Troubleshooting OZ811 dc-dc controller - showing negative voltage on output?

[gboone]

I'm troubleshooting this circuit which powers a dc-dc controller. It's on a HP dv7 motherboard with AMD chipset.



I believe there is a short and I'm unsure whether it is a downstream component (like a bad chip) or a problem with this controller. I found the specs for this controller: http://www.datasheet-pdf.com/datasheet/O2MicroInternational/697173/OZ811.pdf.html

I have verified the following, with the computer powered off, and power adapter plugged in:

• 19.2v is supplying the pin 2 VIN
• 5.2v is supplying both pins 5 and 16
• BST on pin 8 is showing 5.2v
• LDR on pin 7 is showing 5.2v
• nothing on HDR and nothing on LX
• All caps and resistors seem fine

When I start the computer, it will show a -.38v on LX measured to ground, taken at terminal 5 on MOSFETs PQ66 and PQ73.

Question 1: without powering the computer, should LDR be showing any voltage at all? It drives the gate for the low side. It makes the MOSFETs PQ66 and PQ73 seem shorted as with the power adapter plugged in, the gates are energized. Also, these two MOSFETs are getting hotter than they should when the circuit is energized. The coil right after it (on the other side of the board) gets hot as well and I can feel it through the board.

Question 2: why is LX negative? It is supposed to power the VGA core circuit after this and obviously is not the proper voltage. I could understand a short downstream causing problems later...but why would it be negative voltage? I could not figure out what voltage it should be.

Circuit may be sensing overload and not allowing a turn-back-on scenario. I have to wait awhile and then it will turn on again, not boot, and then totally turn off after 15 seconds or so.

I have checked all components downstream and removed many and put them back on. Seems like a short in the chip. BUT - it turned on yesterday and was just fine. If it does turn on, the computer is fine unless I try to restart it - it won't even POST. So, it appears to be having voltage issues, and this controller - how do I know if it's working right? What do I check?

Thanks!

 

[kellys_eye]

Can you 'scope LDR? It shouldn't be a 'fixed' voltage (PWM) but you won't be able to tell with a simple multimeter anyway. Then again, if it WAS a fixed DC then the switching FETs would be 'hard on' and you'd be seeing the magic smoke!

Simplest method to test is to break the supply line and fit your own, known value, load resistor.

As with many SMPS supplies these days, the smoothing capacitors are the weakest link and intermittent operation is symptomatic of this (PC226/PC227?)

 

[gboone]

I checked the smoothing capacitors...they were fine. Unfortunately, I don't have an oscilloscope.

I also just checked with no adapter in, and the two subsequent MOSFETs both show continuity to ground from Source and Drain. So, not just when adapter is plugged in.

With adapter then plugged in again, I also took measurements again of LDR with a different multimeter, and this time it was either zero, or on a different time showing up to 0.2v. Seemed to be steady.

I do believe it is steady voltage either way because I replaced both MOSFETS already and one of the two old ones checked out to be fine, but the other was toast. They get too hot to touch and the coil on the other side is even hotter. The whole thing then shuts down.

To break the supply line as you suggested, are you saying to take off the chip, and jump a resistor between +VIN and LX?

Also, the multimeter has a Hz setting...could that be used?

 

[kellys_eye]

NO! Do not link +Vin to LX - disaster will ensue!

I was referring to you breaking the circuit (track) between PL14 and the point marked 22.8A and connecting a fixed load resistance to the junction of PL14/PR242 and ground.

With a fixed load in place the output of PU14 should be measurable and you can vary the load resistance to check where the cut-off point is (overload).

It may be simpler to just replace PU14 and the two FETs though...... the FET/inductor overheating points to a shorted FET (as you initially report) and/or non-operational driver chip (PU14) which will make one or both FETs go short anyway. Note that the inductor may also require replacement - recommend you do.

The multimeter would need to be able to read up to a couple of hundred kHz to be of practical use.

 

[gboone]

With a fixed load in place the output of PU14 should be measurable and you can vary the load resistance to check where the cut-off point is (overload).

That's why I ask! LOL I totally get what you're saying. And if it IS something downstream, that component won't be getting voltage anymore and the output from PU14 should be smooth when measured, right?

The multimeter would need to be able to read up to a couple of hundred kHz to be of practical use.

It can do up to 5MHz. Also has duty cycle measuring capabilities. I tried getting the cycles measured but it was different each time.

 

[kellys_eye]

The kHz should be relatively stable but the duty cycle will vary according to load.

Seriously - just replace the control IC and two MOSFETS. If they blow 'again' then you've got problems downstream of that regulator that may result in condemning the equipment!

 

[kellys_eye]

...oops. The control IC is one of those that requires specialist desoldering equipment (a QFN package). Not much chance of successful replacement I suspect.

Sadly, with the original fault making one of the switching FETs go 'hard on' (giggidy) it really does point to the IC being dud.

 

[gboone]

I've got a hot air station...thanks for the tips I will report back with update.