Fried Digispark

[chipman78]

I was troubleshooting why the backlight to my LCD kit wasn't working (the jumper fell off... d'oh!), and I think I inadvertently shorted / fried my Digispark.  

When testing some of the pins with my multimeter, I may have created a short, or arced a connection.  The outcome: I can no longer load programs onto my Digispark.  

Is it fully fried, or did I possibly just kill the base software (bootloader??).  Can anyone help in troubleshooting / repairing?

[digistump]

chipman - to keep cost and size down the Digispark doesn\'t have any fuses (not that the Arduino has many either) - so you very likely have fried it, in my experience though it takes an extended short - I\'ve shorted some many times over and they\'ve recovered.

Killing just the bootloader is very unlikely (I\'ve yet to see it happen) but if that was the case you\'d need a HVSP Programmer to reprogram it. Another Digispark or Arduino can be used to do this, but I haven\'t had the time to write up the circuit yet...
You\'d still then need an Arduino or Atmel programmer to reprogram it (or another Digispark, but again the software for that still needs to be written) - do you have an arduino as well?

[mikeyR]

@digistump Now when you say you fried some boards and were able to recover them.. what do you mean by that?


I had my 'spark working with an RGB LED shield and think I shorted the 5V and GND pins on the 'spark by accident. It was not a prolonged occurrence at all. Now, I can't get the LED on the board to come on when plugged into USB. When I use another Digispark on the same USB port, it works. Using a 5V supply will get the LED lit, but I don't know if anything else is functioning. Any suggestions? Is she a goner? Thanks

[digistump]

mikeyR - sounds like you fried the Schottky diode the protects the USB port form the external power if both are connected at once. Since it is fried in the off state your USB port should be safe if you want to connect either 5v to the 5v pin or 7-12v to the vin pin and gnd to the gnd pin and then plug it in to a USB port, it will then use the external power and if the chip is still ok it should connect to the computer. Of course, do so at your own risk, I'd use a hub for a line of safety between it and the computer, but you might find that it still is usable.

[vmods]

@digistump,

Thank you.  I just apparently did a very similar thing, and your suggestion made it very easy to troubleshoot.  Apparently the rev 2 Digisparks are not forgiving what-so-ever, if you short the 5v pin to GND. It seems like a very common thing to test with a multimeter - just make sure you don't touch the leads (now very obvious.)  I believe I've done this to 2 of the 3 boards I purchased.

It's too bad the Schottky diode is so sensitive - makes it pretty easy to render the product useless, at least as it's intended to be used with a USB port.


Any suggestions on replacing the diode? 


Maybe the next revision can have a small optocoupler?
I might also suggest separating the ground pin, unless of course, that is part of the plan to sell more products =)

[Bluebie]

You can just solder a wire from the 5v pin to the edge of the USB connector - the strip of gold plated PCB closest to the power pins on the digispark is the 5v positive power supply from the computer. It means if you hook up an external power supply via VIN or 5V you should not plug the digispark in to a USB port, because it could push electricity in to your computer, and that could be bad, but if you don't externally power the digispark while using USB it should be fine to just wire it across. I've done that on the one digispark I have which blew out the schottky diode.

[Mark]

Don't you guys realise.
The electronic bits are the fuse, and always react quicker...

They work on smoke, and if you let the smoke out, they won't work.

You should be able to do what Jenna suggested but use a Schottky diode inside of plain wire.

You aren't the first, and won't be the last to blow up something while testing it.....


mark

[vmods]

Definitely not the first, nor the last time, I'll burn up something.  It just seemed too fragile IMO, to the point I'd like to blame the design more than my admitted negligence.


It's not like I hooked up a lab's PSU and provided lot's of power, in the wrong direction, with the wrong components - like so many of us did our first year of Electronics Fundamentals.


Guess I'm just happy to know it was a Schottky diode and not the µC.

** I found this schematic: https://s3.amazonaws.com/digispark/DigisparkSchematicFinal.pdf

Looks like a less-than-ideal Zener diode could have been chosen for voltage regulation - which potentially failed.
Although, I did look up the Schottky diode, MBR0520LT1, and I'm willing to bet it is in fact the problem.
** Just confirmed with a multimeter, and by bypassing the component, which made the board work again.

Conveniently located in the corner, should be easy to swap out with a better one.
I can post results if anyone else has the same problem, or would like to know what I replace it with.

http://www.datasheetarchive.com/MBR0520LT1-datasheet.html#

[digistump]

The schottky we are using isn't particularly sensitive but it still isn't made to handle a short -


We use a MBR0520LT3G or equivalent - which is the same used on the Arduino nano's if I remember correctly (which also doesn't have a fuse last time I checked).



Could we have added short circuit protection? Yes, but our goal was to make a cheap, small, device and short circuit protection would have added size and cost.


Is the design flawed? could be - I'm here to help, not to say our design is perfect, so I'm certainly open to suggestions for future revisions.


If the short truly was momentary - then most polyfuses like those on the Arduino would not have helped the situation. If anyone has a suggestion for a same form factor Schottky that can handle more abuse - please speak up!


I'll do some testing on a few and see if I can blow them with a momentary short.

[vmods]

Cheap and small is why I bought it =)  -I do appreciate this design very much, just frustrated I broke 'em so quickly.


Did a side by side comparison:
MBR130T1G ($0.35 for 1 or $192 for a reel of 2000)
http://www.mouser.com/Search/ProductDetail.aspx?qs=3JMERSakebo%252bJOSYnNXaYQ%3d%3d


vs


MBR0520LT1G ($0.34 for 1 or $122 for a reel of 2000)
http://www.mouser.com/ProductDetail/ON-Semiconductor/MBR0520LT1G/?qs=sGAEpiMZZMtQ8nqTKtFS%2fCKUxMvjsmGzA9NL%252bHfdx4I%3d


The voltage drop doesn't seem to be too much more on the 130T1G, although it "may" be forgiving enough (@ 1A) to survive a momentary short.
These specs on 'paper' are so relative, temperature pretty much makes them invalid.


I could take measurements and run calculations, or I could just order these and see what happens.  -This time I'll for sure solder on the headers and be careful when stepping through with a multimeter.

[digistump]

I don't think the continuous current is what makes it sensitive (or at least not the main limiting factor) and a short even on USB will probably see more than 1A - I think the limiting factor here is the max surge current of 5.5a - which is the same for both parts vmods listed.


This might be a worthy replacement:
http://www.mouser.com/ProductDetail/ON-Semiconductor/MBR120VLSFT3G/?qs=sGAEpiMZZMtQ8nqTKtFS%2fCKUxMvjsmGz%2fCigMfHc%252bQw%3d


It also has a slightly lower vdrop - as that is something we don't want to increase as we want things as close to 5v as possible so that 5v designs work as well as possible.


It also comes on a 10k reel for cheap - a big bonus for us because we produce about 10k at a time.


Of course, this is all off-the-cuff speculation and I'm not sure if it would make it more robust or not.

[vmods]

I agree continuous current is not the metric to gauge robust against a surge.  USB 2.0 with 5V @ 500 mA IIRC, does seem to beg the question if a surge would pull over 5.5A - but  then again, something gave out, twice.


Taking the 1A spec over the .5A spec at face value could be a mistake comparable to taking a 3 GHz old CPU over a 2.4 GHz new CPU.


The ratings taken at 25°C do show the expected dramatic  performance decrease with ↑ temps.


I do like the MBR120VLSFT3G you pointed out;
Good Vdrop(s) too;  Interesting to compare each with regards to the different temperatures and Amp draw.
 45A sounds ludicrous, but I'll take these specs with a grain of salt.


Then again, it is nice when a diode takes a bullet for the rest of the circuit.  Suppose it can act as a fuse in this case, maybe I should just make a socket it for it so I can replace it every time I impatiently take an overly hasty reading =]


Think I might go with the one you pointed out.  Very much appreciated.  Thanks again.

[digistump]

I guess it is worth noting that when I've fried them by direct 5v with short - not growing through the schottky they are dead forever because the attiny is fried - so certainly it might be desirable for this part to fail.


On a similar note - we will be releasing a "programming tool" which is a fancy name for two USB ports (one male, one female) and a switch in between allowing you to put it inline when programming a digispark and disconnect/connect the digispark with the switch instead of unplugging/replugging - basically a cheap added tool for those programming a lot - without adding cost to the actual Digisparks. This thread got me thinking about adding a resettable fuse to this board:


http://www.mouser.com/ProductDetail/Schurter/PFRA025/?qs=sGAEpiMZZMsxR%252bBXi4wRUPAUULLzmI6SAI7rgO0%252bA0M%3d


They are cost and size prohibitive to add to the Digispark (even in SMT form) and certainly aren't a cure all, but they are cheap to add to the programming tool board - since you only need one per person instead of per Digispark - which already has room to spare.

[Bluebie]

I still think it'd be really really awesome if digistump or someone else came out with a digispark shield which 'ruggedizes' the header connectors using techniques from the open source Ruggeduino project: http://ruggedcircuits.com/html/ruggeduino.html


This way we could have a little man-in-the-middle shield that protects our digisparks without increasing the unit cost of digisparks, so you could have just one ruggedizer stacking shield to use during development then swap to direct wiring once you're done with your project and feeling confident in your wiring. If someone made this, I'd buy one ^_^

[digistump]

Bluebie - that one is definitely on the drawing board for the next round of new products - coming out soon is - screw terminal shield, rtc shield, coin cell shield, programming tool - next round (maybe a month out) is temperature shield, rugged shield, hotplate/oven controller.