LM317 pins short circuited

[Xenon Photon]

The LM317 is driving me crazy!  I do the same circuit I did before but it blows up the potentiometer when it gets near the minimum voltage range. After disconnecting the LM317 found that output pin is directly shorted to the input pin of the IC!!! I tried with different circuit configurations and destroyed a lot of ICs.

some data and photos:
Supply voltage: 26V DC (From 18-0-18 50VA transformer and regulated with a bridge rectifier and a 4700uF cap)
C1: 0.1uF.
C2: 10uF, one try with a 47uF resulted in instant destruction of the IC without moving the pot form its middle range at start!
C3: 1uF, 10uF.
Values of R1 and R2 tried: 1k & 10K, 280 ohm & 5K, even tried with a 1K fixed resistance in series with the pot.
Load: LED in series with 1k resistor.


Schematic

https://www.dropbox.com/s/x30m3l9wj5gd0uh/20150303_001021.jpg?dl=0

Resistance measurement between in and out pins of one of the dead ICs. all other ICs in this photo are destroyed 

https://www.dropbox.com/s/bwrgm6o38rosbl8/20150303_001607.jpg?dl=0

Close view at the dead ICs

https://www.dropbox.com/s/wwhl7htu5x89mcu/20150303_002054.jpg?dl=0

I think it is CMOS SCR latch up not sure what is happening here!!!!!!!!!!

[Andy Watson]

Are you sure they are genuine LM317 ?  The logos and part number on some of those devices in your picture looks almost hand-written!

[RJFreeman]

The first question the cynic within me asks: what is your source of these Devices?

OK next observation:

Supply voltage: 26V DC (From 18-0-18 50VA transformer and regulated with a bridge rectifier and a 4700uF cap)

by this do you mean you have both 18V windings in series? if so this will give you a voltage of (18x2x1.414)=51V*  which is over the LM317 maximum rated input to output voltage of 40V.

This could explain why you suffer failure when you set output voltage to minimum since minimum voltage would be 1.25V (or so) giving input  to output voltage difference of 51-1.25= 49.75 Volts

* A rectifier and filter will charge up to the peak voltage, while AC voltage is specified in RMS

[rdl]

Often the problem is the difference between the schematic and what you actually built.

[max666]

Those markings and logos are all over the place 

[Xenon Photon]

Are you sure they are genuine LM317 ?  The logos and part number on some of those devices in your picture looks almost hand-written!

I'm not sure if they are genuine. I tried from all the local electronics stores and all of them were destroyed.
btw, good quality products in Egypt are rare and very expensive and I don't think those IC are genuine but they worked when I used them before. Same store and same circuit but 2 years before. Maybe those stores all today sell fake parts that does not work?

The first question the cynic within me asks: what is your source of these Devices?

OK next observation:

Supply voltage: 26V DC (From 18-0-18 50VA transformer and regulated with a bridge rectifier and a 4700uF cap)

by this do you mean you have both 18V windings in series? if so this will give you a voltage of (18x2x1.414)=51V*  which is over the LM317 maximum rated input to output voltage of 40V.

This could explain why you suffer failure when you set output voltage to minimum since minimum voltage would be 1.25V (or so) giving input  to output voltage difference of 51-1.25= 49.75 Volts

* A rectifier and filter will charge up to the peak voltage, while AC voltage is specified in RMS

No, the 18-0-18 transformer is for positive and negative supply. This part of the circuit is in the positive is supplied from 18-0 volts AC. The 26 volts is the DC output after the rectification at no load measured between positive supply and ground.

[RJFreeman]

No, the 18-0-18 transformer is for positive and negative supply.

ok, well that discounts the most likely design error then, which gets us back to:
1) dodgy counterfeit component
or
2) construction error
Noting that you do have the multimeter happily measuring between the tab (which would be output, the same as the middle pin) and the input pin suggests you probably have this part correct, but would it be possible to post pictures of the circuit?

[rdl]

... would it be possible to post pictures of the circuit?

 This would be useful because...

often the problem is the difference between the schematic and what you actually built.

[Alex Eisenhut]

btw, good quality products in Egypt are rare and very expensive

That's kind of sad, how is the postal service there? Will packages from overseas make it, or will they disappear?

[RJFreeman]

This would be useful because...

often the problem is the difference between the schematic and what you actually built.

Yep....

[andrew_c]

btw, good quality products in Egypt are rare and very expensive

That's kind of sad, how is the postal service there? Will packages from overseas make it, or will they disappear?

Good question. Pictures of the circuit would be interesting.

Andy

[Xenon Photon]

Photos of the circuit:
https://www.dropbox.com/s/mvhuiq26ydzn5y4/20150303_044041.jpg?dl=0
https://www.dropbox.com/s/xolginq584sv2lq/20150303_044002.jpg?dl=0
https://www.dropbox.com/s/743jzgqam5upjkj/20150303_043954.jpg?dl=0
https://www.dropbox.com/s/xvgug6xqct46l10/20150303_043938.jpg?dl=0
https://www.dropbox.com/s/k40p8tg69k19d06/20150303_043917.jpg?dl=0
https://www.dropbox.com/s/hgwf2g897wztzdm/20150303_043828.jpg?dl=0

the yellow crocodile is +26VDC
the black is 0VDC
the 100 ohm resistor was used in some tests to protect over current and burning the pot.

btw, good quality products in Egypt are rare and very expensive

That's kind of sad, how is the postal service there? Will packages from overseas make it, or will they disappear?

Packages with tracking number arrive but they cost a lot of money. without tracking number it has 50% chance especially if shipping to a home address.
If there are Egyptians here I'd advice them to ship things to their work address and don't try to order anything that contains a wireless module in it because it will be taken at the customs.

[amyk]

Even if they are fake I don't think you'd be able to consistently kill them like that - unless they were already dead before you used them. A bad batch, or perhaps ESD damage?

[macboy]

Those LM317 are 100% certainly fake. There is no way that ST would have the markings all over the place like that... one of them is even off the left edge. Also the "ST" logo is not clear and sharp. Fake 100%. I haven't seen your circuit, so I can't say that a genuine part will work with your circuit, but you will almost certainly have better luck with the real thing. Your schematic is good, having the reverse biased protection diode from input to output, and appropriate capacitors. I'd reduce the R1 from 240 Ohm to 120 Ohm. This ensures that the load is always above 10 mA, which is the specified minimum for all but the higher grades of "real" LM317. Never use with R1 less than 240 Ohm!

[Alex Eisenhut]

Packages with tracking number arrive but they cost a lot of money. without tracking number it has 50% chance especially if shipping to a home address.
If there are Egyptians here I'd advice them to ship things to their work address and don't try to order anything that contains a wireless module in it because it will be taken at the customs.

I see, but what about all the eBay specials? 2$ for 10 LM317s, is that too expensive, or are small packets from China "lost" very often?

[Xenon Photon]

Even if they are fake I don't think you'd be able to consistently kill them like that - unless they were already dead before you used them. A bad batch, or perhaps ESD damage?

The pins were not shorted before using them. I did continuity test to most of them before using. it is only shorted after the circuit is connected to the supply. one thing that is interesting is the 100 ohm resistor seems to protect the ICs from complete destruction but that is not a good solution because the regulator should supply 1A current (it is a power supply project for my little brother).

Those LM317 are 100% certainly fake. There is no way that ST would have the markings all over the place like that... one of them is even off the left edge. Also the "ST" logo is not clear and sharp. Fake 100%. I haven't seen your circuit, so I can't say that a genuine part will work with your circuit, but you will almost certainly have better luck with the real thing. Your schematic is good, having the reverse biased protection diode from input to output, and appropriate capacitors. I'd reduce the R1 from 240 Ohm to 120 Ohm. This ensures that the load is always above 10 mA, which is the specified minimum for all but the higher grades of "real" LM317. Never use with R1 less than 240 Ohm!

do you mean "Never use with R1 more than 240 Ohm"?
I'll test this and I hope to have some luck

Packages with tracking number arrive but they cost a lot of money. without tracking number it has 50% chance especially if shipping to a home address.
If there are Egyptians here I'd advice them to ship things to their work address and don't try to order anything that contains a wireless module in it because it will be taken at the customs.

I see, but what about all the eBay specials? 2$ for 10 LM317s, is that too expensive, or are small packets from China "lost" very often?

yes, they are lost very often. Packages lower than 50$ from china does not have a tracking number which means 50% chance they wont arrive. they also take about 40 to 60 days sometimes I forgot what I ordered after this long time. This project has to be done in this week.

[Agent24]

Going by the awful marking on the ICs, they are most certainly knockoffs or otherwise inferior product. And if they are not genuine, all bets are off.

Interesting that you say it died as soon as you used a 47uF capacitor. I wonder if there was some oscillation going on?

Otherwise, when you turn the voltage down to lowest, the IC is dropping the most voltage across it and will run hottest. I do notice you have no heatsink on these parts, but if it overheats it should shut off. How much current are you drawing during testing?

Since they work until you push them a bit, it's quite possible they are clones with some or all of the protection circuitry omitted, so instead of shutting down etc like a genuine one would, they just break instead.

[BlueBill]

From AG
The LM317 uses a 120 ohm resistor from the output to the ADJ pin, the more expensive LM117 can use a 240 ohm resistor.

[dannyf]

I think it is CMOS SCR latch up

It is always CMOS SCR latch up - remember to always put the "SCR" there or it doesn't sound sophisticated.

Last week, my hammer broke and sure enough, due to CMOS SCR latch up too.

Damn!

[macboy]

Those LM317 are 100% certainly fake. There is no way that ST would have the markings all over the place like that... one of them is even off the left edge. Also the "ST" logo is not clear and sharp. Fake 100%. I haven't seen your circuit, so I can't say that a genuine part will work with your circuit, but you will almost certainly have better luck with the real thing. Your schematic is good, having the reverse biased protection diode from input to output, and appropriate capacitors. I'd reduce the R1 from 240 Ohm to 120 Ohm. This ensures that the load is always above 10 mA, which is the specified minimum for all but the higher grades of "real" LM317. Never use with R1 less than 240 Ohm!

do you mean "Never use with R1 more than 240 Ohm"?
I'll test this and I hope to have some luck

YES. R1 must be 240 ohm or LOWER. In fact, 240 ohm is only for the higher grade LM317 with a 5 mA minimum load current; most have 10 mA specified/tested minimum, so you should use 120 ohm or lower for R1. The alternative is to ensure that the load is never less than 10 mA over the entire adjustable range. When building a "bench" type supply, you can't know what kind of load is connected, so the most sensible way to get that 10 mA load is via R1 (voltage across R1 is always 1.25 V, so then the constant current through R1 is always 1.25/R1, regardless of output voltage).

[max666]

YES. R1 must be 240 ohm or LOWER. In fact, 240 ohm is only for the higher grade LM317 with a 5 mA minimum load current; most have 10 mA specified/tested minimum, so you should use 120 ohm or lower for R1. The alternative is to ensure that the load is never less than 10 mA over the entire adjustable range. When building a "bench" type supply, you can't know what kind of load is connected, so the most sensible way to get that 10 mA load is via R1 (voltage across R1 is always 1.25 V, so then the constant current through R1 is always 1.25/R1, regardless of output voltage).

Just so I get this straight, R1 = 120 ohm should guarantee at least 10 mA load. But how can that work with a pot R2 of up to 5 kohms? 37 V / 5120 ohm = 7 mA (and that's best case scenario, with lower input voltage you wouldn't even be able to get 7 mA)
Also, the datasheet only mentions a rise in output voltage if minimum load isn't met. It doesn't say the device will be harmed by that, will it?

[macboy]

YES. R1 must be 240 ohm or LOWER. In fact, 240 ohm is only for the higher grade LM317 with a 5 mA minimum load current; most have 10 mA specified/tested minimum, so you should use 120 ohm or lower for R1. The alternative is to ensure that the load is never less than 10 mA over the entire adjustable range. When building a "bench" type supply, you can't know what kind of load is connected, so the most sensible way to get that 10 mA load is via R1 (voltage across R1 is always 1.25 V, so then the constant current through R1 is always 1.25/R1, regardless of output voltage).

Just so I get this straight, R1 = 120 ohm should guarantee at least 10 mA load. But how can that work with a pot R2 of up to 5 kohms? 37 V / 5120 ohm = 7 mA (and that's best case scenario, with lower input voltage you wouldn't even be able to get 7 mA)
Also, the datasheet only mentions a rise in output voltage if minimum load isn't met. It doesn't say the device will be harmed by that, will it?

The device should not be harmed by having R1 too high (more specifically, the load too low), but as you said, the regulation may be lost.

With a pot (R2) of 5k... you get 5000/120*1.25 = 52 V. This is a valid output voltage as long as the input - output voltage difference is within limits (min ~4 V; max ~40 V).

The damaged chips could simply be very badly made knock-offs, or there could be something else at work. A low ESR output capacitor, for example, can cause oscillation, and this can destroy the chips. Try adding a resistor of a few ohms (1 to 5 ohm) in series with the output capacitor.

[max666]

The device should not be harmed by having R1 too high (more specifically, the load too low), but as you said, the regulation may be lost.

With a pot (R2) of 5k... you get 5000/120*1.25 = 52 V. This is a valid output voltage as long as the input - output voltage difference is within limits (min ~4 V; max ~40 V).

The damaged chips could simply be very badly made knock-offs, or there could be something else at work. A low ESR output capacitor, for example, can cause oscillation, and this can destroy the chips. Try adding a resistor of a few ohms (1 to 5 ohm) in series with the output capacitor.

Ahh yes, of course input voltage could be higher than 40 V. I guess my mind ignored that possibility, because you get all sorts of new problems when Vin > 40 V.
Thanks macboy!

[RJFreeman]

Ahh yes, of course input voltage could be higher than 40 V. I guess my mind ignored that possibility, because you get all sorts of new problems when Vin > 40 V.
Thanks macboy!

Input voltage is 26V (we covered that earlier)

[max666]

No ... Yes ... never mind. Sorry I strayed off topic
Bottom line, a too big value of R1 should not be able to destroy the regulator. Ergo, majority vote is on not genuine ST parts, but we still haven’t seen a picture of the circuit Xenon Photon actually built up, soooo ... human error is a bitch

[Xenon Photon]

Thank you very much everybody it is now working. the same circuit but better quality IC. I had to travel 50KM to the Cairo the capital of Egypt to just buy this IC.

Here the photo you can find the difference.
also do u think the IC from Cairo is original or high copy?

[SeanB]

Middle and right are fakes, the left one is a possible real product.

[RJFreeman]

Just so I get this straight, R1 = 120 ohm should guarantee at least 10 mA load. But how can that work with a pot R2 of up to 5 kohms? 37 V / 5120 ohm = 7 mA (and that's best case scenario, with lower input voltage you wouldn't even be able to get 7 mA)

R1 = 120 ohm will mean that the regulator circuit itself will draw at least 10mA (vref/R1 = 1.2/120 = 10mA) if however you have a load drawing 10mA already (such as an indicator LED) then R1 can be a higher value.