LM386 VB 1/2V (White) Noise

[lars-musik]

LM386 Half Supply VB
In this post I described troubles with my KingOfTone built. It seemed like the VB formed by a single voltage divider for the whole circuit was the culprit. In the first build a reduction of the resistors from 47K to 10K helped, but now I have a second built at hand and it showed some of the  behavior again. So I looked for a different solution (I had the boards fabbed, so adding more voltage dividers isn't really an option) and found RG's hints at using a self-biasing LM386 to provide a stable VB.

I have done this now (using a LM386N-4, because of the voltage doubler and the resuting V+ of up to 17V that wouldn't work with the LM386N1-N3 that have a max. supply of 12V) and tzhe circuit is stable now.

BUT is it possible the LM386 as an VB introduces a huge amount of (white-ish) noise? It redners the pedal nearly impossible to use.

Has someone of you came across that phenomenon? Any ideas what I could do to keep the LM386N-4 as VB supply and de-noise it somehow?




Thanks a lot in advance!

Lars 




PS: WOW, THERE'S AN IMAGE EMBED FUNCTION NOW! GREAT! THANKS! A LOT!

[Plexi]

When I work with LM386, it's usual to get some noise from a lot o power supplys.
I usually use, in this order from 9v/18v in:
1) diode
2) 100R resistor
3) in parallel to ground the usually 100nf and 47uf caps

NEVER hear a noise again.

[anotherjim]

Are the 386 inputs actually floating about unconnected? You know what we'll say about that if they are

[lars-musik]

Aye, Jim, you got me there....I thought that if I use this chip not as an amplifier but only as a self-biasing-thingy I could just get away with floating inputs.... Could I possibly?


And Plexi: I see what you mean. But I really do not use the thing as an amplifier but just as an alternative voltage divider. I am not sure if the same rules apply?

[PRR]

> floating inputs

And amplify ALL the hiss and buzz in the chip and in the area. And inject that to every stage of your circuit.

Unlike many other chips, the inputs of the LM386 can be just grounded.

[FiveseveN]

Just a thought, but it (also) might not like that electrolytic as a load and oscillate at hypersonic freq. Try a much smaller film cap.

[lars-musik]

> floating inputs

Unlike many other chips, the inputs of the LM386 can be just grounded.

All right, I grounded both inputs - still no luck. It still shhhhhhes like before.

Just a thought, but it (also) might not like that electrolytic as a load and oscillate at hypersonic freq. Try a much smaller film cap.

I will try that next. 10n? 100p? How much smaller - or none at all?

[anotherjim]

Start with none. I'm not sure how the 386 fares with high capacitive load. Ordinary op-amp would oscillate, so it's a good thing to try. Usual buffered op-amp reference has the cap on the reference input side? 386 will depend on it's supply pins having plenty of C as the reference is internal where you can't slap a C on it.

[lars-musik]

Start with none.

I took the capacitor out but it didn't help anything either.

So now I went with the standard buffered reference voltage via opamp. And the thing is dead quiet, no hiss, no hum no problems at all.

The only thing I wonder is: Why am I the only one with these problems? There must be a million copies out there follwing the same schematic. The two 47K voltage divider for the two circuits just can't handle it. So is everybody else just living with the volume drops?


So thanks again all for your input! Problem solved. Next PCB run (if there will be one) will have another opamp on it.

Lars

[duck_arse]

hmmm, the datasheet circuits show the pin 7 (bypass) w/ a 10uF to ground. I wonder if that might help?

[anotherjim]

If you're asking "why me", then you might have a noise issue (radiated or power cable borne) in your place that either needs dealing with or at least knowing about. If there is, then the plus is that in making stuff that can cope with it, you are making very good stuff indeed.

[lars-musik]

If you're asking "why me", then you might have a noise issue (radiated or power cable borne) in your place that either needs dealing with or at least knowing about. If there is, then the plus is that in making stuff that can cope with it, you are making very good stuff indeed.

Hi Jim,

the "why me" adressed the issue of a possibly under-dimensioned reference voltage (currentwise) in the King of Tone. The noise issue is another thing and arose with the usage of an LM386 as an alternative bias reference voltage generator, because the 47K-47K voltage divider proclaimed in all schematics I found on the web resulted in an instability in the VB rail, leading to tone drop-outs.

I based my builds on the madbean schematic of the "Aristocrat" (that has been removed from all websites for unknown reasons) but the Rullywow "Queen of Bones" is basically the same. So there should be lots of builds with this problem. I was just thinking aloud why nobody reported this before?