I hope it is oscillating somehow because otherwise I'm going to struggle to explain my weird DC readings
Thanks for the tip about creating an RF loop antenna with my scope probe. Agreed, I can't do much without a working oscilloscope which I won't have until the end of next week at the earliest.
You aren't mistaken about the circuit. It has too much gain and too many transistors. Needs simplifying.
Well. The report is a simulated phase margin not the real circuit. There are multiple feedback paths so what does this margin mean? In any case, phase margin is not sufficient to guarantee stability. And the real circuit is an unlicensed radio station by the sound of it. A scope will reveal all.
I think the amplifier has too much gain to have the stated phase margin. If you try to calculate the gain, it must be at least above 80 dB, and a single Cdom around the second differensial pair can not be enough to stabilize the amplifier.
Is open loop gain above 80 dB excessive? Sounds all right to me. Closed loop gain is 27 dB.
Let's see what the scope shows before we jump to conclusions.
Last edited:
Try putting 100 ohms in series with the bases of Q22 and Q23 (base stopper resistors).
Well, parasitics seldom help avoiding Murphy's law applied to amp (oscillator) building.
Maybe a quick picture would help knowing how (I hate to say) bad is the situation...mine has been awful too
One trick that helps is referencing directly the zobel to a good star ground...maybe that will apply to many of the ground references...
Good luck,
M.
Maybe a quick picture would help knowing how (I hate to say) bad is the situation...mine has been awful too
One trick that helps is referencing directly the zobel to a good star ground...maybe that will apply to many of the ground references...
Good luck,
M.
UPDATE:
I noticed a few different oscillation frequencies all in the MHz range. I got rid of all but one of them with some rail decoupling: 100nF from the collectors of the output transistors to ground, 100pF between collector and base of the driver transistors. I also tried base stopper resistors for the driver and output transistors but they didn't have any effect.
I have one oscillation left: a sine wave at 235 MHz (measuring 100 mVp at the output) so I think that must be coming from the 2N390x transistors since they are the only ones with an fT high enough. I used my oscilloscope probe as an RF loop antenna to check which part of the circuit might be causing this but I don't get any additional information with this method: the magnitude of the oscillation looks to be the same wherever I point the probe.
I already tried removing some of the feedback topologies I used, like changing the input tail current source to an LED-biased one, and changing the Wilson current mirrors to simple ones, but it has made no difference. I still see the 235 MHz.
Any tips on how to remove the 235 MHz oscillation would be appreciated.
I noticed a few different oscillation frequencies all in the MHz range. I got rid of all but one of them with some rail decoupling: 100nF from the collectors of the output transistors to ground, 100pF between collector and base of the driver transistors. I also tried base stopper resistors for the driver and output transistors but they didn't have any effect.
I have one oscillation left: a sine wave at 235 MHz (measuring 100 mVp at the output) so I think that must be coming from the 2N390x transistors since they are the only ones with an fT high enough. I used my oscilloscope probe as an RF loop antenna to check which part of the circuit might be causing this but I don't get any additional information with this method: the magnitude of the oscillation looks to be the same wherever I point the probe.
I already tried removing some of the feedback topologies I used, like changing the input tail current source to an LED-biased one, and changing the Wilson current mirrors to simple ones, but it has made no difference. I still see the 235 MHz.
Any tips on how to remove the 235 MHz oscillation would be appreciated.
Last edited:
Did you use a 1:1 or a 1:10 probe for the loop antenna trick? I usually get the clearest results with 1:1.
If you haven't already done so, you can switch off the amp to see if the 235 MHz disappears, just to make sure it isn't a nearby DAB or television transmitter.
You could try adding base stoppers to all the high-fT devices, starting with the ones that have the highest bias currents (highest gm -> most likely to oscillate).
Other tricks: if the supply voltage is low enough, and if I remember well it is, you can touch nodes with your finger while measuring the 235 MHz at the output to see what nodes are sensitive.
If you haven't already done so, you can switch off the amp to see if the 235 MHz disappears, just to make sure it isn't a nearby DAB or television transmitter.
You could try adding base stoppers to all the high-fT devices, starting with the ones that have the highest bias currents (highest gm -> most likely to oscillate).
Other tricks: if the supply voltage is low enough, and if I remember well it is, you can touch nodes with your finger while measuring the 235 MHz at the output to see what nodes are sensitive.
I used a 1:10 probe. I'll try again with 1:1 (if I can find one!).
That was the first thing I tried. It's definitely coming from the amp unfortunately.
Thanks! I'll give that a go.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.