There are sometimes, that I can't understand why could the simulation turn out like that, which looks like oscillation but i don' think so. There are few attachements, showing the simulation transient of it.
May someone tell me what phenomena is this ? and what problem normally exist ?
May someone tell me what phenomena is this ? and what problem normally exist ?
Attachments
Is there feedback? It looks like ringing (just before oscillation). Is something happening at the points where it starts? A circuit would help. There are some LTSpice parameters/preferences that might be causing this. It also looks like you have it set to a "thick line" or "show data points" which dosnt make the response graph as clear.
This post should be in the software forum, you will get more responses there.
This post should be in the software forum, you will get more responses there.
This time something different. The attachement is the third screenshot, which is different from 2nd result, but THIS is the SAME schematic..... just do it the second day...... well... they resemble the same phenomena anyway.
THe weird thing for me is that why it isn't appearing at every same place of every wave cycle, but like every few cycle it will happen.
I personally don't think posting schematic will help, as same schematic will produce this and NOT produce this by little tweak (increase bias seems like delay/varies the phenomena). When this happen, even the front end of the amplifier will show eradict result like that.Is there feedback? It looks like ringing (just before oscillation). Is something happening at the points where it starts? A circuit would help. There are some LTSpice parameters/preferences that might be causing this. It also looks like you have it set to a "thick line" or "show data points" which dosnt make the response graph as clear.
This post should be in the software forum, you will get more responses there.
THe weird thing for me is that why it isn't appearing at every same place of every wave cycle, but like every few cycle it will happen.
how much is small time step you meant ? i never change any value of that.It looks like LTspice is trying to tell you something. Try forcing a smaller time step. It might make the oscillation a bit more continuous, and therefore convincingly real, as opposed to something that you might think is a simulator quirk.
Attachments
Try setting the following parameters in the Edit Simulation Command:
Stop Time: 2m
Maximum Timestep: 200n
That will force at least 10K steps.
You could also try the other solver...
Tools=>Control Panel=>Spice=>Solver=>Alternate
Seeing the schematic can also be helpful. Someone on the forum might recognize the thing that's giving the simulator fits.
Stop Time: 2m
Maximum Timestep: 200n
That will force at least 10K steps.
You could also try the other solver...
Tools=>Control Panel=>Spice=>Solver=>Alternate
Seeing the schematic can also be helpful. Someone on the forum might recognize the thing that's giving the simulator fits.
I have had plenty of my simulations, and a few of my real circuits behave like that. You need to run an ac analysis and check the phase margin. Looks like marginal stability to me.
see C:\Program Files\LTC\LTspiceIV\examples\Educational\LoopGain.asc and LoopGain2.asc for examples of how to evaluate stability.
see C:\Program Files\LTC\LTspiceIV\examples\Educational\LoopGain.asc and LoopGain2.asc for examples of how to evaluate stability.
Last edited:
So should I assume that this is a simulation error ? I check the phase margin at output, doesn't seems to be much different with my other simulation. Phase : 5m degree to -1.5 degree at 100kHzI have had plenty of my simulations, and a few of my real circuits behave like that. You need to run an ac analysis and check the phase margin. Looks like marginal stability to me.
see C:\Program Files\LTC\LTspiceIV\examples\Educational\LoopGain.asc and LoopGain2.asc for examples of how to evaluate stability.
I have tried the setting you told me, and the first attachement is the result.Try setting the following parameters in the Edit Simulation Command:
Stop Time: 2m
Maximum Timestep: 200n
That will force at least 10K steps.
You could also try the other solver...
Tools=>Control Panel=>Spice=>Solver=>Alternate
Seeing the schematic can also be helpful. Someone on the forum might recognize the thing that's giving the simulator fits.
second one is using the previous simulation setting. it is weird that it show different things ..
Attachments
The circuit's unstable. It's oscillating. "Weird" looking results like that are common with spice simulations. The smaller the timestep, the more accurate the results, but it's never perfect. To see the ringing/oscillation better you could use a low-level 100KHz square wave input, and just run it for a few cycles. Doing an AC analysis of the loop gain would also give you a much better idea of what's going on.
edit:
edit:
It sounds like you don't know what "phase margin" means. What happens at 100KHz is irrelevant. What matters is behavior around the unity loop gain frequency, which is probably somewhere around 1MHz, but who knows, since you don't want to show the circuit.I check the phase margin at output, doesn't seems to be much different with my other simulation. Phase : 5m degree to -1.5 degree at 100kHz
Last edited:
Apparently, this looks like a class-B power amp and that the ringing is occuring at the zero-crossings.
Try the sim with higher quiescent current.
If it is just an initial 'coming up', thry to look at it after 1 second so that all switch-on transients have passed.
jan didden
Try the sim with higher quiescent current.
If it is just an initial 'coming up', thry to look at it after 1 second so that all switch-on transients have passed.
jan didden
It may be so, but I tried to vary the Iq, and at some value, it delay the effect (after 7 waves only show eradict noise), at another value, it show nothing within 8 waves.Apparently, this looks like a class-B power amp and that the ringing is occuring at the zero-crossings.
Try the sim with higher quiescent current.
If it is just an initial 'coming up', thry to look at it after 1 second so that all switch-on transients have passed.
jan didden
However, at those value which I mentioned before, it have insane Iq.
Found that increasing the Iq will delay the effect, but mostly will appear, example of putting 16mA gets the effect only after 20 cycle. still experimenting now.
About 11A going through the OPS..... (no eradict reading across 100 waves)
Hm... then could you explain to me, since i'm just beginnerThe circuit's unstable. It's oscillating. "Weird" looking results like that are common with spice simulations. The smaller the timestep, the more accurate the results, but it's never perfect. To see the ringing/oscillation better you could use a low-level 100KHz square wave input, and just run it for a few cycles. Doing an AC analysis of the loop gain would also give you a much better idea of what's going on.
edit:
It sounds like you don't know what "phase margin" means. What happens at 100KHz is irrelevant. What matters is behavior around the unity loop gain frequency, which is probably somewhere around 1MHz, but who knows, since you don't want to show the circuit.
However, one of the reason I don't want to show is because my schematic is messy.... but don't care about that much. It was SSA front end
R10 & R12 is for vary the power deliver by IPS to rearward. (primally for balance, also partially responsible for Iq), R15 is for OPS biasing.
Attachments
I don't see any kind of compensation in this amp.
Also, the value of R24 indicates a driver current on the low side of things. Could make the output stage needlessly slow. Given this plus no compensation, oscillation is hardly surprising.
R16/17 are a bit high. Not sure about thermal stability of the biasing setup, but that's another story. (D. Self likes to use 0.1 ohms even, so as to minimze gm-doubling distortion, but that does require a bit more effort for good thermal stability.)
BC550/560 seem inadequate in terms of Vceo.
Also, the value of R24 indicates a driver current on the low side of things. Could make the output stage needlessly slow. Given this plus no compensation, oscillation is hardly surprising.
R16/17 are a bit high. Not sure about thermal stability of the biasing setup, but that's another story. (D. Self likes to use 0.1 ohms even, so as to minimze gm-doubling distortion, but that does require a bit more effort for good thermal stability.)
BC550/560 seem inadequate in terms of Vceo.
Last edited:
Hm.... If R24 and R16/17 value is not too appropriate, then why would this happen when i change Q7 & Q9 from MJE 13035 to BC550/560 ?I don't see any kind of compensation in this amp.
Also, the value of R24 indicates a driver current on the low side of things. Could make the output stage needlessly slow. Given this plus no compensation, oscillation is hardly surprising.
R16/17 are a bit high. Not sure about thermal stability of the biasing setup, but that's another story. (D. Self likes to use 0.1 ohms even, so as to minimze gm-doubling distortion, but that does require a bit more effort for good thermal stability.)
BC550/560 seem inadequate in terms of Vceo.
However, I didn't exceed the Vceo of those BCs when driving(setting peak output voltage as 12V). Actual situation i'm using 25V supply.
Tried again +-25V supply, the MJE schematic show nice result, and BCs still give those weird result even adjust all the bias to be same a MJEs.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
- Home
- Amplifiers
- Solid State
- Understanding SPice Simulation ?