It is 2021, almost 5 years after this video: Silly me shorted the rails of my breadboard design. Result: - A spark occured during the short - An opamp got fried (to be expected) - Channel 1 on my DP832 didn't work anymore (not to be expected) I remembered this video from years ago and rewatched it with regard to my case: - Same error indications - Same faulty FET and fuse - Replacement of both parts repaired my DP832! Thanks Dave that you shared this video which helped me out with the troubleshooting and a cheap repair. As if you had known that I was going to goof up my project five years later :-).
Dave as a technician for over fourth years the saying in the trade was.... "Engineers can design, but have no idea of repair." The Fuse blew, that was the protection, what caused it to blow? You simply replaced the fuse without tracking down the cause. You never even measured the FET so you had to go back in again.
Dave, I had the same issue with a 2N3055 supply I built years ago. The issue was caused by the response time for the current limit circuit. A quick short caused the device to operate out of its safe area of operation VI region for a short period of time. It was quite device dependent, my channel 2 device never had the problem but I killed 3 channel 1 pass transistors on one week. Craig
you should always screw transistor on a heatsink before soldering to avoid putting stress on the solder joints, so it makes sense if the transistor was soldered on manually after the big heatsink and transistor had been mounted
Was just thinking the same; and Dave said it when he replaced the faulty FET. They can't run the board through soldering with screws and heat sinks on. Another thing: What's the fuzz about the flux? Could be an issue in high impedance design, but the gate looks into the output of a low impedance op-amp (or the like). No worries from my point of view.
Will be interesting to see what the folks at Rigol will have to say about this documented component failure or the "dodgy" solder job on that component. Keep us updated, Dave!
sorry to disturb the dead but, they use black and red labels because the underpaid Chinese assembly workers don't know anything about electronics. they simply just put the square peg in the square hole and the round peg in the round hole; likewise the red cable in the spade labeled "red"
I bet since that heatsink was screwed to the board, they chose to hand solder the pass MOSFET and heatsink assembly as a later step as I bet the heatsink-to-board screws wouldn't like wave soldering. Not to mention that without the screws that heatsink would put all of the mechanical load on the MOSFET leads. The other heatsink (the BJT) has solder tabs for mounting, and since it's a "wrap around" flush-to-board type heatsink (and a smaller one too), it can be safely placed on the board before wave soldering. Did the old board also have hand-soldering evidence before you pulled the MOSFET? Also, did the relay have a flyback diode/resistor built in? If not, that could be the issue. A coil (relay, solenoid, etc.) without a flyback/snub diode/resistor will blow all sorts of things to bits, even BJTs (or so I was taught in college).
Except that the only inductance would be in the leads between the relay contacts and the power supply. It was not the channel powering the relay coil but the relay contacts alone that blew, so no coils involved, just a few feet of wire for the alligator clips, so not much inductance there. Plus there is several hundred microfarads on the power supply output, so even if there were an inductive spike, that is a lot of capacitance to overcome before reaching harmful voltage levels. At this point I would be more suspicious of inductance in the power supply's own transformer, though, again, there is an awful lot of input filter capacitance to overcome, which ought to make that an unlikely source as well. It is very possible, even highly likely, that the MOSFET was previously damaged by ESD in manufacturing or assembly (prior to final incorporation into the PCB) and had already been heavily damaged and the repeated shorts just finished off what good silicon remained.
+Larry Bolan As Ethan said. It's the contact channel that blew, only lead indictance involved. It's most likely a control loop response time thing taking out a MOSFET that couldn't handle it.
+Ethan Poole +EEVBLOG Doh! Wrong side of the relay... I feel really stupid. Knowing that, I concur with dave that the control loop wasn't designed to handle such spikes. That's the trap with digital "advanced" DC supply design; if you just use a plain feedback 1-transistor (or FET) supply circuit, you have no short circuit protection and rely on the speed of your microcontroller to perform current limiting. There is a three-transistor circuit that I remember from school that does have short-circuit protection built in; however, in a case like this that may not be the desired functionality.
Having worked on many audio amps, lots with MOSFET output stages, I have witnessed the occasional failure from previously stressed components. Possibly, with the clue here that the soldering was done by hand, this could have been a dodgy FET that found its way into a production unit? However these devices became ''weakened'' , occasionally, they seem to perform ok until asked to do a bit of heavy work ie, a fast transient in a difficult load whereupon they catastrophically fail. Maybe the gate had partially broken down or almost had a hole blown through, but survived to a few atoms - who knows. Maybe it was just one of those.
I'm guessing they did factory testing of a known issue. Sure, half of this batch was dodgy, we'll factory replace it if it fails the stress test. Oh, it failed? Replace it. Test it again. It passed? Ship it! Only now you have a component that is known to be borderline by nature of the batch, and has been stressed by the factory testing, and now Dave puts sparks on it. It's kinda no wonder why it failed.
Have killed MOSFETs in a power supply before, seems the problem is less shorting, but shorting indecisively: on, off and then on again in short succession. My understanding being that the turn off causes a smallish back emf that causes the freewheel diode inside the transistor to conduct, but because it's a diode, and has a reverse recovery time, if the output is shorted again during that period, it's gonna conduct a lot of current till the diode recovers fully, which can be too late to save the transistor. So maybe the relay bounced/chattered? To protect it you'd probably need a freewheel diode that was simultaneously really fast, but with a lower voltage than the body diode, so it never gets conducting Schottky?
I always put a 15v zener diode on those MOSFETs between Gate and Source. I put it right at the transistor also. I didn't see one on that board. All it takes is a fast voltage spike to punch through the G-S insulation and powee, there goes your transistor! Maybe a small value gate resistor is also a good idea too?
I believe that this fault has nothing to do with the relay or the load you applied because I have seen a pretty similar problem with an other PSU. The source of a problem was just an unwashed rosin-based flux believe it or not. It looks like after some time rosin tends to recrystallize and absorb moisture. I have seen like a similar blob of flux between pads like this apeared to measure only a couple of kOhms. In your case I think the MOSFET just got fully opened by the leakage to the gate loosing any means of regulation and producing too much heat and in the end it just died before it could be saved by the fuse. And the worst part is that the damn thing will work just after production. It will pass any automated testing. And once you reheat the flux on the joint with a soldering iron the conductivity vanishes and the problem goes away again. I had to cut the device I had issues with of its pins co confirm that it really was the case (I was "lucky" to have a couple of similar failed chineese PSU modules). The moral of this story is that leaving crap and junk like this on a board contaning any high impedance circuits is just asking for a troube to happen after a couple of moths. So just wash your boards and everything should be fine.
For those playing at home. Have the save PS. Same issue... but it was only the fuse that crapped the bed. Transistor was ok. Tested ok from min current to max current of channel1 using same 8500 electronic load. Thanks Dave!
As a matter of course, thou shalt not replace fuses before investigating what made them blow in the first place. Otherwise, you risk blowing the new one, and possibly even causing (further) damage to the circuitry downstream.
Watching this reminded me just how amazing my PSU is. It's one of the $50 eBay 30v 5A linear power supplies. I tried leaving it with a 5A load to see how it reacted, I saw a thermal switch on the heat sink (a single plate of metal) and assumed that would shut off the output. It actually turned on the fan when the plate got above 80 degrees C or something. Not that that kept the temp from rising further. My guess is the thing would let out the smoke after 15 minutes at full load. *Now that's quality*
Oh and my first PSU was a switching one of similar quality. I killed it by powering a DC-DC converter. Somehow. I'm pretty sure I managed to repair it by replacing a blown MOSFET, but then it died again and it had to go in the bin.
I have a "LONG WEI" 30V/5A supply I got for 45€. It had the fan that ran forever, so I put an 80°C thermal switch on the heatsink and a 55°C switch on the transformer (so if either of these get hot it will turn on the fan). I also put a 95°C switch on the heatsink so that if it ever gets THAT hot, it will blow the fuse (so that I am force to troubleshoot why it got so hot). I also partially covered the vents on the side leaving about 2cm open towards the front on both sides so it passes all the air around the transformer, and changed the JWCO caps with Elna 105°C ones. Best 45€ ever spent. This thing is EXTREMELY robust. I used it to charge a 250Ah/12V battery bank (with the current pot set to maximum!!!). Of course it took 4 days to do that, but during those 4 days it was on 24/7, running at 5A and 14.4V and it never blew the fuse! I have also abused it in several other ways, including using it to power self-oscillating relays, shorting the output accidentally too many times, using it to overrev DC motors, powering "dirty" circuits like flyback inverters etc. It is still going strong after 4 years of abuse. I usually also leave it on overnight to charge batteries at low currents (so the fan doesn't come on) without any worries. The only thing that happens is that when it gets hot the voltage display drifts a bit. The regulation is rock solid however, so I usually set it when it's cold and then leave it like this.
If your going to use your power supply to trip mechanical relays at least only do so by applying and removing the lead instead, do not deactivate the relay by simply hitting the power switch on the power supply as that sends a massive voltage spike back through the output of the supply which is what happened here. Or you can simply buy mechanical relays with flywheels built into them.
I believe this is the first video of yours I've watched where a repair was successful. Good job Dave! Love it. Also, I'm glad I'm not the only one that has equipment fail randomly and for no good reason.
I used two TO-247 MOSFET for my 30V5A/15V10A 150W-max bench power supply to get better thermal performance. With multi-tapping transformer, the max dropout power was < 50W.
I'm a bit surprised by your comments about "a lab supply of this price and quality". If you had worked in any electronics design team (...and I mean any, including the most prestigious military ones too) you'd know that now a days there is no thing as careful design. It is all driven by cost and schedule and the over riding paradigm is get it done yesterday for 10% less than tomorrow. The only reason things are expensive is mostly due to branding mark-up. This $1200 unit most likely cost around $100 or less to produce. So there should be no surprises when it fails due to (most likely) inadequate product and/or production testing.
Odds are the series pass transistor was hand soldered because it was bolted to the heatsink in a jig, and then the assy is bolted to the PCB, then the transistor is soldered in place. Very normal for that kind of thing that I've seen. And odds are they used no-clean flux.Not that that makes it ok....
At 12:40, the board has no protection diodes on the pass transistor, but at 28:40, when measuring the pass transistor, the protection diodes magically appear on the board. It appears to me that he was measuring the replacement board, not the failed one.
Dave, Chinese built unit likely has some counterfeit Chinese transistors used in it.Therefore the pass transistor might not have met spec. Heat from dead short was just too much for the FET.Hand soldering is not such a surprise as the Heat sink has to go on after Wave Soldering and the transistor gets mounted on the heat sink then soldered.
At first I was worried because the video update at 15:33 says that the newer Power MOSFET was introduced after Nov 2015. Good thing I went to the forums because everywhere else says Nov 2013. I didn't want my DP832 that I got last Summer to be prone to transistor failure!
One of the best videos I have enjoyed from you. Glad to hear your opinion on the benefits between the BJT and the MOSFET. In school I spent most of my time playing with 2N3904, and very little time with the MOS. Thanks!
I'm curious about the voltage rise during the test starting at 18:45. It seems to rise for a couple seconds, then pause, then rise some more, then pause, etc. I would have expected it to be smooth (then of course slow down as it reaches its maximum). Any idea what was going on there?
It's probably the sense/negative feedback causing the voltage to leak out onto the output when it's output transistor was blown. Why it isn't smooth, god knows. It could be the multimeter lagging as well.
Still weird, even more weird that the stand alone DC load suddenly blew it's fuse. Makes me think, what if it's not the output what's caused the issue with the legendary killer relay But the input, how stable is the AC outlet, maybe it spikes every now and then like crazy?
I have one of these in the A version. The current limiting is quite slow. I was using it to test some LEDs with the limit set to 10 mA and 5V. The first one I tested was fine, but then I blew 3 in a row before I figured that the current limit was just a suggestion for the initial few milliseconds. Unfortunately, Rigol was what I could afford. The control software over the network is also very bad, although they did update it and it is improved.
You do a great job with these videos Dave. I love all your additional bits of info, as they can certainly help with repair of unrelated items. I learn so much from your efforts, and really appreciate you taking the time to pass on your knowledge.
to me, it makes sense to have the 30 V channel paired with the 5 V channel on the same PCB instead of going for symmetry and putting both 30 V channelss on the same PCB. Spread the heat-load (or however you might call it).
Collapsing magnetic fields from relay coils/solenoids can be very destructive. I learned the hard way years ago. :-) Probably better off replacing the 5A fuse with a 5A polyswitch.
+electronicsNmore No, he was apply 30V 3A to contacts side - not coil side so no magnetic field - spikes happen. I would not go for polyfuses in that case, he have 5A fuse so it's way up 3A limited. Only scenario when this fuse will will go open is hardware fail - not done by user - so the best scenario is cut power. Poly will turn power again after getting cold. They are good for protecting output when You can accidentally short leads without current limiting.
Dave Blane Well he mentions in the video "possibly" back EMF, talks about a blocking diode in the PSU, so clearly it was not back EMF if he did not connect the relay coil to the Rigol PSU. The PSU mosfet simply did not like the 30V/5A shorted output. Maybe he connected up the relay coil off camera to quickly test it, then later discovered the problem. Who knows.
+Electronic Noob Blog Incorrect. The wire to the contacts is still an inductor, and it can create very fast, high frequency transients because of the low inductance.
I read about this recently...it has something to do with the collapsing field and the voltage/current generated within the inductor and sends it backwards?
If gear keeps failing, then I would be tempted to do some mains power quality monitoring. It sounds like you may be getting some surges propagating through gear causing failures.
Did the same thing to a Mastech HY3005F-3 power supply, was shorting out channel 1 and it took a crap on me. I took it apart to see if I could find anything, not knowing anything about a series pass transistor (still learning) and I found nothing obviously wrong, when I powered it up again it worked fine.
Single most common failure in a linear supply (aside from filter caps after a certain number of years) is indeed the series pass transistor. Pretty much the first thing to check on a dead linear supply is a shorted series pass transistor.
That´s why i prefer the old, calssic, vintage, all analog power supplies as my standard lab supply. If they fail....i know why and it would be easily repairable.... If these fancy-pancy with nonsense bells and whizzles added USB/eth, digital, 7segment, all uC controlled rubbish fails...naah...too much hassle. If one of these fails due to a ground rail spike, i´ll bet the whole thing will fail and went imediately BER.
I got a new Rigol DP 832 in 2021 and I am enjoying using it. Hopefully this problem that Dave mentioned will not happen in the newer units. One criticism that I have of Rigol is the lousy user guide (manual). They put everything under the sink in the guide without clear steps on how to quickly get to a particular feature and set it up. It would have been nice if they gave examples for setting pulsed waveforms. I need to pulse the current to a PCB heater circuit for a sensor application. If there is any UA-cam video on this I would appreciate the info. Thanks.
check that power cord,if it blew the fuse in the other piece of equipment,mabe it caused a spike or short in your first event,also check your outlet,power strip or ground. the chances of having two problems with the same cord makes me suspect the cord.I have had that happen before,turned out the cord had an internal short in the plug,strange too,it wasnt molded in thats why it shorted,it allowed the wires to move and contact each other,but only occasionally.
+scott firman That's what I'm thinking. If not the strip and gremlins are still showing up in more than one piece of equipment, I'd also look at what's going on with the mains coming from the wall. Could even be Dave's solar panel stuff switching, if not coming down the line from the power company.
+scott firman My guess would be that the fuse is rated too low, or it isn't a time-delay fuse. The power supply in that thing will surely draw much more than half an amp, which may unexpectedly blow a weak fuse upon turn-on. (I've had this happen before :P) Some things have some serious inrush current. In fact, I thought I had a dead toroidial transformer, but it turned out that the thing had such a low impedance that the magnetizing current spike was enough to trip the circuitbreaker in our breaker panel!
Thanks for the video, Dave. I was thinking about buying one of these and the information will help if the issue happens to me in the future. I will still definitely buy one of these power supplies as they are within my price range and do the job that I want.
Quite typical to have hand soldering flux residue on heat-sink-mounted pass transistors. You didn't mention how you were controlling that relay or how fast. You may have cycled that relay fast enough to generate back EMF to blow the FET. ALL FET junctions have capacitance on them. You pick a correct frequency and you could pop the FET.
Maybe you could PAT test the IEC cable if you have a tester. It could be interesting to see if it passes, I check about 30 a day and get 1 or 2 failures a week
+Gadget Addict Power supplies usually include reverse diode protection on the output. And the relay coil was on CH2, which was *not* the one that blew.
I bet that transistor was soldered on after the fact because the big heatsink was probably installed later and they attached the transistor to the heatsink to make sure it's positioned properly before soldering it.
SPOILER, Dave! 21:05 You tell us the pass Transistor has failed before ever having measured it. Unless correcting mistakes in your footage, please use YT annotations, so we playing along at home can turn them off.
1975 while serving in the U.S.Navy overseas a USO band on tour came to our little military base. While playing a set the bass guitar amp blew up. It was an Acoustic 360 if I recall correctly. One of the guys who lived in the barracks across the parking lot from the club brought his bass amp over to the club so the band could continue that night. We hauled his amp out to the workshop and sure enough the output transistors were blown. Yes, the 2N3055 came to save the day! popped them in and were jammin'!
2N3055 was the go-to transistor in many instrument amps then. And in some hifi gear too... The famous J. Linsley Hood class A amp, and the NAD 3020 (3055/2955). Their limit is really the Safe operating area, so above appr. 50 W output, the 2N3442 was a better choice. Kind of miss the days when an amp was 5-6 transistors, an output stage and a huge power supply...
+Bo Helsted Yes I know exactly what you are talking about. My Pioneer SPEC 2 power amp is nothing but huge caps, a massive torridal transformer, TO-3 (6 per side) output transistors and massive heat sinks. 60+ pounds! I bought it while overseas in the PI in 1975 and it is still going strong. Probably be an item contested in my will when I pass on. And to think that we have Bob Carver and the Phase Linear 400 power amp that started it all.
Bad day at the office. Dave was super lucky he did not blow his replacement board fuse when he was looking around for heat and whatnot. I guess he had another sort of non-functional board somewhere and he salvaged the mosfet?
Also, the "mystery power" might have been current flowing through a capacitor on the output that was only rated for 35V. (Dunno if there is one, but it's something to check)
the relay has a bobin which can generate high voltage, and MOSFETS under excessive voltage can have current carriers punch through and short to the gate. ALso by exceeding the output supply voltage you forward biased the MOSFET internal source to substrate diode.
I have a question... Why would they drive a mosfet with a BD139? I mean a MOSFET is voltage drive, you could just drive it directly from an op-amp, I usually find those BDs driving some power BJTs where you actually need some serious current to drive it...
+RPBCACUEAIIBH a lot of mosfet drivers use half-bridge or totem pole to overcome the capacitive gate. This is when the mosfet is used in PWM switching when you want it to swing quickly. The size of that mosfet heatsink tho suggests this is being used in the analog reqion so I dunno . if the bd139 is buffering the feedback, the opposing pullup/down resistor could be selected for for good damping rate.
+MattOGormanSmith Aham... So it can switch but not instantly because of gate capacitance, and higher power driver is used to switch rapidly in PWM applications, but this is clearly linear, then maybe it is perhaps used here to achieve quicker rise/fall time... Right?! After all this supposed to be a quality product given the brand...
I think that was this particular transistor problem, not its control circuit. In circuit is fuse 5A but transistor should stand continuous drain current 55A in 100 °C and 76A in 25°C. Voltage at filter capacitors is 3x lower than 150V which transistor should stand. First, the transistor was damaged, and then the fuse blown, because there was no more current limit in the circuit. There is also a danger that if the transistor fails (breakdown), the full voltage from the capacitors (53V) will appear at the output. There is probably no protection in the system that disconnects the output voltage in such a situation? Or selfoscilation in control loop ...
If I was Rigol, I'd add a snubber network across source and drain, and then a low-pass filter on the gate pin right next to the output MOSFET to prevent self-oscillation (long traces can act as antennas). Mains power surges are unlikely in this specific case, as the supply is linear and uses a large (laminated iron) toroidal power transformer. This means that it's horrible at passing high frequency noise from the power input jack to anything that cares about it, including the digital logic.
Dang, I just bought one a month ago. I figured if it was good enough for Dave, it was good enough for me. Maybe I should have purchased a used HP on ebay.
Wouldn't it be a good idea to add the missing diodes around the output MOSFET? If they added them in the next revision of the board it is probably to prevent this kind of problems.
total shot in the dark here, but could the material on the actual contacts be causing some kind of piezo effect sending a voltage spike back to the PSU? Would be interesting to put a scope just to see what happens when it comes on/off. Or perhaps it's best not to taunt the legendary killer relay, maybe it's next victim will be a scope.
+Red Squirrel No, I don't think so. I'm certain that either one or both of two things has happened: 1. The dead short from the relay contact caused the series pass FET to exceed it's SOA (Safe Operating Area). 2. Don't be fooled by the low inductance of the test leads - there is still enough inductance to have very high frequency ringing, especially with high current pulses like this. The voltage may have risen so fast that the FET's body diode didn't have enough time to turn on, thus allowing a high voltage spike to kill the FET.
recently blew my lab psu, which contains a 2n3055 lol. replacing was a quick job.. i like these to-3 packages. not sure what i did.. definately had something to do with back emf like on all my failed psu's
The 15 megs you measured on the pastrans - might it come form leakage from all that flux on the bottom side? or did you clean it up? Cause on the other version board it clearly did not measure so "low".
superdau He measured GS and GD more then once on the first board too - and nothing there, so that shouldn't cause that. Aside that the measuring should not be able to affect it that much.
ABaumstumpf Try it yourself. A measurement can easily affect it that much. Charge the gate on a MOSFET to 2 or 3 volts (or whatever the multimeter puts out in resistance measuring mode), disconnect and then measure SD. A MOSFET can stay "on" for a long time with a charged floating gate (flash memory is based on that principle and keeps data for years). I did not check what polarity he used last when measuring. Of course only the "right" one will open the transistor. The flux was on the broken MOSFET. How should that influence the measurement on the spare board or after soldering? There's also additional diodes on the second board. Will change the reading as well.
Just watched this; maybe as the relay contacts were just about to connect, the probes on those contacts were too close also, resistance was lower at that instance, and maybe the higher current caused by the relay contacts and the probe contacts in parallel caused the failure, maybe.
I see something red peaking out from under the right side of the heat sink. My guess is that it is a wire. If so, and it is pinched, there might be a short. See 25:17
Did Rigol put the current-sense resistor on the positive output? If so, that was kind of dumb because it means the common-mode voltage is orders of magnitude greater than the differential voltage across the sense terminals. So their current sensing would be wildly inaccurate during voltage transients, like when you short the outputs.
If you were powering the relay from one of the other channels is there any way the relay could have acted like an inductor so when the voltage was cut a high voltage could have been induced in the coil? Perhaps it effected the supply as a high voltage flowing back into the the other channel. But relays are found everywhere and i have not heard of them causing problems by acting as inductors.
That SPT seems quite small with its TO220 case. In the siglent where 2 TO247. My 30V 5A tapchanging powesuply uses 3 TO247. That coud bee the reason why it failed
~40V 3A = ~120W. Can a TO-220 dissipate that much? I was under the impression that a 220 is 25-35W only. So, my guess is: huge design fault that allow the power dissipation to excede the thermal dissipation limit on the package, which resulted in the part to overheat and melt internally. If allowed, you need to derate the transistor probably for the temperature. Obiviously, in this case it is again a big issue.
+thephantom1492 cheapoo...cheapoo....that´s why....no one does things like that...only chinese do this.... Even in the old days they run some 3055 in parallel if the power dissipation was around 70W. Just to be sure it will not blew off, regardless of conditions. Of course i´m aware of the second breakdown...;)
+TubiCal Exactly! My feeling is that the SOA of the FET was exceeded. That can and will kill a FET! I would like to build a linear 0-60V 0-15A (or maybe 0-5A for starters) (15A would be at the full 60V, mind you - that's 900 watts!) power supply sometime. I've figured out that I'll probably need at least 4 or 5 pass transistors for 5A, and even more for 15A :P (The power transformer is the hard bit ... I've never made one before. It would need lots of custom voltage taps. This is because the taps are switched in and out to reduce power dissipation in the bypass transistor(s). Or I could use a switching pre-regulator, but those can be tricky to design as well.
Dave - that was amazing! Two questions: 1.) when you replaced that soldered fuse, why didn't you replace it with replaceable fuse? (I'm a noob - I'm not questioning your judgement, I'm trying to learn) 2.) more basic question: would you say, if I'm not capable of fixing my power supply were it to fail as yours did, I shouldn't own this power supply?
What is your AC line voltage? There is linear transformer at the input, that is probably designed for 220-240VAC + margin for chinese transformer mfg. When you are on high side with input voltage, then output may be close to component limits. That would also kind of explain blown fuse on BKP load.
Gah, now you've got me going down the rabbit hole of trying to understand bleeder resistors and why that one doesn't affect the current sense. The fact that it's on the negative terminal doesn't really explain it for me.
Ah, I get it now. When you said it's on the other side, you meant it's *before* the sense resistor. And it's not in series with the load as it looks in the video, it's across the two rails (which you said, but which I didn't appreciate at first). Got it.
Is it me or are modern electronics more liable to blow an internal fuse? I've had this happen on multiple occasions. I recently got a flatscreen LCD from Goodwill that didn't work. What was the culprit? You guessed it. A blown fuse. So, what's the deal?
i had this problem on mine ... i replaced the transistor and still nothing so i stuck it up on ebay as broken ... should have checked the damn fuse! ... im not used to fuses of that style i might have just thought it was a resistor and glanced over it checking everything else it was a real nice unit and when i get the money again i will be happy to buy one again and just not abuse it again i think i was testing a design for a flyback DC/DC with way too high of a duty cycle when it popped
+Alyx BioHaz I didn't see how the fuses were laid out but I suppose when the pass transistor shorted it dumped the unregulated current back into the driving transistor. We saw it was pulling 7 amps so that then exceeded the rating of the fuse and the fuse popped. (Aww! He popped it.) Normal worst case scenario should be between .2 and .5 Watts through the driver transistor based on the resistance measured and a range of 30 to 53 volts (the unregulated input voltage) but the short kicked it up to 210 watts. (30V @7A). I'm surprised that it held up with that much heat being dissipated through it.
my best guess is that the relay somehow switched the current faster than the device could react to the change in current and popped the fuse causing an inductive spike on the fet popping it ... makes more sense in my failure but simply being shorted by a relay would not really explain it, you do have a nice theory hey dave, grab a few hundred of these and test them out! XP (added points for photonicinduction reference)
Blue smoke technology is what runs all solid state components. i just finished building a power supply today with four 2n3055 transistors. handles 20A. i am using a rewound MOT which gets nice and toasty as you would expect.
Stupid Questions from a beginner: @26:21 he says that Gate-Drain should be open. I get that. Then he checks Drain-Source: we see a 4.5k resistance - as we would expect. Why would we expect 4.5k resistance Drain-Source? When the Gate is above RDSon, I'd expect 0.0x Ohms or so... Wenn the Gate is below RDSon, I'd expect Megaohms. I'm a complete beginner at fets, so where is my mistake?
The excess voltage triggered the crowbar circuit (the BJT etc) and blew the fuse as designed. Wonder why it didn't blow the second time? Was the fuse replaced in-kind? How clean is your mains voltage?
2N3055 are a lower base voltage and do tend to blow. I threw out all mine and used BUX20s for years with never a one blown, heck, but I'm an old school goose.
haha im just at the bit where his second supply failed. 'winner winner chicken dinner! oh what the hell??' such a quick turn around. i love this channel.
It is 2021, almost 5 years after this video:
Silly me shorted the rails of my breadboard design. Result:
- A spark occured during the short
- An opamp got fried (to be expected)
- Channel 1 on my DP832 didn't work anymore (not to be expected)
I remembered this video from years ago and rewatched it with regard to my case:
- Same error indications
- Same faulty FET and fuse
- Replacement of both parts repaired my DP832!
Thanks Dave that you shared this video which helped me out with the troubleshooting and a cheap repair. As if you had known that I was going to goof up my project five years later :-).
Dave as a technician for over fourth years the saying in the trade was.... "Engineers can design, but have no idea of repair." The Fuse blew, that was the protection, what caused it to blow? You simply replaced the fuse without tracking down the cause. You never even measured the FET so you had to go back in again.
Love that FAIL button!
+Fran Blanche Oh its Fran! Again. In the comment box. :D Good on you Fran!
+BloodySword Everyone loves Fran!
Dave,
I had the same issue with a 2N3055 supply I built years ago. The issue was caused by the response time for the current limit circuit. A quick short caused the device to operate out of its safe area of operation VI region for a short period of time. It was quite device dependent, my channel 2 device never had the problem but I killed 3 channel 1 pass transistors on one week.
Craig
you should always screw transistor on a heatsink before soldering to avoid putting stress on the solder joints, so it makes sense if the transistor was soldered on manually after the big heatsink and transistor had been mounted
Was just thinking the same; and Dave said it when he replaced the faulty FET. They can't run the board through soldering with screws and heat sinks on. Another thing: What's the fuzz about the flux? Could be an issue in high impedance design, but the gate looks into the output of a low impedance op-amp (or the like). No worries from my point of view.
Will be interesting to see what the folks at Rigol will have to say about this documented component failure or the "dodgy" solder job on that component. Keep us updated, Dave!
At 13:22 the chinese words are 黑 and 红 which means black and red respectively.....
To tell the assemblers how to assemble :D
3:18 黑：black 红：red
It's probably good for assembly.
It is strange that they decided to use chinese caracters instead of something more universal (+ , -). I guess whatever works.
sorry to disturb the dead but, they use black and red labels because the underpaid Chinese assembly workers don't know anything about electronics. they simply just put the square peg in the square hole and the round peg in the round hole; likewise the red cable in the spade labeled "red"
I bet since that heatsink was screwed to the board, they chose to hand solder the pass MOSFET and heatsink assembly as a later step as I bet the heatsink-to-board screws wouldn't like wave soldering. Not to mention that without the screws that heatsink would put all of the mechanical load on the MOSFET leads. The other heatsink (the BJT) has solder tabs for mounting, and since it's a "wrap around" flush-to-board type heatsink (and a smaller one too), it can be safely placed on the board before wave soldering.
Did the old board also have hand-soldering evidence before you pulled the MOSFET?
Also, did the relay have a flyback diode/resistor built in? If not, that could be the issue. A coil (relay, solenoid, etc.) without a flyback/snub diode/resistor will blow all sorts of things to bits, even BJTs (or so I was taught in college).
Except that the only inductance would be in the leads between the relay contacts and the power supply. It was not the channel powering the relay coil but the relay contacts alone that blew, so no coils involved, just a few feet of wire for the alligator clips, so not much inductance there. Plus there is several hundred microfarads on the power supply output, so even if there were an inductive spike, that is a lot of capacitance to overcome before reaching harmful voltage levels. At this point I would be more suspicious of inductance in the power supply's own transformer, though, again, there is an awful lot of input filter capacitance to overcome, which ought to make that an unlikely source as well.
It is very possible, even highly likely, that the MOSFET was previously damaged by ESD in manufacturing or assembly (prior to final incorporation into the PCB) and had already been heavily damaged and the repeated shorts just finished off what good silicon remained.
+Ethan Poole yep, and they didn't even bother to clean up the flux.
+Larry Bolan As Ethan said. It's the contact channel that blew, only lead indictance involved. It's most likely a control loop response time thing taking out a MOSFET that couldn't handle it.
+Ethan Poole +EEVBLOG Doh! Wrong side of the relay... I feel really stupid. Knowing that, I concur with dave that the control loop wasn't designed to handle such spikes. That's the trap with digital "advanced" DC supply design; if you just use a plain feedback 1-transistor (or FET) supply circuit, you have no short circuit protection and rely on the speed of your microcontroller to perform current limiting. There is a three-transistor circuit that I remember from school that does have short-circuit protection built in; however, in a case like this that may not be the desired functionality.
Having worked on many audio amps, lots with MOSFET output stages, I have witnessed the occasional failure from previously stressed components. Possibly, with the clue here that the soldering was done by hand, this could have been a dodgy FET that found its way into a production unit? However these devices became ''weakened'' , occasionally, they seem to perform ok until asked to do a bit of heavy work ie, a fast transient in a difficult load whereupon they catastrophically fail. Maybe the gate had partially broken down or almost had a hole blown through, but survived to a few atoms - who knows. Maybe it was just one of those.
I'm guessing they did factory testing of a known issue. Sure, half of this batch was dodgy, we'll factory replace it if it fails the stress test.
Oh, it failed? Replace it. Test it again. It passed? Ship it!
Only now you have a component that is known to be borderline by nature of the batch, and has been stressed by the factory testing, and now Dave puts sparks on it.
It's kinda no wonder why it failed.
Have killed MOSFETs in a power supply before, seems the problem is less shorting, but shorting indecisively: on, off and then on again in short succession.
My understanding being that the turn off causes a smallish back emf that causes the freewheel diode inside the transistor to conduct, but because it's a diode, and has a reverse recovery time, if the output is shorted again during that period, it's gonna conduct a lot of current till the diode recovers fully, which can be too late to save the transistor. So maybe the relay bounced/chattered?
To protect it you'd probably need a freewheel diode that was simultaneously really fast, but with a lower voltage than the body diode, so it never gets conducting
Schottky?
Inductive kickback from the collapsing field of the coil. Sounds like a trap for beginners. Lots of volts there.
+Doug McArtin the coil was on channel 2, which didn't fail
I always put a 15v zener diode on those MOSFETs between Gate and Source. I put it right at the transistor also. I didn't see one on that board. All it takes is a fast voltage spike to punch through the G-S insulation and powee, there goes your transistor! Maybe a small value gate resistor is also a good idea too?
Damn you Dave. It's 4 AM, I was just going to sleep but you had to upload new video.
I'm at the 15 minute point in the video, the close-up shot of the transistor. It looks like a poor solder joint on the rightmost pin.
I believe that this fault has nothing to do with the relay or the load you applied because I have seen a pretty similar problem with an other PSU. The source of a problem was just an unwashed rosin-based flux believe it or not. It looks like after some time rosin tends to recrystallize and absorb moisture. I have seen like a similar blob of flux between pads like this apeared to measure only a couple of kOhms. In your case I think the MOSFET just got fully opened by the leakage to the gate loosing any means of regulation and producing too much heat and in the end it just died before it could be saved by the fuse.
And the worst part is that the damn thing will work just after production. It will pass any automated testing. And once you reheat the flux on the joint with a soldering iron the conductivity vanishes and the problem goes away again. I had to cut the device I had issues with of its pins co confirm that it really was the case (I was "lucky" to have a couple of similar failed chineese PSU modules).
The moral of this story is that leaving crap and junk like this on a board contaning any high impedance circuits is just asking for a troube to happen after a couple of moths. So just wash your boards and everything should be fine.
+Кирилл Рагузин I agree and have seen this happen to me.
I've seen the same thing happen.
In fact, the flux conducted enough to start burning the circuitboard, creating a carbon track! O.o
+Benjamin “Ozias” Esposti Gee, carbon tracks are pretty dangerous once we are talking about mains. I hope it was not under mains.
+Krisztián Szirtes
Nope, it was 12V!
+Benjamin “Ozias” Esposti Well that's a lucky case then. If it becomes graphite it can do some nasty little ( and lethal) things.
This was one of my favourite videos. I really enjoy watching (and learning) your process of troubleshooting.
27:35 the MOSFET (and the heatsink with it's screws) was installed after the solder bath. Hence the hand soldering.
For those playing at home. Have the save PS. Same issue... but it was only the fuse that crapped the bed. Transistor was ok. Tested ok from min current to max current of channel1 using same 8500 electronic load. Thanks Dave!
As a matter of course, thou shalt not replace fuses before investigating what made them blow in the first place. Otherwise, you risk blowing the new one, and possibly even causing (further) damage to the circuitry downstream.
Watching this reminded me just how amazing my PSU is.
It's one of the $50 eBay 30v 5A linear power supplies. I tried leaving it with a 5A load to see how it reacted, I saw a thermal switch on the heat sink (a single plate of metal) and assumed that would shut off the output. It actually turned on the fan when the plate got above 80 degrees C or something. Not that that kept the temp from rising further. My guess is the thing would let out the smoke after 15 minutes at full load. *Now that's quality*
Oh and my first PSU was a switching one of similar quality. I killed it by powering a DC-DC converter. Somehow.
I'm pretty sure I managed to repair it by replacing a blown MOSFET, but then it died again and it had to go in the bin.
I have a "LONG WEI" 30V/5A supply I got for 45€. It had the fan that ran forever, so I put an 80°C thermal switch on the heatsink and a 55°C switch on the transformer (so if either of these get hot it will turn on the fan). I also put a 95°C switch on the heatsink so that if it ever gets THAT hot, it will blow the fuse (so that I am force to troubleshoot why it got so hot). I also partially covered the vents on the side leaving about 2cm open towards the front on both sides so it passes all the air around the transformer, and changed the JWCO caps with Elna 105°C ones.
Best 45€ ever spent. This thing is EXTREMELY robust. I used it to charge a 250Ah/12V battery bank (with the current pot set to maximum!!!). Of course it took 4 days to do that, but during those 4 days it was on 24/7, running at 5A and 14.4V and it never blew the fuse! I have also abused it in several other ways, including using it to power self-oscillating relays, shorting the output accidentally too many times, using it to overrev DC motors, powering "dirty" circuits like flyback inverters etc.
It is still going strong after 4 years of abuse. I usually also leave it on overnight to charge batteries at low currents (so the fan doesn't come on) without any worries.
The only thing that happens is that when it gets hot the voltage display drifts a bit. The regulation is rock solid however, so I usually set it when it's cold and then leave it like this.
If your going to use your power supply to trip mechanical relays at least only do so by applying and removing the lead instead, do not deactivate the relay by simply hitting the power switch on the power supply as that sends a massive voltage spike back through the output of the supply which is what happened here.
Or you can simply buy mechanical relays with flywheels built into them.
I believe this is the first video of yours I've watched where a repair was successful. Good job Dave! Love it. Also, I'm glad I'm not the only one that has equipment fail randomly and for no good reason.
Single TO-220 device for a 90W power supply? Seems a bit wimpy to me.
I used two TO-247 MOSFET for my 30V5A/15V10A 150W-max bench power supply to get better thermal performance. With multi-tapping transformer, the max dropout power was < 50W.
I can dissipate 250W in such a package, though its back was soldered to piece of copper before going onto the heatsink.
I'm a bit surprised by your comments about "a lab supply of this price and quality". If you had worked in any electronics design team (...and I mean any, including the most prestigious military ones too) you'd know that now a days there is no thing as careful design. It is all driven by cost and schedule and the over riding paradigm is get it done yesterday for 10% less than tomorrow. The only reason things are expensive is mostly due to branding mark-up. This $1200 unit most likely cost around $100 or less to produce. So there should be no surprises when it fails due to (most likely) inadequate product and/or production testing.
Odds are the series pass transistor was hand soldered because it was bolted to the heatsink in a jig, and then the assy is bolted to the PCB, then the transistor is soldered in place. Very normal for that kind of thing that I've seen. And odds are they used no-clean flux.Not that that makes it ok....
If you revisit the relay video attempt, put a charged big-ass capacitor across the normally open contacts. Get a nice spark!
At 12:40, the board has no protection diodes on the pass transistor, but at 28:40, when measuring the pass transistor, the protection diodes magically appear on the board. It appears to me that he was measuring the replacement board, not the failed one.
Dave, Chinese built unit likely has some counterfeit Chinese transistors used in it.Therefore the pass transistor might not have met spec. Heat from dead short was just too much for the FET.Hand soldering is not such a surprise as the Heat sink has to go on after Wave Soldering and the transistor gets mounted on the heat sink then soldered.
At first I was worried because the video update at 15:33 says that the newer Power MOSFET was introduced after Nov 2015. Good thing I went to the forums because everywhere else says Nov 2013. I didn't want my DP832 that I got last Summer to be prone to transistor failure!
The problem is possibly the speed of the short and whether the short protection circuit is fast enough to respond.
your right, but I think the point here is that it should be fast enough.
Glad it’s not just me that starts to fix some thing and end up fixing two other things I’m using before the finish ! Great to watch thanks
One of the best videos I have enjoyed from you.
Glad to hear your opinion on the benefits between the BJT and the MOSFET. In school I spent most of my time playing with 2N3904, and very little time with the MOS.
Thanks!
I'm curious about the voltage rise during the test starting at 18:45. It seems to rise for a couple seconds, then pause, then rise some more, then pause, etc. I would have expected it to be smooth (then of course slow down as it reaches its maximum). Any idea what was going on there?
This is just the meter refresh rate
It's probably the sense/negative feedback causing the voltage to leak out onto the output when it's output transistor was blown. Why it isn't smooth, god knows. It could be the multimeter lagging as well.
Dave, have you checked to see if the other transistor is still drawing excessive current? Perhaps it caused the main transistor to fail.
Still weird, even more weird that the stand alone DC load suddenly blew it's fuse.
Makes me think, what if it's not the output what's caused the issue with the legendary killer relay
But the input, how stable is the AC outlet, maybe it spikes every now and then like crazy?
Low Quality Mosfet ?
I have one of these in the A version. The current limiting is quite slow. I was using it to test some LEDs with the limit set to 10 mA and 5V. The first one I tested was fine, but then I blew 3 in a row before I figured that the current limit was just a suggestion for the initial few milliseconds. Unfortunately, Rigol was what I could afford. The control software over the network is also very bad, although they did update it and it is improved.
You do a great job with these videos Dave. I love all your additional bits of info, as they can certainly help with repair of unrelated items. I learn so much from your efforts, and really appreciate you taking the time to pass on your knowledge.
Dave would you consider doing o video on VLSI concepts?
What do SRAM and DRAM look like from a schematic view? How do they work?
to me, it makes sense to have the 30 V channel paired with the 5 V channel on the same PCB instead of going for symmetry and putting both 30 V channelss on the same PCB. Spread the heat-load (or however you might call it).
Maybe you have transient spikes on your line voltage, Something upset the BK 8601.
Big Thumbs Up.
+Mike James Maybe the power plug was wiggling. I once killed the MOV and the auxiliary rail switcher IC of a non isolated SMPS that way.
True, I have seen arcing primary side dry joints due to a loose live connection in plug.
Looks like there might be a need for a snubber acros the pass transistor too. This will stop high transients appearing when you power up.
I killed many power transistor to learn about this concept :)
Collapsing magnetic fields from relay coils/solenoids can be very destructive. I learned the hard way years ago. :-) Probably better off replacing the 5A fuse with a 5A polyswitch.
+electronicsNmore LOL. that would have affected the 12 volt source! NOT the 30 v.
+electronicsNmore No, he was apply 30V 3A to contacts side - not coil side so no magnetic field - spikes happen.
I would not go for polyfuses in that case, he have 5A fuse so it's way up 3A limited. Only scenario when this fuse will will go open is hardware fail - not done by user - so the best scenario is cut power. Poly will turn power again after getting cold. They are good for protecting output when You can accidentally short leads without current limiting.
Dave Blane Well he mentions in the video "possibly" back EMF, talks about a blocking diode in the PSU, so clearly it was not back EMF if he did not connect the relay coil to the Rigol PSU. The PSU mosfet simply did not like the 30V/5A shorted output. Maybe he connected up the relay coil off camera to quickly test it, then later discovered the problem. Who knows.
+Electronic Noob Blog
Incorrect. The wire to the contacts is still an inductor, and it can create very fast, high frequency transients because of the low inductance.
I read about this recently...it has something to do with the collapsing field and the voltage/current generated within the inductor and sends it backwards?
If gear keeps failing, then I would be tempted to do some mains power quality monitoring. It sounds like you may be getting some surges propagating through gear causing failures.
+Zadster Never had another item in the lab blow a mains fuse just doing nothing.
Did the same thing to a Mastech HY3005F-3 power supply, was shorting out channel 1 and it took a crap on me. I took it apart to see if I could find anything, not knowing anything about a series pass transistor (still learning) and I found nothing obviously wrong, when I powered it up again it worked fine.
Single most common failure in a linear supply (aside from filter caps after a certain number of years) is indeed the series pass transistor. Pretty much the first thing to check on a dead linear supply is a shorted series pass transistor.
That´s why i prefer the old, calssic, vintage, all analog power supplies as my standard lab supply. If they fail....i know why and it would be easily repairable....
If these fancy-pancy with nonsense bells and whizzles added USB/eth, digital, 7segment, all uC controlled rubbish fails...naah...too much hassle. If one of these fails due to a ground rail spike, i´ll bet the whole thing will fail and went imediately BER.
I got a new Rigol DP 832 in 2021 and I am enjoying using it. Hopefully this problem that Dave mentioned will not happen in the newer units. One criticism that I have of Rigol is the lousy user guide (manual). They put everything under the sink in the guide without clear steps on how to quickly get to a particular feature and set it up. It would have been nice if they gave examples for setting pulsed waveforms. I need to pulse the current to a PCB heater circuit for a sensor application. If there is any UA-cam video on this I would appreciate the info. Thanks.
check that power cord,if it blew the fuse in the other piece of equipment,mabe it caused a spike or short in your first event,also check your outlet,power strip or ground. the chances of having two problems with the same cord makes me suspect the cord.I have had that happen before,turned out the cord had an internal short in the plug,strange too,it wasnt molded in thats why it shorted,it allowed the wires to move and contact each other,but only occasionally.
+scott firman That's what I'm thinking. If not the strip and gremlins are still showing up in more than one piece of equipment, I'd also look at what's going on with the mains coming from the wall. Could even be Dave's solar panel stuff switching, if not coming down the line from the power company.
+scott firman
My guess would be that the fuse is rated too low, or it isn't a time-delay fuse. The power supply in that thing will surely draw much more than half an amp, which may unexpectedly blow a weak fuse upon turn-on. (I've had this happen before :P)
Some things have some serious inrush current. In fact, I thought I had a dead toroidial transformer, but it turned out that the thing had such a low impedance that the magnetizing current spike was enough to trip the circuitbreaker in our breaker panel!
Do you think the mosfet was hand soldered because they attach it to the heatsink first the put on?
Thanks for the video, Dave. I was thinking about buying one of these and the information will help if the issue happens to me in the future. I will still definitely buy one of these power supplies as they are within my price range and do the job that I want.
i bought a power supply a few days ago and thanks to your reviews it was neither the siglent nor the rigol.
Quite typical to have hand soldering flux residue on heat-sink-mounted pass transistors. You didn't mention how you were controlling that relay or how fast. You may have cycled that relay fast enough to generate back EMF to blow the FET. ALL FET junctions have capacitance on them. You pick a correct frequency and you could pop the FET.
that is why there is always a diode used at the input terminal of relay to short the inductive voltage.
Maybe you could PAT test the IEC cable if you have a tester. It could be interesting to see if it passes, I check about 30 a day and get 1 or 2 failures a week
The reason why the soldered in fuse blew is because the pass transistor failed shorted for a split second, likely due to an internal arc.
i think dave you need to get a local preist to do an exorcism of the lab just in case
+Bruce Miller Yep. Need to summon the spirit of Bob Pease to do battle with the dark forces.
Did you use a diode when testing the relay? Otherwise it could have sent a very high voltage back into your power supply when it turned off.
+Gadget Addict I haven't watched the video yet but I'd be surprised if he didn't mention the back emf being a possible cause!
+Gadget Addict He talked about this and having not used one in the video but the PSU should have protection against it even a cheaper one would.
+Gadget Addict Power supplies usually include reverse diode protection on the output. And the relay coil was on CH2, which was *not* the one that blew.
I bet that transistor was soldered on after the fact because the big heatsink was probably installed later and they attached the transistor to the heatsink to make sure it's positioned properly before soldering it.
SPOILER, Dave! 21:05 You tell us the pass Transistor has failed before ever having measured it. Unless correcting mistakes in your footage, please use YT annotations, so we playing along at home can turn them off.
1975 while serving in the U.S.Navy overseas a USO band on tour came to our little military base. While playing a set the bass guitar amp blew up. It was an Acoustic 360 if I recall correctly. One of the guys who lived in the barracks across the parking lot from the club brought his bass amp over to the club so the band could continue that night. We hauled his amp out to the workshop and sure enough the output transistors were blown. Yes, the 2N3055 came to save the day! popped them in and were jammin'!
2N3055 was the go-to transistor in many instrument amps then. And in some hifi gear too... The famous J. Linsley Hood class A amp, and the NAD 3020 (3055/2955). Their limit is really the Safe operating area, so above appr. 50 W output, the 2N3442 was a better choice. Kind of miss the days when an amp was 5-6 transistors, an output stage and a huge power supply...
+Bo Helsted Yes I know exactly what you are talking about. My Pioneer SPEC 2 power amp is nothing but huge caps, a massive torridal transformer, TO-3 (6 per side) output transistors and massive heat sinks. 60+ pounds! I bought it while overseas in the PI in 1975 and it is still going strong. Probably be an item contested in my will when I pass on. And to think that we have Bob Carver and the Phase Linear 400 power amp that started it all.
I'm curious to see what the harmonics look like on your mains. I'd bet that may be an explanation for the blown fuse in your BK electronic load.
Bad day at the office. Dave was super lucky he did not blow his replacement board fuse when he was looking around for heat and whatnot. I guess he had another sort of non-functional board somewhere and he salvaged the mosfet?
TO220 on that huge heatsink? Why not using a TO247 instead?
+marco56702 C-grade MOSFets....cheapooo -- get what the supplier have on hand for cheap...ever heard of that CET-brand on that MOSFet...?
+TubiCal never heard of this brand, I strongly think you're right, it's way better to buy an used but name brand power supply
I think you had to have a diode on the coil as you are using a common power supply for the relay and the load
Also, the "mystery power" might have been current flowing through a capacitor on the output that was only rated for 35V. (Dunno if there is one, but it's something to check)
the relay has a bobin which can generate high voltage, and MOSFETS under excessive voltage can have current carriers punch through and short to the gate. ALso by exceeding the output supply voltage you forward biased the MOSFET internal source to substrate diode.
Dave did you put the demon relay near the BK electronic load before it blew up?
I have a question...
Why would they drive a mosfet with a BD139? I mean a MOSFET is voltage drive, you could just drive it directly from an op-amp, I usually find those BDs driving some power BJTs where you actually need some serious current to drive it...
+RPBCACUEAIIBH a lot of mosfet drivers use half-bridge or totem pole to overcome the capacitive gate. This is when the mosfet is used in PWM switching when you want it to swing quickly. The size of that mosfet heatsink tho suggests this is being used in the analog reqion so I dunno . if the bd139 is buffering the feedback, the opposing pullup/down resistor could be selected for for good damping rate.
+MattOGormanSmith Aham... So it can switch but not instantly because of gate capacitance, and higher power driver is used to switch rapidly in PWM applications, but this is clearly linear, then maybe it is perhaps used here to achieve quicker rise/fall time... Right?! After all this supposed to be a quality product given the brand...
I think that was this particular transistor problem, not its control circuit. In circuit is fuse 5A but transistor should stand continuous drain current 55A in 100 °C and 76A in 25°C. Voltage at filter capacitors is 3x lower than 150V which transistor should stand.
First, the transistor was damaged, and then the fuse blown, because there was no more current limit in the circuit.
There is also a danger that if the transistor fails (breakdown), the full voltage from the capacitors (53V) will appear at the output. There is probably no protection in the system that disconnects the output voltage in such a situation?
Or selfoscilation in control loop ...
Were the pass transistors hand soldered on both boards ?
If I was Rigol, I'd add a snubber network across source and drain, and then a low-pass filter on the gate pin right next to the output MOSFET to prevent self-oscillation (long traces can act as antennas).
Mains power surges are unlikely in this specific case, as the supply is linear and uses a large (laminated iron) toroidal power transformer. This means that it's horrible at passing high frequency noise from the power input jack to anything that cares about it, including the digital logic.
Dang, I just bought one a month ago. I figured if it was good enough for Dave, it was good enough for me. Maybe I should have purchased a used HP on ebay.
Eh, any component can fail at any time but pass all tests before shipping. It's why there's warranties!
Wouldn't it be a good idea to add the missing diodes around the output MOSFET? If they added them in the next revision of the board it is probably to prevent this kind of problems.
Dave sounds like an old nice grandma in voice intonation, good integrity.
total shot in the dark here, but could the material on the actual contacts be causing some kind of piezo effect sending a voltage spike back to the PSU? Would be interesting to put a scope just to see what happens when it comes on/off. Or perhaps it's best not to taunt the legendary killer relay, maybe it's next victim will be a scope.
+Red Squirrel
No, I don't think so.
I'm certain that either one or both of two things has happened:
1. The dead short from the relay contact caused the series pass FET to exceed it's SOA (Safe Operating Area).
2. Don't be fooled by the low inductance of the test leads - there is still enough inductance to have very high frequency ringing, especially with high current pulses like this. The voltage may have risen so fast that the FET's body diode didn't have enough time to turn on, thus allowing a high voltage spike to kill the FET.
recently blew my lab psu, which contains a 2n3055 lol. replacing was a quick job.. i like these to-3 packages.
not sure what i did.. definately had something to do with back emf like on all my failed psu's
The 15 megs you measured on the pastrans - might it come form leakage from all that flux on the bottom side? or did you clean it up?
Cause on the other version board it clearly did not measure so "low".
+ABaumstumpf
No. Most likely he charged the gate of the MOSFET by measuring the resistance, so it wasn't completely "off".
superdau He measured GS and GD more then once on the first board too - and nothing there, so that shouldn't cause that.
Aside that the measuring should not be able to affect it that much.
ABaumstumpf
Try it yourself. A measurement can easily affect it that much. Charge the gate on a MOSFET to 2 or 3 volts (or whatever the multimeter puts out in resistance measuring mode), disconnect and then measure SD. A MOSFET can stay "on" for a long time with a charged floating gate (flash memory is based on that principle and keeps data for years).
I did not check what polarity he used last when measuring. Of course only the "right" one will open the transistor.
The flux was on the broken MOSFET. How should that influence the measurement on the spare board or after soldering?
There's also additional diodes on the second board. Will change the reading as well.
Just watched this; maybe as the relay contacts were just about to connect, the probes on those contacts were too close also, resistance was lower at that instance, and maybe the higher current caused by the relay contacts and the probe contacts in parallel caused the failure, maybe.
So this wouldnt be something that would be covered under warranty?
+Jacob Allison That would be totally uninteresting!
+EEVblog good point! I had fun watching this.
I see something red peaking out from under the right side of the heat sink. My guess is that it is a wire. If so, and it is pinched, there might be a short. See 25:17
+t1d100 You can see it even better at 27:09
Did Rigol put the current-sense resistor on the positive output? If so, that was kind of dumb because it means the common-mode voltage is orders of magnitude greater than the differential voltage across the sense terminals. So their current sensing would be wildly inaccurate during voltage transients, like when you short the outputs.
+Steve Robbins High side current sensing is quite common in PSU's.
I was loading down on a 10W resistor the other day glad mine didn't let the magic smoke out.
If you were powering the relay from one of the other channels is there any way the relay could have acted like an inductor so when the voltage was cut a high voltage could have been induced in the coil? Perhaps it effected the supply as a high voltage flowing back into the the other channel. But relays are found everywhere and i have not heard of them causing problems by acting as inductors.
That SPT seems quite small with its TO220 case. In the siglent where 2 TO247. My 30V 5A tapchanging powesuply uses 3 TO247. That coud bee the reason why it failed
~40V 3A = ~120W. Can a TO-220 dissipate that much? I was under the impression that a 220 is 25-35W only. So, my guess is: huge design fault that allow the power dissipation to excede the thermal dissipation limit on the package, which resulted in the part to overheat and melt internally. If allowed, you need to derate the transistor probably for the temperature. Obiviously, in this case it is again a big issue.
+thephantom1492 cheapoo...cheapoo....that´s why....no one does things like that...only chinese do this....
Even in the old days they run some 3055 in parallel if the power dissipation was around 70W. Just to be sure it will not blew off, regardless of conditions.
Of course i´m aware of the second breakdown...;)
+TubiCal
Exactly!
My feeling is that the SOA of the FET was exceeded. That can and will kill a FET!
I would like to build a linear 0-60V 0-15A (or maybe 0-5A for starters) (15A would be at the full 60V, mind you - that's 900 watts!) power supply sometime. I've figured out that I'll probably need at least 4 or 5 pass transistors for 5A, and even more for 15A :P
(The power transformer is the hard bit ... I've never made one before. It would need lots of custom voltage taps. This is because the taps are switched in and out to reduce power dissipation in the bypass transistor(s). Or I could use a switching pre-regulator, but those can be tricky to design as well.
Is there any mileage in adding another series pass transistor in parallel? Being FETS they should load share automagically.
+EEVblog maybe your main power line was buggy that day and fried the PSU and the electronic DC load?
Dave - that was amazing! Two questions: 1.) when you replaced that soldered fuse, why didn't you replace it with replaceable fuse? (I'm a noob - I'm not questioning your judgement, I'm trying to learn) 2.) more basic question: would you say, if I'm not capable of fixing my power supply were it to fail as yours did, I shouldn't own this power supply?
What is your AC line voltage? There is linear transformer at the input, that is probably designed for 220-240VAC + margin for chinese transformer mfg. When you are on high side with input voltage, then output may be close to component limits. That would also kind of explain blown fuse on BKP load.
Dave, you need a free-wheeling diode across the coil if you switch it like this. I bet you had ch1 on the coil, ch2 on the contacts.
Gah, now you've got me going down the rabbit hole of trying to understand bleeder resistors and why that one doesn't affect the current sense. The fact that it's on the negative terminal doesn't really explain it for me.
Ah, I get it now. When you said it's on the other side, you meant it's *before* the sense resistor. And it's not in series with the load as it looks in the video, it's across the two rails (which you said, but which I didn't appreciate at first). Got it.
I'd check your mains supply voltage if you have multiple devices popping fuses...
Is it me or are modern electronics more liable to blow an internal fuse? I've had this happen on multiple occasions. I recently got a flatscreen LCD from Goodwill that didn't work. What was the culprit? You guessed it. A blown fuse. So, what's the deal?
i had this problem on mine ... i replaced the transistor and still nothing so i stuck it up on ebay as broken ... should have checked the damn fuse! ... im not used to fuses of that style i might have just thought it was a resistor and glanced over it checking everything else
it was a real nice unit and when i get the money again i will be happy to buy one again and just not abuse it again
i think i was testing a design for a flyback DC/DC with way too high of a duty cycle when it popped
+Alyx BioHaz So I'm not the only one, interesting!
not sure if it was overloading or a serious EMF spike ... the whole DC-DC converter was designed poorly on my end
+Alyx BioHaz I didn't see how the fuses were laid out but I suppose when the pass transistor shorted it dumped the unregulated current back into the driving transistor. We saw it was pulling 7 amps so that then exceeded the rating of the fuse and the fuse popped. (Aww! He popped it.) Normal worst case scenario should be between .2 and .5 Watts through the driver transistor based on the resistance measured and a range of 30 to 53 volts (the unregulated input voltage) but the short kicked it up to 210 watts. (30V @7A). I'm surprised that it held up with that much heat being dissipated through it.
my best guess is that the relay somehow switched the current faster than the device could react to the change in current and popped the fuse causing an inductive spike on the fet popping it ... makes more sense in my failure but simply being shorted by a relay would not really explain it, you do have a nice theory
hey dave, grab a few hundred of these and test them out! XP
(added points for photonicinduction reference)
Blue smoke technology is what runs all solid state components. i just finished building a power supply today with four 2n3055 transistors. handles 20A. i am using a rewound MOT which gets nice and toasty as you would expect.
Stupid Questions from a beginner:
@26:21 he says that Gate-Drain should be open. I get that.
Then he checks Drain-Source: we see a 4.5k resistance - as we would expect.
Why would we expect 4.5k resistance Drain-Source?
When the Gate is above RDSon, I'd expect 0.0x Ohms or so...
Wenn the Gate is below RDSon, I'd expect Megaohms.
I'm a complete beginner at fets, so where is my mistake?
The excess voltage triggered the crowbar circuit (the BJT etc) and blew the fuse as designed. Wonder why it didn't blow the second time? Was the fuse replaced in-kind? How clean is your mains voltage?
2N3055 are a lower base voltage and do tend to blow. I threw out all mine and used BUX20s for years with never a one blown, heck, but I'm an old school goose.
haha im just at the bit where his second supply failed. 'winner winner chicken dinner! oh what the hell??' such a quick turn around. i love this channel.