Amilo M1425 Notebook repair - how to locate root cause on shorted DC input?

[dicky96]

You are talking about 19V rail (PL2 is still in place)?
I injected 2,5V @19V rail (DC-in), current is 1.5A.
GPU is getting warm. On higher voltage/current it's getting warmer (and faster), so I did this only for a few seconds.
I know, it's getting warm at normal operation, but I don't know how far can I go with removed heat sink.

No I meant with PL2 removed and drive the load direct from a bench PSU with CC mode.  If you drive it from the 19V side you are making the buck regulator on that rail drive intro a (nearly) short circuit.  It won't like that.

I heated CPU, GPU and the third "CPU-like" (bridge?) IC using a heat gun (not too hot, ~60-70°C):
The only one that showed a reaction was GPU (currrent increase from 1.5 to 1.8A @Vin=2.5V ).
I think changing load over temperature would be relative normal (it's a semiconductor).
I also think the problem is a permanent one, not a temperature problem.

GPU supply, I measured at one of the bigger ceramic caps at the backside of the GPU, is about 1.85V @Vin=2.5V (2.1V @Vin=3.0V).
That's the same voltages (1.85V, 2.1V) I can measure at PL12.
That's 1.2V rail, the voltage is to high.

Found shorted high-side switching MOSFET (PQ49) !



Removed PL12 to inject 1.2V from external supply.
Current is 450-500mA. GPU stays cool, so I hope, it's not damaged.

I will remove MOSFET (PQ49) and try to get replacement...

Ah I should have explained more the idea of this fault finding technique. It wasn't that I thought that the fault was heat sensitive.  When a component fails and goes low resistance it often becomes very thermally unstable as well - so heating and  or cooling can have a noticeable effect on the resistance of the faulty component if you monitor the resistance of the 'short' and your DMM is sensitive enough to read .1 or even .01 ohm resolution 

The s/c mosfet is an interesting find.  Does the short circuit from PL1 or PL2 inductors go away with the mosfet removed?

Regards a GPU getting warm in normal operation - yes it would.  But that is with the GPU being clocked at high frequencies which is most likely not the case at all with your current s/c fault on at least 2 voltage rails. And heating the other semiconductors didn't have much effect 

The fact it reacts to you warming it seems more likely to indicate this is where your short is.  If you are saying the s/c mosfet is related in some way to a 1.2V voltage rail which supplies te GPU and is too high (which could be the case if the mosfet was supposed to be regulating that rail) it is possible you have a s/c mosfet and a faulty GPU because of that. 

The short reacting to external heating specifically when you heat the GPU and the fact the GPU is the thing generating heat proportional to the amount of current when you feed into the voltage rail from your bench psu tell us quite a lot I think.  Having said all this I am not in anyway an expert on motherboard repair, in fact I am a learner.  I just have lot of experience on electronics repair techniques in general.

Rich

[carl_lab]

No I meant with PL2 removed and drive the load direct from a bench PSU with CC mode.  If you drive it from the 19V side you are making the buck regulator on that rail drive intro a (nearly) short circuit.  It won't like that.

I don't see a significant difference with or without PL2 populated.

I unplugged original 19V-PS and injected voltage from laboratory PS at the solder joints of the jack.
No problem at all. I set I=1A (later 1.5A) max and V=2.5V, so I found GPU is warming.



Then I measured voltage at GPU (1.85V) and found out, from which rail it comes (PL12, normally 1.2V).
I removed PL12 and injected 1.2V at load side. I = 450 to 500mA so I think (hope), GPU is not shorted.

Does the short circuit from PL1 or PL2 inductors go away with the mosfet removed?

There was no real short.
Shorted MOSFET is removed now, and I'm looking for a replacement.

I'm not quite sure, if I should operate the PCB without replaced MOSFET for testing, maybe it's not a good idea to switch it on, without 1.2V supply (GPU and more?)...

[dicky96]

Ahh  I see on the schematic  PL2 is not part of a buck regulator circuit.  It is an input filter inductor or something like that on the 19V rail.  I agree it makes no difference whether you power using the bench supply from the DC jack or PL2.  Sorry I didn't look before. 

PL12 is part of a buck regulator however.

I would certainly replace the MosFET first :-)  Where is PQ49 on the schematic?

Rich

[carl_lab]

Where is PQ49 on the schematic?

[dicky96]

Cheers mate that explains why I couldn't find it.  You mis-typed PQ49 not PQ59 

That isn't good is it, for anything that was on the 1.2V rail. A drain-source short on PQ59 is going to connect Vin directly to the 1.2V rail so 19V is likely to go to where it shouldn't. Until of course the PSU trips out or whatever.

This is also a good example why you shouldn't power the 19V with your bench supply to see what gets hot on the lower voltage rails. You should power the short circuit rail directly with the inductor removed and your bench supply set for the maximum voltage of that rail. Which was my original posts suggestion.  Having said that you already had a low resistance on this rail before you started powering with the bench supply so any damage done was already done.

So even before you change the s/c FET PQ59 you need to find out why you have a 5.1 ohm low resistance on the load side of PL12 other wise you will quite likely just blow the replacement FET anyway.  If you still have 5.1ohm on PL12 that is. I assume you still do?

I would connect a lab supply to the load side of PL12 set to 1.2V maximum and with current limit on.  Then wind up the current and see what gets hot - but from various observations so far it is going to be the GPU. The schematic says this 1.2V 7A rail is the VGA supply.

Alternatively you could try my trick of warming and cooling the GPU and at the same time measure resistance PL12 load side to ground.  If you have a sensitive DMM and can read 0.01ohm or 0.1ohm increments and the resistance changes when you warm the GPU that is another good indicator you have located the underlying problem. 

What part number is the GPU anyway? Is it a BGA package, or something more easily workable?

Rich

[Rasz]

5.1 ohm low resistance 
this 1.2V 7A rail

why do you think 5 ohm is low resistance on this(10W gpu) rail?

[carl_lab]

Cheers mate that explains why I couldn't find it.  You mis-typed PQ49 not PQ59 

Sorry...

Having said that you already had a low resistance on this rail before you started powering with the bench supply so any damage done was already done.
So even before you change the s/c FET PQ59 you need to find out why you have a 5.1 ohm low resistance on the load side of PL12 other wise you will quite likely just blow the replacement FET anyway.
If you still have 5.1ohm on PL12 that is. I assume you still do?

I measured now 6.5 Ohm, what is not excessively low for a 1.2V/7A rail.

I=U/R=1.2V/6.5Ohm=185mA, or in other words 2.6% of full amperage.
The MOSFET (4404) is able to drive 23A.

So I hope, hickup protection mode of original DC adapter prevented a real short to GND.
I just measured a few ms-short 5V pulses at DC-in, repeated every few 10ms. 

I would connect a lab supply to the load side of PL12 set to 1.2V maximum and with current limit on.  Then wind up the current and see what gets hot - but from various observations so far it is going to be the GPU. The schematic says this 1.2V 7A rail is the VGA supply.

That's exactly, what I did already:

I unplugged original 19V-PS and injected voltage from laboratory PS at the solder joints of the jack.
No problem at all. I set I=1A (later 1.5A) max and V=2.5V, so I found GPU is warming.

Then I measured voltage at GPU (1.85V) and found out, from which rail it comes (PL12, normally 1.2V).
I removed PL12 and injected 1.2V at load side. I = 450 to 500mA so I think (hope), GPU is not shorted.

I forget to add: GPU stays cool at 1.2V/500mA.

What part number is the GPU anyway? Is it a BGA package, or something more easily workable?

It's an ATI Mobility Radeon 9700. BGA, about 30mm x 30mm size.

[dicky96]

why do you think 5 ohm is low resistance on this(10W gpu) rail?

@rasz
Because carl_lab is measuring with a DMM on resistance range so that should not even turn on semiconductor junctions in the GPU, plus the GPU only draws a lot of current when it is working at high frequencies.

5 ohm on a DC resistance test seems very low to me.  But as I mentioned earlier I am no expert on motherboard repairs so I could be wrong.

@carl_lab
As you also said you have low resistance on Vcore it is probably worth checking high side FETs PQ2 and PQ4 for short circuits as well


Rich

[carl_lab]

...probably worth checking high side FETs PQ2 and PQ4 for short circuits as well

They seem to be OK.

[carl_lab]

Today I got the MOSET(s). I soldered one in and injected voltage at the 19V DC in (rising slowly from 1V to about 16V). Below 10V current was very low, but then rised up to 1A @ 16V). Suddenly smoke came from the backside of PCB.
I switched off and found a cracked PU10 = ISL6224CB (SMPS controller of 1.2V rail) that released the magic smoke.



Tried to order a few ISL6224 but they seem to be obsolete and not available anymore.
Intersil named ISL6269 as replacement, but wrong casing (QFN 4x4 instead of needed SSOP-16).

Found some ISL6224 on ebay.
Waiting again...

[Rasz]

PU10 was already damaged?
your replacement PQ59 was DG shorted?
you soldered it in wrong orientation?

even diagram is wrong ;] look at the arrows on PQ58/59, its picasso FET, someone knew all the parts, just didnt know the order?

why do you think 5 ohm is low resistance on this(10W gpu) rail?

@rasz
Because carl_lab is measuring with a DMM on resistance range so that should not even turn on semiconductor junctions in the GPU, plus the GPU only draws a lot of current when it is working at high frequencies.

5 ohm on a DC resistance test seems very low to me.  But as I mentioned earlier I am no expert on motherboard repairs so I could be wrong.

its best to experiment, take any graphic card and measure vGPU rail, it will be <1ohm on modern 100W chips.

[carl_lab]

PU10 was already damaged?

Probably it was.

your replacement PQ59 was DG shorted?

No.

you soldered it in wrong orientation?

No.

[Rasz]

recheck PQ58/59 and PD11 for shorts just in case

[carl_lab]

recheck PQ58/59 and PD11 for shorts just in case

I have removed the broken controller IC and PL12 to re-check following components:

PD9+11:
reverse direction: high-impedance
forward voltage: 0.2 V -> OK (schottky)

PQ58+59:
Rgs: high impedance
Rds: ~40 kOhm (in-circuit), reverse voltage 0.5 V (body diode)
Rdson: <0.1 Ohm @Vgs = 5 V -> OK

PR70: 1.5k Ohm (OK)
PC56: 100 nF (OK)

Resistance (load side) 1.2V rail to GND: ~6 Ohm (OK)

So everything should be OK.

Still waiting for some chinese electronic parts...