revisiting the good old 386.

[deadastronaut]

hi guys, i was bored whilst waiting for the postie...

and i decided to bread up a nice simple 386 distortion..(reallly quite good actually if you like a metal flavour)

now in all 386 dirts i see they always use the 'input 3'

well ive been using ''input 2''...why?..well as i have it in this setup the input 3
just does not like it at all...a constant drone is there with '3' ...but with '2' its ok..

the usual 1k pot across 1/8 was naff....(with pin 3)

so i moved the gain to a 'pregain' instead, which is much more useful...(pin 2)

but now and then i get a little motorboating noise...

so why the differences in '3' '2' inputs with the gain setups,

should be ok right?

[anotherjim]

Move C1 to the +supply pin of the 386. Should make it much more stable.
Using the - input makes it inverting, so much more stable if there is any accidental feedback from out to in via wiring/breadboard pickup.

[slashandburn]

Hey Rob! I could never find any rhyme nor reason to which input to use.  I usually tend to go in to pin 2 as its ever so slightly easier and tidier to tie pin 3 and 4 together. But yeah, sometimes I find it doesn't like one or the other input and I'd need to swap them over. Not a clue why!

I like your circuit, particularly the pregain. I'm hopeless with transistor circuits. One thought though, why not keep the gain pot on pins 1+8 and have, you know, like, 2 gain pots?

I thought I would've outgrown my 386 amps but they're great for OTT metal tones at low volumes. Basically all I need sorting around the house.

No cap on the output? Is that due to the tone circuit placement?

The motorboating issue might be solved by moving the 9v bypass cap as close to pin 6 as you can get it if you haven't already. Or just adding another one on there, beside the chip. Im guessing here though!

Edit: cheers Jim!

[dschwartz]

I love 386 distortion. It has a very "plexi" like breakup ..
Always wondered why is that.. anyone knows?

[R.G.]

Hidden inside this picture lie all the secrets of the LM386. Well, most of them.

[antonis]

I was stuck on R2/R3/C2 for a while till I regognized the D2 anti-pop arrangement..

Is C7 bypass cap really needed, Rob..??

[deadastronaut]

Cheers guys ,

antonis, not sure really,  just seen it in every other 386 arrangement....

that discrete diagram is interesting, ...

daniel, yes the 386 definitely has a certain bite to it...its nice with the pre as it is,
goes from a tynbe screameresque break up to full on metal....nice,..

more tinkering...



jum, yeah i have the 100uf across pos/neg...the anti pop isnt on it...
i just add that as a matter of course nowdays...

[slashandburn]

I'm sure the pin 7 cap is to filter out noise. iirc I got loads of junk on the output without the cap and the amp was generally more unpredictable.

Its probably power supply related, clean power I'd say the cap is less necessary. A big noisy wall-wart and you might notice the need more?
Don't quote me on that though.

[antonis]

<Hidden inside this picture lie all the secrets of the LM386. Well, most of them.>

Some designs create "mystery"..
(at least to ignorant guys, like me..)

i.e. the reason for using single & Sziklai transistors with 2 diodes on push-pull out instead of using 2 single (or 2 Darligton/Sziklai) with 3 diodes...

[Passaloutre]

What is the sort-of diode-with-a-circle symbol near the top of that schematic?

[EBK]

It's a current source, not a diode.
Most likely mirroring the current flowing down that left hand branch.

Compare it to this diagram of the lm380:

[Passaloutre]

Cool, so it's just a short-hand space-saver on the schematic?

And is it those diodes on the output that cause the clipping character of the 386?

[EBK]

Cool, so it's just a short-hand space-saver on the schematic?

And is it those diodes on the output that cause the clipping character of the 386?

Yes, just a space saver.
No.  Those diodes aren't actually on the output if you look carefully.  They are there to keep the transistors biased properly to minimize crossover distortion.

[dschwartz]

Played a bit with an lt spice model of the 386.
It clips very nasty. One side flat squared, the other clips and then goes to the opposite rail, dividing the wave in 2... Weird!!!

[PRR]

> Is C7 bypass cap {on pin 7} really needed

"Yes".

Look at the gut-shot R.G. helpfully posted. From power rail we find a 15K, 15K, 1.5K (2 parts), and 15K to Out.

This is a combination DC bias and AC gain network.

For *design* AC action, we want 15K on each side of the 1.5K. Keeps things nearly symmetric. Hanging a large cap on pin 7 makes that side act like 15K, not 30K.

For design DC bias we want 30K and 15K, so the output sits near half the supply.

Since the supply is usually crap, we do want to filter crap from coming down the 30K side where it will refect-over to Out at half level. A fat cap on pin 7 cleans this side path. "Supply crap" is not just 100/120Hz wall-power trash, it can be sneakage from other circuits on the same power supply.

Yes, I am 99% sure the LM380 plan is the LM386 plan, except device sizes and the small difference in feedback/bias resistors.

> using single & Sziklai transistors

At that time, "PNP" devices on ICs were really lame. They could not handle more than a few mA, so could not be a Power Output device. They often had quite low hFE. Typically >1, but sometimes not much over 10. So the PNP is mainly a turn-around so the big NPN can work-like a big PNP. And yes, 2 diodes is appropriate, and are actually custom-designed (may even be strapped transistors) to match the output stage.

On the LM380 plan, Q11 is clearly a larger-area device than Q10. Resistors pass only a part-mA to Q10, but Q11 must drive Q7 Base, which will be many mA. If Q11 is 10X or 20X the size of Q10, it works out.

[PRR]

> spice model of the 386.

Never believe SPICE too much.

We know the actual LM386 does not have much nasty, not even super-abrupt clipping. The transistors are much "softer" than discretes. Particularly the lateral PNPs. Which bleed to substrate even in normal operation. Camenzind's book  says "For a lateral PNP transistor the Spice bipolar transistor model alone is woefully inadequate" and has some models for older lateral PNPs.

[pinkjimiphoton]

another reason to use the other input is the phasing of the effect. every stage the signal goes thru it gets flipped in phase.
the old fender amps with reverb had enough stages to put the output out of phase with the input, so the speakes actually would "suck" instead of blow. easy fix we did was just reverse the wires.

but yeah, the phasing aspect also makes it much more stable.,

cheers on another cool circuit rob!!