EEVblog #512 - Rigol DP832 Bad Design Investigation

hi I was just going to have a quick play around with this Roy Goldy p83 to power supply this is just a trace out the output circuit and probe a couple of waveforms and stuff like that maybe see if I could possibly find out what's going on investigate that power on spike a bit further but got the guts hanging out and one of the first things I noticed was that suddenly like I couldn't see the display but the feeling would suddenly start revving up and then going crazy and I found out what was happening without even probing the thing is that the power supply would actually reset after a little bit of time and I'm going to leave it here and just see if it reset so I switched on all the outputs there and hopefully if I won't touch it because I think maybe you know the reset might have something to do with you know there might be an EMI issue the board's all hanging out here all the wires the you know that some of the digital stuff going here is all loosey goosey hanging out in the breeze and maybe that has something to do with it but anyway I'm just going to leave it here there's no load on it of course and I found that the damn thing reset so I'll just leave the video running and see if we can capture that usually doesn't take too long known takes like a minute or two know as Murphy would have it it's not going to reproduce the problem the old white coat syndrome strikes again I've left it for a couple of minutes and ah annoying anyway what I was doing is I was adjusting the output voltage in steps just as a first thing just to see where it you know got the transformer taps and stuff like that so maybe if I turn the voltage back up or something like that anyway I'll leave it for another couple of minutes I don't remember moving it or touching I didn't move my multimeter from here out of the shakhter to here but I don't know better hold my tongue at the right angle hang on whoo-hoo did I get it did I get it I got it see it reset it took bloody what a nine minutes on the camera here but I got it I was just in the other cubicle and heard the fan rev up past it so the thing just resets itself let's turn that back on see if we can get it to reproduce even faster I don't know why I have no idea I'm not probing anything not touching anything the outputs up still seem to work but I've got it under no-load and the thing just reset itself there we go there we go got it bingo what was that like a minute there we go it just switched off and reset why I have no idea my only guess is because it's you know it's hanging out here it's not in the system so maybe an EMI issue getting in somewhere that's resetting this sucker I don't know all I can do really is fold this back in screw it back in place and run it as a bench supply for you know an hour and see if it see if it resets now there's one thing I found is that this regulator in here is tiny little LEDs an lm317 it's actually delivering the that isolated are 5 volts for the main logic that we saw in the teardown and I noticed that was getting really really hot Jen look at that I mean we're talking you know this isn't going to be super accurate but look I mean it's not going to read over though so you know I was getting close to 90 or something at one point there we go 95 it's crazy that thing is getting super hot and I'm not sure if that's normal or not surely it can't be running that hot unbelievable have no idea why now I've actually measured this thing and there's the 5 volt so it's actually supplying and the input voltage if I can actually get in there correct ground yet then we got 12 volts now I'm reading well over a hundred degrees on that now and I can should be able to get in there with my couple and even confirm that look at that that is ridiculous a hundred and ten degrees on the heatsink and this thing's only been on for like 10 min undred and fifteen hundred and twenty unbelievable no wonder something that regulators just going to shut down and no wonder the things going to reset so clearly something is drastically wrong there because you know like an lm317 is only got an operational temperature range to like a hundred and twenty one hundred and twenty-five degrees or there abouts and it's got built-in thermal overload protection so I think that's why these things are resetting actually as I think that even though I haven't measured it um I think the regulator's just getting too hot and shutting down for some reason and of course you can't have it operating at that temperature it's just ridiculous let alone the like I was measuring 110 degrees plus on the case after a minute or two let alone the actual dye temperature itself so yeah it's going to thermally protect itself and shut down and I think that's what's happening there because this supplies the main five volt rail which powers all of the logic board of course so you know it'll just shut off the output and it'll reset like that and of course you can't run this thing at too hot a temperature anyway being right next to these two large filter caps here even though they're 105 degree C I rated they're just going to die you know in very short order so there is something wrong here surely this cannot be normal so I'm just wondering if there's anything that I've done in the teardown to cause increase power consumption to me most supply seems to work just fine so I don't know I'm I'm at a loss here I wasn't you know I wasn't expecting to troubleshoot this thing I thought I'd um you know just be able to probe some are waveforms and and you know do a little bit of reverse engineering and stuff like that but know now I've got this bloody problem to contend with what that and there's also a software bug in this thing which are people on the forum have actually reported and confirmed and I'll just confirm it here what it is is if you set the current limit here below 10 milliamps and actually shows even with no load nothing connected at all no trickery going on here whatsoever it will actually show 30 milliamps reading currents that let's try that let's lower it 14 13 12 10 here we go bang look at that it jumps up and reads 30 milliamps crazy why unbelievable of course when goes down to zero it's it reads zero but anywhere from 1 to 10 and to nine milliamps there are 30 and then it just goes back to zero over that ah what's going on there now what I've done is put the lid back on here and it done well it's not actually resetting now presumably because it's got more airflow actually you know it's sucking in the air through here and out the back so we're getting some forced airflow there where we didn't have that before it was just still in the air and I can still get in here and and probe the heat sink in here and yeah it it rises it takes some time to get the correct contact on the thing but it's still running incredibly hot check this out look at that we're already up to 90 and remember that heat sink is right next to those caps as well 90 to 93 we're still rising that's just crazy and you know I'm not only making the best contact there with the heat sink of course so you know it basically you know round it to at least a hundred degrees you've got to be kidding me and just to make sure I wasn't hallucinating this thing or that I'd there was some bizarre fault that happened in the teardown or something like that I decided just to make sure I'd put this to the forum and I've left it overnight and sure enough there's a whole bunch of responses from other people I asked if they could test their units and they have and they have absolutely 100% confirmed exactly what I'm getting this heatsink some people are even getting a hundred and ten degrees in the case like this poking their thermocouple through somebody showed a fleur our thermal image of the heat sink up to 130 degrees Celsius of course to do that they have to take the case off and then there's no air flow and all that sort of stuff so you know much higher temperatures than what you're getting the case here but it's absolutely confirmed it the this thing the design of this Roy Goldy p83 2a is totally flawed I have no idea how it even made it past the first design review meeting with a bloody five volt oh sorry yeah five volt regulator for the main digital logic operating at a nominal around about a hundred degrees Celsius it got to be me it's a it's one of the worst design oversights I've ever seen it is absolutely bread you know awful this is bread and butter stuff for a power supply one of the first things you can look at is the bloody thermal design of this thing verify the thermal design when you're designing this product unbelievable anyway definitely confirm but what I'm going to do is I'm just going to I've left this overnight I haven't powered it up today so I'm going to switch it on and see if I can get the heatsink through here so it's got the proper air flow coming out the back of course the fans on a minimum when you first turn it on because it's no it's not loaded so potentially the heatsink could actually cool down even further when I ironically when the outputs are loaded because then it will turn the fan on greater at a greater speed you'll get greater airflow over that heatsink and potentially could actually cool that heat sink down ironically but anyway I'll switch it on see what I can get as you can see my ambient temperatures around about 23 degrees C which is you know a typical office ambient temperature I've only had it on for a couple of minutes and it's difficult to probe I've got it going straight through there I've actually took out some of the silastic between the the two filter capacitors there so I can actually get through and I am probing the heatsink but you know it's it's not ideal your contact at all but as you can see I mean you can never actually read too high on this thing so it's not like you know I can get bad contact and accidentally read high so I'm getting 75 degrees C on that heating I've got to put a bit of pressure on the thermocouple just to get you know I'm right on the side of the heatsink trust me it is that's not making good contact at all so I think the true temperature of the heating is greater than that but we're up to 77 and at the moment and still climb in 78 this is not looking good folks here we go look at this 91 degrees that is just insane really I mean you know a lot of people might think okay what's the problem right but it's all about design margin and you saw what happened when we'll verify this again when we open it back up but I'm sure that this is causing that 5 volt reset problem now they to run something at ninety to a hundred degrees to run a you know and a three one seven regulator or any heat sink at that sort of temperature for something dumb ass like a nominal you know five volt rail to power you know the digital circuitry in this thing is absolutely insane because there is no design margin in there yeah okay maybe this thing might work and might continue to work for most people for a couple of years or something until those caps dry out of course had not even mentioned in the cap shed okay without resetting at all right there might just be adequate air flow in there but what happens if you stick this damn thing in a rack you know they I think they even sell a rack mount kit for it if you've got you know a rack can easily get 40 or 50 degrees ambient in there something like that raises even if you raise the ambient temperature here in the lab by an extra five degrees that could be enough to actually reset to trip the thermal overload in the die in that lm317 and who knows what will check which try and check which lm317 they're using but that will vary based on batch it'll vary air they might have you know they might declare that they can source that from any manufacturer so it's not going to be consistent your entire production run of these units all sorts of stuff it's just it's just crazy there is no way in hell that anyone could sensibly make a decision that says to run that heatsink at 90 degrees would be a good idea for anything let alone a production unit like this it's disgusting cigar and I've managed to get almost a hundred and ten degrees look at that hundred and eight depends where I wiggle it disgusting who the hell designs a five volt regulator that runs quiescent at a hundred and nine degrees C what a now I'm going to see if I can actually verify that the reset problem I'm seeing on the on the unit is actually the five volt regulator going into thermal overload and dropping out so to speak so I've set up the scope nominally it's there it is we've got our five volt output there so pretty clean triggering at about four and a half volts on- going so we'll just leave it there and I've got my outputs I switched on so I'll definitely be able to tell when it's done that and well let's see if it correlates if we trigger anything on the scope or drop out in that five volt rail when this thing resets could take a while though and look at the temperature that thing's running at a hundred and forty-one and climbing when there's no airflow over it unbelievable it's even got some additional little heat sinking on it due to the oscilloscope probe oh I only had to wait a minute or so and it so it has reset here but I didn't get anything triggering on the scope over here so it looks like it didn't drop below that four point five volts r-value I mean I you know maybe I've got to tweak that up a bit maybe there's a voltage uh supervisor or something on the five volt rail inside the main chip that even could be a function inside the main process or something a that you know is actually detecting a smaller dropout than what's there ah let me tweak it No still couldn't get it to do it I'm triggering a four point eight volts and the thing just reset itself but I still couldn't detect any drop in that five volt rail so maybe my Theory's wrong there but jeez I don't see what else it could be I mean now it just as a rough indication as what temperature those caps are running that just by sitting near that heatsink it you know one hundred and thirty degrees or whatever it is there you go they're almost up to 70 degrees just the cans on there of course there's no air flow it's going to help when there's air flow over these things of course but that's how you can get with just you know just the coupling over to those capacitors well I'm at a loss now as to explain exactly what the mechanism is for resetting this unit I could have sworn it must have been the drop out of that regulator but I cannot seem to capture any drop out AC or DC coupled at any time base of this regulator so I I don't know but anyway what I've done is this board if when it was sitting here before it was resetting every one minute or two minutes absolute tops with monotonous regularity then I just put this sup pace fume exhaust sir I've got air blowing over it like this it's been probably more than after I'm pretty sure it's more than what was getting inside the case and all of a sudden BAM I've left for ten minutes and it's not resetting at all so it definitely looks like it has something to do with that the later that regulator but the exact mechanism ah still eludes me now have even got of the trouble to set up a window trigger here oh oh hang on yeah I just saw that I just sort of switch off I don't know if you thought there was some dip in the waveform there there was something there was something there I had damnit I hadn't had the trigger on I was too busy shooting this video but anyway I looks like we may have finally got it let me switch this back on anyway I've set up a window trigger here so that it can trigger anything outside of that those two window there so I'm going to put it into single mode again and see if I can capture that I've got it back on but let's let's see what happens let's just leave it running well no it's switched off and reset but we didn't capture anything even with that tiny trigger window there but no I still couldn't get anything there so I've gone back to AC mode and there it is there's our AC coupled mode 50 millisecond time-based there and I set up a window trigger mode just around that so let's up there we go look at that just see that look at that there we go so we've got some salt yep yep I think we're getting very close to getting this anyway you can see that right there looks like that's normal you know in quote marks right that's normal five volt AC coupled output we're only 100 millivolts per division there now let's trigger off that and see it and watch it and see what happens if we can trigger anything when these switch off and it resets bingo there we go we finally got it yes it triggered you saw it there you go this waveform started I don't know maybe a second before this thing reset itself so there you go but my theory was right as well you know it could not be wrong really it had to be that voltage regulator in this case just doing you know just some subtle drop out there it's as you saw we couldn't trigger on the 5 volt any very significant variation on the 5 volt rail itself but when you get down you know we're only talking like a hundred millivolts per division so just that sort of noise or ripple that regulator's doing something it's not regulating as it should anymore it's yeah it's still regulating the 5 volts but the AC component of it has actually changed so that is causing something on the digital board I don't know why I'm not going to go into the digital board and trying you know dissector why as you know it doesn't matter the fact is that regulators overheating and is causing reset on the digital board somehow so that's what we captured there and up that is normally so you can see that it was you know double or triple in amplitude before when it actually fails and we can no doubt capture that again it'll be fully repeatable you watch bingo too easy there it goes switched off I hopped barely even had to turn around for you know ten seconds and we captured it whoa so actually this is a really good example of a little bit of a tricky will real-world troubleshooting scenario where I had a theory okay this regulator is overheating it was dropping out causing resets on the on the processor in side this thing in some manner but you know I my theory was almost blown and blown out of the water I expected the five volt rail to just you know it plummet down to zero or drop down to three volts or I don't know do something stupid allow you know a couple of volts or ripple to come through or something horrible like that and I couldn't capture it even with a tight couple hundred milli volt window triggering around the five volt rail that I had there before I couldn't do it on that 5 volt scale so I had to skip switch to AC and I originally couldn't even find it on AC as well I've done that before but it turns out I wasn't setting my trigger point narrow enough and in this case I switched to the window trigger and you can see it's probably just you know got over that top one so I had to set that I had to use window triggering to go outside of the normal operational window to capture I really what is quite a small variation and most circuits would tolerate that quite well you know if you've got an additional 3.3 volt local regulator when you're really just powering some 5-volt logic it's going to tolerate this sort of ripple generally no problem whatsoever but there is something subtle on the particular process inside processing circuitry inside this Rygel that is causing that thing to reset so if I wasn't absolutely confident that that regulator was dropping out then I would you know I could have thought okay well I've checked that and that's not the problem you know not an issue at all it must be something else you go I go away you chase red herrings until the cows come home but no we nailed it because I finally got down to a point where I could trigger off something that was causing this to drop out now a couple of people have already been a little bit confused by this issue so I'll make it very clear this five volt regulator that we're looking at here it has absolutely nothing to do at all with powering these outputs or what loads you put on the output whether you have a load or not it's a completely isolated circuit with its own tap on the transformer and its sole purpose is to power all the digital circuitry in the front here so that applications processor the LCD some of the i/o stuff here at the back that's all there is to it you can load down this all the outputs to the 495 watts and the dissipation on that heatsink is going to remain exactly the same although as I mentioned before when you do load down the outputs the firmware knows that or it's measuring the temperature of the main heatsink but I can't see any thermometer on thermistor on there at all measuring the temperature that it does increase the flow rate of the fan and as I said the increased flow rate of the fan could actually have the effect of actually cooling down a little bit that 5 volt heating but it's got nothing to do at all with loading the outputs now the question is how much load does this thing actually take how much does all this digital stuff in here take well let's have a look at it let's measure our power let's switch it on shall we let me put my probe in here and we'll switch that on here we go and it's powering up it's powering up it's not much at all uh-huh silly me I figured out what's going on well not silly me silly Rygel this connector here which you know and obviously carrying all of that five volts over the board they've actually got the colors are the wires back to front I assumed who silly me that the positive wire would actually be positive it's not it's actually negative negative relative to that regulator so now I've pulled out the black why are there because the reason why we're only measuring like you know 20 30 minutes before is because this ribbon cable here was taken that returned current so here we go now we should be able to get it measure the actual current here we go if we break into the positive wire here which is actually black bingo there we go we're getting three hundreds booting now probably can't see the screen there we go it's just booted up there we go and now after it's booted we're getting you know let's call it say 700 milliamps something like that let's switch the outputs on doesn't make any difference of course but yeah you know it's jumping around as you'd expect but let's call that 700 milliamps and the input voltage about eleven point eight volts which will of course vary with the line voltage because it is bridge rectified with just some fielder caps coming from the transformer so you know that could vary but let's just call it 12 volts now if we get in 12 volts in and 5 volts out well we've got a delta or a voltage drop across this regulator of seven volts and because it's a linear regulator it's got a drop seven the power is going to be seven volts times the current flowing through it which is who of course the output current which is 0.7 amps seven volts times 0.7 hours 4.9 let's round it to five watts this thing is dissipating five freaking watts now anyone with any electronics design experience knows that no way in hell you're going to use a heatsink of that size for five Watts even if you've got a fairly high airflow you know going through your design and good thermal managers just ridiculous five watts did no one even stick their bloody finger on this or even think about it I'm flipping the finger that's for sure let's just go to a representative heatsink you haven't wasn't able to find the exact one but this is going to be fairly close it's an avid thermal a one to-220 free free standard well it's actually a PCB mount one it's got a PCB mount tab this one looks to have two PCB mount tabs so I'm not sure if there's any heatsink on the copper at the bottom side of this board I haven't taken it out but anyway we're going to be easily able to get some ballpark stuff here and if we have a look at it let's go in here it's taut we're talking twenty four point four degrees C thermal resistance there per what but this is what we're interested in down here let's have a look at the graph shall we and what we've got here is we don't need to really worry about the thermal resistance what we're talking about here because this is a power dissipated we know we're dissipating five watts look it's four heatsink of this size it's off the graph already that should be ringing alarm bells right and this is the mounting surface our temperature rise from 0 to 100 degrees C above ambient and that's the key of course and what were we measuring on this thing well with no air flow you know we were getting basically you know 100 and you know well over 100 hundred 30 or something like that and this is the ballpark that were operating up here at 5 watts with this sized heatsink I mean it's going to be very similar we're just talking ballpark calculations here we're looking at a hundred degrees C rise above ambient and that's exactly what we're getting it's ridiculous I could go into there you know draw the thermal graph of all the things in there and the heat sink compound and the bloody you know everything the junction the case and all that and imagine what the if this is what the heat sink temperature is I imagine what the junctions at well we actually don't even need to guess what the junctions are getting at because let's look at the data sheet for the lm317 just take a typical one from Fairchild for example let's go down here let's get the thermal characteristics but here it is there we go thermal resistance Junction to case we already know that let's assume that there's no loss between the case and the heatsink right let's assume that that's just fine well the case there it is five degrees C per what we're trying to dissipate five watts in this thing the junction is going to be 25 degrees C at least above that's above the already measured and quantified temperature on that heat sink which even in the case with the proper air flow and everything else is over a hundred degrees C you got to be kidding me ah facepalm hang on double facepalm well enough of that Fiasco I may as well do a little bit of poking around of what I originally do before I bloody discovered this ridiculous issue anyway I was just going to have a look at the little just a little bit at the output circuitry here and see exactly what we got and it is very easy you've probably already guessed it but I've drawn a simple dive card here and this is basically what we got on the output we've got the thousand mic output filter cap right on the front panel terminals as you saw I haven't shown the sense wise going back out there obviously going back to a sensor but yeah there's nothing in this at all really there's no output relay switching of course or any sort of electronic well there's electronic switching but it's done by the series pass MOSFET that we've got in here but we basically got a big Schottky diode in there there we go we've got some Schottky diode protection across the output as you're typically fine then we've got a couple of moles here going to mains earth here and remember this output is not means worth reference it's actually floating so these things are going to chazzy earth ground and then we've got to get another couple of mods on our high side current sense resistor here so there's our high side currents resistor we saw that close up the tracers go off there - there's our high side current app they've rubbed the number off that the bastards but and of course the output of that will be tied into the constant current circuitry which then controls the gate so that'll all be analog loop stuff going on in there and then we've got a bleeder resistor across here that's that one down in there there we go it's a fairly large one and then we've got another bleeder resistor across the filter cap says our main three filter caps up here there they all are boom boom boom they're twenty two hundred Mike each 63 volt and we'll measure some we'll get the scope out mostly measures some things on in and have a look at the gate waveform there but and basically the input here as I mentioned in the teardown that they actually well I mentioned that there were tracks in there but didn't mention that they're actually switching the secondary taps on the transformer here the transformer taps coming in here and there are two tracks in there they're our two track drivers down in there as we saw in the teardown but that's what they're using instead of your more traditional relays to select the secondary taps because because this is a linear supply you want to you need some sort of tap on there imagine you're delivering only 3.3 volts out of here and you're getting you know 40 volts out of your transformer that's a lot of power to dissipate in your linear regulator like that in your part series pass transistor there it's called so really you want to choose that they've got a couple of selections on Mayer's taps there so that's basically I'm what they're doing is nothing in here at all there's no output relay switching to swim when you press that on/off button on the front all it's doing is just effectively grounding that gate and pull in the output down to zero so it's not actually isolating the output at all it's just switching off the output series pass transistor so we'll just see where those are voltage taps actually occur you can see I've got the full were 31 volts output voltage I'm measuring the voltage and also looking at the waveform so this is identical to that you can see the voltage down there 54 we're getting yeah basically our 54 volts out of those filter caps with 31 volts output voltage now a low or my voltage in 1 volt our output voltage in one volt increments and see where the tap drops and of course in there you'd normally hear a real a click and a regular power supply but this one uses our track switching and bingo there it goes it looks like it when it goes from 2021 from 22 once we get down to 21 it drops down to 35 volts so we're getting you know a 15 volt Delta there so it's got to dissipate 15 volts best-case there 14 13 12 11 10 there we go we're going to drop down to 10 volts and then we drop down to our next tap which is 7 about 17 and a half volt so there you go and they will be the 2 taps we won't find any more than that because we've only got our two Triax on this thing's and as we saw in the teardown there is only a single to-220 MOSFET in there and you know a lot of people are probably going to argue well you know they're 90 watts is this particular channel this is one of the 30 volt our 3 amp channels so 90 watts are dissipation is this enough heat sink and air flow I don't know you could try and get the datasheet and stuff like that but what we should probably do is measure the temperature on that heat sink at a full 90 watt load and see what it gets to yeah you know some designs would actually parallel up the MOSFETs they're just too so you're not actually stressing just the one individual MOSFET spreading the power against a couple of them but why go well now I'm doubting their design decisions after that ridiculous lm317 fiasco but anyway they've assumed they've done it right their vibe determined that well no we can get away with a single MOSFET on there so anyway let's let load it down and see if we can measure temperature on this thing somehow but yeah there's no real easy way for me to stick my thermocouple on that and get a really good connection I'm afraid and just as we saw in the reviewer can't actually deliver the full 90 watts on that channel but I was able to just you know shuts the voltage down I was able to get a you know 85 can do Li a couple of Watts more than that but let's just that'll do 85 watts output so I'm drawing 85 watts I've got the lead kinda sorta close so we're getting an extra airflow over the top instead of through here but it's going to be you know it's going to be reasonably close and I am probing the main filter cap as well and as you can see there we've got 10 volts per division 10 20 30 40 so you know forty six volts minimum the plenty of margin in there for the ripple and some initial probing of the heatsink there it's at least two forty five degrees so you know as I said it is quite difficult I'm not going to get in there and actually but you know I can't really probe the MOSFET itself it's getting up there but I wouldn't call that particularly hot and of course you probably can't hear it but the fan actually has turned on louder and we have been able to hit 70 there so there you go I don't know you know I'm not going to go into the full thermal calculations with that you know that's not too bad that's what you'd expect really you know that's a ballpark of what you'd expect for a full load on this thing so you know not a problem and I just took that back out and with no air flow that yes jumped up to 84 it's a bit hot but you know there's no air flow so it expected okay I have a quick probe of the gate of the series past MOSFET there and as I said that actually control switches the output off and on and we'll actually see that here what I've got dual channels here channel one which is the yellow waveform is the gate voltage there we're at 5 volts per division on both channels so 5 10 15 20 25 30 30 volts output by the way set 30 volts there the output is actually switched on at the moment and as you can see if you switch it off up run it there we go we can see our output and channel 2 is our output wave form so I output voltage so we're smack on 30 volts there and if we just single-shot capture that being there we go we can see the rise there there's got some little something happening down there I'm not sure what but anyway that's it still there's no overshoot on that output at all it's ramping up the output ramps up to 30 volts and of course directly controlled from the gate voltage so if you actually bring that up you'll find that those two waveforms are virtually perfectly superimpose there because there's you know there's nothing else I'll switch in the output it's just that actual I gate series pass transistor via the gate there and I'm afraid I'm no real closer to that turn on glitch when you actually power the thing on so I don't know that was the aim of this when I started this yesterday was maybe to get down into that detail but who cares now I mean we've got that showstopper which is the bloody lm317 and here's another thing the lm317 they're using check out the ridiculously thin tab on that and it's just well is really pissant thin tab ones absolutely horrible and you can see the difference on these two devices over here whereas this has got like a normal thickness tab on it that's what a you know a proper well-designed tio2 e20 these ones little pissant thin tabs on them look hopeless so that lm317 I mean you've just got less thermal mass right there and for the record that looks to be an ST brand 301 lm317 t I check the datasheet for the s T brand lm317 and just like all the others you know maximum Junction operating temperature is you know around about that 125 degrees C mark and we've found weather you know proven by measurement and based on the datasheet values the Junction to case and that sort of stuff that they're operating above that they're operating above the recommended Junction temperature it's just a complete fail right there so there's some I wasn't expecting and when I started probing this thing yesterday I noticed an issue where we reset I tracked it down to a bloody overhead in piss-poor designed five volt regulator for the main logic can you believe it bloody ridiculous I'm pissed off this is a huge serious design oversight no one in their right mind would deliberately design a little lm317 with that piece ant heatsink to run at five bloody watts and think that you can get away with it right next to the output filter caps these power supplies I got to fail in the field no doubt about it probably even see the resetting feature I had our in a couple of years timers caps are going to dry out it's just absolutely shocking how this thing even made it past the bloody designer of you meeting let alone into production let alone in people's hands how many months has this thing been on sale for now and well yeah okay nobody's found it okay maybe a few units I don't know it might have experienced issues that that we haven't heard of but this is a huge very serious design oversight and rygar need to explain what they'll happen here and I don't think they can explain this away how like you know apart front we missed it or we swept it under somebody found it and they'll told to no it's not a problem shut up go back to your bench you know and it's ridiculous a hundred plus degrees C on a on a you know on a regulator at it just quiescent static current driving the thing as measured by multiple people through the case with the proper airflow yeah they'll probably do like a firmware upgrade oh we can fix that firmware upgrade make the fan run move all the time ah man unbelievable and it's probably going to be some people who say well what's the problem right nobody's had not look at just reset again oh bloody hell people are gonna have a few people say I'd you know why God might even say I had no problems in the field no return units that's beside the point the point is is that this is has no design margin in it whatsoever you mount this thing in a bloody rack right sits in a rakia ambient goes up by 20 or 30 degrees Celsius is going to be screwed those regulators are going to shut down there you know who knows what regulator what the thermal cutout is in that particular type of regulator and you got variations in your Junction to case I don't even think no you know put some heat sink in there is going to be variations in that it's going to be variations in the airflow on the fan the huge variations going to be variations in the life of those caps and all sorts of stuff it's just yeah it's just bad engineering it's not going to work needs to be fixed anyway my god please explain because this power supply it now gets a huge thumbs down until this problem is fixed I think it needs to be fixed before you sell any more of these bloody things unbelievable what can you do well you know it's tempting to sort of move that 5 volt reg over to you know an existing heatsink on here but then you break your isolation stuff between your isolated 5 arch powers of digital and your output so that's not really going to work Mike you've got there's a couple of mounting holes in there they probably could solve it by manufacturing some sort of custom heatsink or something which goes in there and yeah that'll probably be a fix maybe you know that well it probably would be a half reasonable fix if they upped the size of that heatsink by you know 4 or 5 times or something like that but definitely involves something custom other than that it probably have to relay out the board or something like that it's clearly unacceptable ah anyway let's see I definitely let Roy go know about this and let's see if we hear back from him cuz this is just complete not a catch you next time you you

Channel: EEVblog

Views: 96,940

Rating: 4.9202061 out of 5

Keywords: rigol, dp832, dp832a, power supply, rigol power supply, rigol dp832, troubleshooting, repair, schematic, oscilloscope, oscilloscope triggering, bad design, fault, faulty, themral management, thermal design, heatsink, heatsink design, temperature measurement, temperature

Id: y-KkPLWZJko

Length: 44min 3sec (2643 seconds)

Published: Thu Aug 29 2013