Since I'm still working on the preamp for my composite LM3886 + LME49710 amplifier (thread here), I ordered a preamp using the venerable NE5532 thinking it will be good enough to test things.
I was surprised to see how noisy it was (hiss) and how much it distorted the signal - smth that looked like crossover distortion and slew rate limiting.
I changed all the resistors, reversed engineered the schematic and made a new one but the problems were still there.
Then I finally tested the opamps and realized they were fake: a slew rate of around 0.4V/uS and unable to drive even a 2KOhm load...
I made a video about this here: YouTube
From what I could test, these were actually LM358's.
I contacted the preamp manufacturer, I'm curious what they will say...
I was surprised to see how noisy it was (hiss) and how much it distorted the signal - smth that looked like crossover distortion and slew rate limiting.
I changed all the resistors, reversed engineered the schematic and made a new one but the problems were still there.
Then I finally tested the opamps and realized they were fake: a slew rate of around 0.4V/uS and unable to drive even a 2KOhm load...
I made a video about this here: YouTube
From what I could test, these were actually LM358's.
I contacted the preamp manufacturer, I'm curious what they will say...
There's not a whole lot of money to be made by faking 5532s. You'd have to sell thousands to make it worthwhile at all.
I guess if they go through this effort they plan to sell more than thousands.
Still, despicable practice.
I'm also curious why manufacturers don't do some random tests for the chips they buy. That would seem sensible, especially knowing that fake chips are very common.
The chip is decoupled with 100nF caps between the rails and to ground.
I tested decoupling with bigger caps as well and no difference. Not in the video though, apparently it's hard to focus and film stuff
Thanks for the advice!
I, too, find it surprising that someone would be counterfeiting one of the cheapest audio op-amps in the world. I wouldn't think the ROI would be that good.
IF it were counterfeit, I'm not sure how you could conclude by measurement that it's specifically an LM358. The only way to tell what it is is to carefully open one up and look for the identifying labeling and logo on the die.
While you're at it, just order a few new NE5532 units from a reputable distributor, or directly from TI's web site, pop them into your board and see if the performance changes.
IF it were counterfeit, I'm not sure how you could conclude by measurement that it's specifically an LM358. The only way to tell what it is is to carefully open one up and look for the identifying labeling and logo on the die.
While you're at it, just order a few new NE5532 units from a reputable distributor, or directly from TI's web site, pop them into your board and see if the performance changes.
Already ordered, they are on their way.
Regarding the conclusion that they are LM358's, there are other tests that I didn't show on video: the distortion when loaded is identical (on the oscilloscope) and there is another weird behavior: they both show very bad (and identical) cross-over distortion when using any gain >= 2 and >10KHz and more than 3Vpp input signal.
However you're right, opening up the chip is the only way to be sure.
Fakes can be factory rejects, so every one makes profit. Factory reject 358's packaged as 5532's would wreak havoc, the 358 hasn't enough slew rate for audio at all even when upto spec!
Most production lines weed out the completely broken dies at die level, but those test machines only test each die for milliseconds, so cannot measure things like LF noise or do full functional tests, they are just to avoid the cost of packaging dud dies.
Second level testing after packaging can be more thorough, typically also combined with marking/binning of parts by performance level. Its the step after these tests and before package marking I think where the unscrupulous step in. Rejects presumably are supposed to be ground up and disposed of, presumably at a cost.
And yes the reject bin will have 1000's, or 100,000's of devices after a run for common chips. 12" wafers have a lot of dies on them, and you'd do a whole stack of wafers in one run for volume parts.
Last edited:
The crossover distortion of the LM324 is so distinct that I would have no doubt.
LM358 is just a dual of the quad LM324.
324/358 die was specifically designed to be CHEAP, a utility industrial part. Even at current prices of '5532, '358 is cheaper.
Part-marking machinery is very-very common in asian factories. Any semi-responsible worker can stay late and re-mark thousands of chips in an evening.
LM358 is just a dual of the quad LM324.
324/358 die was specifically designed to be CHEAP, a utility industrial part. Even at current prices of '5532, '358 is cheaper.
Part-marking machinery is very-very common in asian factories. Any semi-responsible worker can stay late and re-mark thousands of chips in an evening.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.