Defective/ fake LM317T

[Maarten]

Originally Posted by Jolly 7

My attempts to build a 1.2 to 37V voltage regulator have been frustrated by what I can only guess is a fake LM317T. It was part of a kit purchased from eBay. The kit did not come with the insulated washer or mica sheet either. Removing the IC has caused damage to the board and it looks like it could be a write off.
I wonder what other members' experience is in relation to LM317Ts purchased from China ?

A picture is always helpful in case of fakes.

[Radio Wrangler]

Setting your pot to too high a resistance value does not park the output at max volts.

The current coming out of the 'adj' pin is what powers the voltage reference and regulation amplifier inside the chip. Rob it of this minimum current, and the reg shuts down. If you're unlucky, it may go silly for a moment.

Note the diode reverse-connected across the reg. If the input gets shorted by some fault elsewhere, this protects the reg from being destroyed by the charge stored in the output cap. That datasheet diagram shows 1N4002 which is fast enough. The board is marked 1N4007 which, I found out a long time ago, isn't always fast enough. The 1N400x diodes aren't just different grades of the same thing. The higher voltage ones are totally different designs and are glacially slow. They aren't PN diodes, they are PIN diodes. They even make lousy rectifiers giving a burst of RF noise every time they turn off. They can dominate the noise of a switch-mode PSU!

Note that your circuit, if the pot slider skips while moving, will give a burst of max output before the reg shuts down. This may destroy whatever you are powering if it can't take the full output voltage.

David

[ajgriff]

It looks to me as if the 'flying pot' might have been wired with the track ends connected to the negative rail and the the slider to the Adj pin. If so this would generate some strange effects. In any event it sounds like the regulator has now fried.

Alan

[Argus25]

Originally Posted by Jolly 7

For some reason the LM317T did not like it, the heatsink got very hot and let off some smoke till no voltage was obtainable. ��

Very suspicious of RF instability. Add the o.1uF caps mentioned previously.

[Biggles]

That definitely sounds like oscillation/instability to me. Does it all quieten down and behave with the preset refitted? Could just be poor pcb design. Personally I have never had a problem with using linear regulators, even when constructed with veroboard. I do always use the decoupling on input and outputs though, maybe 0.1 and something like a 10 microfarad in parallel?
Alan.

[Andrew2]

Another vote for 1uF caps directly on the in and out pins. The only regs I've had real RF instability problems with was the low-dropout 28xx types, which would hoot like Billy-O if not firmly nailed down.
Oh, and the big T0-3 adjustable ones too.

[MrBungle]

LM317 only goes unstable if the dropout voltage and the minimum load isn't respected. It'll quite happily run without caps in most cases believe it or not under load but best to put 1uF across output and 100n on input. It doesn't need any more than that if's DC in and out.

See my earlier post on the things wrong here. Main issue is not enough volts going in for the volts coming out.

[Argus25]

Usually, for most linear regulators a 0.1F 50V monolithic ceramic capacitor will do on the input and output directly to the ground pin of the regulator and if that is not ground you need the same bypass caps over that to to ground.

The minimum spec though for some regulators is 0.22 to 0.33uF. You cannot go wrong of course of you fit 1uF monolithic ceramic capacitors.

Some designs just use 2.2 or 4.7uF Tants, however the quality of these is variable and some have an unsuitably high ESR, especially 35V rated ones, which leads to regulator oscillation if they are the only bypass there.

[MrBungle]

Yes I've given up on tants now. There are decent enough X7R ceramics which are more predictable and fractionally less explodey and a lot cheaper now the capacitor drought appears to be in decline.

[ThePillenwerfer]

I've only ever had one duff LM317 and honestly don't know it's origin as it was in a drawer with ones off e-Bay/AliExpress as well as some from 'blue-chip' suppliers. I certainly couldn't swear I didn't damage it either.

The only things I've had from China which probably were fakes were a lot of ten NE5534s, none of which worked.

[Radio Wrangler]

Be careful of HF decoupling of regs. Capacitors affect stability and some reg designs rely on the ESR of Al or Ta electrolytics to keep them stable. Some chips take off with low loss ceramic capacitors.

Check the datasheet very carefully.

David

[Jolly 7]

Thanks everyone for the replies. The regulator is well and truly fried, but different manufactured and hopefully better ones are on the way from a local seller. Once they arrive, I will connect the monolithic caps as advised and also look to connect a larger transformer, the details of which I will post shortly.

[Jolly 7]

Originally Posted by Maarten

A picture is always helpful in case of fakes.

Here are the two regulators which didn't work. At this stage the one on the right is definitely gone, if you look closely you can see where the smoke escaped from in between the body and the mounting tab. The manufacturer logo markings are also slightly different.

[Argus25]

That one in the photo on the left is probably a fake. If you look at the recess on the side of the epoxy package its shiny and in a recess on the top. But the top surface has a satin look and I think has been through a machine that laser ablated the surface and the original labels and laser engraved new ones. It would have been a bulk re-labeling procedure on some cheap generic TO-220 part. The other one melted its junction with excess dissipation, and if that happened with a low load it must have been RF oscillation.
Be very suspicious of IC's that have that characteristic satin surface. The ones with a gloss surface usually have not been interfered with.

[Argus25]

Originally Posted by Radio Wrangler

Be careful of HF decoupling of regs. Capacitors affect stability and some reg designs rely on the ESR of Al or Ta electrolytics to keep them stable. Some chips take off with low loss ceramic capacitors.

Check the datasheet very carefully.

David

Most snubber networks do work better with a combination of a series resistor and capacitor to improve the damping and avoid tuning up oscillations in inductive links. So some resistance in the capacitor is helpful and without it it could tune up some instability like you say.

However, the exact ESR is a very variable feature with real life Tants and electrolyics, and not necessarily stable over time either. I have had most luck with 0.22uF 50V monolithic ceramic capacitors for the task on analog regulators, at least that value is not too excessive and the ESR probably about right and fairly stable..... most of the time. Probably the ESR will be lower on the 1uF monolithic ceramic cap, I would have to look that up.
With all the generic parts and fakes out there now, one wonders how relevant the original MFR's data sheet really is, so it does pay to hunt around for a genuine part.

[Maarten]

They look like random ST parts that may have been pulled, sanded down and relabeled.

If only one is fake, I agree with Argus25 it's definitely the left one, if only for the reason that it comes from both Morocco and China at the same time... Or else because the laser engraving looks off.

[MrBungle]

Both look fake to me. None of my ST ones (from RS) look like that. I tried to take a photo of some but failed miserably.

On the ST ones the numbering is a different font (more "robotic") and there's a G next to the (e3) bit at the bottom right. There is no exposed metal in the recesses on either side either.

[Electronpusher0]

All this talk of fakes has got me worried.
About a month ago I purchased 5 off LM317 kits from e-bay to keep in stock but I have not tried them yet.
So this morning I built one.
The pcb is very good and has the recommended capacitors and protection diodes.
I admit I tested it from a bench power supply connected across the output of the bridge rectifier so that I could current limit it in case the smoke got out.
I connected a metal clad resistor of 82 ohms as a load.
It works fine!
The voltage varies from 1.24 volts to about 2v below the supply. I tested it up to 25V input.
I attach photos, the LM317 may be fake, in any case I have a supply of quality 317s so will use them when I use the kits in anger.

Peter

[sortedradio]

This is a picture of an old stock, working ST LM317 that I have

[Terry_VK5TM]

R1 at 510 Ohms is more than double the recommended value of 240 Ohms and the pot only needs to be 5k to give a 2 - 30 v range.