EEVblog #861 - Rigol DP832 PSU FAIL & REPAIR

[BravoV]

Maybe there's a follow-up video to be had here?

Hook up a scope to the MOSFET and measure Vgs and Vds.

Then, apply various conditions at the output terminals, and see what the MOSFET is subjected to in each case. Maybe stick a differential probe across that current sense resistor too, so we can see the drain current and calculate the instantaneous power dissipation in the die.

Then, see if there are, say, transient spikes on the gate which exceed its Vgs rating. See what the maximum Vds is, and what the power dissipation would be. Try and identify if there are load conditions which place undue stress on the MOSFET, and perhaps come up with a modification to protect it better.

+1 

[bktemp]

The worst case is setting the power supply to max voltage and current, leaving the MOSFET heat up by shorting the output for some time, then disconnecting the output and shorting it again. When disconnecting the load, the power supply will change the transformer tap to the maximum voltage. When shorting the output again, all the caps are charged to the max voltage.
The MOSFET has to dissipate a huge amount of power (at least >30V*3A= >90W, probably much more for a shorter time because the current regulation needs some time to respond). Because the MOSFET is already warm due to the previous load the silicon may reach a temperature where it fails.
That is probably exactly what happend when Dave did the relay video.

[McBryce]

Dave,
     did you notice that Channel 1 was displaying "UR" as the mode instead of "CC" or "CV"at the start of the video? Could this have been a factor in the failure or was it a symptom of it having failed? I haven't seen this before and the DP832 Manual doesn't say much about it, just a one-liner.

McBryce.

[EEVblog]

Dave,
     did you notice that Channel 1 was displaying "UR" as the mode instead of "CC" or "CV"at the start of the video? Could this have been a factor in the failure or was it a symptom of it having failed? I haven't seen this before and the DP832 Manual doesn't say much about it, just a one-liner.

I didn't notice that.
I presume that UR means Under Regulation?
In that case I would expect that based on the fault. I'm surprised it would even display anything for that, that's a good thing though.

[station240]

Don't just turn it on, blow it apaaarrrrtt!

By the way I've killed a 3055 series pass transistor in a PSU, the back emf protection was on a seperate board and the wire got disconnected. 
Thing it this was a stupidly high current PSU so it wasn't one 3055, it was nine! 
White smoke everywhere (heatsink compound), had to evacuate the shed.

[Monittosan]

Dave,
     did you notice that Channel 1 was displaying "UR" as the mode instead of "CC" or "CV"at the start of the video? Could this have been a factor in the failure or was it a symptom of it having failed? I haven't seen this before and the DP832 Manual doesn't say much about it, just a one-liner.

I didn't notice that.
I presume that UR means Under Regulation?
In that case I would expect that based on the fault. I'm surprised it would even display anything for that, that's a good thing though.

UR Sounds like Agilent Unreg status in their power supplies. That mite have been why it's blown the 5amp fuse.
Have you measured your incoming line voltage? Strange to be pooping 2 fuses so close to each other.

[mikerj]

Dave Says:
"...I would not expect a lab power supply of this price and grade to blow a pass transistor..."

Really?  It's a $450 power supply from Rigol.  That is pretty much the supply price and grade I would expect to randomly blow a pass transistor.

You likely have very different views to most people then.  For $450 I would expect something engineered well enough to do it's job.  If this was one of the $50 piles of junk being sold eBay then you just might have a point.

I'm dumbfounded at your annoyance by the hand-soldering around the pass transistor. How else are they to attach it to the heatsink?

Your dumfoundedness confuses me   It sounded (and looked) to me as though someone has previously been inside this and replaced the transistor before Dave got it, hence his suspicion.

[Synthetase]

So what benefits do MOSFETs have over BJTs in this application?

[nowlan]

Lower RdOn usually? compared to VCE.

[matkar]

Name one.
I've shorted countless cheap linear supplies over the years, never killed one that I can recall.
Only exception would be the Korad, but that wasn't a short with leads, it was oscillatory constant power mode on a load.
I wouldn't say you can't of course, but IME it's not common.

I repaired a 3055 based 20A chinese power supply last year. I had to change 6 transistors. Nothing else was wrong with it. It was destroyed when making repetitive shorts in a lab experiment.

I wouldn't say MOSFETs are more prone to fail in power supplies. It all comes down to how good the protection is done.

[mcinque]

Hook up a scope to the MOSFET and measure Vgs and Vds.

Then, apply various conditions at the output terminals, and see what the MOSFET is subjected to in each case. Maybe stick a differential probe across that current sense resistor too, so we can see the drain current and calculate the instantaneous power dissipation in the die.

Then, see if there are, say, transient spikes on the gate which exceed its Vgs rating. See what the maximum Vds is, and what the power dissipation would be. Try and identify if there are load conditions which place undue stress on the MOSFET, and perhaps come up with a modification to protect it better.

 

[EEVblog]

I wouldn't say MOSFETs are more prone to fail in power supplies. It all comes down to how good the protection is done.

Of course it's about the design.
But all things being equal, MOSFET's do have an inherent probability of being partially damaged by ESD perhaps leading to more field failures.

[bktemp]

But all things being equal, MOSFET's do have an inherent probability of being partially damaged by ESD perhaps leading to more field failures.

Do you have any reference for that?
I have never heard it. The gate is sensitive to ESD damage, that is well known. But that should be no problem with a mosfet mounted on a board with a zener/TVS diode across gate-source.
From my experience a MOSFET is much more tolerant to overvoltage conditions because of its build in diode across drain-source. The brekdown of those diodes is even specified in most datasheets (avalanche rating).

[tszaboo]

High density means it is less suited for linear operation, because the die is small.
It looks like the designer of the power supply simply trusted the specs in the datasheet, without thinking about if it is plausible.
There is nothing wrong with MOSFETs for linear operation, but you need to understand the limits, and it looks like the Rigol engineers did not.

It is called forward bias safe operating area. It is a local thermal runaway effect, which is the plague of high voltage FETs. Almost none of them handle high temperature, high voltage and high current at the same time. For this 30V-ish region, I think it is safe to assume that this is less likely to happen.
If you see a FET datasheet, where the DC ends with a line at above 100V, assume that DC= something like 100ms.
I dont think this Rigol supply is rated correctly. 30V 3A is 90W dissipated in that FET. It is very on the limit. I did not allow more than 75W in a TO220 package, when it was put on an anodised water cooled heatsink. With a 220, you have best case 0.5K/W thermal resistance junction to case, plus case to sink. With a black anodised heatsink like this, with screw, that is like another 0.5K/W. So the FET is running 90+ degrees above heatsink temperature. Not a safe margin. I bet if Dave would short the output, put it on the sun, the supply would die on its own after a while.
They should just upgrade it with a TO247. That has approximately twice the surface area to the heatsink, and 0.2K/W junction to case for the best devices.
And dont get me started on the lack of output relay.
So this is how many thermal issue in this PSU? Already 3? I guess chinese watts are smaller.

[McBryce]

Dave,
     did you notice that Channel 1 was displaying "UR" as the mode instead of "CC" or "CV"at the start of the video? Could this have been a factor in the failure or was it a symptom of it having failed? I haven't seen this before and the DP832 Manual doesn't say much about it, just a one-liner.

I didn't notice that.
I presume that UR means Under Regulation?
In that case I would expect that based on the fault. I'm surprised it would even display anything for that, that's a good thing though.

According to the Rigol DP832 User Manual:
"DP800 series power supply provides three output modes: constant voltage output (CV), constant current output (CC) and critical mode (UR). In CV mode, the output voltage equals the voltage setting value and the output current is determined by the load; in CC mode, the output current equals the current setting value and the output voltage is determined by the load; UR is the critical mode between CV and CC."

Which doesn't say much. This is the only sentence in the Manual that mentions UR at all.

McBryce.

[Swemarv]

I would think its another example of poor thermal design.
Lets do the math..

Dave measured the unloaded voltage at the filter caps to be about 53V. This would correspond to about 37.5V RMS. With a shorted output , all this voltage will be over the transistor at 3A
so the total power dissipated in the transistor would be roughly 110W:

Looking at the datasheet of the transistor we can see that the thermal resistance from junction to case is 0.5 C/W. A typical thermal resitance from case to heatsink is 0.5 C/W. We dont know the figures of the heatsink but I would be surprised if its below 0.5 C/W. Assuming 25C ambient, the junction would be at  25+(0.5+0.5+0.5)*110 = 190 C. This is 15 C above reccomended and it is therefore not surprising that it couldn't handle it.

The situation for channel 2 is worse since it shares its heatsink with channel 3.

[VK5RC]

Dave, how often/hours have you used this supply? Is it still in its 'running in ' period?
I would be leaning toward a pre-insertion static damage, (?hand soldered).

[bktemp]

I dont think this Rigol supply is rated correctly. 30V 3A is 90W dissipated in that FET. It is very on the limit. I did not allow more than 75W in a TO220 package, when it was put on an anodised water cooled heatsink. With a 220, you have best case 0.5K/W thermal resistance junction to case, plus case to sink. With a black anodised heatsink like this, with screw, that is like another 0.5K/W. So the FET is running 90+ degrees above heatsink temperature. Not a safe margin. I bet if Dave would short the output, put it on the sun, the supply would die on its own after a while.
They should just upgrade it with a TO247. That has approximately twice the surface area to the heatsink, and 0.2K/W junction to case for the best devices.
And dont get me started on the lack of output relay.
So this is how many thermal issue in this PSU? Already 3? I guess chinese watts are smaller.

It looks like the power supply switches transformer taps, so the MOSFET only has to dissipate the full >90W for a short time. But even 45W are on the high side for a TO220 package if you want a reliable design. As I said before: The static power dissipation is only one part. The MOSFET must also be able to handle short circuits after it has been heated up by the static load.
When I did design my 50V 6A power supply with a SMPS preregulator I had so select quite a big MOSFET to make it short circuit proof for this case because of the high peak power (330W for a couple of 100ms + >1kWs for some 10us).

[EEVblog]

But all things being equal, MOSFET's do have an inherent probability of being partially damaged by ESD perhaps leading to more field failures.

Do you have any reference for that?
I have never heard it. The gate is sensitive to ESD damage, that is well known.

It's an infant mortality thing. You can partially damage/weaken FET's with ESD and they still work, but lead to early failures in the field.
In serious products and systems they burn-in parts to weed out infant mortality due to ESD and others factors.
https://books.google.com.au/books?id=Vo0PFNielQkC&pg=PA41&lpg=PA41&dq=infant+mortality+due+to+esd&source=bl&ots=BOqYOv6xUm&sig=uhi7ODOmvnODPWBmyG_9Ig0VtJA&hl=en&sa=X&ved=0ahUKEwjc7s226MfLAhWiGqYKHVfkBRIQ6AEIPjAJ#v=onepage&q=infant%20mortality%20due%20to%20esd&f=false

[EEVblog]

Dave, how often/hours have you used this supply? Is it still in its 'running in ' period?
I would be leaning toward a pre-insertion static damage, (?hand soldered).

Couldn't say. A few hundred hours maybe.

[EEVblog]

According to the Rigol DP832 User Manual:
"DP800 series power supply provides three output modes: constant voltage output (CV), constant current output (CC) and critical mode (UR). In CV mode, the output voltage equals the voltage setting value and the output current is determined by the load; in CC mode, the output current equals the current setting value and the output voltage is determined by the load; UR is the critical mode between CV and CC."
Which doesn't say much. This is the only sentence in the Manual that mentions UR at all.

Interesting. Surely it would only be in UR "mode" during the very short time it's transitioning CV <> CC. So basically just the response time of the regulator loop.
I'm surprised they even detect and display this. Perhaps that shows they know it's loop response is pretty poor?

[madires]

First the overheated voltage regulator, now a poorly choosen pass transistor? And this also provides insight into Rigol's R&D capability. With a little bit more effort they really could make their products better.

[amspire]

If the load was sourcing voltage back to the supply that was greater then the set voltage, then the supply would not be in CV mode and also not in the CC mode. I would guess they would use UR mode to indicate this state.

[McBryce]

According to the Rigol DP832 User Manual:
"DP800 series power supply provides three output modes: constant voltage output (CV), constant current output (CC) and critical mode (UR). In CV mode, the output voltage equals the voltage setting value and the output current is determined by the load; in CC mode, the output current equals the current setting value and the output voltage is determined by the load; UR is the critical mode between CV and CC."
Which doesn't say much. This is the only sentence in the Manual that mentions UR at all.

Interesting. Surely it would only be in UR "mode" during the very short time it's transitioning CV <> CC. So basically just the response time of the regulator loop.
I'm surprised they even detect and display this. Perhaps that shows they know it's loop response is pretty poor?

I just looked back at some of your videos (as I'm nowhere near my DP832 at the moment), but if you take a look at video #549 from around 11 minutes, where you were experimenting with the settings, the display momentarily displays UR several times.
I can't think of any other situations where it might be displayed for longer?

McBryce.

Edit: What if you set it to CV 30V and limit the current a few mA. Does it show UR while the capacitor is charging?

[bktemp]

Maybe it displays UR when it detects a repeated toggling between CC and CV mode. Instead of randomly displaying CC or CV it shows a stable UR?
There are several ways to detect CV/CC mode. One way is to compare the control loop outputs and use the lowest one. Another is to look at the output of the control loop and whenever it is not at its limit (one of both loops typically requests full power while other loop actually regulates the output voltage) it can be considered as active. When it is near current or voltage limit both control loops may be active at the same time.

Any idea why the power supply showed 0.7A with no load? Did there really flow 0.7A thru some discharge path or was it only the current sense circuit being out of common mode range?