Rigol DP832 power supply repair - U18 replacement

[giosif]

Hi,

I am trying to repair a DP832 power supply which has had its U18 IC burned to a crisp.
This is one of the 3 ICs on the top board which have had their marking rubbed off from factory (the one closest to the series pass transistor for channel 1).

Would anyone have any idea what make and model this IC might be (or some hints on how to identify it)?
Alternatively, would someone happen to have a spare such board they would be willing to extract the IC from? (cough... Dave? cough...)   

Thank you!

[Kleinstein]

My first guess for the parts doing the regulation would be an OP.  The resistors / caps and the power connection could give a hint. Chances are there are a few OPs with still some making intact, so one could identify the supply rails.

The other possible chips to expect are DACs and ADC - these should have a live to the digital part, usually not found with an OP. ADCs / DACs are usually less exposed to enough power to really burn the parts.
So chances are high to have an OP of some kind. I would also expect the power MOSFET to be blown as well and this way get higher voltage to the gate drive.

One may have to do some local reverse engineering to get an idea on which part would match. 

[giosif]

Thank you for the hints!

I will investigate and try to determine at least the type of component (i.e. op-amp, DAC, other).
I will worry about finding a suitable replacement after that.

Interestingly enough, the power MOSFET seems to be ok, on a preliminary check.
I will check again, just to be sure.

[m k]

Any step by steps somewhere?

After clearing the area check possible supply lines.
Then pins that are directly connected to the supply lines.
Then pins that have a single helper component in between.
Something can also be quite directly connected to the other chip or clear other circuit.

Other possible approach is to start listing parts it surely is not.

I was once digging up MCU types by comparing prints.
It was finally quite easy up to a point.
Help there was that I knew the maker and its/their habits.

What kind of a company Rigol is, own ways or by the example circuits?

[TurboTom]

It's a single opamp in standard pinout, used as a current sense amplifier (its inputs are routed to the current shunt resistor, see Dave's teardown photo here:

https://www.flickr.com/photos/eevblog/9607771682/

Since the shunt is located in the positive output rail, and the OPAMP's input is divided only to 9/10ths, it would have to run on a fairly high voltage (which may explain its pyrotechnical demise) and needs to have a decently low offset voltage. I'ld say an OP27 or OP37, possibly of the better grouping (A) if available, should do the job. To be sure, I'ld measure the supply voltage with the damaged IC is removed. If it's more than 40V, rather go for something more exotic like the ADA4522-1. Also, check the resistors in its vicinity to be free of damage.

You will probably have to run a re-calibration of the affected channel after replacing the chip.

[giosif]

Thank you!
And sorry for the lack of update: I am waiting for a few replacement components, before I move forward with the investigation (mainly, a 7A fuse).

[giosif]

It's a single opamp in standard pinout, used as a current sense amplifier (its inputs are routed to the current shunt resistor, see Dave's teardown photo here:
[...]

Just to check: in Dave's teardown photo, are you referring to the IC in the middle of the picture just above the 4 SMD resistors and between C51 and C52?
I'm asking since the current shunt resistor is connected to that IC, which is U16.
I am talking about U18, which is the IC at 1 o'clock from U16.

[Kleinstein]

That chip looks like a dual OP. There is something connected to pins 1 and 5, so it should not be a single OP.
I would expect it the OPs for the main regulation loops for current and voltage. U16 is likely just some amplification for the current signal.

With an already amplified current signal, this OP should not be so critical - chances are a simple TL072 can do the job. An RC4558 would be another possible candidate.
One point hard to tell is whether the circuit needs or maybe can not tolerate the back to back diodes found between the inputs of some OPs.

[giosif]

Yes, it most likely is a dual OP.
Interesting you should mention TL072, as one of the nearby ICs which has not been rubbed off (i.e. U24) is a TL072.
Which also kind of excludes our IC as possibly being a TL072 as well (otherwise, why would they rub off one but not the other).
But, as you say, it may not matter that much what dual OP it is, so I am going with that.

Thank you!

[TurboTom]

It's a single opamp in standard pinout, used as a current sense amplifier (its inputs are routed to the current shunt resistor, see Dave's teardown photo here:
[...]

Just to check: in Dave's teardown photo, are you referring to the IC in the middle of the picture just above the 4 SMD resistors and between C51 and C52?
I'm asking since the current shunt resistor is connected to that IC, which is U16.
I am talking about U18, which is the IC at 1 o'clock from U16.

Indeed, I've been refering to the chip between C51 and C52. The designator wasn't clearly readable to me and I mistakenly identified it as U18. The "real" U18 surely looks like a dual OPAMP, supply and bypassing caps on 4 and 8 appear to make this obvious. Yet, I'm wondering a little about the three BAV99 / single BAS16 diodes in its direct neighbourhood. I'm quite certain it's a TI part (the bar that's still visible at the pin 1 / 8 end is a clear indication), but instead of OPAMPS, several comparators would also fit the pinout, like the LM2903. On the other hand, just next to it, there's a National Semiconductors LM393 which is basically the same, so if U18 would really ba a comparator of that flavor, I'ld expect these two to be identical... To give a "shot from the hip", I'ld try an OPA2990... 

[m k]

(otherwise, why would they rub off one but not the other).

Motive, I've thought that too.
Maybe they are fogging the area.

Included picture has a bit bad focus and lighting but my uneducated guess is that sanded ones are not NatSemis nor TIs.
What else there are, AD?

Closer look of the bottom of the package mold should also reveal something, corners will differ and stuff between their legs.
Overall color also, of course.

[m k]

What the area is generally,
I/VMON, C/V_REF and C/V_STATE.

1st and 2nd stage pin 4, equal.
1st and 2nd stage pin 7, equal.

R60 and R61 possibly equal.
1st and 2nd stage in and out mostly equal.

Is a trace rounding under R67, R77 and back under the 2nd stage?
Is CV_REF continuing left under C63 and 2nd stage and right to the D21?
Is VMON one leg of D21 through Rx3?

My educated guess is that 1st and 2nd stage chips are equal.

If C/V_REF goes to the right then logically C/V_STATE goes to the left.

Is the 2nd chip a driver with a mode selection?

If you take 1st stage out and put it into a chip tester, can it be identified?

[TurboTom]

Sorry I don't quite get what you are refering to with "1st and 2nd stage".

It appears quite obvious that U18 is driving the base of the BD136 PNP transistor (Q15 as it seems as per Dave's teardown video) that's probably a combined gate driver / pulldown device for the power stage. Both double diodes D9 and D22 are wired so the outputs of U18 can sink current from the base of Q15, provided pin 7 of U18 is connected to the "combined" terminal of D22. So I'ld assume that the two sections of U18 are the CC and CV drivers, respectively. Since U18 went in a pyrotechnical manner (...), I'ld have a good look at all the diodes around and as well at Q15. Hope you'll get the defect sorted soon  .

Cheers,
Thomas

[m k]

Even without their fuctions those two are like parts of one, so stage1 and 2.

Ie. support component numbering.
Follow up numbers are not around one chip but between two chips.
Like C50 - 53, pins 4 and 7.
Or input C48 and C49.

AD Current Sense Amplifier of 8pin SOIC they can't be, wrong pins.
Pins 7 and 4 not supply or one n.c.
Supply is 4 and 8.

[giosif]

The above conversation goes a bit over my head, so not sure what to make of it, at this stage.

At the same time, an update on my side:
- U18 was replaced with an RC4558 and, powering things on, the Q16 series pass transistor was turned fully on (i.e. about 48V DC on the output) all the time.
- checking with my FlirONE, I could see that U3 and U24 were running hot, so I went ahead and replaced them.
Now, Q16 is off when powering on the supply, but there is no output even when enabling CH1.
Right now, I suspect U7 might be dead, but I don't have a replacement for it (placed the order and waiting for it to arrive).

[TurboTom]

Could you possibly check the supply voltage of U18? This is between pins 4 and 8. It may be easier to test at the vias where the traces connect to. Also, it would be good to know how the supply voltages of U18 are referenced to ground, i.e. measuring the two voltages with reference to output ground.

In general, replacing components arbitrarily, or just because they are getting warm, without a better understanding of the working principle (and, if necessary, partial reverse engineering / extracting the schematic), may lead to success but more often to more damaged components and an instrument that's still defective...

If U18 released its magic smoke in a somewhat spectacular manner, I'ld expect other components in its vicinity to have gone bad, too. Moreover, linear regulators on the internal supply rails may have been damaged, and they may be located elsewhere on the PCB. You may want to check where U18 gets its supply voltage from. It's possible that U18 clamped an internal supply rail that had been fed by a linear regulator that went bad, outputting the full, unregulated supply to the opamps. Now that U18 is replaced, the rail is too high again and the other opamps run hot. Just guessing, though.

The basic operating principle of the channels is like this: The slave "analog" microcontroller of each channel (here U4) is an ADuC7060. This has got a single D/A channel, available at pin 11. This signal is fed to a single OPAMP, U6, that's connected as a buffer (the clear indication for that is the guard trace that connects to pin2 and most likely also to pin6. This buffered DAC voltage connects to the common analog terminal of the analog multiplexer U7 (HC4051). This multiplexer is operated by the slave microcontroller as a sampling switch and alternatingly charges the two sampling caps C14 and C16 that are buffered by some OPAMPs of U3. These will hold the presets for voltage and current that are supposedly present at CC_REF and CV_REF. It's well possible that Rigol uses some "dithering" of the DAC output since the intrinsic resolution is only 14 bit and this seems a little low when looking at the possible output increments. I guess I_MON and V_MON are the current and voltage signals to be fed back to the ADC channels of the slave micro. All the other stuff is just "housekeeping", necessary to complete the output, drive and discharge circuitry. The other half of the PCB houses the AC rectifiers and TRIAC switches with their opto-isolated drivers to select the proper transformer winding for the selected output voltage (again, controlled by the slave microcontroller). If I'll ever tear down my DP832, I may just sketch the schematic of this section, doesn't actually seem to be too complex. Yet, my PSU isn't giving me any reason to crack it open...fortunately! 

[giosif]

Could you possibly check the supply voltage of U18? This is between pins 4 and 8. It may be easier to test at the vias where the traces connect to. Also, it would be good to know how the supply voltages of U18 are referenced to ground, i.e. measuring the two voltages with reference to output ground.

Sure:
- pin 4: -15V (referenced to circuit GND)
- pin 8: +15V (referenced to circuit GND)
- circuit GND is different from the output GND, as well as from the chassis GND (which is connected to Earth ground)

In general, replacing components arbitrarily, or just because they are getting warm, without a better understanding of the working principle (and, if necessary, partial reverse engineering / extracting the schematic), may lead to success but more often to more damaged components and an instrument that's still defective...

Right. I should have explained this more: the decision to replace U3 wasn't based solely on the fact it was running a bit too warm.
Since U18 had such a spectacular demise, I suspected other components around it in the circuit were most likely damaged as well.
And, since U3 has a couple of connections to U18 via some resistors, it was already a suspect.
Furthermore, with my (albeit limited) understanding of the circuit, the series pass transistor Q16 is being controlled by U18 via Q15 and, U18 is, in turn, driven by U3.
And, since Q16 was being fully turned on all the time (and U18 is now good, after replacing it), the next component to suspect was U3.
The fact it was running hot was more like the last aspect that lead me to the final conclusion.
For U24, I admit I replaced it mainly because I could. 

If U18 released its magic smoke in a somewhat spectacular manner, I'ld expect other components in its vicinity to have gone bad, too. Moreover, linear regulators on the internal supply rails may have been damaged, and they may be located elsewhere on the PCB. You may want to check where U18 gets its supply voltage from. It's possible that U18 clamped an internal supply rail that had been fed by a linear regulator that went bad, outputting the full, unregulated supply to the opamps. Now that U18 is replaced, the rail is too high again and the other opamps run hot. Just guessing, though.

Apologies: like above, I didn't mention all checks I've done, in addition to the changes I mentioned in the previous post.
I did check all major components, in preparation of replacing U18 with a new component, starting from the AC inputs (connecting to the transformer), following with the bridge rectifiers, linear regulators, capacitors, etc.

[pedro48]

Hi Giosif ....

did you manage to recover the DP832. I have just started down the similar path having shorted channel 1 like so many before me...

Reverse engineering the circuit, has me following somewhat in many of your footsteps

cheers...

[giosif]

Hi,

No, still working on it.
So far, I've replaced U18, U24, U53, U3 and U7.
And the current behavior is that the series pass transistor for CH1 does not get turned on at all, no matter the settings on the display (and enabling the output, of course).
And the transistor itself is definitely good.
If I feed some voltage into CH1 (e.g. from a different power supply), the display is showing the correct voltage applied to the terminals.
So, at least the voltage feedback part of the circuit is working.

[giosif]

Hi,

Coming back to trying to fix this power supply, would someone with a working CH1 be so kind to confirm some measurements for me, please:
* CV_REF = ??
* CC_REF = ??
* pin 3 of U7 = ??
* pin 9 of U7 - does it ever change state (on mine, it always stays high which, I suspect, is wrong)

Thanks!

[giosif]

Anyone, please?