low value feedback resistors on LM3886

[jarthel]

Is there any disadvantage if my feedback resistors are: 75ohm and 1.5Kohm (for a gain of 21).

Thank you

 

[Juergen Knoop]

temperature...power dissipation could reach 0,3W
which increases noise or maybe even cause thermal modulation.
Stick with the usual suspects 1k+22k! Those work fine.
Regards
Jürgen

 

[Sangram]

^^ I'd like to know more about this, if you can help me find some references.

I've used 22/1 and 4.7/220 (on different chips, though) and I'm still to figure out the effects and differences of these arrangements. I was going for lowest impedance to lower the output offset, and it did help. That was the only thing I could measure, really.

Would love to read more, though I'm beginning to get a hint of why this is true.

Thanks.

 

[AndrewT]

^^ I'd like to know more about this, if you can help me find some references.

I've used 22/1 and 4.7/220 (on different chips, though) and I'm still to figure out the effects and differences of these arrangements. I was going for lowest impedance to lower the output offset, and it did help. That was the only thing I could measure, really.

22k/1k0 and 220k/4k7 ?
or
22k/1k0 and 220r/4r7?
or
22k/1k0 and 4k7/220r?

 

[Sangram]

Sorry - 22K/1K and 4K7/220E.

Edit: Both the combinations were used on two separate amps at two separate points in time, the 'regular' one on all the 3886 ans 4780 amps I built, and the low resistance one on a (now exploded) LM4766.

 

[Juergen Knoop]

the usual values are just convenient. Small enough for noise, offset and parasitic effects.
Large enough for thermal effects and if you want to add some capacitors.
I think the LM chips are sweet tempered and one could choose the feedback resistors over a broad range.
regards

 

[jarthel]

temperature...power dissipation could reach 0,3W
which increases noise or maybe even cause thermal modulation.
Stick with the usual suspects 1k+22k! Those work fine.
Regards
Jürgen

what if I used 2W resistors? I suppose I shouldn't be concerned with heat?

 

[Juergen Knoop]

those have larger temperature coefficients, larger resistance tolerances and larger dc related noise.
I don't know, if all these theoretical disadvantages matter in a real world application.
But this is also true for the advantages.
What do you expect from lower value feedback resistors?
Regards
Jürgen

 

[AndrewT]

Hi,
a pair of double value paralleled resistors gets upto 1.2W. One to each side of the PCB helps keep them in the air stream.
One gets 50ppm/Cdegree using standard 600mW metal film.
There are sub 20ppm resistors available without going exotic.

 

[Juergen Knoop]

Hi,
yes, one can fight the disadvantages of lower value resistors.
But what justifies the effort?
Regards
Jürgen

 

[Sangram]

The only thing I can think of is lower offset at the output, if that thing worries you (I was worried). I managed to get it down to 5 mV with 4k7/220 ohm combination. I used 1/8 watt resistors and didn't face any dissipation problems.

Later I have reverted to 22K/1K plus a 47uF FC cap, and find this combination sonically acceptable - plus around 0.3 mV output offset. TBH, I haven't done any AB tests for sound quality, plus the chips are all different.

LM4780 BPA - 22k/1k, no cap
LM4766 - 4k7/220E, no cap
LM3886TF - 22K/1K, plus 47 uF Pana FC