Subtracting error correction

[HAYK]

With fine adjustments, decreasing the gain 3db to 23db, readjusting the corresponding corrector gain, I reached -70db@20khz, minimum is -75db.
The Monte Carlo test cannot be applied as the corrector gain, about 1.5 needs to be adjusted for the corresponding gain of the power amp.
The saturation is similar to the corrected amp alone.
I will try this circuit on my 2x300w PA amp that once I bought for less than $200 to power a subwoofer that I conceived and built the 4x8" drivers. It is ordinary Lin type 5551s in LTP 2SB647 VAS +CCS and 3EF with 3 pairs of Toshiba outputs. I chose it for its bass but for large jazz orchestras, large vocals even 16 voice of "Happy Together "of the Turtles, it is bad.

 

[ihan]

Hi Hayk,

You might find the blog archive at this site useful: https://blog.easychipamp.com/
Perhaps, start at April 2022: https://blog.easychipamp.com/2022/04/hawksfords-error-corection-and.html
and then follow the 'High Precision composite op-amps" series from May 2022 onwards: https://blog.easychipamp.com/2022/05/high-precision-composite-op-amps.html

Cheers, Ian

 

[HAYK]

I made a quick simple 2 stage MOSFET amp, 90ma bias, with gain 18. It has 0.2% THD for 18v 1khz. I applied my error corrector and adjust the gain for minimum THD to get 0.0075% for 1khz. The story is not telling me, why I need more gain to get minimum by error generator.

 

[HAYK]

I gave some extra gain with CFP differential, it is now 0.0005% for the same output.
The stability is not yet exemplary.
I am still exploring why with error generator I can subtract down to -80db@10khz but not with input signal.

 

[HAYK]

It worked. I mounted on an LM3886 and by ear adjusted it to get clean Mozart's Requiem. If I over do, it is creating a grain in multivoice.
I must find a procedure to adjust precisely.