Another 'Fake' Part Found?

[MrAl]

Hello there,

This is sort of a review, but more like a report on a product being sold in different sites on the web. It's a small blue board buck converter supposedly with a part LM2596.

The part is definitely stamped "LM2596", but there is a somewhat older part LM2576 that comes in the same package with the same exact pinout, so if you laid them side by side without seeing the markings, they would look both identical in the TO263S package.
This means that if one is marked wrong, it will work similar to the other so it will be impossible to tell if it is marked wrong without doing some measurements.

Someone on another site suggested measuring the switching frequency, because one is 150kHz and the other is 50kHz (or near those). Measuring the circuit, it looked like 57 to 57kHz but there looked like there could be other switching points in the main cycle. The waveform however showed that the output cap must have had too much ESR so it was difficult to be sure. However, recently i connected a 1000uf cap wth low ESR to the output, and it cleaned up the waveform, and guess what, it was 57kHz to 58kHz but no higher.

This indicates that the real part on the board is really a LM2576 not a LM2596. I dont see how it could be any other way because the switching frequencies are too different to be confused.

Also, i wondered all along why there was no compensation capacitor on the board, which is recommended on the LM2596 data sheet. The LM2576 however has no compensation capacitor, so that explains that too.

I'll take a picture of the board if anyone is interested, but it's that small blue board about 1 inch wide and 3 or 4 inches long with a trim pot on top and a SM part marked LM2596 and some caps and diodes. It's a buck converter.

It works, but it would be better if it was the real LM2596 and after all it is being sold as that kind of regulator board, and has that part marked as an LM2596.

This means there could be some LM2576 parts marked as LM2596 parts out there too.

 

[ronsimpson]

Where did you get the board from?

 

[MrAl]

Hi,

I got my boards from Amazon, a seller called "Rio Rand", but i have known this seller before and have gotten other products that were quite nice and worked well. Others got theirs from eBay.
The product (board) does work, but it's not the chip they say it is which means it's not as good because of the 3x lower frequency. The LM2596 is a somewhat newer part while the LM2576 is older. I think the efficiencies are similar, it's just the switching frequency difference i think.

 

[debe]

Is this what you purchased? Iv bought about 12 of them & they were cheaper than buying 3 term linear regs. Don't know if they are genuine chips but don't realy care as they work fine in my aplications. Purchaed on Ebay from the cheapest supplier in China.

 

[Willen]

Haha funny, little off topic- if there's a word 'fake', there I found another word 'China'! Cannot government controll such tradition?

 

[MrAl]

Hello,

debe:
Yes that is the board, but it is slightly different than mine, and there are at least three or four different versions of that same board. Mine is a little longer for example, but has the same part values: 33uH, 220uf caps.
Any way you can check the switching frequency on your boards? That's the way to tell if they are LLM2596 or the older LM2576 chips. I dont see the compensation capacitor on your board either, so it could be a fake chip too.
But yes, i did say that they do work, even if a fake chip, and are still useful. The main difference is that the frequency is lower for the older chip so the output ripple will be higher for the older chip and lower for the newer chip, them both having the same main components (33uH, 220uf caps).

There is also another problem however. See that Schottky diode? That one is rated for 3.0 amps while some of the boards have a diode rated for 5.0 amps. The data sheet for the chip recommends using a diode rated for MORE than 3 amps because of the peak currents when the output average current is 3 amps. This means you should not really use it at the full 3 amps. For me this doesnt matter because i dont need it to work at 3 amps, but for some people it might matter and the diode might overheat.

But one more problem on all of the boards seems to be the heat sink area. The heat sink does not look big enough to handle the full 3 amps output at 50 percent duty cycle, or possibly other duty cycles too.
Thus i would only use the board up to 2 amps output to be safe. I think the heat sink area is good enough for that, if just barely.

If you could measure your switching frequency with a scope that would be great. We can then compare.

 

[schmitt trigger]

In addition to what you said....
An aluminum electrolytic capacitor is next to useless at frequencies above 100 khz. Its ESR and ESL is simply too large.
One way around this shortcoming is by "brute force", paralleling many electrolytic capacitors.

But the better solution is either a tantalum cap, or better yet MLCC caps.

So yes, this device is indeed fake or poorly engineered.

 

[debe]

In some aplications I also felt the heat sinking wasn't quite good enough, so I cut up a heat sink & glued it to the top of the chip with Arctic Silver epoxy 99% silver. Seems to work ok. Will get around to checking the freq & report back.

 

[MrAl]

In some aplications I also felt the heat sinking wasn't quite good enough, so I cut up a heat sink & glued it to the top of the chip with Arctic Silver epoxy 99% silver. Seems to work ok. Will get around to checking the freq & report back.View attachment 90678

Hi,

Wow interesting idea with the heat sink. Did you check the before and after temperature?
I ask because the tab is on the bottom on these SM chips, so a heat sink on the bottom would be very good.

Thanks for checking the frequency when you can too, and i hope soon
Since this thread i have now read other reports of other fake LM2596 chips too, that other people purchased. Apparently it's a bigger problem then i thought at first. There are also other variations of the 'fakedness', because one user reports that the HV version blows out at about 45 volts, when it should make it up to 60 volts. A real HV version was tested up to 63 volts at least for a short time with no problem, but the fake version blows out immediately once it gets an input that is around 45 volts, which makes it a non HV version (HV stands for high voltage here, which means 60v for this chip).

 

[debe]

Don't know how to read the freq. These are no load & a small load waveforms on P2. CRO set at 2us per div horz, & .1v per div vert with 10-1 probe set on 10. The other pic is the same chip adjustable with a voltmeter & the wave form are the same.

 

[MrAl]

Hi,

Very interesting pics, thanks.

Yours appears to be operating at the correct frequency near 150kHz.

Any chance you can take one more scope pic of the output voltage maybe on the AC scale?
That would show us the ripple.

Thanks a lot for taking those pics.

 

[debe]

Will look in to that tomorrow here. Bear in mind I purchased the cheapest I could find from China & Hongkong. $1.00 each Aust for the ones with out the voltmeter. The ones with volt meters were $3.35 each Aus & that was free freight.

 

[debe]

First picture is no load & 5mv/div. Second is loaded about 300ma 20mv/div. Third is with a 1000uf cap across the rail, 5mv / div. The probe is set at x1.

 

[debe]

Did have a little problem with interference using one as a powersupply for a VHF tranciever, fitted inside a battery pack to run of 12v dc. By fitting a couple of toroid rings & capacitor it fixed the interference. The tranciever runs on 8v.

 

[MrAl]

Hello again,

Yes yours appear to run at the right frequency.
The ringing could be what is interfering with the radio. That's a higher frequency. You could try a snubber on the output of the chip which might help.

 

[dr pepper]

Yes that ringing looks really bad.