Junk Parts Gainclone Build Thread (BPA Chipamp)

[benb]

This is a neat project, I just wanted to make some comments on measurements and stuff, I hope they're not too obvious!

I haven't been keeping up with all the changes in audio electronics in the last decade or more (I DID used to read AudioXpress/Audio Amateur regularly), so I hadn't heard of paralleling chip amps for more power until getting on DIYaudio a few months ago. This appears to be quite a popular technique.

I was astonished with while paralleling 3 lm3886's (I built a bpa300.) How close the resistors should be matched for good operation. The output 0.22ohm resistor doesn't really matter that much but the other do a lot!
I did use 1% resistors but did not test them. After about a month of use the one chip went up in smoke and put -35V on the output.

After replacing the broken one I measured the temperature one each chip while putting it under high load just to see one chip shooting up to 55C while the other two stayed at 35C.

While temperature is certainly important, I'm thinking the best way to check imbalance between the amps would be by measuring the current through (by measuring the voltage across) each output resistor. This would be both a DC measurement for quiescent current and difference between DC offsets, and AC (with a sine wave input) for checking gain balance. The (obvious?) idea comes to mind of of putting trimmers on two of the chip amps to tweak offset and gain.

I now have expensive 0.1% resistors in my parallel circuit and all stay within 5C from each other even at high load.

oh yes

+10 for the radiohead album lgreen 🙂

If you've got a quantity (say, 20 or more) of the needed value of "regular" 1 percent (or even 5 percent) resistors, a DMM and a few minutes of time, you can easily measure and sort them to get two or three resistors within 0.1 percent of one another. Even a "3 1/2 digit" DMM may be adequate, as you're looking at precision (the relative difference in values) and not exact accuracy, but it may be marginal. At this level you might need a more accurate meter. My BK 878 displays a full four digits, and so is good for 0.1 percent precision, though a "true" high-accuracy meter would be even better.

A methodical way I've used is to number each resistor (or put them in a line on a breadboard/plug-in protoboard), enter each measurement in a spreadsheet vertically next to a number series column, sort both columns based on the column of measurements, then you have the number sequence of resistors in increasing value. Pick the three with the closest values, compare the values, and they're almost surely within 0.1 percent. If not, repeat with a larger batch of resistors.

 

[tangmonster]

lgreen

I used the bpa300 circuit diagram. Voltage offset is around 2mv at the output. I can set each indivitual chip's dc output to less than 2mv

DIY BPA300 6x LM3886 300W audio Amplifier

thanks benb

I agree with your measurements idea ,but after that chip blew up and nearly took my speaker with i was kind of fed up and opt for the expensive easy way out to solve the problem. 😉

 

[jimbo1968]

If anyone fancies doing a similar project, I've got 2x chipamp lm4780 kits for sale. See swap meet. I'm taking offers.

Maybe use transformer for bridging like tangmonster BPAmk2 project.