Guitar practice amp TIP31/32 w/ lm386

[Clint Eastwood]

The noise probably comes from the switch mode power supply. I use a laptop brick a lot for my projects, and they are perfect for class A amplifiers wich draw a lot of current. But with class AB amps I have ran into the same noise problem as you, and the cause is the low current draw at idle. The easiest solution is to make the power supply deliver a certain amount of current all the time by, for example, inserting a 100 ohm 10 watt resistor between + and -.
You could also of course make a low pass filter, the most elegant I guess would be an LC filter.

Good luck!

[Matthew Sanford]

Ok, sounds good. I think all my resistors are 1/4W so I'll need to get some to handle that, or an LPF instead. I'm moving it on my breadboard currently as the 1875 was kind of loose ish, and should be able to mock up an LPF for now...targeting something like 10KHz? And placed on the output before the 2200u BFC?

[Matthew Sanford]

I reduced it for now by twisting the power and the inputs and adding a 330R on the input to the 1M log volume pot, now it's only there with pot turned all the way up, and pretty quite even then. I don't have much of a heat sink on it, but I'm going to veroboard it and call it functional, then go back to it later.

One question: do you think the ferrite on the power cord could cause any of it? I figure it calms it and smooths issues with a supply like that but just checking.

[Clint Eastwood]

And placed on the output before the 2200u BFC?

No, I mean a filter at your dc input. I am not an expert on this at all, and SMPS's can be a pain to get quiet, a search on the net will tell you. The best solution seems to be an LC filter. If you don't have a power resistor for the solution I mentioned before, you could try the LC filter option, but then you need a suitable choke.

[Matthew Sanford]

Ok. I'll look into LC DC power filters, the more you know...but may just box it up for now, and come back to it with better understanding later. The noise isn't too bad, and a little extra resistance (turning the pot down) clears it out completely.

Thank to Clint (et al) for all the advice, it's invaluable, and one day I'll try designing a discrete AB amp, but fast track works for now

[Matthew Sanford]

Quick update (and cautionary tale/reminder): so excited for my soldering station, I soldered all the outboard stuff for the 1875 data sheet single supply circuit (worked great on the breadboard) and put sockets to plug the speaker in, fired it up and...pop! Bye-bye 1875. I'd gone over my layout a bunch and knew it to be right, was going to post the pics for more eyes, but realized I'd put the other side of the speaker not to ground but the other side of the output BFC. I'm assuming it went to the feedback and that's what killed the IC?

For removing multiple pin ICs, is it best to use a blade tip and try to heat them all at once?

Also my Aunt told me when working electronics (60s or 70s) and organizing diodes and resistors hanging them on a piece of paper, then shaking them to make them dance, they were Diana Diode and the Resistor Sisters...sounds like a good electro band!

Pics for posterity

[PRR]

I'd put the other side of the speaker not to ground but the other side of the output BFC. I'm assuming it went to the feedback and that's what killed the IC?

Other side of this life... so, like this?


That won't hurt anything. Won't play, but won't hurt.

What makes you think it is dead? If you did kill it, you musta done something cleverer than this.

[Matthew Sanford]

Yes, just like that. I only know as I opened it expecting a cap popped, I've become familiar with the sound a couple times, but it was the lm1875 that was deformed. I should note the power supply I used was different than on the breadboard, though both 18v laptop style power cords...this one's connector fit nicely in a boss power Jack, with ground on sleeve, so I used it. I did wire two 9v In series with ground to the battery pin and power to power, both to the jack, but no batteries in. That'd be correct wiring though right?

I'll go over the layout and see, must be something there...possibly in the 22k coming off the power in pins trace.

[ElectricDruid]

For removing multiple pin ICs, is it best to use a blade tip and try to heat them all at once?

The best way I've found without having specific tools for the job is to cut the body of the chip out by cutting all the legs close to the body with wirecutters. Then you have a lot of individual legs sticking out of the board, which is lots of little problems instead of one big problem! You can then turn the board upside down, heat each one, and with a bit of knocking and poking, they should fall out.

HTH

[Matthew Sanford]

K, makes sense, or tweezers to pull each individually. I'd had issues trying to pull a good 7805, but this is a bad chip now, so chop the legs it is.

[Matthew Sanford]

Poor lil legless LM1875...




So reading all over I'd seen if you remove the reinstall the speaker the BFC will discharge and fry the chip...no ground on the - side but maybe the ac signal feedback through the 25ohm speaker made it discharge....

Current conundrum, no sound....except when using the DMM to check pin 4 Dc, get some crackle, though it may have tapped the pin 5 v+. I've checked continuity and did reflow solder (mix of no lead and 60/40 Xtronic gave).

Pin voltages:
1 (+in) 9.68
2 (-in) climbs to 17.2
3 (g)
4 (out) 19.56
5 (v+) 19.62

I'm guessing pin 2 should match the pin 1 voltage, but here it's almost the full supply. I've looked over my part placement (too cramped!)a bunch so I'm sure it's not going to pop out to me, here's side pics if it helps. Any ideas what I should focus on here?










Tomorrow I'll go at it with the audio probe, got a kid's senior choir concert tonight

[Matthew Sanford]

So audio probe shows it gets to pin 1, not on 2 or 4. I am guessing it is a bad layout issue, I crammed it all in and so better to have separate ground traces that jump to join at pin 3? Currently it goes chip -> power caps -> bias divider resistor/cap -> power ground -> output zones cap ->input resistor -> speaker ground. Reading the data sheet again (more fully) I feel like this layout is causing power issues, nearly doubling pin 2 voltage compared to pin 1, etc.


Am I on the right track?

[Matthew Sanford]

I had issues with the LM1875 vero layout not working right and it sat to the side of my workstation. I decided to get back to it and make the original idea work by thinking through what Rob was telling me, and came up with something to make it work! Volume goes 0-50%-100% switch off gain then 50%-0, nothing much fancy but open and low up to 15th on high E.

On the breadboard it was more clean through the volume increase, breaking up towards the top end, with the gain switching to 200 on the LM386 by placing the cap in. Once on vero it was doing a put-put motorboat till I moved the TIP ground to the ground near the decoupling part instead of through the LM386...so I need to work on better grounding ways. That gave me oscillations till I put a 2200u to ground from the TIP power and a 100u from the LM386 power. Now all boxed up I get this, which makes me happy!

So this is the scheme I went with, which needs revisions but I have a couple more waiting for circuits so this works for now... plus these are findable on eBay (like this cheapest one) I'm thinking I should have a 2.2M to ground on the input, might bring more treble?...for simplest.



More to fix would be to limit the input signal on the LM386 and put something inbetween it and the push-pull, maybe a phase splitting 2n2222 or a darlington or something. I'm sure my value choices here could be better even as is too. Plus add a 78L09 for the LM386 (I do like how it distorts!) and use 18v for the rest.

Odd thing with this is for the gain on the LM386 I soldered the 10u from pin 1 through the spst to pin 8, but for the video I have it switched on in the beginning then put it off at the end and it kicks up the distortion...I just realized it's pin 8 I have on the center, so it goes to cap->pin 1 or hangs with pin 1 -> cap blocked in the switch.

I went with this for my layout, put the push/pull portion a row low but did the original cuts before noticing so new layout drawing (this) with new cuts and jumpers.



Kind of drew in red lines for the new ground routing from the push/pull and the lm386 on my phone and the cut so I wouldn't forget when doing it. I made a foolish decision initially to just run the 9v through the push/pull to the LM386 and ground through the LM386 to the push/pull. I have been a slow learner on ground and power routing, very likely the reason my LM1875 didn't work out so well. So for this one, I tied the ground after the power decoupling stage to another row, then speaker ground, then push/pull ground, then all the grounds from the LM386 stuff and input. I figure that is as close as I am getting this to a star ground - but for it, that all should return on the other side of the power decoupling caps, or is it better to run them directly to the jack side of it?

Last thing I'm thinking on are the "Talk" and "Listen" 12 pin momentary switches. They run to different wire connector screws, for different portions of the house intercom system I'm sure, but I'm thinking they should go to flip flops to turn on/off different effects, if even only a DRM or something, or PT verb, other distortion choices, bassballs, hmmm...

Anyway here is the vero, before wires and mini heatsinks, with everything including the caps stuck in to pin sockets, and the backside.