CubeRevo HD PSU repair

Attachment 66292Attachment 66293Attachment 66294Attachment 66295
A few things to report, but I’ll mention the main points first.
1 We are back to where we were two weeks ago. I’ve fitted a replacement fuse at F81 and I’m now getting the same voltages at the output end of the PSU board as before i.e. all except 3SV and 5V, and no SATA to power a HDD. With a link between R98 and R95, I get them all, and SATA works. So the readings in the pic I posted on Jan 7th are still valid.
2 With the main board connected, we still get +5VS.
3 I haven’t cut continuity to the transformer on the front panel board yet in order to do this check. Connecting both sets of wires to the front panel again kills +5VS.
4 However, joining only the longer connector, which carries +5VS produces a different result. First, I get an LED flashing slowly in the standby switch, but nothing on the display. Also we still have 5.1V on +5VS on the PSU Board (J7).
Pics attached of the new fuse, and repaired solder on L81. Also rest of PSU in case its useful.
A few other points.
Voltage from hotside heatsink to TH61 is 333V.
I checked all the diodes I could find and all were OK except:
D21 about 0.09V in both directions
D24 0.4V in one direction, 1.4V in the other.
D41, D42, D43 infinity in both directions, but I’m not confident about those readings. Once D43 showed OK.
To practice I took a capacitator off a redundant PCI card from an old computer. As I don’t have a capacitance meter, I did a resistance check and watched the readings start low and steadily climb; switched the leads and same again. I don’t know its rating – it’s just labelled SM1106. There’s another one labelled KM1106. On another card, I’ve got one rated 222µF and 16V.

This is great news, all supplies up again and a flashing LED. At long last a sign of life!
We can ignore the meaning of the flashing for now since we are powering on in an abnormal way, I assume the R98/R95 link is still present.

I was a little concerned by your readings of D21 but after checking there is a 100R resistor (R21) across C22 which would probably explain that.
I don't see anything too worrying in your pics so I guess we need to remove C21 and check it using the resistance check method.
I notice from your pics that C22 is a different value to C21.
For completeness I would suggest removing and testing C22 as well.

I need to point out that these capacitors are polarised so when replacing them they need to be fitted the correct way around.
Just to confuse everyone it is usual for the negative lead to be marked rather than the positive so be careful.
You can always check the orientation in you pics before fitting them.

If C21 is faulty and you don't have the right value you may be able to use the 220uF 16V you found for now but we cannot be sure if this will work properly.
Of course you can always order a new one - perhaps I am just getting a little impatient!

While you have C21 and C22 removed I suggest you remove R21 as well you should then hopefully be able to test D21. You will need to unplug the motherboard. This way we can be sure of everything in the 5VS supply, the only other component is L21 which is not a worry.

I looked at some old caps I had removed from a PSU a while ago.
They seemed OK using the resistance test, the tops were not bulging but the bottoms were!
The PSU at the time was working OK but I noticed that the ends of these caps felt quite hot so I replaced them anyway.

OK, I’ve taken C21 and C22 off the board.
C21 resistance starts in the 8MΩ range and then GOES DOWN. Same with probes reversed. Repeated the check a couple of times after shorting with a pair of pliers to discharge. Same patterns with different numbers.
C22 shows resistances going up. Unfortunately, I’m not well equipped here and just used an ordinary G-Clamp to hold the capacitors in place, and C22 is a bit too small and got squashed in the process – pic attached. So it may be damaged.
As you know, C21 is rated 370µF and 10v, and I’ve got one rated 222µF and 16V. C22 shows 22µF and 50v. I’ve got one rated 22µF and 16v. Shall I go ahead trying these if only on a temporary basis?
I’ll remove R21 later today
Attachment 66297

Both caps should only have 5V on them so the caps you have are worth a try.
Your squashed C22 may be ok but it's safer to use the 16V one you have.

C21 doesn't seem too healthy - don't know why it should behave like that!
So chuck the caps in, observing polarity of course, I don't think we have anything to lose.

If you feel the need too check D21 remove it and check it off board, you may find it's a bit tight in the holes so be careful.
I suggested removing R21 as it would be easier but you need the caps out to test D21 in circuit properly.

Feeling optamistic about this, if 5VS is ok with everything plugged in remove the link on R95/98 and see if it powers up from standby button.

Good luck!

Supplementary info:
I disconnected R21, and D21 showed 1.5v and infinity, albeit the latter with a an occasional flash of voltage.
R21 reconnected and reads 95Ω. While R22 shows 3.28kΩ. Hope nothing to worry about there.

Will try other caps tomorrow - probably late afternoon.

D21 looks OK although I would expect less than 0.6V but infinity one way is always good.

R21 is 100R resistor and R22 is 3K3 (3.3K) but is on the -25V line.

The 220uF you are going to try is half the original value (470uF) but I will be very surprised if does not work or at least show an improvement.
So tomorrow is the big day!

Sorry, I should have said 0.15v.
And, yes I was puzzled by 50V given thAttachment 66299at the largest output voltage we've got is 30, but what do I lnow?. But just for you to double check, here’s a better pic. 5, small space (no decimal point), bigger space V; next line 2, same small space, 2, bigger space, µF. Yeah, not easy to understand these Chinese!

That's more like it 0.15V these diodes are fast recovery diodes and have lower forward voltage drop than normal silicon diodes (0.6V).
Switch mode power supplies run at something like 40 to 70KHz and hence the need for fast recover diodes

I cannot understand why they used 50V cap for C22 unless they just had a stock of them hanging around.
I agree it looks like 50V in your pic.

You have reminded me of something that happened to me a long time ago.
I needed a cap for a 36V supply I was making and found what I thought was a 63V cap.
When the supply had been running for a while I could smell something getting hot but could not work out what it was.
Then all of sudden there was a load bang and there was silver paper everywhere around the room.
The case of the cap had popped off causing the bang and the insides of the cap were scattered around the room.
When I found the case of the cap I looked closer and found there was a very faint decimal point between the 6 and the 3. It was only rated at 6.3V

Lol - that wouldn't have happened to me stroker! I wouldn't even have known what a 63v supply was if it hit me in the face - come to think of it a few other things did hit you in the face or got close.
Anyway, I'll get back to you tomorrow.

I’ve found I also have a cap rated 300µF and 25v. Would that be better than 222µF and 16V?

And now I've found on that is actually rated 22µF and 50v - on a simple modem card.

The 300uF 25V would be better.

Use the 22uF 50V (just because it's the same as the old one) unless you want to keep the modem card intact.

Sorry for the delay in answering I've been out and about.

Unfortunately, I’ve hit a snag here. I’ve now spent an hour and a half, maybe two hours making five attempts to get old capacitors off. They all have pins not folded over but vertical. I clear the solder off the underside, but the components are not loose. I wiggle a bit – wiggle the capacitors that is – and I can lift the 1mm or 2. With the first I tried to pull it up, fearing that the leads would simply pull away from the cap, and in fact that was what happened. Again unfortunately that was the 22µF and 50v one, and it was the only one I have. Blast! Should have tried one of the others where I have several to play with.
The best I’ve been able to do is remove the solder and then gouge a hole through the board with a bradawl. That partially worked with two, but perhaps inevitably one of the leads got partially broken in each case.
Any tips someone?

Just a thought, should we get a replacement C21 anyway, because if we find a solution, we might as well do a like for like? And if it comes to C22, should we go for a 50v version?

Damn I was hoping you won't have this problem!

I believe you have some electrical knowledge so you will understand that high currents need heavy conductors or cables.
If they are not heavy enough you will have a large voltage drop along them or at the extreme they will burn out like fuses.
There are heavy currents involved in electronics especially PC motherboards.
You won't see the heavy conductors because they are buried in the internal layers of the board.
These are usually referred to as power planes and ground planes.
Heavy conductors will conduct heat as well as electrical current and act like a heatsink absorbing the heat from your soldering iron.
So to solder on these power planes you need much more heat than usual, so you need a high power soldering iron.

I think you have a large soldering iron, perhaps you could use this to heat both leads at the same time while adding more solder.
Once the solder is fluid in both holes you then pull the component out - be careful it will be bloody hot!

Another possibility which will sound quite extreme is to use a hot air gun which is normally used for paint stripping.
Again add fresh solder to the solder joints and perhaps some flux as well, apply heat with the gun.
As above pull the component out when the solder is flowing freely.
Using this method a lot of nasty fumes may be produced so do it in a well ventilated area.

If all this seems a bit too much I suggest you use the 222uF and the lower voltage 22uF caps you already have.
Fit the caps lifted a little off the board so they are easier to remove if/when you need to.

It won't hurt to get the correct values for the caps - they don't cost much.
You could order them anyway if you don't use one or other of them it won't be a big deal.
I was just hoping to see the thing running sooner rather than wait for new caps to arrive.
I guess a few more days won't be such a big deal if you'd rather go down that route.

Actually my 22µF and 222 µF caps are on a PCI card and fixed the same way. Anyway, before trying the extra heat methods, I decided to do the bradawl gouging method another go and succeeded. Bet you they don’t teach that in college! So I had a 330µF (not 300 as I said before), showing OK in the resistance test, for C21. The 22µF ones weren’t up to the job because the leads were too short. The one I took out of C22 has long leads which are bent out to span the gap. So I put the old one back, along with the 330µF one in C21.
Powered up with the R98/R95 link still in place and just the one connector to the front panel. OK, same as where we were – flashing LED at the front, SATA powered.
Removed the link, and still OK – LED and SATA working.
So, joined the other connector to the front panel. Up to now, that would have killed off +5VS – but not this time! We still have 5.1V between J7 and GND, SATA working and the flashing LED. We also have 2.4v at P/S – J9 - and a steady 11.9v at the fan header – it’s just the fan doesn’t turn
So great work Stroker – looks like your deduction that C21 was faulty has been proved right.
We’re not there yet though. Nothing else shows on the front panel and the standby switch has no effect. I haven’t finished cutting the track between +5VS and the transformer on the front board.

Don't bother with the track cut now, I wanted to do that because I thought the fault causing the 5VS problem was on the front panel. It looked like you almost cut it anyway, check if 5VS still gets to the middle connector of the transformer on the front panel. If there is no continuity bridge the track-cut with a solid piece of wire you will have to scratch the solder resist from the track each side of the cut first so you can solder the wire to it. The track should appear as copper with the solder resist removed.

I think some boxes start up fully when first powered up and not in standby which would explain the HDD running, but this did not happen before, this could be because we now have a stable 5VS. Then why does't the fan run - does it comes on at a certain temperature or run continuously?

I noticed some time ago there is a coin cell on the motherboard like used on PC motherboards. I would expect this to support the RTC (real time clock) when the box is powered off but it could also power the memory containing the boxes firmware if the box use volatile memory. They usually use non-volatile memory which retains info when power is removed. Anyway easy thing to do is check the cell/battery should be 3V I guess - if it's dead you will need to replace it.

The flashing LED may suggest a firmware problem or so I read somewhere.

Do some searching on how to flash the firmware/software, I am sure I read that you can do it from a USB stick and you press certain buttons when powering up and you get a response from the display saying USB or FLSH or something. I will try to find where I read this and let you know but I expect it's easy enough to find this on satpimps. Anyway what I am looking for is some reaction from the box that would suggest flashing may be possible. Don't rush into flashing just yet - we need to consider the risks if there are any. You will need to find software to flash anyway but lets make sure we know what we are doing before we try.

Perhaps someone else could give us some advice on the best way to flash the box assuming we need to.

I'm going to get some grub now so may be gone for up to an hour.

Trying to post a PDF file - software update is on page 30.

The "Hard Turnoff" by pressing OK and Standby may be interesting to try.
Then press and hold standby until display reads booting.

Of course we don't know yet if the display is working!