Feedback and Input resistor value

Hi Everyone,
I am building a solid state power amp project (approx 60~80W), and need some help and opinion on the feedback and input resistor values.
See below image, I read somewhere that ideally the sum of R1+R2+R3 should equal or closed to the value of sum of R4+R5.
The values given to me were as Option1, but I do not have 22kR and 1.8kR in my box but have a bunch of 47kR and 4kR (all CMF55).
Is Option2 bad in anyway? My concern is if increasing the feedback resistor (R4) value too much will have any unfavourable effect, but maybe not since R4+R5 are just voltage divider?
Let me know you think and I wish everyone a wonderful Christmas and a peaceful 2024.
 

Attachments

  • NFB Question.png
    NFB Question.png
    9.6 KB · Views: 227
Hi Lineup,
I only have one source which is DAC, currently using Rega DAC (output voltage is approx 2v, output impedance <1kOhm)
Saving up to move into a streamer eventually but may take another year🙁
Love vinyl but the are expensive in this part of the world, and the climate here is not the best for them.
 
The inverting input of the amplifier has a tiny bit of capacitance. Together with a large value of R4, this forms a pole and is a good reason to make R4 as low as practically possible within reasonable power dissipation limits. 5...10kOhm would be good, 47k is a bit high.
 
I’m afraid I must take issue with some of the advice I’ve seen offered.

Assuming the objective is to minimize DC offset error at the amp output, the short answer is: R1+R2+R3 = R4. R5 does not enter the picture because C2 blocks DC currents.

To illustrate with an example, assume the amplifier has 0 input offset voltage error, has a NPN differential input stage, and that input bias current is 5uA to both the + and - inputs. And assume R1+R2+R3 = R4 = 47k.

The + input will see 5uA * 47k = minus 235mV, since the NPN input bias current will pull the voltage negative.

Since input offset error is 0V, very large loop gain, feedback will drive the - input voltage to -235mV to equal the + input.

5uA * R4 produces 235mV across R4.

This voltage across R4 adds to the -235mV present at the - input, and delivers 0V at the amp output.

In virtually all cases, R3 the serves no useful purpose and only adds undesirable thermal noise. Typically, R2 is small compared to R1, so for most cases R1 ~= R4.

I agree that with FET inputs, matching bias source resistances is less of an issue.
 
  • Like
Reactions: DavidT
Thank you for all the valuable input, I have now update taking out R3.

R1=20kR
R2=1.5kR
R3 - eliminated
R4=22kR
R5=1.6kR
C1=1nF
C2=100uF

This would fulfill the requirement for R2 = R4//R5 (closed enough I hope)
R1+R2 very closed to R4
and R2 and C1 form low pass filter at 106kHz

The gain would be -23.4dB, though generally lower than most solid state amp but I this should be fine for DAC input and I'm not looking for high power stuff

I hope I am getting these right..............
 

Attachments

  • NFB Question v2.png
    NFB Question v2.png
    16 KB · Views: 94
These look reasonable to me. It’s conventional practice to add a series blocking cap at the input to remove any DC present in the input signal. Another way of saying this is that the circuit has a gain of unity at DC.
 
Thank you Nelson, I read earlier that it has to be so to minimise any DC offset at the output, I guess I have to read up a bit more again.
DC offset is usually of no consequence in audio as everything is ultimately AC coupled (or should be). However with high impedances and bipolar input opamps you may have to be a bit careful.

Another reason to match the impedance seen by both inputs is with devices with bias current cancellation circuitry as the lowest current noise is only obtainably with equal impedances - however usually adding the relevant resistance just increases the voltage noise much more.

Most modern opamps have low input offset voltages anyway so mismatched impedances is less of an issue than it used to be.

Also keeping resistances low in the 1k to 10k range reduces any offset issue due to bias current as well as being lower noise.
 
These look reasonable to me. It’s conventional practice to add a series blocking cap at the input to remove any DC present in the input signal. Another way of saying this is that the circuit has a gain of unity at DC.

Thanks, I intentionally left out the input DC blocking cap as I think the output from source should have no DC. Also frequently I read that different input caps have different sound characteristics, so I think it will be best if I can make do without one.
On my PCB layout I'll probably leave some space for one just in case (maybe 10uF or smaller should do I think), but will try with and without. Ideally I would not want to use one as much as I can.
 
  • Like
Reactions: BSST