Dell complaints

[Rocke_T_Sinetist]

http://www.aplusperfect.com/articles/fix-dell-cx305n-power-supply

All of Dell's Chinese vendors are using short-lived capacitors, deliberately. They're cheap, and they guarantee the product will not last beyond 2 years. Capacitors can easily be made to last 10-30 years, so there's little excuse for this connivance.

Motherboard capacitors are challenging to replace, because the board has 7 layers and it's hard to get them all hot at once for unsoldering. Power supply boards are much simpler. The box the supply is in requires disassembly, but if you're comfortable with that, you can fix a Dell power supply for a few bucks and a few hours.

The link above is to a computer repair service in Connecticut, which has kindly posted instructions with photos of what the bad caps look like and how to replace them. This particular supply is in the 5150/E510, but again, if you're comfortable with disassembly/reassembly, other supplies are similar.
_________________
Rocke T Sinetist
as in, 'it doesn't take a...'

[Doghouse]

Thanks!

I just sent that to our on-site IT person. Unfortunately, the company I work for uses Dhell computers company wide. There are nearly 300 of these little 'gems' in the building.

I heard her comment some time ago that she was constantly replacing the power supplies.

hmmm, Dhell probably pays $20 for them and sells them for $90...

[Rocke_T_Sinetist]

It will probably help, if your IT folk are the solder-and-screwdriver types; not all are. And not ALL failures are capacitors, but most are. Reason being, the government mandated efficient power supplies. You may have noticed almost no heat comes out of them. So they don't often burn out the active/semiconductor devices.

Little technical background on why capacitors kill supplies:

Capacitors (caps) are short-term storage devices. The supply works by filling them about 100,000 times a second, then shutting off and resting at least as long as it was on. When they fail to store, next time the supply turns on the cap is very empty. That appears to the supply as overcurrent and it turns itself off or folds back to its preset current limit, resulting in a drop in voltage. The motherboard won't tolerate that and continue operating.

A little deeper, just for fun and if you're still with me. Up to Pentium 3, the motherboard did little if any post regulation. It used the 5V for all the works, just as it came from the supply. That all changed with P4. To get speeds above 1GHz and still stay within a survivable temperature range, voltages had to be reduced. 'The works' now runs on 1.3-1.8V. The supply still produces the standard voltages, +12, +5, and their minus counterparts though those are not widely used.

Voltage is much less a limitation to semiconductor switches than is current. If you regulate 12V down to 1.3V, it takes less than half as much current than if you started with 5V. So now the 12V supply runs almost the entire core, and there's another set of those switching regulators on the motherboard, complete with another set of capacitors to go bad and cause the same foldback as the supply can do.

That plus the cheap capacitors is why P4 and above computers are so much less reliable than P3 and before were.
_________________
Rocke T Sinetist
as in, 'it doesn't take a...'

[Altair75]

A little more info to add.
Capacitors, like all electronic components, have a slight bit of resistance and inductance. This is called parasitic resistance/inductance and it limits how much current can flow into and out of the capacitor. The resistance causes the capacitor to heat up internally when high currents flow, this breaks down the internal insulation and eventually causes the capacitor to fail.
Wet electrolytic capacitors will 'dry out' and stop acting like capacitors and more like open circuits, this means more noise and less ability for the power supply circuit to supply power in the form of current.
Typical capacitor lifetimes are 2000 hours at 100C, run them cooler and they last longer, hotter and they fail much earlier, IE a 2000Hr cap might only last 500 hours at 120C. 2000 hours is 40 hours a week for a year.
Adding a 2nd cap might mean adding 5 years of life, but it adds cost.

[Rocke_T_Sinetist]

Note that at more typical temperatures (~40C), quality capacitors can be expected to last beyond 50,000 hours. Fer gawdsake, I've got a $15 clock radio from 1982 that's been plugged in for 245,000 hours and still works fine.

BUT, those were 1982 capacitors. About 10 years ago a cut-rate class of capacitors emerged and of course priceline resellers like Dell gravitated to them. The good ones are still out there but you have to ask for them specifically and not grumble about the price difference.
_________________
Rocke T Sinetist
as in, 'it doesn't take a...'