LTspice is giving me a -3dB of over 1MHz
all models are b.cordel and Ian Hegglun (i get the same results with other models)
with the real circuit, i measured a -3dB of around 500KHz
my 2N5xxxx are coded "B331" so i replaced these with bc559c/bc549c with the same results
mje340g/350g were replaced with old bd139/140 with the same results
(outputs are from profusion)
any ideas why the circuit -3dB is so off compared to the LTspice?
all models are b.cordel and Ian Hegglun (i get the same results with other models)
with the real circuit, i measured a -3dB of around 500KHz
my 2N5xxxx are coded "B331" so i replaced these with bc559c/bc549c with the same results
mje340g/350g were replaced with old bd139/140 with the same results
(outputs are from profusion)
any ideas why the circuit -3dB is so off compared to the LTspice?
Attachments
Maybe C6 or C9 is way off.
Or the amount of open loop gain is way off.
In any event, that's excessive bandwidth for audio.
Or the amount of open loop gain is way off.
In any event, that's excessive bandwidth for audio.
While I like LTSPICE a lot it does have its limitations.
I did a valve circuit running off 12V B+ and it simulated great.
In practice it failed to work. I have to add positive bias to grid to get it to work on the pcb.
I did a valve circuit running off 12V B+ and it simulated great.
In practice it failed to work. I have to add positive bias to grid to get it to work on the pcb.
Capcitance in Mosfet models vs actual devices. I've seen a lot of over optimistic mosfet models in LTspice.
C6 and C9 are 5pF only. Chances are that parasitic PCB capacitance has significant impact. Estimate how much capacitance your PCB adds and correlate that with your measurements.
i checked a spare board and measured less than 2pf for the various actives junctions.
adding these capacitances to the simulation made no sigificant change.
Last edited:
You can also feed an amplifier stage in SPICE with an input signal that causes the output to go above the supply rails, and the output merrily puts out a clean signal exceeding the rail voltages
I disagree. It's a feedback amplifier with no low-pass filter at its input or output, so its bandwidth has to be much greater than 20 kHz if it has to have a decent amount of loop gain at 20 kHz.
Hi ctrlx,
Nice circuit!
On a real pcb its not only parasitic capacitance youre getting.
Try simulating power supply impedance. Add some resistance and inductance to your tracks. If your amp is high bandwidth PSU impedance could have a big impact I found in my ltspice vs prototyping experience.
Also try adding some ESR and inductance in your feedback electrolithic. Try bypassing it on pcb with a ceramic. Also bypass psu with capacitors that have low impedance at high frequencies.
Cheers,
Ruben
Nice circuit!
On a real pcb its not only parasitic capacitance youre getting.
Try simulating power supply impedance. Add some resistance and inductance to your tracks. If your amp is high bandwidth PSU impedance could have a big impact I found in my ltspice vs prototyping experience.
Also try adding some ESR and inductance in your feedback electrolithic. Try bypassing it on pcb with a ceramic. Also bypass psu with capacitors that have low impedance at high frequencies.
Cheers,
Ruben