As a couple of other posters already said I think the problem is simply bad thermal design.
The designer must have thought 'ok I have a pass Nfet
and a preregulator which limits the max voltage across my nfet to lets say 12V and I have a max I of 3A so my max pass fet power dissipation is only 12*3= 36W' . But he has not taken into account of the slow speed of any pre-regulation due to the large filter input capacitor. Which means that in cases where the output is going from max V out then shorted to 0V that the input to the Nfet will be at max 30V and only slowly ramp down to toward lower tap Voltages as the energy in the large large filter cap is used. If the output is continually shorted and opened at a high frequency then the mosfet basically has to dissipate maxtap V *3 for shorted periods (~ 120W) . So this means designer has made a very poor choice for a pass mosfet . A to220 case really? a Nfet optimized for switching ? . If I was you Dave I would upgrade that Nfet to somthing at least with a to264 or to3p case style .(choose something with a nice lowish Cgd and higer rdson (so in other words not a highly switching optimized Nfet )
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?
Judging by the temperature of the BJT when Dave viewed it through the Flir, I'd say the 0.7A were flowing through that.
McBryce.
But the current shunt is at the output of the power supply, only connected to the output terminals. So it should not measure the current flowing inside the power supply.
It looks like the BJT drives the MOSFET: There is probably a current source somewhere supplying the gate voltage. The BJT pulls the gate towards GND to regulate the output voltage.
Oh, so it is. Then I've no idea where they were going?
McBryce.
Nice! At my university they ordered a few of these for the mechanics lab (where freshmen do their first basic physics experiments). Luckily, the electronics lab technicians decided to stick to Agilent (*cough* Keysight) power supplies.
Best redesign would be to put 2 of those pass transistors on that heatsink, with 2 0R1 5W ballast resistors to share the load between them during transient conditions, along with some 18V zener diodes and 100R series resistors right by the 2 gate terminals. That at least minimises the chances of cooking the pass devices.
As the unloaded voltage in the filter caps is in the 55 V range, the poor little MOSFET has to cope with 55 V at 3 A - so one rated to 3 A up to 37 V is just not good enough. So if you really want to keep this design the MOSFET should have a rating up to 60 V 3A in the SOA curve. Reducing the transformer voltage might ease things a little (the 240 V are on the high end for the 110/230 V transformers) - I don't think you need more than 50 V to get 30 V out - something like 40 V should be enough.
There are two more power devices near the rectifier - so there might be a kind of preregulator that failed and thus giving the rather high voltage. I am afraid the problem is not yet reliably solved.
Anyway I am surprised to see just one TO220 size power transistor and no preregulation - using 2 transformer taps and 2 power devices in series is not that difficult.
A 5 A fuse at AC to get 3 A DC out with a simple rectifier is allready quite close to the edge (especially with UL rated fuses), the RMS current is expected to be in the 4.5 A to 5.5 A range (depending on the caps) for 3 A DC.
Murphy makes an appearance on the blog....LOL. Indeed Murphy does not discriminiate. When that happens Dave should know better than to continue to troubleshoot or test, unless he loves the smell of magic smoke. Dave, I hope you wore your seat belt on the way home.
BTW the Chinese characters next to the output voltage connectors mean "black" and "red"
. Assembly instructions in other words.
Did anyone else share my pain watching the video?
[03m] Initial hypothesis of a failed pass transistor.
[15m] About to test the pass transistor.
[16m] Tests the current shunt resistor.
[17m] Tests a bleeder resistor.
[19m] Finds a blown fuse.
[22m] The unit is still faulty.
[23m] Gets out a thermal camera and finds a hot component that is NOT a pass transistor.
[25m] Finally measures the pass transistor, the hypothesis was correct.
Give the guy a break! Its an adlib video, not scripted nor instructionally precise and edited for continuity and educational value. Its entertaiment for Nerds. And as far as many of us are concerned, and I have been doing electronics design for over 50 years, 40 of those years professionally....thats how it goes in reality. I loved the "what the ****? moments " I maybe new to the eevblog but no newbie to the field. This is reality TV for techies. Take it all in. Its better with popcorn and beer
The CEP80N15 TO220 MOSFet has been replaced by the IRFP260 TO247 MOSFet in board revision 02.20 (dated 5 Nov 2013). Dave's spare board is probably 02.00 and the fixed board 02.10.
See
https://www.eevblog.com/forum/testgear/rigol-dp832-firmware-updates-and-bug-list/msg518139/#msg518139 for picture of the latest and current revision 02.20 of the topboard.
The IRFP260 is more robust, so my guess is that Rigol already knew the problem and fixed that in 02.20. Since the date code of the 02.20 board is 5 Nov 2013, issue has been long solved. My (one year old) DP832A has a 02.20 board like on the pictures on the link above.
Did anyone else share my pain watching the video?
[03m] Initial hypothesis of a failed pass transistor.
[15m] About to test the pass transistor.
[16m] Tests the current shunt resistor.
[17m] Tests a bleeder resistor.
[19m] Finds a blown fuse.
[22m] The unit is still faulty.
[23m] Gets out a thermal camera and finds a hot component that is NOT a pass transistor.
[25m] Finally measures the pass transistor, the hypothesis was correct.
That's "Info-tainment" for ya!
Where have I seen that damn fuse before?
https://www.eevblog.com/forum/testgear/rigol-dp832-smoked-channel-1/msg580527/#msg580527
Good info all around though for future reference when I blow up something else on it...
That thread describes that the dp832 failed in the same way, when loaded with a reversed electrolytic capacitor. Evidently the sequence is the same: at the transient, somehow, the gate puts the pass transistor in full conduction, its rdsON and the shunt resistor are almost the only limits to all the energy stored in the filter caps. Destroying the mosfet and blowing the fuse in the time that it takes the reversed cap to explode.
My guess is that the driver under some conditions does not switch to CC fast enough and the CV part is trying to get 30v into a short circuit.
Many power supplies have an additional "hardwired" current limit protection, perhaps rigol didnt include that one.
The DP832 might be ripe for a few quality hacks to improve it. Particularly in the area of protection.
A simple reverse-biased diode across the output terminals was usually enough to protect against a reverse polarity injection taking out circuitry.
I haven't checked the circuit topology, but it also sounds like a lack of Vgs protection as well on the FET, which was probably not considered relevant in a normal linear operating mode by the designers, but absolutely necessary when doing any switching into inductive loads (eg SMPS design).
Did anyone else share my pain watching the video?
I've learned far more watching Dave go down the rabbit-hole than when he gets it right first go.
I also don't think he really did this time. Even had he changed out the pass transistor first, the fuse would still have been blown...
I would call this one a very realistic (and successful) debugging session.
I've learned far more watching Dave go down the rabbit-hole than when he gets it right first go.
I also don't think he really did this time. Even had he changed out the pass transistor first, the fuse would still have been blown...
I would call this one a very realistic (and successful) debugging session.
I'm not saying the other steps were not necessary or that the video was too long. I just think that if he did the transistor first, there may have been some additional "mystery" value in further debugging (i.e. "Hey, I replaced the MOSFET, but the PSU is still faulty").
And, yeah, good to see EEVblog repair curse gone
Did anyone else share my pain watching the video?
[03m] Initial hypothesis of a failed pass transistor.
[15m] About to test the pass transistor.
[16m] Tests the current shunt resistor.
[17m] Tests a bleeder resistor.
[19m] Finds a blown fuse.
[22m] The unit is still faulty.
[23m] Gets out a thermal camera and finds a hot component that is NOT a pass transistor.
[25m] Finally measures the pass transistor, the hypothesis was correct.
Somewhat yes. It felt more like a "Made for TV" special then a repair video. He waited to the very end to get to the part we all knew was broken all along.
Did anyone else share my pain watching the video?
[03m] Initial hypothesis of a failed pass transistor.
[15m] About to test the pass transistor.
[16m] Tests the current shunt resistor.
[17m] Tests a bleeder resistor.
[19m] Finds a blown fuse.
[22m] The unit is still faulty.
[23m] Gets out a thermal camera and finds a hot component that is NOT a pass transistor.
[25m] Finally measures the pass transistor, the hypothesis was correct.
That's "Info-tainment" for ya!
And would have taken all of 5 minutes real time if I wasn't shooting video.
Many people complain about this, but they (understandably) fail to understand the mindset of a video blogger when they are shooting a video. I'm always thinking of stuff to say and do, little hints and tibbits etc, and if I think of it them I shoot it. Sometimes that doesn't always make sense in a methodical troubleshooting scheme of things.
You probably can't appreciate this unless you make videos like this.
Some made a comment in another thread that I have an old revision board that hasn't been made since 5 Nov 2013. Revision 02.20 of the TopBoard uses a IRFP260N in TO247.
So seems like Rigol knew about this problem and have potentially fixed it.
In that case I think I might scrub the video I planned to do on measuring the transistor parameters.
Or maybe not... I can always make a meal out of scraps
Some made a comment in another thread that I have an old revision board that hasn't been made since 5 Nov 2013. Revision 02.20 of the TopBoard uses a IRFP260N in TO247.
So seems like Rigol knew about this problem and have potentially fixed it.
In that case I think I might scrub the video I planned to do on measuring the transistor parameters.
Or maybe not... I can always make a meal out of scraps
May as well make the video and see if the new FET they selected is really up to the task. Then find a nice way to shoe-horn the new FET into the older rev PCB.
May as well make the video and see if the new FET they selected is really up to the task. Then find a nice way to shoe-horn the new FET into the older rev PCB.
I notice those diodes seems to be magically gone again in the new version!
May as well make the video and see if the new FET they selected is really up to the task. Then find a nice way to shoe-horn the new FET into the older rev PCB.
I notice those diodes seems to be magically gone again in the new version!
Maybe they knew the FET was a weak point on the old board? If that's the case I wonder why they added the diodes rather than just go for the beefier FET in the first place.