Maybe if you were a little more specific or included some sort of reference as to what that statement refers to someone could help.
I would blindly guess that the gain is approximately equal to the feeback resistor divided by the input resistor 😉
Yep... and that resistor must have silver leads!
And they should be Kobe resistors from Japan. These are special resistors massaged daily and fed beer.
😉
And they should be Kobe resistors from Japan. These are special resistors massaged daily and fed beer.
😉
I asked that question in the thread on UcD / LtSpice.
Conceptually, it is the ratio between the resistor / cap network and the base resistor (1K in the case of the original UcD).
so you would think increasing the upper resistor would increase the gain. It does, but only to an extent as the impedance of the resistor / cap side comes dominant and at that point increasing the resistor doesn't help.
In other words, the maximum gain is capped by the resistor / cap network.
Increasing that will change the behavior of the feedback system and may cause the oscillation to stop.
The oscillation frequency is determined by the output filter. a filter with high cornor frequency will cause the amp to oscillate faster.
Somehow, I hope someone will come out with a set of formula that will tie the two together.
Conceptually, it is the ratio between the resistor / cap network and the base resistor (1K in the case of the original UcD).
so you would think increasing the upper resistor would increase the gain. It does, but only to an extent as the impedance of the resistor / cap side comes dominant and at that point increasing the resistor doesn't help.
In other words, the maximum gain is capped by the resistor / cap network.
Increasing that will change the behavior of the feedback system and may cause the oscillation to stop.
The oscillation frequency is determined by the output filter. a filter with high cornor frequency will cause the amp to oscillate faster.
Somehow, I hope someone will come out with a set of formula that will tie the two together.
PS: that was meant to be humorous. I always enjoy your comments and provocative point of view. -- a.s.
Sorry, I was just messing with ya.
Seriously, it's probably best to determine the oscillating frequency by starting with something you know that works, change part values, see how it affects it, and make a note of it. Or you can run simulations and change anything you want! That way is obviously faster and there have been LT spice simulations posted in this forum before that you could use.
Seriously, it's probably best to determine the oscillating frequency by starting with something you know that works, change part values, see how it affects it, and make a note of it. Or you can run simulations and change anything you want! That way is obviously faster and there have been LT spice simulations posted in this forum before that you could use.
BWRX said:
thanks for your kind suggestions.
I believe the ask is to come up with formulaic solutions, not empiric or simulational.
Let me know when you find something in that direction anywhere, google or not.
Again, many thanks.
fokker said:
Your question is unclear (actually, it's left out entirely), but if you are interested in knowing how one determines the nominal operating frequency of a self-oscillating type class-d amplifier design, it is done by calculating the frequency at which the phase shift around the total feedback loop (forward amplifier gain together with feedback gain) just reaches 360 degrees. It is assumed that normal output saturation will automatically limit total loop gain to precisely unity.
Regards -- analogspiceman
- Status
- Not open for further replies.