I've a cheapo eBuyer Value router/modem that works fine, excellent spec, but the smoothing cap on the on-board power circuit gets stinking hot. System runs from a 9V AC wall-wart, bridge rectifier and smoothing on board before 5V and 3.3V regulators, both low drop-out types. The original was a 1000uF, 16V 105C type, no-name brand, and I found it bulged and leaking when I opened the box a few months after buying it. Swapped it for a larger 1000uF 25V 105C no-name - gets too hot to touch for more than a few seconds. Started running the device on 9V DC input (a SMPS wart), which will lose about 1.2V going through the bridge, still plenty of overhead for the regs. The cap is a bit cooler but still hot. Swapped again for an even bigger 1000uF 35V 105C Dubilier with 100n ceramic across it for good luck, still too hot to touch for long. Why does it get hot on an already smooth DC input? Do the bigger motherboard caps also run hot?
excessive ripple current though it is the only thing I can think of. can you measure the current though it? Preferrably with a shunt resistor and an o-scope
That's the usual reason, but the cap spec is 1430mA max, the router is supposed to draw under 1A (from the rating of the original adaptor which only gets warm.) They've all been low-ESR caps.
can you measure the leakage current on the 25v cap to see if it has become excessive? possibly indicating some sort of overvoltage Maybe some sort of high frequency or high voltage is making its way back to the cap from the circuit somehow. I'm really grabbing for stuff here... An O-scope capture of the voltage, and the current would probaby give some clue, if you have one.
That's a bit of an odd one, especially with DC. Hot capacitors are either due to not being low enough impedance, overvoltage, reverse current or just being a bad capacitor. Can't think of anything other than finding some computer grade capacitors and trying those. Any pictures of the board?
You're grabbing at the right straws, that's why I added the 100n in parallel, although there are some SMD caps already around that area. Hasn't changed anything, though. These routers (looking at eBuyer customer feedback) sometimes fail after around 6 months, I'm wondering if this cap is the main reason. Mine's over a year old now but I replaced the bulged cap before it failed altogether, dropped the input voltage and improved the ventilation. At that price (£18 now, I paid £23.50) for an SPI firewall, I may just keep replacing the cap, it's easy to keep an eye on it.
I have the same Safecom router but in the old style case at home (SAMR-4110), but it doesn't get hot at all. I might open it up when I'm back at the weekend to check for a dodgy capacitor. It's now 20 months old.
Interesting thread... Sounds like you've touched nearly every base.. I would be heading there next myself, if only out of curiosity.
throw a heatsink on the cap, or a peltier for good measure Or at least put a bag or something on top of it so if it goes up it doesnt spray foul smelling fibers everywhere...
If its only taking 1A would it be possible to replace the whole regulation circuit with a lm317 or similar? or would that make too much noise? Networking stuff is very well known for having badly designed power circuits. I have a safecom switch that worked well for about a year, but I always knew it was running hot. When it failed I opened it up to find the pcb very dark brown discoloured AND several components had started to desolder themselves (smd caps/resistors moved out of place, and the main IC had some pins start to lift off. Im suprised it would continue to work while this hot, 180c.. or higher
Ive heard of mosfets desoldering themselves, falling off a board, getting soldered back on, and still working
Why not add more caps? If the Cap in question is wire across your positive and negative leads, I agree that its job in life is to reduce ripple from the bridge. The heat can only be caused by the cap having to work hard filling those gaps left by the regulator. By adding capacitance to the filter you will reduce the work each cap is required to do. Before anyone starts spouting off about time constants and filtering in the realm of ac/radio circuits, remember this is DC folks. A filter cap bank's job is resist changes in voltage by either charging or discharging. Increase the cap bank..Use two or three of the caps and see what happens to the heat load. You may also want to be sure the bridge in the wall wart is working properly. Perhaps it is the source of the extra ripple. In our 200-800 amp chargers you will find anywhere from 6-10 39,000uF caps in the filter bank. The same concept should apply here..only on a much smaller scale.
You're missing the fact that it still heats up with REGULATED DC when it shouldn't have to do anything. Something else is amiss.
I've also used four supplies on this so far, with each the cap runs hot. Original 9V 9W AC wart A regulated 1200mA linear block, 12V DC and 9V DC settings. 12V DC 3.8A SMPS block 9V DC 1.6A SMPS wart. Coolest is the 9V linear (100Hz) but not by a lot. And, of course, the regulators run much cooler with the lower overhead. More caps - problem is space inside the case. I may remove the bridge and put an extra cap in it's place; but with SMPS ripple frequencies a large smoothing cap should not be necessary. To avoid confusion, with an external DC psu, the circuit is Code: --+-- +DC ----+--->|---+-----+--- + to regs | | | | ===PSU - Bridge - ===Hot cap | ^ ^ | --+-- -DC ----+---|<---+-----+--- - ground The power supplies don't get overly hot, but I can't check their internal cap, they're all "sealed for life". edit: I've drawn that bridge wrong, but the vertical diodes are always reverse-biased off.
What is after the capacitor? I am assuming there is a step-down regulator after the capacitor. Maybe something is amiss with the circuitry downstream of the capacitor that is causing the cap to get hot.
Gets a bit tricky to read what's what with SMD caps, but seems to go straight to the inputs on a UZ1084 fixed 3.3V 3A and a LM1117 fixed 5V 800mA. It is possible to get regulators oscillating at HF, but then they usually overheat themselves and switch off. Possible, though. But as a modem it works well.
~SteveG A "regulated" voltage does not mean that it is free from ripple. I do agree that the ripple after the bridge should be small. A hot cap suggests that this is not the case. The cap has been replaced with the same result. The cause is not a bad cap. ** ~cp Is this a problem with the bridge? What is your voltage out of the bridge? Do you have a filtered/regulated supply to feed that voltage directly to the cap and secondary regulators. Just as a temporary test to slice the pie. At least you can verify if it's a problem with the supply or loads. Troubleshooting these circuits are tough without prints.
As this cap gets damned hot with both crude full-wave rectified AC and regulated DC (with under 120mV ripple according to the SMPS supply spec) I think it must be some peculiarity of the demand side. With DC supply two of the bridge diodes are always on, two always off. A faulty bridge would either have negligible effect, burn out/shut down the supply, or shut down the modem, depending on which diodes were open or short circuit. The bridge ran warm, not hot. The newer Safecom modem arrived yesterday, runs very cool, the adaptor is now a 9V 1.6A DC wall-wart and that's running far cooler than any of the adaptors tried with the old modem - the new modem is drawing much less current, I'm convinced the old one was drawing more than it should have. It's now packed away for emergencies. Disappointed that the new one came with the buggy original firmware from 2005, when there's a March 2006 revision. Must be very old stock eBuyer have picked up, but wtf, at £15 I'll do my own updating.