HP DV6 Laptop not turning on - Fused mosfet?

[Omegaman]

smanet: If you have PQ802 removed from the board, I would make the following 2 measurements.

1) Measure across the pads of PQ802, say pin 5 to pin 1. You should get a high reading on your ohm meter.

2) Measure both sides of PL802, pin 1 & pin 2 to B+, on either PJP804 pin 1 or FL1 pin 2 (this inductor is not populated). You should get a high reading on your resistance measurement.

3) Measure both pins 1 & 2 to GND on PL802. Now this may be a problem for you. I don't know what kind of meter you have. The measurement you will get will be likely less than 10 ohm, even less than 5 ohm. If your meter can do it, zero it out first by touching the probes together and if your meter supports it zero it by pushing the button RES. Your meter may be labeled something else, or may not have this feature. If it doesn't, note the reading you get when the probes are shorted together. Subtract this reading from the reading you get when do the measurement.

If these measurements check out, it is my opinion that it would be pretty safe to put your new PQ802 on the board. You did say you have 9 of them.

You seem to be dealing with 2 different problems.

1) Problem with VGFX_CORE, i.e., PQ802 getting hot. The above speaks to this problem.
2) Problem with 5VSTBY, & 3.3VSTBY, controlled by PU9. Please see the last post on page 2 of this thread for some information on this problem.

It is probably best to work on one of these at a time. You did not provide information as to if VGFX_CORE has any voltage on it (measurable on pin 2 of PL802). If it doesn't, or possibly if it does but is the wrong voltage, it is my opinion that you might be better off to work on the VGFX_CORE (PQ802) problem first and try and fix it first. The reason for this is that if you fix the other problem, number 2 above, depending on what is going on with number 1 above it may turn on VGFX_CORE and risk damaging something else in this circuit. So, this is why the measurements suggested, and why my opinion is you should work on number 1 above first.

Good luck, and I hope we are helping.
Omegaman

[Monkeh]

1) My understanding is that 5VSTBY is approximately 1.8V & 3.3VSTDY is approximately 3.9V. B+ is now up, and is about 18V. The below is based on this understanding.

No, those are the voltages measured on the LDO outputs - if VREG5 (LDO5) is <2V, there is no way it's managing to drive the gates and bring up the main rails. The LDO outputs do not supply anything else on the system.

I think he meant no standby lines, which are those switchers driven by PU9. LDOs are internal to the chip and they work, just wrong. Or at least Monkeh assumes that standby lines are down because of funky PU9 LDO5 voltage, actually smarnet didn't measure them or didn't share the results.

Indeed, at such a low voltage I'm assuming it cannot drive the gates. Certainly not usefully.

We're definitely tackling multiple faults, the entire power supply is swiss cheese.

[Omegaman]

smanet: Hi, hope you are making progress. For the PU9 problem, the signal/net LDO5 (PU9, pin 17 - name of pin is VREG5) powers the internal analog circuits, and the external gate drives for the two switching supplies, 5VSTBY, & 3.3VSTBY and also does connect externally. It connects back to PU9 on pin 4 (TONSEL), and pin 14 (SKIPSEL). Please see my previous post on the function of these pins.

LDO5 also connects to the battery connector - CN_ADP801, pin3 (signal name on connector is - CHG_LED#) through a 510 ohm resistor (PR862, pin 1 = LDO5, pin 2 = CHG_LED#). It also connects to PQ12, pin 3 through this same resistor. Finally, it connects to cap CB856.

I don't know if you have both the battery and the AC adapter plugged in when you are making these measurements or not. If you haven't done so yet, try and power up with only the AC adapter only (no battery installed). I think the most beneficial thing you could do would be to measure the voltage across PR862 with the battery plugged in, since it should be simple, and with the battery unplugged, or whatever condition you are making these measurements under.

Measuring the voltage drop (if any) across PR862 would allow you to calculate the current being drawn from LDO5 here. PU9 can supply a maximum of 100mA. Please note that even if LDO5 were shorted through this resistor, PR862, directly to GND (on the CHG_LED# side of the resistor), it would only draw 10mA from LDO5. If CHG_LED# were shorted to B+, it would draw 35mA from LDO5. Whatever is drawn through PR862, if any, would need to be added to the current for TONSEL, SKIPSEL (both should be very low, probably uA), and the gate drivers of the 2 standby switchers, likely considerably higher that the 2 just mentioned, but still in the low mA range. Likely only 2 of these will be on at any given time.

What I am trying to do here is to help you determine if PU9 is bad and needs to be replaced or if the issues you are seeing on these 2 standby voltages could be caused by something external. Since these should be quick measurements, it is my opinion they are worth doing even if they do not reveal the cause.

Good luck, and please let us know how it is going.
Omegaman

No, those are the voltages measured on the LDO outputs - if VREG5 (LDO5) is <2V, there is no way it's managing to drive the gates and bring up the main rails. The LDO outputs do not supply anything else on the system.

LDO5, which I believe smanet said he measured at 1.8V or so, and it does connect to other places on the board. Please see my comments to smanet above. Yes, LDO5 may not be able to drive the gates properly, but since there is some voltage being produced on it, it will try. The Gate-Source Threshold Voltage of the FETs, PQ808 & PQ809 is 1.8V typical and 1.2V minimum. So, yes it probably can drive the gates on LDO5. Just to be a little more complete, the Gate-Source Threshod Voltage of the other FET used for 3.3VSTBY is 2.3V typical and 1.0V minimum.

What my intent here is, is to try and determine if PU9 is bad, or the issue is caused by some external source, before smanet changes PU9 (which seems to be bad). There may be no issue whatsoever with the things I have suggested to smanet, but since they should be pretty easy and quick, I think they would be worth doing. Sometimes surprising things are found where they are least expected.

Thanks for your comment. It looks like you have gotten him quite a far along on his daunting task.
Omegaman

[Monkeh]

Connections though moderate value resistors merely for purposes of driving LEDs are highly unlikely to be a concern. Even if PQ12 were short, the current would be negligable, and PR862 is highly unlikely to have become lower in value!

As far as ability to drive the gates, yes, it may reach the threshold - but the drivers will be held off at such a low voltage anyway. See the block diagram.

[Omegaman]

smanet: Something I should have mentioned in my previous post is that you could do a continuity check on connector CN_ADP801. Using your ohm meter, with AC adapter removed and battery removed, check between pin 3 and pin 2 on this connector. You also could check between pin 3 and pin 6. You should not get a low impedance (resistance) reading with either measurement. You certainly should not read zero (0).

What you are looking for here is if pin 3 is shorted to GND (through the 510 ohm resistor), or more importantly if pin 3 is shorted to pin 2 (B+), through the same 510 ohm resistor. I think pins 3 & 2 are next to each other on this connector. IF it were shorted to B+ through this resistor (PR862), then LDO5 would be drawing an extra 35mA from this supply, which has a maximum of 100mA drive capacity. This would not effect B+ at all, but could possibly cause an issue with PU9.

I hope this may help. There may be nothing here, but since it should be fairly easy and quick, it cannot hurt and could possibly help.

Omegaman

[Omegaman]

Connections though moderate value resistors merely for purposes of driving LEDs are highly unlikely to be a concern. Even if PQ12 were short, the current would be negligable, and PR862 is highly unlikely to have become lower in value!

As far as ability to drive the gates, yes, it may reach the threshold - but the drivers will be held off at such a low voltage anyway. See the block diagram.

Yes, the LEDs are not the concern here. I provided the current draw through PR862 if there were a short to GND, in my post. If LDO5 is shorted to B+ through this 510 ohm resistor (at the connector, CN_ADP801, pin 3 to pin 2) then that could add an extra 32mA to 35mA approximately to LDO5, which has a maximum current capacity of 100mA. This is 1/3 its capability, and could effect the operation of PU9 depending on the current draw of the gate drivers (assuming 2 gate drivers could be on at the same time, one from each supply). I don't know if PU9 staggers in any way these 2 standby powers or not.

As to the gate drive, please see page 2 of the FET, Si4134DY data sheet. As I mentioned in my post, the Gate-Source Threshold Voltage is 1.8V typical, and the minimum is 1.2V. So, during normal operation the gates of these FETs will be driven at 1.8V approximately. As far as I know, smanet has not measured the actual gate voltage on either of these FETs (PQ809 & PQ808), and I have not suggested this. My understanding is the values of 1.8V is on 5VSTBY and not the gate of either FET. So, we don't really know what voltage is on the gate of either FET is, and without a scope he most likely cannot measure it. Please correct me if I am incorrect.
 
I mentioned your points in my post, and again the goal here is to try and eliminate an external cause of smanet's issues before he changes PU9. Since these measurements should be very easy and quick, it cannot hurt to make them, and there could be something there, or maybe not. The way to find out is to take the simple measurements.

Thanks.
Omegaman

[Monkeh]

As to the gate drive, please see page 2 of the FET, Si4134DY data sheet. As I mentioned in my post, the Gate-Source Threshold Voltage is 1.8V typical, and the minimum is 1.2V. So, during normal operation the gates of these FETs will be driven at 1.8V approximately.

No, they would be driven right up to the rails to turn them hard on.. Driving them just to the threshold would not work. Which is irrelevant, because if you check the block diagram for the TPS51125, you'll see it has a schmitt trigger set to keep the drivers off if VREG5 is too low.

Neither switching section will be active, and there's no realistic path for enough current to pull VREG5 down that low externally to the chip. Excluding physical damage to the board which certainly does not appear to be present, and is not consistent with the symptoms across the board.

[Omegaman]

Monkeh: Yes, you are correct and I agree with the gate drive level during normal operation. I should have said a minimum of 1.8V, but they will be driven much higher and closer to 5V. I sometimes say one thing in my head but write something a little different.

no realistic path for enough current to pull VREG5 down that low externally to the chip. Excluding physical damage to the board which certainly does not appear to be present, and is not consistent with the symptoms across the board.

Here, I agree and disagree. If a potential of 1/3 of the maximum current capacity of PU9 is not enough current to effect the operation of PU9, and PU9 is not functioning properly in my understanding, then I am not sure what is. If board damage is present or not, at the connector, is what the measurements I suggested would help to determine. Since we do not know what the original root cause of this issue is, we do not know for sure if there is board or connector damage. Board damage cannot always been seen just by using sight. It is consistent with the symptoms IF there were a short from 5VLDO (VREG5) to B+ at the connector, or anywhere else, in my opinion. It all keys on how much current the gate/gates drive there is. If enough more current is drawn from PU9 than it can handle, then it cannot be predicted what PU9 will do.

My troubleshooting techniques, are to eliminate the simple things first, IF they have any chance of causing or having something to do with causing the problem I am working on. This is because, first it is simple and quick, and second, there have been many times when working on a problem one gets caught up on the more complex but more likely causes only to find out a few hours or days later, when they decide to check the simple but unlikely things that they find the cause of their problem. I worked on a problem once for some time and ignored an input to a chip because it just connected to a jumper to GND and the jumper was populated. This was many years ago. So, thinking this is not the cause because this input should be at GND level, I worked on more complex areas of the chip to try and find the cause. After some time I finally decide to, hey, just check that signal at the chip when the failure happens. It glitched up to over 2.5V and ended up being the cause of the problem. The reason for this was that the trace ran all the way across the board (about 16 inches long), and through most of that length next to a high speed signal which cross-talked on to this signal, which should have been connected directly to GND.

So, from experience I have found eliminating the simple things, even if they are not the most likely cause, is very helpful and although they are not more then they are, when they are it sure saves a LOT of time.

Thanks.
Omegaman

[Monkeh]

no realistic path for enough current to pull VREG5 down that low externally to the chip. Excluding physical damage to the board which certainly does not appear to be present, and is not consistent with the symptoms across the board.

Here, I agree and disagree. If a potential of 1/3 of the maximum current capacity of PU9 is not enough current to effect the operation of PU9, and PU9 is not functioning properly in my understanding, then I am not sure what is. If board damage is present or not, at the connector, is what the measurements I suggested would help to determine. Since we do not know what the original root cause of this issue is, we do not know for sure if there is board or connector damage. Board damage cannot always been seen just by using sight. It is consistent with the symptoms IF there were a short from 5VLDO (VREG5) to B+ at the connector, or anywhere else, in my opinion. It all keys on how much current the gate/gates drive there is. If enough more current is drawn from PU9 than it can handle, then it cannot be predicted what PU9 will do.

The gate drivers will not be active due to the undervoltage lockout. If there were a fault to B+ it would be sourcing current to LDO5, not sinking, and the voltage would surely go much higher than 1.8V unless further breakdown occured inside the chip - in which case it's dead anyway.

Anyway, this isn't making progress. smanet can decide how to proceed and let us know what he finds.

[smanet]

smanet: If you have PQ802 removed from the board, I would make the following 2 measurements.

1) Measure across the pads of PQ802, say pin 5 to pin 1. You should get a high reading on your ohm meter.

I have set the multimeter to 200k, I'm getting something like 13 or 30 (depending on the probes polarity) and then straight to 0.

2) Measure both sides of PL802, pin 1 & pin 2 to B+, on either PJP804 pin 1 or FL1 pin 2 (this inductor is not populated). You should get a high reading on your resistance measurement.

I have set the multimeter to 200k, I'm getting something like 6.3 and then straight to 0.

3) Measure both pins 1 & 2 to GND on PL802. Now this may be a problem for you. I don't know what kind of meter you have. The measurement you will get will be likely less than 10 ohm, even less than 5 ohm. If your meter can do it, zero it out first by touching the probes together and if your meter supports it zero it by pushing the button RES. Your meter may be labeled something else, or may not have this feature. If it doesn't, note the reading you get when the probes are shorted together. Subtract this reading from the reading you get when do the measurement.

I'm getting around 2.6 from each side (I don't have the best multimeter but I'm going to borrow a good one soon).

If these measurements check out, it is my opinion that it would be pretty safe to put your new PQ802 on the board. You did say you have 9 of them.

I have in plan to do that when I'm sure that I will not break another

Between Pin 1 and Pin 5 of PQ802, which is off now, I'm getting around 3 ohm from each direction.

Doing the same test on PQ803 and PQ804, I'm getting around 1.7 (which would let me think my meter isn't the best, is it?).

I have also to report something strange that happened during my last measurements. First of all, without changing anything to the PCB, I have provided power to the PCB and I have started to smell something that was way too hot (or even burning), but I cannot see anything on the board itself that looks like it's gone.

That said, I was able to see the 19V for a while and now it's gone again.

The funny note, at least for me, is that when I don't have the 19V on the connector, the led on the power jack is blinking (it looks like the board is in standby) and PQ808 and especially PQ809 are becoming really hot as I provide power.

I don't know what to believe anymore

If you like to bear me and ask me for more measurements, I'm happy to do them (please indicate what PIN to check so it will be simpler to me).
Otherwise, I will understand you

Thank you all again

[Monkeh]

Between Pin 1 and Pin 5 of PQ802, which is off now, I'm getting around 3 ohm from each direction.

PQ802 is, as we knew, shorted. Dispose of it, it will require replacement - later.

Switch your meter to diode mode, this will ensure the correct polarity (some meters are reversed in resistance mode). With the positive lead on drain (pins 5 to 9 on all FETs on page 43) and negative on source (pins 1 to 3), you should read open circuit. Any part which displays a reading in this direction should be removed and the measurement made again on the component directly, if it again conducts, the FET has broken down and requires replacement. Do this for all FETs on page 43, and we'll go from there. Don't replace any components, just remove as needed.

E: I should clarify that a reading of a volt or two in diode mode shouldn't be a concern - only low readings, half a volt or below, are really of interest. If the reading is climbing, you're charging a capacitor, that's not a problem.

I have also to report something strange that happened during my last measurements. First of all, without changing anything to the PCB, I have provided power to the PCB and I have started to smell something that was way too hot (or even burning), but I cannot see anything on the board itself that looks like it's gone.

Components can get extremely hot without looking like anything has happened - the epoxy the ICs are encapsulated in is very durable. PQ820 only made itself visible because an enormous amount of energy was dissipated at once.

The funny note, at least for me, is that when I don't have the 19V on the connector, the led on the power jack is blinking (it looks like the board is in standby) and PQ808 and especially PQ809 are becoming really hot as I provide power.

This is due to the power supply shutting down and restarting due to the short.

I don't know what to believe anymore

You have a multitude of faults and they are going to be hard to track down.

E: Please remove the coin cell while we're working to avoid any oopsies involving it.

[Omegaman]

smanet: The best I can tell from your post the measurements you made look to be ok helped by the fact, as you clearly indicated, you did this measurement by swapping the probes and measuring again, however I do not know what you mean by "...and then straight to 0". You can change your meter down from 200K to the next lower setting, probably 20K and you will get a more accurate measurement. If you mean that when leaving your probes on you first measure the values you indicate but then it drops to zero while they are still connected, then you may be charging up something (cap, etc.) with your meter. I do not understand what you mean here so cannot say.

I see that Monkeh has some suggestions for you so will let him help you with this as you getting suggestions from multiple people may just make things more confusing for you. I will just make one more comment that may help you to not change out FETs that may not be bad.

Make sure when you do any measurements you make on the FETs on page 43 that the CPU is NOT installed. I believe you have it out, but just wanted to make sure you knew that. If it is in, you will measure a very low resistance, almost certainly below 5 ohms on the bottom FETs (those connected to GND on pins 1-3). This would cause you to unmistakably remove FETs that are not bad. This has nothing to do with the FETs and everything to do with the CPU, and these particular FETs cannot be measured in circuit reliably if the CPU is installed.

Good luck!
Omegaman

[Monkeh]

Yes, do not make any measurements with the CPU fitted. Not only will it mess with them, but the open circuit voltage of the meter could damage it (if it's not already wrecked).

[smanet]

smanet: The best I can tell from your post the measurements you made look to be ok helped by the fact, as you clearly indicated, you did this measurement by swapping the probes and measuring again, however I do not know what you mean by "...and then straight to 0". You can change your meter down from 200K to the next lower setting, probably 20K and you will get a more accurate measurement. If you mean that when leaving your probes on you first measure the values you indicate but then it drops to zero while they are still connected, then you may be charging up something (cap, etc.) with your meter. I do not understand what you mean here so cannot say.

You're right, I can see just for a second the resistance and then the multi meter goes to 0, so most likely the multimeter is charging something.

I don't mind to get help from more people, but I need clear directions or I will get lost
Also it's good to see diverse points of view.

I have removed PQ809 and I have got back my 19V.
I have tested it again with my multimeter and it seems to be shorted now.

I have previously tested PQ808 and PQ809 by removing them from the board and, as far as I can tell, they were good, so definitely something bad happened here.

I have kept PQ809 and I'm going to run another test tomorrow with a more professional multimeter, maybe the late times doesn't help me too.

I will remove the battery and check all the mosfet on page 43.

The CPU is not on the board since a while, so it should be safe if it's not already broken of coruse!

Thank you all!

[Monkeh]

Are you absolutely certain you put them back in the right orientation and with no solder shorts underneath?

The specific meter is irrelevant, even the cheapest will work for this. Just use diode mode to be certain of polarity.

[smanet]

Are you absolutely certain you put them back in the right orientation and with no solder shorts underneath?

The specific meter is irrelevant, even the cheapest will work for this. Just use diode mode to be certain of polarity.

Yes, I have actually used the pictures taken and posted here to be sure they were exactly oriented in the same way.

I didn't see any solder shorts underneath after removing it, if I will put it back, I'll make sure to double check everything again.
By the way, my understanding is that if you have a short between 1 to 3 and from 5 to 8 it's not really relevant, as they're all attached together internally. Am I right in thinking this? So basically of course we don't want shorts between source and drain and as well between gate and source/drain.
However, I'm not looking to mess up anything but to improve myself in soldering.

I think I have the cheapest one and I can borrow a good one, so at least I get to compare.

I'll update you all tomorrow

[Monkeh]

By the way, my understanding is that if you have a short between 1 to 3 and from 5 to 8 it's not really relevant, as they're all attached together internally. Am I right in thinking this? So basically of course we don't want shorts between source and drain and as well between gate and source/drain.

Correct, 1-3 are common and 5-8 are common. You're looking for shorts between those groups. Remember if you probe source to drain you'll see a diode drop (something like 0.4 to 0.8V, possibly lower, but well above zero), that's normal. This is for N-channel FETs (which all these buck converter parts are, as far as I can see), P-channel like PQ812 and 813 are backwards - diode conducts from drain to source.

[Omegaman]

smanet: Thank you for your response. I agree that clear communication is essential, especially when doing this kind of remote troubleshooting. I have tried to be very specific, giving specific reference designators (i.e PQ809) and the pin numbers. I don't think the problem is clarity in this particular case. Please let me explain.

What I believe has happened to some degree here is that we are talking two different languages. By that, I am not referring to English versus Spanish or anything like that. What I mean is when someone who is just starting out with electronics, or anything else technical for that mater (I believe you said this is the first time you have done this), is communicating with a design engineer, they are often speaking a different language. This is NOT the fault of anybody, it is just design engineers, and probably most engineers, kind of have their own language. I think some examples will help to explain.

Lets take the measurements you took on VREG5 & VREG3. These are the names inside the IC (PU9) assigned by the chip manufactures. Design engineers never, or almost never, use these names. Unfortunately, this caused me to incorrectly associate these with 5VSTBY & 3.3VSTBY instead of LDO5 & LDO3, and resulted, I am sure, in comments I made that did not make sense. Again, this is no ones fault, or if someone is to blame it is ME. I realized this at some point, but much later than I should have. I apologize for that.

Design engineers refer to net names, or signal names if you prefer, and not the names inside the IC. You will notice in the schematic that almost no net name is the same as the name inside the IC. This is because the engineer assigns his/her own net names that make the most sense to them. Often, the names inside the IC do not make sense to the engineer for what they are using it for, and sometimes it has multiple functions, 4 or 5 or more with a long name inside the IC with / in between each name for each function, such as GPIO10/ICS1/AD2 and so on.

So, it helps if you use the net/signal name (net name and signal name are the same thing). So for example, you measured VREG5 (which is connected on PU9, pin 17) you would say "I measured LDO5 across cap PC60". Because the designer here used their nomenclature for naming components, it helps to say what kind of component it is, cap, resistor, diode, etc. Most Engineering places have their own nomenclature, so this just helps to be clearer. A sudo-standard, quite often not followed, is this - U for digital IC, C for cap, R for resistor, D for diode, etc. Here they use PQ for FETs, PU for ICs, PC for caps, etc. If you are measuring across a 2 pin component, then you can simply say "I measured 1.8V across PC60". This is clear enough for the other person to find it if they have the schematic. If not, it helps to give the signal name as designers are often familiar with those, or can figure them out, and what the component is, cap, resistor, IC, etc.

Please let me stress that this is NOT critical of anyone. It is just an explanation of a little of the language engineers speak, or things they are used to hearing. I will also be as specific as I can, but sometimes too much information can be confusing, such as saying "I measured LDO5, VREF5, PU9 pin 17 across PC60, pin 1 to 2". Just saying "I measured LDO5 across PC60", provides all the information needed.

Let me know if you have any questions or something I can help you with. I suspect you will be a while checking all the FETs on page 43, so you probably don't need any more suggestions right now.

Good luck as always.
Omegaman

[Monkeh]

This thread's gone awfully quiet. Have you popped it?

[poot36]

Here in Canada you can buy that laptop for around $150 used.  I don't think it is worth fixing judging by the fact that it already took out a hard drive and now this problem.

[Jimzz]

Even i have a same laptop with exact "laptop dead" situation. PQ812 & PQ813, PC807,PC808,PC809 overheats.

Could someone send me the updated link to download the same schematic. existing link is broken. Thanks