maybe not - wouldn't assume it without testing.
Here's what Dave put together the other day. Notice the absence of 80% of the parts you included and it still worked and without a slowdown capacitor around the opamp ;)
I would start simple and add gain stages where required if speed is insufficient rather than build something complex and worry on speculative basis that removing components could perhaps worsen transient response. That's the definition of audiophoolery :--

U2B isn't really needed. Reduce the shunt 10x, divide the control voltage 10x, nothing changes. You may need to upgrade U2A if offset plays tricks on you or if it drifts with temperature. U2A will now need the same precision specs as U2B. But you save by removing U2B altogether.
Navigation
Message Index
Previous page
There was an error while thanking
Thanking...

Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod