A little wire here..., a little smoke there...

Alex.Stanfield

I recently fried a Propeller USB protoboard (the old model). The whole thing is Smile, an XBee mounted, RTC, the LDO regulators, I even let some "magic smoke" get out from the FTDI chip...:depressed:

Since the smoke came out from the FTDI I thought it happened because I was powering the FTDI chip from the 5V LDO regulator on board (although I had no USB device connected and I had this working for several months) I couldn't see why this would happen but it was my only clue.

I have been working with it for several months perfectly well. However, after a recent rework on some input pull-ups I must have inadvertently left a tiny piece of wire just sticked (not soldered) between two output pins on an edge connector. After several days something must have changed or just the ambient temperature might have melted the resin and produced the short between two outputs, one high and the other of course low.

So the question is if anyone had seen this happen? I know that the propeller should get fried that's understandable, but if I power the board now I get almost Vin on the 5V and 3v3 rails. This explains why the FTDI, RTC and Xbee got fried too.

I would expect the LDO to lower the output voltage when overloaded (that's what I recall from the datasheet). They just seem to be a short between input and output.

Any thoughts?

Alex

kwinn

I find it hard to believe that shorting 2 output pins together would fry the whole board. It might damage one or both of the shorted outputs, but that would be all. It might blow the whole board if a regulator was to fail by shorting input to output, but this is extremely rare. The 3 terminal regulators are very rugged well protected chips. Typically the only thing that happens if there is a short, or too much current is drawn, is that the regulator shuts down until it cools off and then start up again.

Check carefully and you may find it was caused by something else.

LoopyByteloose

Well, there may have been other debris or solder bridges. When a board is not clean, it is hard to say how dirty it was before failure.

Start with a complete cleaning and check closely for both debris and solder bridges... solder bridges can be just a whisker in width.

Alex.Stanfield

Thanks for the suggestions, I have checked the board and found nothing except for what I said before. Will check again.

What puzzles me is that for that kind of failure I think either the LDOs shorted due to excessive current (unlikely) or some extra large voltage entered the board. The board was mounted with screws on a plastic base (the same I have been using for months) nothing was loose so it came through the power or signal wiring.

Maybe this DC-DC converter didn't startup correctly? I've using them ok for months however.
The output is adjusted to 6.5V and fed to the board. On the input side I have 32V which exceeds the absolute max rating of the LM1086



Alex

cavelamb

Alex.Stanfield wrote: »

Maybe this DC-DC converter didn't startup correctly? I've using them ok for months however.
The output is adjusted to 6.5V and fed to the board. On the input side I have 32V which exceeds the absolute max rating of the LM1086

Alex

Bingo!

That's the one itty bitty problem with those converters.

When they let go, they let go dramatically.

Alex.Stanfield

cavelamb wrote: »

Bingo!

That's the one itty bitty problem with those converters.

When they let go, they let go dramatically.

F...k, sad to hear that but at least I start to have a clue on what when wrong

The converter is now apparently working ok (6.5V with no load), although I measure more (7+V) when powering the dead board for a short time, I assumed this was due to the abnormal state of the board, maybe it's not.

How frequently can that happen? Is there any workaround or prevention technique besides lowering the input voltage?

Thanks
Alex

cavelamb

Alex.Stanfield wrote: »

F...k, sad to hear that but at least I start to have a clue on what when wrong

The converter is now apparently working ok (6.5V with no load), although I measure more (7+V) when powering the dead board for a short time, I assumed this was due to the abnormal state of the board, maybe it's not.

How frequently can that happen? Is there any workaround or prevention technique besides lowering the input voltage?

Thanks
Alex

Oh for Pete sakes, Alex...
You have over 25 volts of difference between input and output.
Look at the data sheet again...

I'm just guessing you added something lately?
Or shorted a wire?

Workaround?
Yeah. Don't short the thing out when it's highly loaded.

(edit)
Sorry Alex. I didn't mean it to sound that bad.

(edit again)
And no, I woudn't trust it any more...

Alex.Stanfield

cavelamb wrote: »

Oh for Pete sakes, Alex...
You have over 25 volts of difference between input and output.
Look at the data sheet again...

I'm just guessing you added something lately?
Or shorted a wire?

Workaround?
Yeah. Don't short the thing out when it's highly loaded.

(edit)
Sorry Alex. I didn't mean it to sound that bad.

(edit again)
And no, I woudn't trust it any more...

that's ok i deserved it.

I didn't add anything, but I usually started first the psu and then toggled the power switch on the board. This time it was on from the start, maybe that was.

I reworked the power lines to lower the input to the dc-dc module to 16v just in case to avoid repeating the experience.

I haven't found many references to problems with these modules. Do you have any pointers?

Alex

Mike Green

The short answer is that it's possible for them to fail, even transiently, and a failure can be catastrophic to the rest of the board. Failures are unlikely though. The same is true of any power supply where there's a relatively high voltage input, even a simple linear regulator like the LM7805. If you're concerned, add a crowbar with a fuse so the power supply will be shorted if the voltage rises above some non-destructive threshold and the fuse will blow. There are other safety circuits that limit the power supply voltage. Remember though that safety circuits fail too. Sometimes they fail to protect and sometimes they erroneously detect a fault condition and shut down the device.

Remember, even "old time" transformer supplies have been known to develop a short between the primary and secondary windings. Very unusual, but it has happened.

Alex.Stanfield

Mike Green wrote: »

The short answer is that it's possible for them to fail, even transiently, and a failure can be catastrophic to the rest of the board. Failures are unlikely though. The same is true of any power supply where there's a relatively high voltage input, even a simple linear regulator like the LM7805. If you're concerned, add a crowbar with a fuse so the power supply will be shorted if the voltage rises above some non-destructive threshold and the fuse will blow. There are other safety circuits that limit the power supply voltage. Remember though that safety circuits fail too. Sometimes they fail to protect and sometimes they erroneously detect a fault condition and shut down the device.

Remember, even "old time" transformer supplies have been known to develop a short between the primary and secondary windings. Very unusual, but it has happened.

Thanks Mike, I'm planning to put a fuse before a 12V zenner on the board's power input. Also just in case I'm using now a lower voltage before the DC-DC module which the LM1086 should withstand.

Alex

Circuitsoft

Alex - The zener may not cut it. What may work is a zener (~12V) with an SCR (like MCR106) and a holdoff resistor (100 ohms or so). When the Zener fires, the SCR will short the circuit out, hard, until the fuse blows.

Alex.Stanfield

Just in case someone stumbles on this thread I finally found the reason for the failure. Blogged here http://forums.parallax.com/entry.php/3770-LM2596-DC-DC-converter-modules-quot-best-practices-quot

And wanting to err on the safe side I also lowered the input voltage to the module and put a crowbar on the propeller board. That would change a spof to 3 failure points.

Alex

Mike Cook

Thanks for posting! I need to check my modules. I've had good luck with these so far.

LoopyByteloose

hmmm. very interesting suggestion (replace the trim pot with some real resistors).

I bought a couple of these adjustible boards about a month ago. One I had to take back for exchange immediately as an electrolytic cap was falling apart. So, I guess one should do a close inspection for all sorts of loose or damaged components.

Personally, I dislike and distrust the tiny trim pots as they tend to be very hard to adjust and not standup well to being tweaked when under load. I can't seem to find a specification but I guess they are rated at 1/10th or 1/20th of watt... not at all rugged.

I try to use 10-turn trim pots that are rated at 1 watt with my own builds. The extra turns and the higher wattage really make the designs work much better... but more expensive. These boards do appear to have the more expensive trim pots... but quality control of the assembly seems to be lacking at the factory. I suppose that is the dilemma of marketing to Ebay prices.