EEVblog Electronics Community Forum
Electronics => Repair => Topic started by: jogri on September 22, 2019, 04:58:07 pm
-
I am currently trying to fix two old Oltronix B703 DT linear power supplys (again) and i have encountered two blown rectifiers that caused a fair amount of pcb damage. When i tried to desolder them (through-hole), the pads on the top of the pcb sticked to the legs of the component and started to lift up when i pulled on the rectifier and the pads on the bottom just floated away. Is there a trick to avoid this and save the pads when desoldering? And can i somehow attach those pads again (never done that before) or should i just use jumper wires?
Also, should i still use a pcb with this sort of damage? Judging from the schematics, both rectifriers (w02m) just provide around 5V for the LEDs/ICs and don't seem to have a particulary big load.
-
The PCB round that rectifier is most likely FUBARed. Its obviously carbonised and appears to be delaminating as you can see the glasscloth weave. IMHO, Its far past simple pad replacement being viable. As burnt PCB becomes conductive, chances are that you would have to cut and grind away the burnt area for a successful repair.
However before you go down that road it would be worth removing enough parts to isolate at least three of the tracks to it to let you do an insulation test without zapping the rest of the circuit. A DMM with a 10Meg FSD or higher resistance range will give an indication if its definitely become conductive, but to be certain its rebuildable without cutting away the burnt area, you really need an insulation tester that uses a higher test voltage.
Although its technically possible to reconstruct the PCB by cutting out the damaged section and gluing in a replacement section, the cost and time to do so is seldom justifiable, except for extremely rare and high value PCBs. Without reconstructing the PCB, it becomes a total PITA to rebuild neatly as there wont be much left to support the bridge rectifier legs.
If you cant find sufficient sound PCB round the existing footprint to re-drill for the legs of a bridge rectifier in a larger package (or for a replacement of the same type with its legs splayed, probably the best option is to fit a bridge rectifier with a mounting hole for chassis or heatsink fixing somewhere nearby, either mounted to the PCB or the chassis, and connect it using stranded hookup wire to the tracks, where they are still on sound PCB.
-
How would i do an insulation test? Remove the next components from three tracks and measure across the burnt patch? I would do that if the solder pads would stay in place, but chances are they are going to peel off as well.
Btw, the patch in the photo is not conductive. I know that because that was the rectifier for a working channel (it seems that it only partially blew out as the control IC behind it still got enough power for normal operation). I don't know if the second burnt spot (no photo bc it looks like the other one and it is difficult to get a good picture of it) under the other rectifier is conductive as that channel wasn't working (that got me starting the repair in the first place). But for safety reasons, i think i'll move both rectifiers to the casing (even if that brings the problem of isolation). There is no chance that i'll find a new spot on the pcb for them as said pcb contains three seperate power supplys that are galvanically isolated from another on a space of 10x18cm...
Is there anything i need to watch out for when cutting pcbs? Or can i just drill a hole through it and connect everything with jumper cables?
-
Odds are the insulation breakdown is greatest between adjacent pins rather than diagonally across the bridge rectifier from + to - or AC1 to AC2, so *IF* there are no other components linking the AC side to the DC side of the circuit, and you are hppy to settle for a low voltage resistance test (e.g. using a DMM) you could simply short + to - and, measure between there and the AC pins, which are almost certainly linked by a transformer secondary. However, if there's a center tap on the transformer secondary for DC 0V, you'll need to disconnect that.
For a low voltage circuit, I'd condemn any PCB with a leakage resistance under 1Meg. The leakage resistance will only deteriorate with more age, heat, humidity, contamination (e.g. from flux) and higher ambient temperatures, so you need a very generous safety margin to avoid any significant risk of the deterioration proceeding to the self-heating stage, resulting in a risk of the PCB catching on fire.
Fibreglass PCBs (e.g. FR4) are particularly hard on drill bits, and small carbide drill bits or cobalt drill bits, which have cutting edges that can stand up to cutting fibreglass are fragile (carbide notoriously so). In small diameters, even an ordinary HSS bit is very easy to snap if you slip and apply significant side loading or if it jams in the hole. That means, if you don't have a precision drill press with a very low runout spindle, you cant effectively use carbide or cobalt bits as you'll snap more than you blunt. A sub-2mm HSS bit in a hand drill may only be good for ten to twenty holes in FR4 before its unusably blunt. Your best chance of a good result using hand tools is with a new HSS bit in a Dremel or similar, with the board secured to the bench so it cant slip, and a very steady hand.
-
I would sacrifice the pads, grind out the burnt board, and drop in a jumper wire along the trace. I have had to resort to this more than a few times.
-
Drilling FR4 is not that hard. Just a reasonable sharp HSS drill would do it. I would consider carbide if it is more than 50 holes. With just 2 or 3 holes the drill would still be usable afterwards.
My guess would be some 3 mm first and than some 8 or 10 mm to remove the hole burnt zone. The board should be tightly clamped and some wood or similar below to avoid the drill to more to fast when going through.
If there are 2 similar burnt rectifiers this looks like there is a fault like a design flaw.
-
It's amazing how tough those little W0x bridge rectifiers are. I've seen many that have cooked away at very high temperatures for decades, de-laminating the PCB as in the OP's case, with very few failures.
P.S. The design flaw is not providing any significant copper area - pretty much the only way those packages can dump heat.
-
When designing a PCB with diodes that need to dissipate, give the cathodes some copper surface. In the case of a bridge like this you end up with 2 cathodes on the "+" terminal, then the repair guy confuses it for a ground plane and does something stupid...
If it isn't bad design in the first place, it's bad caps on old equipment that end up frying the bridge.
-
Well, they couldn't give those rectifiers some copper surface as the board is completely packed with either components or ground traces. Although i don't get why they packed one of those rectifiers between the heatsink of the main rectifier for a channel and a bunch of smoothing caps (about those caps: they are from ROE, i couldn't find a definite answer about the reputation of that brand, do you guys know more about that manufacturer?).
But thanks for mentioning excessive heat as a possible cause of failure, i somehow didn't thought about that (it didn't occur to me that a 100W linear power supply will produce a lot of heat under load). I have no pictures of mine fully assembled to show you how badly cramped it is, but Marco Reps did a video on his unit: https://www.youtube.com/watch?v=Tr0FfZ-A3P4 (https://www.youtube.com/watch?v=Tr0FfZ-A3P4)
I'll try to mount a small 40mm fan somewhere in the case (maybe on the back) to get at least a little bit of active cooling.
Btw, do you know if heatsinks for those round w02m rectifiers exist? Otherwise i'll just make some out of aluminum rods.
Oh, and do you by chance know if combined four digit volt/amperemeters exist (like this one, but with an ampmeter https://www.aliexpress.com/item/32843540086.html?spm=2114.12010612.8148356.2.104df9cc73Vkft (https://www.aliexpress.com/item/32843540086.html?spm=2114.12010612.8148356.2.104df9cc73Vkft))? I have found volt/amp-meters that use two seperate displays, but no units that use the same four digit display for both purposes with a switch to toggle between the two functions.
Yes, i could theoretically use the exact same displays the manufacturer used (they are still available), but the manufacturer of those wants 50 bucks per unit and i am just not willing to spend 300 bucks on my two PSUs.
-
Survival isn't sign that it's conductive. Use an ohmmeter, and stick the probes deep into the burned area if you can.
Best practice will be, add pigtails from whatever pads the traces connect to, using say 0.8mm (20AWG) solid wire if you have it. Preferably, glue the wires down, or the replacement component. Or since it's a FWB, maybe use a bolt-mounted type, and heatsink it -- double insurance against this happening again. Remove connections from the affected area, or as mentioned above, cut/drill it out.
The best best practice is, cut out the affected area, glue in a new patch, glue down new traces, and add plated-through-holes using rivets. (FR-4 / G10 bare laminate is readily available, and epoxy is used to patch. Preferably glass-filled epoxy, usually done by grinding up some PCB and mixing it in.) But, that's more of a... Idunno, military thing? Patching boards on ancient, irreplaceable, high performance kit seems like only something they would do. (I mean, there's modern laptop and cellphone repair that goes to nearly the same lengths, but they don't usually care about exact repair, just as long as it works again, or, long enough to get files off of anyway.) I certainly wouldn't expect you to go that far, just giving you some idea as to what's possible. :)
Tim
-
The best best practice is Tim says, but due to the rapid and cheap availability of PCBs in small quantities nowadays, for double sided PCB repair you can simplify it a bit, avoiding all the faffing about gluing down copper foil and riveting replacements for PTHs, by simply laying out a replacement section with all the tracks and pads you need complete with plated through holes, and silkscreen marks for where to grind back the edge steps to, order a PCB of the replacement section, then when you get it, lift and fold back a short length of the traces going past the edge to have overlapping copper to reconnect them, then cut to size and grind the edge steps, clean and tin the undersides of the lifted trace ends, then all you have to do is glue it in place and solder the track ends back down.
-
Yes, sacrifice the pads, remove the char and fill in with epoxy, then mount the new, larger size and higher rating, bridge rectifier either there, or with some short heavy leads to the existing other pads or tracks. Larger surface bridge for cooling than a tiny 1A part, I use a 3A square type if possible, or a 5A SIL part if space is tight, using short sleeved leads to make the pinout. you can drill new holes for the leads if needed. Have done plenty of track repairs for loose pads (thank you for 1970's era boards with poor adhesion) for non cooked parts using care, and a drop of superglue after soldering in a new part, and a wire link for trace continuity, especially on military boards where a replacement board (if available at all, or if lead time was in the year range) was a six figure sum in dollars.
Some of them were 7 figure prices, and I had to order them, because there is not much you can do if the only part that has survived the toasting session inside the case is the milled aluminium chassis, 6 chassis mounted stud rectifiers, the plug used to connect to the mainframe, and one of the 3 1A Littlefuse solder in fuses on a separate board, that are supposed to prevent this from happening. Caused by a common fault, one of the other rectifier diodes went short circuit, and it carried on operating till a second died from overheating. Came ordinary parcel post......