Fake LM386, how to test?

[soggybag]

I got some LM386 cheap on the internet. They work but I'm wondering if they are real or not. I got the data sheet out and thought I might be able to measure the resistance across some of the pins but the results don't seem right.

I measured the resistance across pins 1 and 8 expecting 1.35k but it measures as a diode.

I tried the same across pins 6 and 7 expecting but I get a diode again.

Is it possible to test this way? Or is there something else going on internally that makes this type of test not work?

[cspar]

There are different Lm386's.

Lm386n1, lm386n3, lm386d, ect...

They're interchangeable but all tuned to different operating points.

It's kind of like transistors having b and c versions that reflect hfe.

Check your cheap ones and look at the datasheet.

[Clint Eastwood]

I have some LM386n1 that are known to work well and did the same measurements out of curiousity. Between pin 1 and 8 I get 1.38k, as expected, but between 6 and 7 it measures like a diode. Interesting. I also would like to know what goes on in there.
But for me, the criterium wether to use them or not is do they work, not are they 'real'.

[anotherjim]

Are they tested in or out of a circuit?
There are usually various "parasitic" PN junctions littered throughout an IC chip that are seldom shown on the internal schematic. When power is applied they become reverse biased and do nothing. With reversed power polarity, the parasitics can even form NPN or PNP transistors or more complex structures like an NPNP SCR or Thyristor. That's why revered power usually destroys the IC. The same equivalent circuit built from discrete parts might just do nothing and survive reverse polarity since there are no parasitic junctions.
You'd expect these parasitics to only test like diodes in one polarity. When reverse biased it ought to read the internal resistor. However, the parasitic junctions might happen both ways so only when power is applied to the IC will they shut off.
Another point is that a DMM applies voltage in resistance/diode tests and without a complete circuit, it injects charge into the IC which may not bleed off for a while without an external circuit path.

[tonyharker]

The trouble with that schematic it is only representative. Different manufacturers use different ways of obtaining the same result.

[PRR]

It's ALL diodes inside.

At Normal Working Condition many of them are reverse-bias and 'dead'. But just poking a meter at it is bound to be confusing.

What is an 'un-real' '386? If it works, it is real for practical purpose.

[soggybag]

Thanks for all of replies.

I got these from Aliexpress along with some other op-amps. I read about people getting fake/counterfeit op-amps. I was wondering if I could confirm if these were real or not.

I plugged a couple into a Smash drive, Ugly Face, and PWM. They seem to work fine in these circuits.

Sounds like its all diodes internally, and the actual internals are more than what's shown in the diagram from the app notes. I'm getting the impression if I powered the chip normally I could test the resistance?

I suppose it doesn't matter much if the chips seem to work. I'm wondering if the real chips sound different. There might be applications where the real chip might be preferred.

For example Ugly Face and PWM don't really care about about the sound they just need some amplification to drive the rest of the circuit.

Smash drive on the other is distorted and we like that but the real chip might behave differently and have a different sound. If I had a confirmed legit LM386 I could compare them.

[antonis]

But they DID work for your Rat, didn't they..??
https://www.diystompboxes.com/smfforum/index.php?topic=129232.0

[ElectricDruid]

I'm getting the impression if I powered the chip normally I could test the resistance?

No, because then the voltage in the circuit would screw up the multimeter's attempt to measure the resistance by feeding a known voltage through the "resistor". You can't do resistance tests with the power on, and they're often difficult in circuit at all.

[antonis]

the multimeter's attempt to measure the resistance by feeding a known voltage through the "resistor"

I should say current instead of voltage..
Current injected -> Voltage drop measured -> Resistance calculated..

[Rob Strand]

I have some LM386n1 that are known to work well and did the same measurements out of curiousity. Between pin 1 and 8 I get 1.38k, as expected, but between 6 and 7 it measures like a diode. Interesting. I also would like to know what goes on in there.

The internal circuits aren't exact.

You can try measuring pin 6 and 7 in both directions and see if you get a diode both ways.  Maybe one way will show a diode and the other way 15k.

Another angle is to power up the device normally, then measure the voltage on pin 7.   Then add a 150k to ground on pin 7, and measure the voltage on pin 7 -  it should be lower than the previous measurement.   The idea is the 150k loads down pin 7 you can use the change in the pin 7 voltage to estimate the 15k resistor value.

The current load on pin 7 due to the 150k is,

           I = V_pin7_with150k / 150k

Then the resistance between pin 6 and pin 7 is,

          Rpin6 to pin7  = (V_pin7_without150k - V_pin7_with150k) / I

You can do all sort of tricks like that to workout what's in the box.

If the voltage on pin 7 is approximately 0.65V below the voltage on pin 6 in both cases you might expect a diode in there.

Instead of using hand calculations you can enter the circuit into spice.    Set-up the standard circuit on the bread-board measure the real voltages then do the simulation and compare the voltages.  To have a good starting point you would want the two sets of voltages to match-up.  Next you add small changes to the original circuit, like adding the 150k in pin 7,  measure the voltages.  Now make the change to the spice and see if the simulations show a similar behaviour.   Unfortunately, it's can be a touch job for newbies to play with the spice models and perhaps even harder to interpret the results.

[Mark Hammer]

There's not many OTHER things a 386 could be, given the rather unique pinout.  Now, they COULD be lower-rated versions, marked as higher rated ones (e.g., a 386-1 marked as a 386-4), but if they work, they work.  And since your application is for a drive circuit rather than amplifying and powering a speaker directly,  I don't think you need to be concerned that you MIGHT be pushing it so hard that it exceeds the amount of current it can put out and go poof.

[ElectricDruid]

the multimeter's attempt to measure the resistance by feeding a known voltage through the "resistor"

I should say current instead of voltage..
Current injected -> Voltage drop measured -> Resistance calculated..

Yeah, true, that's probably closer to how it actually works. The basic principle is still R = V/I, good ol' Ohm's law, and you still can't do it if there's voltage/current in the circuit screwing up your measurements.

[PRR]

> you still can't do it if there's voltage/current in the circuit screwing up your measurements.

You can. We just all forgot how.

[antonis]

> you still can't do it if there's voltage/current in the circuit screwing up your measurements.
You can. We just all forgot how.

We prefer to forget instead of implementing  KCL & KVL with all those Norton & Thevenin equivalents..

[ElectricDruid]

> you still can't do it if there's voltage/current in the circuit screwing up your measurements.
You can. We just all forgot how.

We prefer to forget instead of implementing  KCL & KVL with all those Norton & Thevenin equivalents..

I didn't forget. I never knew that stuff! 

[PRR]

https://worldradiohistory.com/UK/Practical-Electronics/70s/Practical-Electronics-1975-03.pdf
In Circuit Ohmmeter
Magazine page 228
PDF page 46-51