Three-resistor LM3886 chipamp

[Mick_F]

I have just finished building a very minimalistic LM3886 based chipamp, which has just three resistors in the signal path, nothing else. From the design point of view it is nothing spectacularly new, but I have spent some thoughts building it very compact and avoiding closed loops. I have posted a detailed and illustrated building report on my website, which may be very instructive especially for beginners.

Here is the schematic and some pictures giving you a first impression. (As usual, I constructed the circuit on a jig and transfer it to a case after extensive testing).

The amp works fine, DC offset is 30 mV for each channel and it sounds very good.

Note: Take care if you build this amp, it has no input capacitors and no zoblel

Enjoy,
Mick

[davidlzimmer]

I've made two pairs of the exact circuit. Very easy to P2P. I agree they sound very good, even with cheap components.

Note: Your schematic does not show the jumper from pins 1 to 5.

[Mick_F]

David, I dont have a jumper from pin 1 to 5. Pin 5 is disconnected. I dont think it is necessary, do you?
I have a jumper from pin 4 (V-) to pin 11 (nc) in order to get a more symmetrical pinout, but this is indeed not in the schematic.

Mick

PS. I forgot to give the component values in my previous post.

Rg 22k
Rf 22k
Ri 680R
Rm 10k
Cs 2200uF || 100nF

[davidlzimmer]

Ya know, I never thought about it.

I'll try it next time. Thanks.

[davidlzimmer]

I wonder why National shows 1 and 5 jumpered?

[alitaido]

I noticed that you have no Cap to ground in your NFB. What does this mean for your low pole roll-off, since 1/(2*pi*Ri*Ci) equals the -3db pole. If Ci is only the capacitance inherent in the wires (and that would very low) then wouldn't the pole frequency be ridiculously high?

[Mick_F]

Originally posted by alitaido
I noticed that you have no Cap to ground in your NFB. What does this mean for your low pole roll-off, since 1/(2*pi*Ri*Ci) equals the -3db pole. If Ci is only the capacitance inherent in the wires (and that would very low) then wouldn't the pole frequency be ridiculously high?

That capacitor is optional and I left it out as I wanted no capacity at all in the signal path. Note that the cap, if it is there, forms a low-pass filter with the feedback resistor. Without the cap, Ci=0, which means that the cutoff frequency goes to infinity.

Mick

[traw]

curious if anyone can chime in as to any problems encountered leaving the Ci out... i etched boards once with an oops where i had to leave it out (didn't want to dremel up the board). seems i've had better results with no Ci or setting it to 33 or 47 uf rather than the datasheet recommended 22 in concert with 1k (instead of 680). datasheet mentions to prevent gain at dc and just wondering when that comes up or what side effects/glitches would be present

[alitaido]

Without the cap, Ci=0, which means that the cutoff frequency goes to infinity.

Yes, that is what I thought. I'm curious, how does this effect DC stability and the overall sonic quality? Wouldn't the DC offset at the output increase (if there were any at the input, that is) ?

[Mick_F]

Well, you should only use this design if you are sure that you dont have any DC at the input. The previous stage should take care of that.

As for the sound quality, I am sure that less capacity in the signal path increases sound quality. Moreover, if you use that cap, the gain is frequency dependent in the low range, which I dont find very appealing.

Afaik this cap has no effect on stability. If stability is an issue, consider a Zobel or caps between V+ and V- and/or in+ and in-, respectively (see datasheet).

Mick