LM3886 feedback resistor value selection.

[tempoct]

Hi,
I wonder what are the real world effect of different values of Rf1 and Ri for non-inverting design. Going through data sheet and old threads points out that keeping it low to provide enough feedback current, thus less noise. Make it too low will increase too much current and heats up the resistor (leads to more noise). Also, Rf1 and Ri value are used to calculate Rb for input bias current. Too much current will also increase the noise, if I understand this correctly. Most people seems to use 20k/1k, which seems to be "magic numbers" from datasheet.

Nevertheless, I haven't found data "how much" less or more noise for different values of these resistors. Some manufacturers (op-amp) provide data of how much noise different, but I can't find for LM3886. Occasionally, I come across some design with higher value says 47k/2.7k or somewhere in between. At the end of data sheet (p22-23), TI/National also show case how to design a 40W @4 ohms amp, which, by example's requirement, they used 100k/8.2k.
https://www.ti.com/lit/ds/symlink/lm3886.pdf?ts=1736072004540

Please share your thoughts on this.
AP

 

[NanoFarad]

Don't think too much, just try the 'magic numbers' or play around that for desired voltage gain. Trust me you're not going to hear any noise, only good music!

Best Regards

 

[tomchr]

I find 20k/1k to work quite well.

As you say, higher values -> higher noise. Lower values -> more power dissipated in Rf, which leads to higher distortion. If you use Cf and Cc as well, I would stick with the values in Test Circuit #2 in the data sheet.

Tom