LM386 minimum count circuit

Christopher Howard schrieb:

After you having ommitted that 10k pot is there a DC path to ground?

It’s probably oscillating. The 386 has a surprisingly large gain bandwidth iirc. That and the crappy grounding and bypassing you get from those white plastic slabs of misery cause a lot of problems of that sort.

Perfboard with no ground plane is almost as bad.

Dead bug construction is the ticket.

Cheers

Phil Hobbs

Those things should be illegal.

Insect abuse!

Live bug on gold-plated copperclad gives really good grounding.

Lately, when there's no rush, we've been doing 4-layer protos from JLCPCB.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Per fig 03:

With Av=20 and 100mV rms input, you're asking for 5.65Vppk on the output.

RL

There is no external path from the input pin to ground. Is that important? I tried adding some kilo-ohm resistors, trying out various values, from the input pin to ground, but this didn't seem to have any beneficial effects. Of course, pin 4 and pin 2 are grounded, as per the schematic.

I am planning to try out the dead-bug construction, as suggested.

Christopher Howard schrieb:

The 10 k pot would have been that path.

For posterity: after further experimentation, I figured out that the two following adjustments were key to solving the problem:

(1) a 47 uF capacitor from the signal input pin (pin 3) to ground.

(2) a 47 uF capacitor from power supply pin (pin 6) to ground.

I also tightened up the connections some, reducing the length of wires, and trying to bring the ground paths closer and to a common point.

With 9V supply, an input of 100 mV rms gets me output of 1.86 V rms. The signal is beautifully clean at various frequencies I tried from 1 kHz to 20 kHz.

I'm still using the breadboard. I tried deadbug but was having a lot of trouble with connections from the tiny DIP leads to the ground plane and elsewhere. Guess I need more practice. The other components (resistors, etc.) were easier to deal with.

One other possibly helpful thing—when building a circuit that you don’t fully understand, it’s generally best to build it as described and then hack it up afterwards.

Cheers

Phil Hobbs

Good!

Cheers

Phil Hobbs

How do folks breadboard with surface mount devices? Deadbug looks difficult if the bugs have no legs and even simple ICs need a lot of them.

It won't be long before components with lead wires go the way of vaccum tubes, if they haven't already.

Thanks for reading!

bob prohaska

When DIPs and TO-92s really started to go away, I bought a career’s worth. (Of course I was about 60 at the time, so that wasn’t that that many.)

The good thing about dead-bugging with DIPs is that you get at least 8 nice strong standoffs per chip. (When he’s not busy badmouthing physicists, JL likes to bend up the leads of DIPs so he can read the markings, but that weakens them so they break easily.)

You can get chip breakouts cheaply—try out the Bellin ones from Newark.

Cheers

Phil Hobbs

John isn't actually bad-mouthing physicists - he's just complaining that they don't flatter as much as he feels he deserves. This isn't a problem with just physicists alone - few professional groups go in for generous flattery. Donald Trump's staff may be an exception, but they aren't all that professional.

It would probably make more sense to design directly onto printed circuit board. Now that KiCad can let you do it for nothing, and PCB houses can turn a digital file into a printed board in a day or two, it should be entirely practical.

You can use these:

SOIC, SOP etc. packages are pretty straightforward to solder to them, even without using a hot air station. Apply flux, tin the pads, place the component (I like to use tweezers with a curved end), tack two opposite corners with a hand iron with a thin tip sorta like:

in one hand, while keeping the chip from moving by holding it down from the top with the tweezers in the other.

Once it's secure you can kinda "slide" the iron down each side applying a bit more solder along the way to ensure a good joint, and with a bit of practice you'll rarely bridge the pins. Use some magnification as necessary..I'm "young" enough that I can usually get by with prescription safety glasses with +3 magnification for SOICs and stuff.

This doesn't load the input signal?

RL

I barely remember DIPs. But I don't recall leads breaking.

It is annoying that a lot of new parts are BGAs or leadless or just too tiny to work on a Dremeled board, or to hand solder at all.

Those are pretty good. We've made some of our own too, like for US8. Or brerakouts with bypass caps.

Physicists who actually have real jobs are usually fine. The ones I have had bad vibes from were academics, or worked at CERN.

Remember Rick, the co-founder of a laser company? He was a physicist. He was cool. But he was an awful circuit designer. I think Daniel was a physicist too.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Really fine pitch, I flood it all with a big solder blob and then wick most off.

My Mantis magnifier is fabulous.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Donald Trump has made millions of people literally insane.

Cool.

John Larkin Highland Tech Glen Canyon Design Center Lunatic Fringe Electronics

Ok, I get it. Kinda. That still looks a lot harder and more expensive that using components with leads for one-off stuff and beginner's projects. I'm thinking of really, really simple things, for example a demonstation one-transistor amplifier. The sort of stuff people begin "electronicking" with.

When the answer to "What does a transistor do?" becomes "Read the SPICE manual" or "For that you'll have to learn KCad" it seems a bad sign for students and teachers alike.

Thanks for writing,

bob prohaska

And you're actually measure 100mV rms across the 47uF capacitor?

At 1KHz, 47uF looks like < 4ohms.

Audio signal sources normally look to work around a 600 ohm line. The source impedance will normally be lower than the load.

RL