Additionally to the above: The paper describes a -120dB noise limit for the used AP 2722, that's why 60dB noise gain was used (by adding a resistor between + and - input) to raise the limit from -120dB to -180dB.
Hans
They seem to be rather insufficient and you also haven't included the input noise
Give me the circuit diagram and I will do the sim in LTSpice.
No problem to agree a NDA.
Hans
I‘m using a speaker cable that has some capacitance, no issues.
1 uF in sim pulls down the phase and gives a small peak in hf response.
Transient analysis still works, thd not much affected.
I could include the capacitance of my cable in the final hf compensation.
1 uF in sim pulls down the phase and gives a small peak in hf response.
Transient analysis still works, thd not much affected.
I could include the capacitance of my cable in the final hf compensation.
Attachments
While loop gain might hide the issue of nested feedback loops, phase margin tells a more detailed story as when using opamp(s), already at fairly low frequencies, the value goes to 90 degrees and from then only gets less. Already ~50 years at Elektor, checking stability with a 1uF cap was standard procedure. So 1st thing with amplifier design is to look at stability, and correct it in a way that loads cannot cause instability.
Don´t worry, I built my monoblocks during covid and never had any problems.
My speakers are ACR RP200 with modified network.
Your amp with resistive load already has a considerable amplitude drop at 100 kHz.
I will hf compensate to have highest possible bandwith while beeing stable with normal speakers.
My amp is not intended to drive electrostats.
IMHO compensating for 10µF is overcompensating and that is a compromise.
My amp was designed to use the absolute minimum of components while providing 50W @ 4R below 1 ppm THD @ 20KHz. That has been achieved and was defined as "good enough".
I added some RLC to ground because the midrange has some peak in the response.
So speakers + network form a bit more complex load. Later I will make braided cable with 16 wires, capacitance will go up...
Good and expensive speakers usually have complex networks to achieve a linear frequency response, that makes them more difficult to drive, they need powerful low impedance stable amplifiers.
So speakers + network form a bit more complex load. Later I will make braided cable with 16 wires, capacitance will go up...
Good and expensive speakers usually have complex networks to achieve a linear frequency response, that makes them more difficult to drive, they need powerful low impedance stable amplifiers.
Due to the minimalist design, there is no need to change compensation, whether using purely resistive load or the 10u cap.
Both in sims and in real world I step from .05Uf to 10 uF looking for stability problems. All too often around .1uF to .5 uF there is transient oscillation that goes away at higher and lower cap. And then 6' of audiophile cable + .22 uF and no load can prove to be really bad for amps. You need to check at different drive levels and currents/loads since device fT's change a lot with current.