Some questions on GC design

This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I am currently collecting information for building a Gainclone amp. I have a couple of questions left over, for which I was not able to find satisfactory answers. Maybe you can help me with some of these issues.

1) LM3875 with 4 Ohm speakers: I want to use the CG to drive 4 Ohm speakers and I have read that some of you are doing this successfully. On the other hand, the datasheet describes the LM3875 as an amp for 8 ohm speakers and on page 9 of the datasheet a graph shows the output power as a function of the load impedance. According to this graph, the output power rapidly tends to zero for loads smaller than 8 ohms.
Can you help me with this apparent contradiction and is there any means to adapt the LM3875 to 4 ohm speakers? Or should I resort to the LM3886 for my 4 Ohm speakers?

2) I have noted that for inverted usually a DC-blocking input capacitor is used while this is not the case for non-inverting designs. Why is this so?

3) According to the National data sheet the local bypass condensators of the power supply should be not too large but they do not provide any numbers. In the amp designs I have found here and elsewhere on the web, usually values of 1000 to 1500uF are used. Compared to condensator values usually used in PSUs this does not seem a particularly small value. On some of the PSU designs presented, the caps used are even smaller since there seems to be a consensus that, unless you use a snubber to decrease the output impedance, a PSU with too much capacitance affects the sound negatively . Can you comment on this?

Thanks for your help.

If you are still choosing components, then go for the LM3886 and be sure that you can drive those 4 ohm speakers! The LM3875 will drive 4 ohm speakers but not as well.

If you have DC on the input of the GC, you must use a DC blocker. AFAIK that applies to both inverting and non-inverting types.

As regards your last question Mick, it is a subject that has been discussed at length here on this forum. If you want the best sound, then stick to smaller caps, ie 1000-1500 uF per pin, or use one of the snubberized supplies. ;)
snubber and small caps

Hi slightly other thread/topic, I know, but tossing a question into that 35 page thread is daunting...:(

What I don't quite get is that in the whole Unreg snubber PSU thread, you leave the 1000 - 2200uf caps from V+/V- to GND and use the 10000 uf, with parallel 100nF, makes a lot of sense... if you follow the argument. One problem I have is that in Carlosfm's 3886 amplifier stage version 3 (which both of you stated was for use with snubber PSU...), you still have the 2200uf with 100nf in parallel from V-... only with a lot of what I guess is bypassing? circuitry added... what gives?:xeye: If I fit the 100nf's, do I chuck the 2200uf's?

And another question: I have a mono (300VA 25-0-25 tfm) power supply for the six LM3886's, but with six pairs of (10000uf) caps feeding them all. And 2200uf at the chips, no 100nf though. Surely my Impedance and Inductance and ripple must be very low after this lot? I think the sound is cool, no harshness that I can hear, but why be satisfied...

P.S. Nuuk love your site, is good for touching base...

Hi, sorry for garbled thread... it just indicates a garbled head... :cannotbe:
In the pictures edits of carlos' schematics I try to show what I think... of output stage...


  • ithinkb.jpg
    45.5 KB · Views: 291
And here is what I think are the snubbers, I gather they are completely different things even though they sit right next to each other, should be as close to the pins as possible and have more or less the same layout, some different values... :xeye: The 4700uf and big R is a cheaper Inductor or filter... u could use a big ferrite core coil, to do the same... would be faster that way...


  • ithink.jpg
    50.4 KB · Views: 264
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.