Generic LED par repair question

[amcsx4]

I'm having an interesting issue with a generic LED "36x3w" par can. I know the best answer to my issue is not to buy cheap chinese fixtures, but with that aside I thought you guys might have some advice on this on.

At the beginning of a recent show one par kicked all LEDs to full and the LCD went to "0000". I thought for sure an IC was cooked and put the fixture in the back of my jeep and forgot about it. This week I pulled it an another par out and decided to separate and mark the bad light. When I went to plug the two fixtures in they both worked as they should. I came back a bit later and tried again and sure enough the one light had stopped working again. The only thing I could think of was that the lights where cold when I pulled them from my jeep so I put the bad fixture in the fridge and came back 3 hours later. When I pulled it out of the fridge it worked as it should. As it approached room temperature it started back to the "0000" issue.

Any ideas? I'm thinking a bad capacitor, though none of them look swollen. I've never seen an IC malfunction in this manner but haven't ruled it out. It may be worth noting that when the fixture is malfunctioning the LCD and LEDs come on instantly to the malfunctioning state. Usually, when working, the LCD will blink "par" in what seems to be a boot sequence and then comes on as it should. I could get a full replacement from my importer, but the geek in me won't let this go. Any advice is appreciated.

-Eric

 

[dj41354]

An intermittent caused by temperature cycling is most likely an IC. On the inside of those chips are a die (the silicon logic/smart part) but there are also very fine wires that connect spots on the die to the legs of the IC. Those wires are "bonded" ( Wire bonding - Wikipedia, the free encyclopedia ) to spots on the die, as well as to the legs. If one of the bonds breaks, it can still be in physical contact such that it will still conduct electricity & work, but heat it up a little and it can move ever so slightly, and no longer be in contact, causing the operational failure. A way to try and find the IC causing the problem is to alternate between using "freeze spray" ( TECHSPRAY | Products ) and a heat gun. The freeze spray has a small straw/nozzle that'll let you direct the cooling to an exact IC.. then warm it it back up with the heat gun to see if the "bonds" inside that IC were affected... if so, that's the IC to try changing. Freeze spray is not the kind of thing you are likely to find easily. Compressed air in a can, like from Radio Shack, can also be used for "spot cooling" it just doesn't work as well.

 

[JD]

An IC can be the problem, but also be aware that IC's use a ceramic resonator for their data clock. It is usually blue or orange, has two or three flat leads, quite small, and is located near the chip with a couple of very small (10pf) capacitors near it. They are very prone to intermittent temperature variation failures and will (of course) immediately crash the processor IC if they stop working. (Replaced hundreds of these back in the E-service days.)

 

[VCTMike]

Temperature faults can also be manifested by a simple poor solder joint. Been there done that. If you have the time and a circular magnifying bench light, just look over the back of the control board to see if there might be some questionable solder joints.

 

[amcsx4]

The freeze spray worked like a charm. I was able to narrow it down to either 2 33Ω resistors or a 15pf cap all of which fed the master Eprom. A quick check with my multimeter showed the resistors to be within tolerance which left the cap. I happened to have one in an open cordless phone from 1985 on my work bench. Threw it in and the light works perfectly. Thanks for the killer diagnostic trick!