LM2575S weird frequency

[mrkev]

Hi everybody,

I have a problem and someone here might have some answer, so - in one of our company designs, there is an old voltage regulator with LM2575S (design is about 10yo and it was in production for that time, working absolutely fine). Recently, when we connected EM4095 (RFID reader), entire regulator went into oscillations (there is about 2kHz and 500mV ripple on the output voltage).

I discovered, that old LM2575 we used had about 55kHz internal osc. freq., but new batch has 64kHz. And since the RFID chip works at almost exactly second harmonic (125kHz), my theory is that this causes unstable feedback. Regulator works fine with resistive loads and RFID has no problem working with old 55kHz regulator.

Does anyone know, if they changed frequency of the internal oscillator for LM2575S? I can't even find datasheet that would say anything about frequency above 60kHz. I am going to try some things (probably notch filter), but so far, just adding capacity seems to make it worse.

[stj]

do a forum search, the chinese are re-labelling some of those to use slightly different parts

[schmitt trigger]

We have suffered some counterfeit issues with similar National Semi "Simple Switcher" devices.

My theory is that these older products have become commodities and for that reason Texas Instruments (who bought National) no longer promotes them.
Therefore some far-east "entrepreneurs" have filled up the void because there is still a demand for the components.

Your new batch, what datecode it is? Does it have a TI or National logo? If the latter, can you compare the logo and/or the labeling with older components that do work?

[mrkev]

I had the same suspicion, but I am not sure. We buy them from some other wholesale reseller and they claim that they got them from the same supplier as before.

I also found out, that RFID has nothing to do with this problem, it just makes it manifest earlier. If I use load close to 1A, it's ok. But when I generate some random pulses (by switching load on and off), it gets into this oscillating mode and you can measure 2,15kHz at about 0,5Vpp on the supposedly regulated output.

Your new batch, what datecode it is? Does it have a TI or National logo? If the latter, can you compare the logo and/or the labeling with older components that do work?

They are branded National Semi, with 64RA3YUE3 code (I don't know what is date there). Old I found looked like they have thick printed label and those new ones have shallow laser edged one. Logo and everything else looks fine, but I don't know...

[schmitt trigger]

Do you have the actual reels?  There is a lot more information on the reel labels. Including some straight datecodes.
Assuming of course, that the reels in stock are similar to the your suspect units.

According to Wikipedia, Texas Instruments acquired National Semi on September 23, 2011. I would assume that datecodes from 2012 onward would for sure show Texas Instruments on the reels.

[exmadscientist]

My theory is that these older products have become commodities and for that reason Texas Instruments (who bought National) no longer promotes them.
Therefore some far-east "entrepreneurs" have filled up the void because there is still a demand for the components.

Simple Switchers are popular to counterfeit because they're popular, low-tech, high-margin products. NSC and now TI pushes them heavily because they're quite expensive for what you get. There are many older designs locked in to them and newer designs done by less experienced SMPS engineers that call for these devices. Both groups will pay a premium for Simple Switchers; TI knows this and prices them accordingly. Counterfeiters see popularity and high margin and jump in.

Your new batch, what datecode it is? Does it have a TI or National logo? If the latter, can you compare the logo and/or the labeling with older components that do work?

For what it's worth, TI still marks brand new parts with the NSC logo. I'm not sure if it's internal or external change control they don't want to bother with, but it does mean the logo doesn't tell you much. I do agree that they've probably changed the reel labels, though I never get to see those so I can't be certain there.

Posting pictures of the "new" and "old" devices might be useful, if you can get some macro shots.

[mrkev]

Do you have the actual reels?  There is a lot more information on the reel labels. Including some straight datecodes.
Assuming of course, that the reels in stock are similar to the your suspect units.

Sorry, but we make just small batches of several hundreds. We bought only 150 of these problematic devices, so I don't have whole rail.

Posting pictures of the "new" and "old" devices might be useful, if you can get some macro shots.

Unfortunatelly, I don't have good macro on my phone and anything alse around. The best I could manage is this: