EveryCircuit - Why is it oscillating

6 years ago

The 1nF cap between the o/p of the op amp and the - input is the culprit. If you take a closer look, you will the a relaxation Oscillator.

PrathikP

6 years ago

The 10k and 1k resistors along with the 1nF cap and 1k resistor cause the oscillations. Presumably, you put the 1nF cap there to prevent CFS?

PrathikP

6 years ago

http://everycircuit.com/circuit/5322830567440384

thebugger

6 years ago

The 1nF cap here seems to remove the oscillations, although in a different circuit I have it actually works to slow the op amp. Without it, higher frequency oscillations seem to happen in the other simulation

PrathikP

6 years ago

And why do you have 1mF of output capacitance? Bit much, don't you think?

thebugger

6 years ago

If you slow down the simulation enough, you'd see that it's trying to oscillate at 65kHz without the op amp capacitor, but the output cap being 1mF filters out most of the oscillations. Still - don't want to oscillate at all

PrathikP

6 years ago

Actually, the 100pF across the feedback resistor prevented the oscillations. Why did you remove it?

thebugger

6 years ago

Everything's still there.

PrathikP

6 years ago

See in the circuit that I linked I have only removed the 1n. Everything else is intact and it doesn't oscillate

thebugger

6 years ago

Problem is that all these capacitors help with high frequency oscillations, where the circuit oscillates in the audio range - so the capacitors don't do much to halt the oscillations.

thebugger

6 years ago

Maybe the open loop gain is too high, but there's some hysteresis somewhere I can't seem to find. It wouldn't oscillate if there's only 1 threshold. It seems to oscillate as a sawtooth waveform, which usually relies on hysteresis

PrathikP

6 years ago

Well that cause of the 1n. And I don't see any oscillations without it in any timebase. Did you see these oscillations in that other simulator?

thebugger

6 years ago

It oscillates without the 1nF as well. Wait it to settle down at 5V and decrease the time resolution to 5uS/s and you'll see it oscillate at 70Khz (or 140kHz). Can't seem to pinpoint it

PrathikP

6 years ago

Apart from initial oscillations, I don't really see anything here

thebugger

6 years ago

Try again. Wait for the output to settle down and decrease to 5us/s and wait a few seconds. You'll see the output of the op amp oscillate. The output capacitor seems to halt the oscillations at a very low level by presenting minimal impedance, but doesn't stop them at the source.

PrathikP

6 years ago

Yes I now see 70khz-104khz oscillations. It's very tiny. Even the oscilloscope vertical limits are 4.84v-4.84v. it's probably some simulator related thing, unless you see it in another sim of course

thebugger

6 years ago

Actually the output is too low because the impedance of the capacitor is far more low than the current gain the transistors can supply. The op amp is oscillating at its full range to try and compensate. It can't, but it tries. Still has a root cause somewhere.

PrathikP

6 years ago

Ok yes I see the op amp is going crazy at its output

thebugger

6 years ago

In fact if you crank up the current gain of the transistors to the maximum, the impedance difference narrows, and the output is free to oscillate in a greater range.

thebugger

6 years ago

Yes, exactly.

PrathikP

6 years ago

One thing I don't understand is why youre controlling the series pass trannys by shunting a constant current source.

thebugger

6 years ago

It creates a less variable environment for the op amp to try and compensate. If I were to use a normal resistor, the voltage variances of the input would cause the output constantly trying to compensate, which will increase the output ripple.

thebugger

6 years ago

It creates a more stable environment, plus I already tried without it. Still oscillating

thebugger

6 years ago

I believe the op amp has a very large local open loop gain that causes this. Substituting the 1nF cap with even a 100k resistor seems to decrease the oscillations significantly. Issue is that with this configuration, it will create a positive feedback loop in the larger scale - thus modifying the input voltage out of proportions. A more complex calculation is needed to determine overall output voltage. Also, I'm not sure if it'll not make the whole thing unstable in some conditions.

swathi2011

6 years ago

Vibration with mimimum frequency is oscillation,with maximum frequency is oscillating

JoeHH

6 years ago

This is a current source with a capacitor, this generates the ramp signal. you combine it with a comparator with histersis and .... ready an oscillator that generates a sawtooth or triangular signal. Adel s. sedra, microelectronic circuits fifth edition, chapter 13, page 1192 (in the Spanish version)

thebugger

6 years ago

But this should be a constant voltage source, not a constant current one

thebugger

6 years ago

Also, I cannot see any hysteresis here

2ctiby

6 years ago

Get rid of the 1nF cap at the opamp and reduce the top right Forward beta down from 500 to 100 A/A ... Both of those aspects are varying (pulsing) the op amp + terminal. Getting rid of just one aspect will not stop those pulses. A 100 A/A is ok with no 1nF cap attached.

PrathikP

6 years ago

@JoeHH in order to generate a ramp signal, the capacitor must be in series with the current source and there must be a circuit to pulse discharge that capacitor. In this circuit, there is a parallel capacitor for stabilization.

2ctiby

6 years ago

@thebugger... An alternative to my suggestion above would be to just replace the 100pF cap which is parallel to the 820k resistor, with a 5pF or less cap if that is feasible? You would then just need to wait until after dozen or so pulses have initially occurred.

thebugger

6 years ago

I think I wi have to restructure the whole circuit. I made it like so to avoid using op amps working with high voltages (up to 36V). The extra voltage amplifier stage would work to step up the voltage instead. However, it doesn't seem too reliable. I believe the op amp has a very high open loop gain in on itself, causing the instability.

thebugger

6 years ago

I would go with a more classic - discrete configuration instead. It has a few drawbacks, such as complexity, but should be a hell of a lot more stable.

2ctiby

6 years ago

Interesting presentation to chew over.

jason9

5 years ago

Changing the 1nF cap to 100nF seems to fix the issue. IDK why it oscillates when the cap is small but not big. I’ll have to study it more.

jason9

5 years ago

I can’t really figure it out. All I know is that 100nF fixes the issue.

fatcat2

5 years ago

@thebugger, why do you use strange feedback configurations in the hifi amps you design? I see no point in it cuz the existing shunt and series feedback configurations also can be used. First of all, why do you use long tailed pairs for making your amps? I think that ICs are way better than discrete implementation. Sorry for being off-the-topic.