GW Instek GDS 2104 power supply failure

[JFerris321]

I have recently acquired a GW Instek GDS 2104 oscilloscope that was not working, when i opened the case up it did not take long to notice at least one problem with it as a number of the aluminum electrolytic capacitors had leaked and some had almost completely disintegrated where they stood. Luckily i could still make out the values of all the electrolytic caps on the board and i replaced them all, leaky or not.
Unfortunately this has only fixed part the supply issues as although the device looks to be booting now it then seems to lock up and when testing for voltages there seems to be a problem with both negative voltages (-5v and -15v) as they are both not there, in fact the negative 5 seems to be sitting at around positive 0.6 and negative 15 at positive 0.9.
Has any one any ideas where i would be best starting to try and find the cause of this fault as i am afraid i am still a bit of a rookie at this type of repair.
The power board is made by Lytec and is a LP5422 if this means anything to anyone and i have attached a pic of the board for reference.
cheers

[Paul Moir]

To trace this thing out it would be really good to have a picture from straight above looking down, and a picture of the bottom side.  It would be really good if the overhead picture showed the silk-screen labels on the output connector so we can see which one is -5v and -15v.

It's probably something simple at this stage.  Since you've got the other sections working it won't be anything in the high voltage section.

[JFerris321]

Attached are some images of the board, one directly from above and one directly from below included as per request.
I have mirrored the image of the traces so they align with the view of the board from above to make it easier to follow.
Also if it helps to know for reference the solder points that look darker on the trace pic are the re soldered caps.

[nctnico]

I'd check the diodes.

[alsetalokin4017]

I'd check very carefully for solder bridges at the new capacitor locations. It's kind of hard to tell from the photo but the ones I've circled look suspicious to me. Clean the flux off with isopropyl alcohol and cotton swabs then look with magnifying glass to make sure there isn't any possibility of shorting between the pads.

[bitshape]

Clean off all brown glue, it's becoming dark and conductive. 

[JFerris321]

I have checked all the solder joints and there are not bridges present, the gaps on them all are easily >1mm wide, its just the way the camera captured the pic that makes them look closer / bridged.
i have also checked all the joints for dry connections, given the crude that was on the board before i cleaned it up this could have happened, but sadly they are all good so i suspect the issue is not with what i have already done to try and repair this.

[nctnico]

Again: check diodes D3 and D12. You can also apply a negative voltage to them from a lab power supply (low current setting!) and see if the -5V and -15V regulators still work. That way you can check the path for the negative rails in the power supply safely (without connecting it to the mains voltage).

[Paul Moir]

I'm with nctnico; start by checking the rectification diodes and then move onto the regulator.
However, while powering up the negative regulators from a bench supply is a good idea, I don't think it'll work here.  I've attached what I think I'm seeing, and rather than using 7905s and 7915s for the -5 and -15 volts they seem to be zener referenced off the +15v supply (which I presume is working).  So they won't work unless the +15v supply is also up and running.

Can you safely access the back side of the supply while it's hooked up, keeping in mind that there are lethal voltages on the high voltage side?

This is what I think I'm seeing.  This is only one channel but the other looks identical:
(EDIT:  Oops, the polarized capacitors are upside-down.  Force of habit.)

[JFerris321]

Thanks for the suggestions so far. i have tried testing the diodes but they look to be fine, i have also removed the 2 N Channel FETS from the circuit and they look like they could be at fault (or at least one of them is) as when testing with a multi meter they both behave differently to each other even though they are both the same component. when applying power to the gates one of them does nothing and the other seems to open but with a high resistance on it.
My next challenge is to match these to a compatible replacement as i can not find any exact match for them locally source-able (currently they are both CET CEP51A3). with out waiting weeks for delivery my two options are Maplins or RS Components. So my next question is will any N Channel FET with the same voltage and current rating do the job?

[Paul Moir]

Oh, they're not PNPs being used for regulators, they're just FETs being used as switches.  What they're doing is preventing the -15 and -5 from coming on until +15v is present using those zeners.

Any similar replacement should work fine for that then.  In addition to Id, Vds, etc, I would be keeping an eye on the Vgs threshold voltage so that the turn-on is similar to the original.

[JFerris321]

Thanks for everybody that replied of this post i am happy to say the fault looks to be resolved.
I have replaced the two N Channel FET's on the negative rails with a pair of STP95N3LLH6's and both are now back and working at their respective voltages, i have then plugged the cables back in on the oscilloscope and i am even more pleased to say that it is now booting and seems to function as normal, the only remaining issue i can see is a lot of ripple on all 4 input channels when the scope was turned on but not had a chance to play with it yet so this might be a set up or probe issue.