Repair.IT – Overheating Presario SR5120AN motherboard

Remember two and a half years ago, approximately, I repaired my Compaq Presario SR5120AN motherboard which had a number of failed capacitors?  Ok, it was some time ago, so here is the link.

At the time, five capacitors had failed, but there were still four others of the same brand and size on the motherboard. I checked them with an ESR meter and they all passed. Fast forward to a month ago – I noticed that my computer cpu fan was getting louder, sometimes normal then suddenly high speed and this kept happening. I ran a utility to check the CPU temperature and it was …  99 degrees, wow! No wonder the fan was on turbos a lot of the time.

I shut down and took the computer apart to reveal the motherboard.


Two of those same original capacitors were showing the symptoms of failure – see the bulging top and black spots. After some effort, I was able to remove these two capacitors, then replace them with the ones in that little bag – I originally bought ten of these 1800uF 6.3V electrolytic high temperature capacitors. I checked the remaining two on the board and they check out fine.

So, reassemble the computer, and power on – leave it on for a while, and I can see that the CPU temperature is sitting reasonably stable at about 80 degrees. This is still quite hot and would appear to be still abnormal. Since I still have three spares left over, why not just replace the remaining two capacitors and be done with it.

That is what I did – took out the other two capacitors, replaced them with new ones. While I was doing this, I checked the capacitors with my ESR meter, which showed that these two were still ok, but anyway I have new ones in the motherboard now. Once the computer was up and running again, leave it for a while and then it was looking good so I decided to run the Passmark Performance Test, which stresses out the computer.


This is what the CPU and graphics card temperatures were during the test and then the cooldown period afterwards. The CPU maximum went to about 93 degrees but then back down and sitting stable at 54-60 degrees. This is amazing and shows that even though the ESR meter indicated that the capacitors were ok, replacing them reduced the average temperature dramatically. Why is that – maybe the capacitance has changed? Wait, I can check this!

Just over a year ago, I had bought from eBay, a Mega328 Transistor and component tester. I can connect the parts I have removed and compare with new parts.

These two are the failed capacitors. They appear to be back to back diodes with differing forward voltages.


Here is a new one – my final remaining capacitor. The value is 1829uF, ESR is good, with Vloss of 2.5%.

Here are the two apparently good ones that I replaced. These also appear to be good as far as the tester is concerned, however replacing these two also brought down my average temperature of the CPU. Why? I don’t really have an answer for this, but maybe someone has.

Right now as I am writing this, my CPU temperature is sitting at around 80 degrees, with CPU load at about 85% since my antivirus scan has been running for some time, but certainly nowhere near the 99 degrees at idle that it was a month ago. It has been a few weeks now, and all seems to be well.


Repair.IT – Sedco 24V / 12V Power Supply

Just over a week ago, I was asked if I repair power supplies – I said sure, so this Sedco 24V / 12V Power Supply was carried out to my car and placed in my boot. It certainly looked heavy, and when I got it home and opened the cover – I realized why.


The insides of the Sedco power supply

There are two 12V batteries inside, obviously looking very much the worse for wear – the acid corrosion on the negative terminal of the right side battery contact is a sure sign of battery failure. Certainly when I measured both batteries, I was getting only 6V on the left one, and about 4V on the right.

Ok, the usual checks are the fuses – are they intact? Yes, they measured fine with my multimeter. The actual power supply is in the metal box at the top left – the one with the five multi-wire connectors . The left connector is the incoming mains power, next one is the battery connection, then the output voltage, and two with lots of wires going to leds, etc.

Next thing is to remove the power supply. This was a fairly standard switched mode power supply, however it has a couple of plug-in boards that connect to the main circuit board. The incoming mains is rectified and conditioned by the plug-in board on the left – I don’t actually have a photo of it, but it is where the mains connector is attached. The second plug-in board seems to be for the two multi-wire connectors, doing things like driving the leds on the front panel.

This little board did have three electrolytic capacitors that were bulging – we have mentioned this previously, that these have most likely failed. I didn’t have 1000uF 35V 105C capacitors in my stock, so had to arrange to get them. I got these in due course and replaced them together with a 100uF 35V capacitor that had also failed. I checked also the semiconductors – such as diodes, mosfet’s – but they all appeared to be intact. The fuses being intact also indicates no catastrophic failure.

The power supply was reassembled, and everything connected back together, then power was applied, and… nothing – no lights, no bang – basically nothing happened other than the switch clicking on. Ok, so the power supply is not starting up – what could be the problem?

Removed the power supply and started having a closer look at the main circuit board. A fairly standard layout – the startup circuit is the one to look at. I spied a small electrolytic in the startup circuit – when I measured the ESR, it was 8.4 which is very high. Unfortunately, the capacitor was hidden away with other large components near it so I could not read the value of the capacitor. A low value capacitor could potentially have a high ESR so being high is not necessarily a definite sign of failure. This capacitor is connected to a UC3845 SMPS controller – which is a fixed frequency current-mode PWM controller.

I did a bit of research on this controller and came across a little tit-bit of information, that if the Vcc capacitor was faulty, a UC3842 PWM controller likely cannot start. As this UC3842 is in the same family as the UC3845, I thought that this capacitor was likely to be faulty. I then desoldered the capacitor and found that it had a value of 100uF 35V. At this value, definitely the high ESR means that it had failed. I had one in my stock, so replaced the capacitor – then reassembled everything.

Finally, I connected up my power meter to the power supply then turned on, and… it powered up, at least the power light was lit – also the low battery light was on. I turned off and went to have something to eat. When I came back and powered up again, I could now see that the 24V and 12V lights also came on. I measured the power rails on the back of the power supply, 12.2V for the 12V rail, and around 25.2V for the 24V rail.

The 24V rail was fluctuating a little and certainly wasn’t 24V but higher. Why is this? Ok, it has to charge the batteries, of course, so needs to be higher than 24V – at 25.2V, the batteries could be connected almost indefinitely and should not show signs of overcharging (ideally). Of course, the batteries were not in good condition, so maybe replacing the batteries would give a more stable 24V rail. Anyway, it certainly looks like it is now repaired and working again – I can’t do more until the batteries are replaced.

[Note] This power supply is used in nursing homes and hospitals associated with medical care. The power supply will provide 24V and 12V normally – and if mains power fails, it will still deliver this via the internal batteries until such time as they go flat.