yes. aswell as a usb scope on a Laptop thats Not plugged in or a usb scope that had isolation like my owon vds 1022i. (i use it with a laptop so i am 2x Safe@@karlramberg
The first time that happened to me, I was working on an All-American 5. I went to connect the scope's ground lead to the chassis and boy howdy! I got my heart rate back down to nominal and reset the circuit breaker. Then I turned the radio's plug around and tried it again. Success! Looking back on the things I did when I was a kid, it's a wonder I survived. Oh, later on I invested in an isolation transformer.
9:45 when measuring on a switch mode power supply, using differential probes is a good idea. Also: in the schematic there is evidence that gnd and chassis is different. Look at the guard (dashed line) in the transformer. It is connected to chassis which is indicated as a slanted earth connection. See the difference with the earth symbol on the chip and the transformer terminal with the three parallel lines.
If you want to use a scope for testing things like this (power supplies/voltages >30V where high currents are possible), you could consider getting a scope with isolated grounds (or a cheap battery operated scope). Then you don't need to worry about the ground reference of your scope lead connecting things that might not be grounded the same way a normal scope lead is to the scope's earth ground at the wall.
Exactly! It's possible to cause issues if you have one channel on a floating ground and another channel on chassis ground. If the scope doesn't have isolated grounds between channels you can bring the floating ground to chassis ground and cause all sorts of weird issues.
I had two of the 575 Super Eagles on a Sun 3/180 back around 1995. Had to spin up one drive at a time, THEN turn on the CPU else it would trip a 20 AMP breaker.
Glad to have been of help figuring he schematic out :) Ah, those ground shenanigans... Nice to see that music gets you going and helps you work too! Pretty much like me. Bunnies and Moomins - that's double cuteness :)
Mate, SMPS engineer here. either get a couple of decent differential probes, or use an isolation transformer to float your scope, and only ever use one channel. i have a Tektronix TPS2024 battery powered scope that four independantly isolated channels, its absolutely fabulous.
Three days isn't bad going! Myself and another engineer, once spent about two weeks of 'back-to-back' shifts trying to fix a spare power supply for an offshore seismic recording system. I used to walk down to the lab at the start of my shift, praying the other guy had fixed it - and he did the same at the start of his shift. In the end, we conceded defeat, and just posted the damn thing back to France to have the manufacturer fix it. Power supplies are a nightmare!
Depending on how it's constructed and your gear you can isolate and test rails under load separately but it isn't always going to get you all the way there even then. It depends a lot on how well designed the PSU is, and a lot of them are unfortunately generational designs where one guy started with one design and successive generations of guys modified and tacked on more and more sheisse onto it cutting corners and making questionable decisions on among others to tap off reference voltages and feeds for stages. Doesn't take many such iterations to end up with a pile of dung where the behavior is totally unpredictable and troubleshooting it properly becomes impractical because you'd have to deconstruct it layer by layer. Not entirely unlike debugging code where there have been generations of coders who don't care about the structure of the code, only about tacking on some new feature. The client doesn't look at the code and *how* something is achieved, they just test some feature and sign off on it when their special case works. I've spent hundreds of hours rewriting bad code that made zero difference for the actual function of it from a user perspective, but what it did do was make the code actually do what it was intended to do instead of pretending to. Great fun to try to explain to management etc. "This function returns a -1 in this case, not because it correctly checks X, Y and Z against the corresponding values in the remote system, but because the unit mfg number matches the range predefined in table T1 in database DB5 which is not maintained by anyone and not part of what this function should do, as defined in documentation from supplier.. bla bla bla...."
Thanks to Usagi and his discord / github Team for helping out Adrian! His Series about a Server that is slightly older than me was great, and would not have been possible without you!
Hah always throws me off when David shaves clean… I’m like “did he post a video from 5 years ago?” Lol Anyway, looking forward to the next videos on this machine. PDPs are awesome
Dave, remember, you should always be plugging the Power Supply into an isolation transformer and floating the oscilloscope with a ground lift plug. You are "lucky" you didn't fry your scope!
Running the equipment under test through an isolation transformer is relatively safe. You should never disconnect the ground on your scope, though. That is both potentially dangerous, and can lead to increased noise and degraded sensitivity of the scope. They're designed to be grounded for a reason. Also, running something like this behemoth of a machine through an isolation transformer is not necessarily easy, because most of them are not rated to handle high power loads (or if you find one that is, it's usually really quite expensive). Still, a smaller isolation transformer might work if you're just trying to get the supply working and have most of the rest of the computer disconnected... But yes, you always need to be very careful of what you're hooking your scope's ground up to. Getting it wrong is pretty unlikely to actually damage the scope (might fry the probe, though), but it can often damage whatever you're testing, just like this great example. Having a differential probe is a good option, but if you don't have one, with pretty much any dual-channel (or quad-channel) scope you should also be able to just connect the signal probe for one of the channels to the "ground" point you want to use as a reference, and then have the scope display the difference between the two channels (subtraction function), to effectively make a poor-man's differential probe, too.
@@oliverer3 I bought one with 1kVA from eBay for 360€ (approx. $390). In europe you have the "advantage" that you only need half of the current for the same power thus the transformer is smaller but on the other side you have to deal with twice the voltage. And with 230V that is about 325V if rectified you should use a transformer for your own safety.
Thank you for not being in the camp of "replace all the caps". This so annoys me when I see this. Awesome video. I'm sure you'll figure out the missing 12v soon.
16:37 - stacking those capacitors like that blocks the safety venting on at least one of them ;) Depending on the manufacturer, they could be intended to vent from the lead side (there will be what looks like a possible opening when viewed from the lead side), or from the top (perforated thinner metal lines would be on the lid, under any plastic), or even both ways. Capacitors that are unable to safely vent can become projectiles.
The “Air flow sensor” board probably uses a diode (PN junction) as a cheap thermistor to measure temperature, to detect if the fan is running or not. If the fan is running, then the airflow will cool the diode resulting in a change in the Vf (Forward voltage drop); if not, then it will measure a different Vf, and it probably uses a comparator (LM339 or similar) to create a digital “fan is running” signal. This way, the PSU can go into thermal shutdown if the fan isn’t running, but it’s supposed to be. That’s my guess. 😊
That is actually a valid approach. However... The temperature coefficient of a normal PN junction is -2mV/K . So a temperature rise of 30K (30°C 86F) will only give you -60mV voltage change. The application you describe would rather use a PTC or NTC which gives you a more significant change of parameter over temperature. And making it part of a voltage devider you can easily tailor the voltage to temperature curve to your needs. If you are really clever, you can run current through the PTC. It will stabilize on a resistance (current) depending on how good the heat transfer with the surrounding air is. In static air it will heat up more compared to being in air flow... The Mass Airflow Sensor in some cars works on this principle....
When in doubt drag out the diff probe (or rig up differential mode with two probes), or at least make a DMM voltage measurement between earth and whatever circuit node you plan to use as a ground. Also disregard anyone telling you to float/isolate the scope, one of those dangerous half understanding problems, you always float the DUT if anything.
That wasn't what I was getting at @@DirkEibach I know the voltages inside any mains powered scope can kill but I wanted to understand why not use an isolation transformer to power the scope in this situation.
@@MrCHrisfj Every chassis grounded part of the scope will be at whatever voltage you have on your signal ground, which is surely not 0 V WRT earth (otherwise you would just not need isolation/diff probes etc., right?). Think of metal chassis (if the scope has one) and other channels exposed BNC ground. That is not something you want, and that is why the whole thing is grounded in the first place. Big no-no for me if you are not really really really sure of what are you doing.
@@MrCHrisfj There is no difference and both are more or less equally hazardous unless you're talking the high dollar Tek, etc. handheld scopes with plastic shell BNCs and special reinforced insulation probes that are actually intended for that sort of service. The point is a floated scope can be at any potential when they're covered in metal that's supposed to be at 0. It adds a lot of unpredictability that can and does kill to just save a buck or some setup time, it's a dumb way to die.
USE AN ISOLATION TRANSFORMER. When servicing any line operated switch mode power supply always use an isolation transformer on the AC power input. You are absolutely correct about the internal power supply grounds,. Schematic ground is not always chassis or output voltage ground and is often associated the AC line connected circuits. Remember your scope has a three wire grounded power cord. That means the scope probe ground is at chassis ground. Some people are tempted to use a ground lift adapter on the scope or the device, DON'T. That's asking for an early death. Remember that ground and neutral are bonded together in your house main breaker panel You have a knack for acquiring things so finding a 500va or better yet 2kva isolation transformer should be rather easy. Also note that most Variacs do not provide isolation. They are just variable autotransformers.
@@andydelle4509 I would say if you really can't avoid poking around with your probes in a powered device you have to use an isolation transformer for your own safety. In addition to that you then have to use differential probes for every single signal you look at. That protects you from at least directly grounding the isolated side of the transformer. In addition it protects the machine from shorts even with multiple probes attached. But usually also with differential probes it is necessary that the probed device and the probes have some kind of common ground to keep the common mode voltage of the probes within the permitted range. So connecting the secondary side of the transformer with trustworthy and certified high value resistors to the same ground as the scope or bnc connectors of the probes, maybe the only option then.
There seems to be a lot of confusion over reforming caps. Short version: If your capacitor reads OVER it's rated capacitance it's a good candidate for reforming. If it reads UNDER, it's dead and should be replaced. Long version: Electrolytic caps work by forming a thin oxide layer on the aluminium electrode. For the same capacitance a low voltage cap will have a smaller electrode area and thinner oxide layer. A high voltage cap has a larger electrode area with a thicker oxide layer. With long periods of disuse the oxide layer can slowly dissolve, becoming thinner. This increases the capacitance but reduces the working voltage. If you then apply full rated voltage the thin oxide layer breaks down and the cap dies. When reforming you apply just enough voltage and current to gently rebuild that oxide layer and restore the cap's working voltage. After reforming, the capacitance should be about the rated value. If it's below I'd strongly recommend replacing it. If it's a bit above and can withstand rated voltage for a day or so it's probably safe. A cap that reads under can sometimes be restored to a certain extent by running it at rated voltage for a while. With use what remains of the electrolyte will be redistributed and capacitance may come up a bit but you're only putting off the inevitable demise. Realistically not many caps are good candidates for reforming and IMHO it's safer to just replace them. Additionally reforming should be done BEFORE powering up the device. If the oxide layer has deteriorated enough to be a candidate for reforming, simply turning on the power could be enough to kill the cap.
The 11/44 did not use static RAM. The separate voltage rails for memory were for a ‘standby’ mode where power to the memory would be kept on while power to the rest of the machine was turned off, allowing the memory to keep its contents without keeping the entire machine running (and using 120A of 5V)
@memadmax69 Depending a bit on how you define it, some modern platforms sorta do too, just not like this thing does. Being a bit pedant here, but the VRM section of your motherboard has a section responsible for a voltage rail which is only really supplying the RAM and the MMU logic which is partially isolated but in most modern platforms physically inside the CPU chip package. I'm not up to speed on the newest chip generations because I have no reason to, but take for example the Core2 architecture, it has a setup like that. What that means is that there is a PSU and a whole set of logic which is partially isolated from the rest of the *stuff* in electrical terms, and can and usually does run at a different voltage from everything around it. This also means that you *can* damage a memory channel on some modern chips without affecting anything else, which means you can still even run the computer, just half or a quarter of your RAM slots will not work, and you won't be able to run dual or quad channel. Pointless trivia, but there you go.
That fan sound bring back memories --- Worked at a manufacturing plant back in 1987 where we had 6 PDP-11/40s -- the fan noise was impressive -- and constant -- not to mentions the associated A/C required to keep the computer room cool. The PDP-11s ran Baseway under RSX-11S to implement a Shop Floor Gateway. Collected digital data from shop floor for use in identifying and counting parts produced among other data by a upper level application.
Excellent! Our old friend, the PDP-11 makes a reappearance and ends up in a much better place than it was at the start of the episode :) - But, of course, it took "the scenic route" to get there: A diode that was inadvertently helped to stop "dioding"; A good dose of heavy metal, followed by electrolytics being taken to reform school ... and more! - Thanks for making these restoration diaries enjoyable stories, as well as being instructive and informative 🍻
I think all of us have done this one time or another, my first time was on a C64 in early 80's, second time testing an rc servo motor in haste and connected gnd clip and scope tip to the leads of the motor and smoke! quikcly realizing an H-bridge send power to the motor
You learned a valuable lesson about scopes - that ground lead is connected to the scope's power AC ground, which is probably connected to other things in the circuit under test. You really need to be careful about where you connect scope ground. Some people learn it the hard way connecting the scope to a "hot chassis" set and left the magic smoke out of everything including the scope!
When you used that replacement diode with a red lead instead of a black lead, I winced and hoped the polarity was correct when you installed it, but I guess it was.
Don't beat yourself up, I've sent 400V DC through a scope probe ground before. Turns out the negative rail went below ground on that motor controller, and the positive rail was at ground potential.
Thanks for all the great videos! I'm sure you've thought of this, but just in case you haven't or it hasn't occurred to anyone else reading the comments, having a little portable battery powered scope is really great! I have a Zoyi ZT-703S and it's an incredible value! I think around $90 USD? Sure, it's not going to be as good as a bench scope, and there are portable scopes with better features, but this thing I have... wow. It's quick and easy and works so well - well enough to poke around and see what's going on during a repair. It's got a built-in rechargeable battery, that I believe is easy to replace, but because of that you don't have to worry about having an accident like what you had in this video. It's not going have a ground that connected to earth ground when it's running on batteries, so you can't short out something to earth ground by accident. I believe in your situation having an isolation transformer and a bench scope might have helped avoid this, but just having a little battery powered scope has been a game charger. Nobody is sponsoring this comment I just have one and I didn't think it would have the positive impact it has. The convenience of having a battery powered scope just on my bench ready to roll means I'm way more likely to just grab it and use it. That's had huge benefits! I love it and wish I got one years ago!
@@MichaelThwaite Actually that brings up an important point! I just checked and my USB based Hantek oscilloscope and it's BNC connectors show continuity between them and USB ground as well as earth ground! So if you had your USB oscilloscope's ground connected to something like a live chassis vintage television that's plugged in then you'd be creating a connection between your computer's ground and the mains voltage! In this case in the video, the ground of the oscilloscope is connected to USB ground and earth ground via the computer. So when, in the video, he connects the oscilloscope ground to the floating ground, he would again recreate the same conditions as in the original video. The floating ground (or logic ground? virtual ground? whatever it's called in this case) may be at some voltage potential that's different from actual earth ground and connecting it to earth ground might cause power to flow in some way the circuit wasn't designed for. But if the oscilloscope is battery powered, and the device is *NOT* connected to USB or any charger of any kind, then there's no way for there to be an unexpected connection between the circuit being tested and earth ground. Although I still wouldn't work on a live chassis television without an isolation transformer. So don't use a USB oscilloscope on something where the ground connection may not be at the same ground as your computer! In short: The safest way to poke around a circuit is with a battery powered, and not plugged into a charger, oscilloscope. Legal notice: I am not qualified to do, say, or give advice to anyone for any reason.
@@chironbramberger Yikes! Yes, I think my USB scope maybe getting a rechargeable battery pack! In my case the scope I have is powered by USB and is connected to a 5V USB Type-A PSU which, is double insulated (no earth) so, I _don't_ think there will be a connection to mains earth in my case as, it's a 2-pin reversible connection (US 120V) so such a passthrough could be, well, fizzy in 50% of the insertion conditions. All food for thought that I'd never considered before.
Something to be very careful about is that you probably want to use differential probes when looking at this power supply. You can approximate a differential probe by using two channels on your scope and using the subtractive math operation to display the signals. I've done lots of work on complex power supplies and this is a critical part of my approach. Having a precision DC load could also help you. I have a feeling that all of this will culminate in getting the DC supply on the Bendix up and running. It's fun to watch as always.
Great video. Two comments: Firstly: TIc-Tic sound coming from switch mode power supply is almost always overcurrent protection. And please, get a new caps + decent component checker for leaks and ESR. Secondly: Ground loop with Oscilloscope, been there, done that. Can get nasty with high power systems. Using isolation transformers at AC is a good idea. Or Diff-probe ..
Worked on a few SEL minis (their 32/77's and 32/87's). They used separate power supplies for each voltage (+5, -5, +15 and -15) and yes, the main SELBus had several 100A 5VDC power supplies with some pretty massive cables. CPU was five boards alone, nothing but 7400 series chips IIRC. (oh, and the 'memory bucket', a separate card cage, had power supplies of its own)
When using an oscilloscope you should always galvanically isolate the scope with an isolation transfo and not connect earth/PE. This makes it so that you can't short-circuit muliple gnds together and use a differential probe
When measuring and testing big diodes, capacitors, and other silicon, it is very helpful to use an analogue multimeter, instead of a digital DMM. The Analogue meter provides more voltage and current for the large component, and will more correctly show leakage, and conductance in those devices.
That UNI-T meter is a nice asset to have. 🙂 I have both the -E and the newer E-Pro. The -E is on my work as a measuring device for the Trucks I work on, and the E-Pro is my daily goto for electronics since it is quite a bit more sensetive in the mA range, and I have reprogrammed them both to start on DC on all ranges by default, switched off the Auto-Off for Backlight, and also switched off the Auto-Off on the meter entirely. 🙂
You should either use a differential probe or if necessary a differential probe combined with an isolation transformer for the power supply. If you would plan to poke around with probes inside of the supply i would highly encourage you to use a transformer. But you decided (almost completely 🙂) to avoid that and i think most of the time that's a very good decision. And as a little precautionary hint, usually differential probes need some kind of ground connection to keep their common mode voltage in an acceptable range. Together with a floating device that can be a really difficult mess. In the end everything you do, should protect you from injury first and then, if possible your machines.
Tech tip - if your power supply can measure the leakage current of a cap - it's bad!! Your power supply only shows down to 10s of mA. Leakage current (once everything is charged) should be a lot less then that. I'm reading 100nA/rated volt. So at 55V, that's still way, way under 1mA, even allowing for aging.
Always use differential scope probes when working on unknown equipment. But can such a beast of a diode blow up by incorrect grounding with such thin wires? I would expect those thin wires to simply melt. Could it have been coincidence or the repeated power cycling that finally killed it?
9:28 maybe take a look at usb scopes with a laptop to float it or handheld ones, that cant happen on these. my cheap owon vds 1022i even has usb isolation so the usb ground isnt the scope ground
electrolytics WILL have a certain amount of remaining DC current even if good , i'd say 10ma for a high capacitance low voltage is good, but important thing is esr , especially on a switch mode psu, if it regulates by sensing dc off a cap, high esr can result in high dc pulses to the sense circuit, making output low or even shutting it off if it has overvoltage trip out protection, i had this issue with a philips crt tv in the mid 1990s .. reformed caps can seem ok but still have poor esr, its not so critical with linear/ac line frequency power supplies but switch mode circuits can be picky..
Since the module that is "broken" uses 35v and not 300v, it might be easier and safer to power it with a lab generator. Now I don't have the schematics, but it looks like you just need to recreate the 35v (probably it doesn't need to be that high, maybe 25-30v will do), Sync (60khz, 5v 1us pulse), +12v and "MEM DC ON (L)"
David, I have an ESR meter I'd like to send you to aid in evaluating the health of electrolytic capacitors, if you're interested. I'm personally not a believer in capacitor reforming as means to return them into service...Typically, the reason these types of capacitors fail over time is that the electrolyte gradually dries out, and unless it's rehydrated it won't operate as designed. Good luck repairing the third rail!
I have always thought of reforming electrolytic capacitors as just a temporary procedure, just to get the device into an operable state for further testing.
It's good to know that electrolyte capacitors reform, but not good to use that as a repair technique. When an aluminum electrolytic has been used at a voltage well below it's rating you know two things; that it's capacitance is going to measure high, and that it's going to be prone to higher leakage current. It's also going to have a shorter life expectancy, but you can't measure that.
This is why i personally don't reform capacitors. They are not that expensive to replace, even the big ones. And as shown, even after reforming it's still an expired component that could fail at any time. And since all caps in that power supply are of similar age.... The should be replaced. Proven by it's twin being a short. You could keep some of them if they test fine, if you had the equipment to test that properly, which you don't. This is a switching power supply, it's a job that is very hard on capacitors. Unlike many other applications. I doubt i will convince you and i don't need to, but this is my two cents on this subject.
Have you thought about feeding your own +12 into that pin? If you have a variable power supply you should see the circuitry trying to compensate without having to plug it in to the "scary voltage".
It`s an Interesting mashine! On this mashine had been eveloped the C language.So, you can write any software, even server application. But, unfortunately TLS libraries not avaliable for usual people.
Nah, isolation transformer is for the equipment being tested. And a variac provides zero isolation. It's just a variable autotransformer. An autotransformer doesn't have a secondary coil, it's just a variable tap on a primary.
Good going! You have an awful lot of patience. I would have given up about 45 minutes in. A few suggestions: (1). Consider buying a good electrolytic capacitor checker. HP made some good ones and they’re now at reasonable prices. (2). Get some power resistors or light bulbs that will put a typical load on the power supply. Many supplies don’t like having no load and might trip their over voltage alarm with little or no load.
Thanks! I do need a good, proper LCR meter. That's on my list of tools to convince Mrs. Usagi to let me buy, lol. If you look closely, everytime I power up the machine, I have a 12V automotive headlight bulb as my primary load. Looking at the schematics, not all the rails need a big load on them to come up properly, so the PSU is definitely operating with the proper load and as expected (with the exception of the +12V rail which has a regulation problem we'll dig into in the next episode).
@@UsagiElectricthe average LCR meter isn't that useful to test life, they don't test at high enough voltage. Some people claim that leakage is a resistive thing and thus can be checked at low voltage. However that doesn't account for material breakdown. That isn't linear and can be gone at low voltage and appear at high voltage. You would need a still very expensive HP or sencore device to measure that. I believe Shariar from the signal path showed one not so long ago. Miss usagi is not going to approve that expense. The simple LCR meters are fine for a quick test and value measurement.
@@UsagiElectric For something much cheaper but still decently capable get something like the "LCR-TC1" which is based on an open-source AVR transistor checker. It identifies many component types. For capacitors you get capacitance, ESR, and loss factor. It also calculates transistor type, gain, voltage drop, and leakage. The big downside of this $20 device is that it can get fried by charged capacitors. This is something I have unintentionally tested myself. Also you'll need to test out of circuit but this is true of almost any tester.
those big old electrolytic caps can become grenades.Especially in linear power supplies. Ask me how i know. I could not hear anything with my left ear for 3 days.
Ah yes, this is why I have an isolation transformer on my bench. It is incredibly easy to kill isolated electronics with a scope ground, and even if you don't kill anything, it will distort the signals.
I would have thought that if you shorted chassis negative to Earth ground, and the power supply was unhappy with that, there would have been an obvious "poof", lights flickering, and magic smoke being released. But, who knows... You're in the home stretch now. I was thinking the last problem would be a quick fix. Guess not. I guess we'll have to wait for the next episode.
I remember correctly that device only tests at low voltage and that doesn't account for dielectric breakdown at high voltage. Most of the caps Carlson tests are so far gone, any tester would fail them. They're just that old and crappy.
lmao, when I learned the 'an oscilloscope has to be isolated from mains when probing mains powered equipment' lesson, I didn't get the pleasure of just shorting out a diode; I had the probe blow up in my hand and the breaker trip a split second later. oops!
When I would use an o-scope for TV troubleshooting and I needed to connect the scope ground lead to the chassis I would use an AC power cord where I tore out the ground connection. This help prevent the scope ground lead from shorting out the chassis to earth ground. So try using a ground buster power cord for your o-scope so you are not causing problems like you had suggested you had in the power supply.
That will often protect the scope from damage, but compromises the shielding of the cabinet and doesn't improve personal safety. The lack of a grounded cabinet can both show noise that isn't present on the circuit being probed and induce scope oscillator noise into the unit under test.
Ah, the fun of breaking something and spending days getting back to where you were before. This is why I gave up on hardware and went with software... while I was still high school age, back in '79 or so! A few decades later, I finally have the patience to deal with big TTL-level stuff. But not these bonkers power supplies! (RF is right out.)
TBH, i'd actually swap out all these regulators with modern snt modules but that's just me. imho it ain't worth blowing up a rare machine by tacking together slowly failing PSUs. just my $0,02 on, i've seen switchmode PSU going bad in many ways. some just stop working, some start producing ripple and some start running up their output voltage. also some regulators like to get instable when the caps go bad. i had at least two instances of only the crowbar saving a one of a kind piece of equipment At least always test all the (easily reachable) caps, and do that first, if you want to keep them original where possible
On your bench, with black thick sharpie write: "Do not trust their lies" and have a drawing of a ground symbol above it. No, but seriously. Trust no ground to be the kind you probe. This mistake or error of judgement has probably taken down more units than any other. Most of the times, people don't even know what they've done.
I would have replaced ALL of the electrolytics as a first step. And used my battery-powered isolated (floating) Hantek 'scope. I am not "precious" about keeping failure points, merely for performance art.
I’m of this school also. In something this old, if you have any thought of having be a runner at all versus “I got it to boot - great, off to recycling” then they just need to be replaced. No nostalgia value.
❤ Hi, thanks for my special spot in the intro for quite some videos but feel free to put me back in the regular list and maybe put somebody else there ❤
@@UsagiElectric I can understand why, that PSU monopolized all your time! 😁 I'll be patiently waiting for the part 2, where you beat it into submission & get all the volts. 😶🌫️😶🌫️😶🌫️
I have some original manuals you may be interested in. Don't know how to contact you. I reached out to Vintage Geek Aaron to get yours but he has not gotten back to me yet. Docs are original documentation for Dec equipment I had, specifically VAX and RSTS for the PDP-11. Also docs on the pdp-11 6502 cross assembler that was used to assemble early roms for computers. If you can use these before I recycle let me know!
I swear every time you shave I suddenly get jumpscared. Completely different look. Not that either is bad, far from it, but my brain always takes a second to register the total visual 180 lmao.
I said "shorted cap" 5 minutes in. Also you're a dirty reformer. You bad boy, you. I know parts are expensive but just replace 'em. I do agree with your ethos of finding the problem rather than shotgunning everything though.
DEC engineer to another: “how many amps you need?”
Hmm, yes.
“Got it. All the amps.”
I doubt that strongly! Within DEC expect clear specifications for something like power usage and rail voltages
pumping amps on 5V is serious business, I had problems with mere 180W on 5V the other day.
DEC Corporate: "We need a very powerful supply."
DEC Engineers: "I've got this arc welder, we could use that..."
DEC Corporate: "Brilliant!"
"At DEC it's not a real computer unless the lights dim on power on".
Really 120A at 5v is unthinkable kudos to the DEC folks
'Things are breaking and repairing themselves at the same time' ... best description of debugging ever.
Heavy metal is the key ingredient to making progress, 100% can confirm.
I was so confused for a little while there, lol.
True machine learning? From the 70's!
Ah yes the classic unintentional short to earth through the scope, been there too many times…
Can you avoid that with a battery powered scope?
yes. aswell as a usb scope on a Laptop thats Not plugged in or a usb scope that had isolation like my owon vds 1022i. (i use it with a laptop so i am 2x Safe@@karlramberg
@@karlramberg yeah, or a battery powered circuit
The first time that happened to me, I was working on an All-American 5. I went to connect the scope's ground lead to the chassis and boy howdy! I got my heart rate back down to nominal and reset the circuit breaker. Then I turned the radio's plug around and tried it again. Success!
Looking back on the things I did when I was a kid, it's a wonder I survived. Oh, later on I invested in an isolation transformer.
@@karlrambergyes, or a isolated usb scope, or usb scope with Laptop on battery power.
9:45 when measuring on a switch mode power supply, using differential probes is a good idea. Also: in the schematic there is evidence that gnd and chassis is different. Look at the guard (dashed line) in the transformer. It is connected to chassis which is indicated as a slanted earth connection. See the difference with the earth symbol on the chip and the transformer terminal with the three parallel lines.
If you want to use a scope for testing things like this (power supplies/voltages >30V where high currents are possible), you could consider getting a scope with isolated grounds (or a cheap battery operated scope). Then you don't need to worry about the ground reference of your scope lead connecting things that might not be grounded the same way a normal scope lead is to the scope's earth ground at the wall.
Exactly! It's possible to cause issues if you have one channel on a floating ground and another channel on chassis ground. If the scope doesn't have isolated grounds between channels you can bring the floating ground to chassis ground and cause all sorts of weird issues.
Or a differential probe.
I had two of the 575 Super Eagles on a Sun 3/180 back around 1995. Had to spin up one drive at a time, THEN turn on the CPU else it would trip a 20 AMP breaker.
Glad to have been of help figuring he schematic out :)
Ah, those ground shenanigans...
Nice to see that music gets you going and helps you work too! Pretty much like me.
Bunnies and Moomins - that's double cuteness :)
Glad you were there to help out!
Mrs. Usagi says thanks for saying her pajamas are cute!
Jameco carries a large selection of big old smoothing capacitors. I just bought a 90,000 uf one for my Dynabyte DB8/2. 😅
Mate, SMPS engineer here. either get a couple of decent differential probes, or use an isolation transformer to float your scope, and only ever use one channel. i have a Tektronix TPS2024 battery powered scope that four independantly isolated channels, its absolutely fabulous.
another option is an owon vds 1022i usb scopethat has built in USB isolation.
Three days isn't bad going! Myself and another engineer, once spent about two weeks of 'back-to-back' shifts trying to fix a spare power supply for an offshore seismic recording system. I used to walk down to the lab at the start of my shift, praying the other guy had fixed it - and he did the same at the start of his shift. In the end, we conceded defeat, and just posted the damn thing back to France to have the manufacturer fix it. Power supplies are a nightmare!
Depending on how it's constructed and your gear you can isolate and test rails under load separately but it isn't always going to get you all the way there even then. It depends a lot on how well designed the PSU is, and a lot of them are unfortunately generational designs where one guy started with one design and successive generations of guys modified and tacked on more and more sheisse onto it cutting corners and making questionable decisions on among others to tap off reference voltages and feeds for stages. Doesn't take many such iterations to end up with a pile of dung where the behavior is totally unpredictable and troubleshooting it properly becomes impractical because you'd have to deconstruct it layer by layer. Not entirely unlike debugging code where there have been generations of coders who don't care about the structure of the code, only about tacking on some new feature. The client doesn't look at the code and *how* something is achieved, they just test some feature and sign off on it when their special case works.
I've spent hundreds of hours rewriting bad code that made zero difference for the actual function of it from a user perspective, but what it did do was make the code actually do what it was intended to do instead of pretending to. Great fun to try to explain to management etc. "This function returns a -1 in this case, not because it correctly checks X, Y and Z against the corresponding values in the remote system, but because the unit mfg number matches the range predefined in table T1 in database DB5 which is not maintained by anyone and not part of what this function should do, as defined in documentation from supplier.. bla bla bla...."
Took a PDP 11 apart in 1990 with some very smart people. I learned an incredible amount.
Thanks to Usagi and his discord / github Team for helping out Adrian! His Series about a Server that is slightly older than me was great, and would not have been possible without you!
Many hands make light work.
I was so happy to be able to watch the unfolding of the Plexus story in real time over on the Discord!
Plexus
Plexus
Plexus
Plexus
Plexus
Plexus
Hah always throws me off when David shaves clean… I’m like “did he post a video from 5 years ago?” Lol
Anyway, looking forward to the next videos on this machine. PDPs are awesome
Same - was looking at the video date lol
Dave, remember, you should always be plugging the Power Supply into an isolation transformer and floating the oscilloscope with a ground lift plug. You are "lucky" you didn't fry your scope!
Or the less dangerous option, using a differential oscilloscope probe.
@@oliverer3 Yes! If you need too much force, you ain't using the right tool.
Running the equipment under test through an isolation transformer is relatively safe. You should never disconnect the ground on your scope, though. That is both potentially dangerous, and can lead to increased noise and degraded sensitivity of the scope. They're designed to be grounded for a reason.
Also, running something like this behemoth of a machine through an isolation transformer is not necessarily easy, because most of them are not rated to handle high power loads (or if you find one that is, it's usually really quite expensive). Still, a smaller isolation transformer might work if you're just trying to get the supply working and have most of the rest of the computer disconnected...
But yes, you always need to be very careful of what you're hooking your scope's ground up to. Getting it wrong is pretty unlikely to actually damage the scope (might fry the probe, though), but it can often damage whatever you're testing, just like this great example. Having a differential probe is a good option, but if you don't have one, with pretty much any dual-channel (or quad-channel) scope you should also be able to just connect the signal probe for one of the channels to the "ground" point you want to use as a reference, and then have the scope display the difference between the two channels (subtraction function), to effectively make a poor-man's differential probe, too.
@@foogod4237 Wow, no joke on the price tag of hefty isolation transformers. Feels weird when the differential probe is the cheaper option.
@@oliverer3 I bought one with 1kVA from eBay for 360€ (approx. $390). In europe you have the "advantage" that you only need half of the current for the same power thus the transformer is smaller but on the other side you have to deal with twice the voltage. And with 230V that is about 325V if rectified you should use a transformer for your own safety.
Thank you for not being in the camp of "replace all the caps". This so annoys me when I see this.
Awesome video. I'm sure you'll figure out the missing 12v soon.
16:37 - stacking those capacitors like that blocks the safety venting on at least one of them ;)
Depending on the manufacturer, they could be intended to vent from the lead side (there will be what looks like a possible opening when viewed from the lead side), or from the top (perforated thinner metal lines would be on the lid, under any plastic), or even both ways.
Capacitors that are unable to safely vent can become projectiles.
or worse-grenades
@@jwhite5008 - that’s possible, but usually the can just flies off intact
The “Air flow sensor” board probably uses a diode (PN junction) as a cheap thermistor to measure temperature, to detect if the fan is running or not. If the fan is running, then the airflow will cool the diode resulting in a change in the Vf (Forward voltage drop); if not, then it will measure a different Vf, and it probably uses a comparator (LM339 or similar) to create a digital “fan is running” signal. This way, the PSU can go into thermal shutdown if the fan isn’t running, but it’s supposed to be. That’s my guess. 😊
sounds reasonable and practical..... DEC would never do something this simple
That is actually a valid approach. However... The temperature coefficient of a normal PN junction is -2mV/K . So a temperature rise of 30K (30°C 86F) will only give you -60mV voltage change.
The application you describe would rather use a PTC or NTC which gives you a more significant change of parameter over temperature. And making it part of a voltage devider you can easily tailor the voltage to temperature curve to your needs.
If you are really clever, you can run current through the PTC. It will stabilize on a resistance (current) depending on how good the heat transfer with the surrounding air is. In static air it will heat up more compared to being in air flow...
The Mass Airflow Sensor in some cars works on this principle....
@@randomsteve4288 I saw,what looked like,two PTC/NTC thermistors near one end of the board.
When in doubt drag out the diff probe (or rig up differential mode with two probes), or at least make a DMM voltage measurement between earth and whatever circuit node you plan to use as a ground.
Also disregard anyone telling you to float/isolate the scope, one of those dangerous half understanding problems, you always float the DUT if anything.
@Broken_Yugo So how is using an isolation transformer on the scope different from using a battery powered one?
@@MrCHrisfj The voltages inside a battery powered scope can't kill you.
That wasn't what I was getting at @@DirkEibach I know the voltages inside any mains powered scope can kill but I wanted to understand why not use an isolation transformer to power the scope in this situation.
@@MrCHrisfj Every chassis grounded part of the scope will be at whatever voltage you have on your signal ground, which is surely not 0 V WRT earth (otherwise you would just not need isolation/diff probes etc., right?). Think of metal chassis (if the scope has one) and other channels exposed BNC ground. That is not something you want, and that is why the whole thing is grounded in the first place. Big no-no for me if you are not really really really sure of what are you doing.
@@MrCHrisfj There is no difference and both are more or less equally hazardous unless you're talking the high dollar Tek, etc. handheld scopes with plastic shell BNCs and special reinforced insulation probes that are actually intended for that sort of service. The point is a floated scope can be at any potential when they're covered in metal that's supposed to be at 0. It adds a lot of unpredictability that can and does kill to just save a buck or some setup time, it's a dumb way to die.
USE AN ISOLATION TRANSFORMER. When servicing any line operated switch mode power supply always use an isolation transformer on the AC power input. You are absolutely correct about the internal power supply grounds,. Schematic ground is not always chassis or output voltage ground and is often associated the AC line connected circuits. Remember your scope has a three wire grounded power cord. That means the scope probe ground is at chassis ground. Some people are tempted to use a ground lift adapter on the scope or the device, DON'T. That's asking for an early death. Remember that ground and neutral are bonded together in your house main breaker panel
You have a knack for acquiring things so finding a 500va or better yet 2kva isolation transformer should be rather easy. Also note that most Variacs do not provide isolation. They are just variable autotransformers.
differential probes with mains powered stuff.
@@v12alpine That still doesn't remove the safety hazard. It's easy to forget and connect some other piece of test equipment besides a scope.
@@andydelle4509 I would say if you really can't avoid poking around with your probes in a powered device you have to use an isolation transformer for your own safety. In addition to that you then have to use differential probes for every single signal you look at. That protects you from at least directly grounding the isolated side of the transformer. In addition it protects the machine from shorts even with multiple probes attached.
But usually also with differential probes it is necessary that the probed device and the probes have some kind of common ground to keep the common mode voltage of the probes within the permitted range. So connecting the secondary side of the transformer with trustworthy and certified high value resistors to the same ground as the scope or bnc connectors of the probes, maybe the only option then.
There seems to be a lot of confusion over reforming caps.
Short version: If your capacitor reads OVER it's rated capacitance it's a good candidate for reforming. If it reads UNDER, it's dead and should be replaced.
Long version: Electrolytic caps work by forming a thin oxide layer on the aluminium electrode. For the same capacitance a low voltage cap will have a smaller electrode area and thinner oxide layer. A high voltage cap has a larger electrode area with a thicker oxide layer. With long periods of disuse the oxide layer can slowly dissolve, becoming thinner. This increases the capacitance but reduces the working voltage. If you then apply full rated voltage the thin oxide layer breaks down and the cap dies.
When reforming you apply just enough voltage and current to gently rebuild that oxide layer and restore the cap's working voltage.
After reforming, the capacitance should be about the rated value. If it's below I'd strongly recommend replacing it. If it's a bit above and can withstand rated voltage for a day or so it's probably safe.
A cap that reads under can sometimes be restored to a certain extent by running it at rated voltage for a while. With use what remains of the electrolyte will be redistributed and capacitance may come up a bit but you're only putting off the inevitable demise.
Realistically not many caps are good candidates for reforming and IMHO it's safer to just replace them. Additionally reforming should be done BEFORE powering up the device. If the oxide layer has deteriorated enough to be a candidate for reforming, simply turning on the power could be enough to kill the cap.
You know ur machine is beast mode when it has to have a power supply just for its RAM.
^.^
Yeah. Static RAM was/is generally very power hungry.
The 11/44 did not use static RAM. The separate voltage rails for memory were for a ‘standby’ mode where power to the memory would be kept on while power to the rest of the machine was turned off, allowing the memory to keep its contents without keeping the entire machine running (and using 120A of 5V)
@@awesomepuppy404 was this to keep it consistent with earlier pdp-11s that used core memory and therefore retained its contents when powered off?
@@awesomepuppy404 Oh ok, thanks for clarifying. I always wondered where M$ got it's idea for Windows warm standby. 😆
@memadmax69 Depending a bit on how you define it, some modern platforms sorta do too, just not like this thing does. Being a bit pedant here, but the VRM section of your motherboard has a section responsible for a voltage rail which is only really supplying the RAM and the MMU logic which is partially isolated but in most modern platforms physically inside the CPU chip package. I'm not up to speed on the newest chip generations because I have no reason to, but take for example the Core2 architecture, it has a setup like that. What that means is that there is a PSU and a whole set of logic which is partially isolated from the rest of the *stuff* in electrical terms, and can and usually does run at a different voltage from everything around it. This also means that you *can* damage a memory channel on some modern chips without affecting anything else, which means you can still even run the computer, just half or a quarter of your RAM slots will not work, and you won't be able to run dual or quad channel. Pointless trivia, but there you go.
That fan sound bring back memories --- Worked at a manufacturing plant back in 1987 where we had 6 PDP-11/40s -- the fan noise was impressive -- and constant -- not to mentions the associated A/C required to keep the computer room cool.
The PDP-11s ran Baseway under RSX-11S to implement a Shop Floor Gateway. Collected digital data from shop floor for use in identifying and counting parts produced among other data by a upper level application.
Sorry... I can't hear what you're saying over the noise of these fans!!!!
That much hasn't changed. We have by necessity put our A/C in the same room as my husband's modern-day gaming PC.
Excellent! Our old friend, the PDP-11 makes a reappearance and ends up in a much better place than it was at the start of the episode :)
- But, of course, it took "the scenic route" to get there: A diode that was inadvertently helped to stop "dioding"; A good dose of heavy metal, followed by electrolytics being taken to reform school ... and more!
- Thanks for making these restoration diaries enjoyable stories, as well as being instructive and informative 🍻
Impressive progress! When I'm troubleshooting, I tend to listen to artists like Tomita Isao, Nakamura Teruo, The Seatbelts, stuff like that.
I think all of us have done this one time or another, my first time was on a C64 in early 80's, second time testing an rc servo motor in haste and connected gnd clip and scope tip to the leads of the motor and smoke! quikcly realizing an H-bridge send power to the motor
Why are you so enthusiastic and please don't stop
Also check out 8 Foot Sativa if you've never heard them
You learned a valuable lesson about scopes - that ground lead is connected to the scope's power AC ground, which is probably connected to other things in the circuit under test. You really need to be careful about where you connect scope ground. Some people learn it the hard way connecting the scope to a "hot chassis" set and left the magic smoke out of everything including the scope!
I took out a 20A breaker powering my whole lab doing this...thankfully scope and DUT survived...learned my lesson real quick.
When you used that replacement diode with a red lead instead of a black lead, I winced and hoped the polarity was correct when you installed it, but I guess it was.
Don't beat yourself up, I've sent 400V DC through a scope probe ground before.
Turns out the negative rail went below ground on that motor controller, and the positive rail was at ground potential.
Thanks for all the great videos! I'm sure you've thought of this, but just in case you haven't or it hasn't occurred to anyone else reading the comments, having a little portable battery powered scope is really great! I have a Zoyi ZT-703S and it's an incredible value! I think around $90 USD? Sure, it's not going to be as good as a bench scope, and there are portable scopes with better features, but this thing I have... wow. It's quick and easy and works so well - well enough to poke around and see what's going on during a repair. It's got a built-in rechargeable battery, that I believe is easy to replace, but because of that you don't have to worry about having an accident like what you had in this video. It's not going have a ground that connected to earth ground when it's running on batteries, so you can't short out something to earth ground by accident. I believe in your situation having an isolation transformer and a bench scope might have helped avoid this, but just having a little battery powered scope has been a game charger. Nobody is sponsoring this comment I just have one and I didn't think it would have the positive impact it has.
The convenience of having a battery powered scope just on my bench ready to roll means I'm way more likely to just grab it and use it. That's had huge benefits! I love it and wish I got one years ago!
Or one powered by a USB adapter like the FNIRSI 1014D
@@MichaelThwaite Actually that brings up an important point! I just checked and my USB based Hantek oscilloscope and it's BNC connectors show continuity between them and USB ground as well as earth ground! So if you had your USB oscilloscope's ground connected to something like a live chassis vintage television that's plugged in then you'd be creating a connection between your computer's ground and the mains voltage! In this case in the video, the ground of the oscilloscope is connected to USB ground and earth ground via the computer. So when, in the video, he connects the oscilloscope ground to the floating ground, he would again recreate the same conditions as in the original video. The floating ground (or logic ground? virtual ground? whatever it's called in this case) may be at some voltage potential that's different from actual earth ground and connecting it to earth ground might cause power to flow in some way the circuit wasn't designed for. But if the oscilloscope is battery powered, and the device is *NOT* connected to USB or any charger of any kind, then there's no way for there to be an unexpected connection between the circuit being tested and earth ground. Although I still wouldn't work on a live chassis television without an isolation transformer. So don't use a USB oscilloscope on something where the ground connection may not be at the same ground as your computer!
In short: The safest way to poke around a circuit is with a battery powered, and not plugged into a charger, oscilloscope.
Legal notice: I am not qualified to do, say, or give advice to anyone for any reason.
@@chironbramberger Yikes! Yes, I think my USB scope maybe getting a rechargeable battery pack! In my case the scope I have is powered by USB and is connected to a 5V USB Type-A PSU which, is double insulated (no earth) so, I _don't_ think there will be a connection to mains earth in my case as, it's a 2-pin reversible connection (US 120V) so such a passthrough could be, well, fizzy in 50% of the insertion conditions. All food for thought that I'd never considered before.
Something to be very careful about is that you probably want to use differential probes when looking at this power supply. You can approximate a differential probe by using two channels on your scope and using the subtractive math operation to display the signals. I've done lots of work on complex power supplies and this is a critical part of my approach. Having a precision DC load could also help you. I have a feeling that all of this will culminate in getting the DC supply on the Bendix up and running. It's fun to watch as always.
Great video.
Two comments:
Firstly: TIc-Tic sound coming from switch mode power supply is almost always overcurrent protection. And please, get a new caps + decent component checker for leaks and ESR.
Secondly: Ground loop with Oscilloscope, been there, done that. Can get nasty with high power systems. Using isolation transformers at AC is a good idea. Or Diff-probe ..
Worked on a few SEL minis (their 32/77's and 32/87's). They used separate power supplies for each voltage (+5, -5, +15 and -15) and yes, the main SELBus had several 100A 5VDC power supplies with some pretty massive cables. CPU was five boards alone, nothing but 7400 series chips IIRC. (oh, and the 'memory bucket', a separate card cage, had power supplies of its own)
Nice touch that the rivet lines up with the O on memory.
When using an oscilloscope you should always galvanically isolate the scope with an isolation transfo and not connect earth/PE. This makes it so that you can't short-circuit muliple gnds together and use a differential probe
When measuring and testing big diodes, capacitors, and other silicon, it is very helpful to use an analogue multimeter, instead of a digital DMM.
The Analogue meter provides more voltage and current for the large component, and will more correctly show leakage, and conductance in those devices.
That UNI-T meter is a nice asset to have. 🙂 I have both the -E and the newer E-Pro. The -E is on my work as a measuring device for the Trucks I work on, and the E-Pro is my daily goto for electronics since it is quite a bit more sensetive in the mA range, and I have reprogrammed them both to start on DC on all ranges by default, switched off the Auto-Off for Backlight, and also switched off the Auto-Off on the meter entirely. 🙂
The last LSI PDP is probably a good device to bring online.
You should either use a differential probe or if necessary a differential probe combined with an isolation transformer for the power supply.
If you would plan to poke around with probes inside of the supply i would highly encourage you to use a transformer. But you decided (almost completely 🙂) to avoid that and i think most of the time that's a very good decision.
And as a little precautionary hint, usually differential probes need some kind of ground connection to keep their common mode voltage in an acceptable range. Together with a floating device that can be a really difficult mess.
In the end everything you do, should protect you from injury first and then, if possible your machines.
Tech tip - if your power supply can measure the leakage current of a cap - it's bad!! Your power supply only shows down to 10s of mA. Leakage current (once everything is charged) should be a lot less then that. I'm reading 100nA/rated volt. So at 55V, that's still way, way under 1mA, even allowing for aging.
Always use differential scope probes when working on unknown equipment.
But can such a beast of a diode blow up by incorrect grounding with such thin wires? I would expect those thin wires to simply melt. Could it have been coincidence or the repeated power cycling that finally killed it?
Use an isolation transformer or differential probes to avoid blowing rectifiers.
Progress is progress, even if it's one step back, then two steps forward. Can't wait to see this machine up and running!
Man! You really bend my brain explaining all those gimmicks in this PSU! I am so curious, how you will completely fix it and get the bits flow :)
9:28 maybe take a look at usb scopes with a laptop to float it or handheld ones, that cant happen on these. my cheap owon vds 1022i even has usb isolation so the usb ground isnt the scope ground
electrolytics WILL have a certain amount of remaining DC current even if good , i'd say 10ma for a high capacitance low voltage is good, but important thing is esr , especially on a switch mode psu, if it regulates by sensing dc off a cap, high esr can result in high dc pulses to the sense circuit, making output low or even shutting it off if it has overvoltage trip out protection, i had this issue with a philips crt tv in the mid 1990s .. reformed caps can seem ok but still have poor esr, its not so critical with linear/ac line frequency power supplies but switch mode circuits can be picky..
You need an ESR meter. That mallory capacitor could have excessive series resistance, causing the 35Vdc to oscillate instead of filtering/smoothing.
I always enjoy watching your successes! Great job on the troubleshooting!
Thanks so much!
Although this one didn't feel too much like a success, we almost got there, but not quite. Still, progress is progress!
That’s what I enjoy, you never give up….😂
Since the module that is "broken" uses 35v and not 300v, it might be easier and safer to power it with a lab generator.
Now I don't have the schematics, but it looks like you just need to recreate the 35v (probably it doesn't need to be that high, maybe 25-30v will do), Sync (60khz, 5v 1us pulse), +12v and "MEM DC ON (L)"
David, I have an ESR meter I'd like to send you to aid in evaluating the health of electrolytic capacitors, if you're interested. I'm personally not a believer in capacitor reforming as means to return them into service...Typically, the reason these types of capacitors fail over time is that the electrolyte gradually dries out, and unless it's rehydrated it won't operate as designed.
Good luck repairing the third rail!
I have always thought of reforming electrolytic capacitors as just a temporary procedure, just to get the device into an operable state for further testing.
It's good to know that electrolyte capacitors reform, but not good to use that as a repair technique.
When an aluminum electrolytic has been used at a voltage well below it's rating you know two things; that it's capacitance is going to measure high, and that it's going to be prone to higher leakage current. It's also going to have a shorter life expectancy, but you can't measure that.
This is why i personally don't reform capacitors. They are not that expensive to replace, even the big ones. And as shown, even after reforming it's still an expired component that could fail at any time. And since all caps in that power supply are of similar age.... The should be replaced. Proven by it's twin being a short.
You could keep some of them if they test fine, if you had the equipment to test that properly, which you don't.
This is a switching power supply, it's a job that is very hard on capacitors. Unlike many other applications.
I doubt i will convince you and i don't need to, but this is my two cents on this subject.
Can't wait to see an Eagle come up :)
You and me both!
They do sound incredible spinning up. I no longer have my 4, so won't mind hearing that sound again.
Have you thought about feeding your own +12 into that pin? If you have a variable power supply you should see the circuitry trying to compensate without having to plug it in to the "scary voltage".
Some dangerous voltages and currents in this thing, wow!
It`s an Interesting mashine! On this mashine had been eveloped the C language.So, you can write any software, even server application. But, unfortunately TLS libraries not avaliable for usual people.
Came for the clean shave, stayed for PDP-11/44 PSU
You should get yourself a nice LCR meter for a better way to measure those capacitors. You'd probably enjoy a nice HP 4261A or 4262A
You need a 1:1 isolation transformer for you scope in this kind of stuff. Think you can use a variac for that too.
Nah, isolation transformer is for the equipment being tested. And a variac provides zero isolation. It's just a variable autotransformer. An autotransformer doesn't have a secondary coil, it's just a variable tap on a primary.
It's always plugging it in that gets me! No shame there.
Good going! You have an awful lot of patience. I would have given up about 45 minutes in. A few suggestions:
(1). Consider buying a good electrolytic capacitor checker. HP made some good ones and they’re now at reasonable prices.
(2). Get some power resistors or light bulbs that will put a typical load on the power supply. Many supplies don’t like having no load and might trip their over voltage alarm with little or no load.
Thanks!
I do need a good, proper LCR meter. That's on my list of tools to convince Mrs. Usagi to let me buy, lol. If you look closely, everytime I power up the machine, I have a 12V automotive headlight bulb as my primary load. Looking at the schematics, not all the rails need a big load on them to come up properly, so the PSU is definitely operating with the proper load and as expected (with the exception of the +12V rail which has a regulation problem we'll dig into in the next episode).
@@UsagiElectricthe average LCR meter isn't that useful to test life, they don't test at high enough voltage. Some people claim that leakage is a resistive thing and thus can be checked at low voltage. However that doesn't account for material breakdown. That isn't linear and can be gone at low voltage and appear at high voltage. You would need a still very expensive HP or sencore device to measure that. I believe Shariar from the signal path showed one not so long ago.
Miss usagi is not going to approve that expense. The simple LCR meters are fine for a quick test and value measurement.
@@UsagiElectric get a capacitor wizard. Measures ESR in-circuit. ESR is what you should be looking at especially with switching supplies.
@@UsagiElectric Maybe you should add an electronic load to that list as driving with only one headlight is prohibited by law. 😀
@@UsagiElectric For something much cheaper but still decently capable get something like the "LCR-TC1" which is based on an open-source AVR transistor checker. It identifies many component types. For capacitors you get capacitance, ESR, and loss factor. It also calculates transistor type, gain, voltage drop, and leakage.
The big downside of this $20 device is that it can get fried by charged capacitors. This is something I have unintentionally tested myself. Also you'll need to test out of circuit but this is true of almost any tester.
its not always good to use a bulb as a load as they have far lower resistance when could, the current surge could've popped the diode...
those big old electrolytic caps can become grenades.Especially in linear power supplies. Ask me how i know. I could not hear anything with my left ear for 3 days.
14:30
Hey! I got the same scope. I feel like a noob when I use the thing though. lol
That is power. And I thought the old IMSAIs were beasts. You could weld with them.
Ah yes, this is why I have an isolation transformer on my bench. It is incredibly easy to kill isolated electronics with a scope ground, and even if you don't kill anything, it will distort the signals.
I would have thought that if you shorted chassis negative to Earth ground, and the power supply was unhappy with that, there would have been an obvious "poof", lights flickering, and magic smoke being released. But, who knows...
You're in the home stretch now. I was thinking the last problem would be a quick fix. Guess not. I guess we'll have to wait for the next episode.
Loved the post-credit part
Maybe you need to build Mr. Carlson lab super capacitor tester?
I remember correctly that device only tests at low voltage and that doesn't account for dielectric breakdown at high voltage.
Most of the caps Carlson tests are so far gone, any tester would fail them. They're just that old and crappy.
"Spinning rust"... and yes, it fits.
lmao, when I learned the 'an oscilloscope has to be isolated from mains when probing mains powered equipment' lesson, I didn't get the pleasure of just shorting out a diode; I had the probe blow up in my hand and the breaker trip a split second later. oops!
Your age goes up and down by about 20 years depending on your beard. Its honestly impressive
That fan sound brings back great memories of ‘my’ 11/44! That’s how it should sound!
8:27 That is me at work right now...
When I would use an o-scope for TV troubleshooting and I needed to connect the scope ground lead to the chassis I would use an AC power cord where I tore out the ground connection. This help prevent the scope ground lead from shorting out the chassis to earth ground. So try using a ground buster power cord for your o-scope so you are not causing problems like you had suggested you had in the power supply.
That will often protect the scope from damage, but compromises the shielding of the cabinet and doesn't improve personal safety. The lack of a grounded cabinet can both show noise that isn't present on the circuit being probed and induce scope oscillator noise into the unit under test.
Maybe crazy but with the big diodes bolted to the heatsink, the black (old) vs red (new) cable may indicate reverse polarity?
Interesting that as crazy as a power supply providing 5V @ 120A sounds, it's really only 600W. Exactly the same as 120V @ 5A.
Good job! Treats for you and the wabbits!
Ah, the fun of breaking something and spending days getting back to where you were before. This is why I gave up on hardware and went with software... while I was still high school age, back in '79 or so! A few decades later, I finally have the patience to deal with big TTL-level stuff. But not these bonkers power supplies! (RF is right out.)
So interesting. It comes from a time where mechanical tech meets digital tech. When it was still heavily dependent on the mechanical.
Good work! When in doubt, replace the caps ;)
TBH, i'd actually swap out all these regulators with modern snt modules but that's just me. imho it ain't worth blowing up a rare machine by tacking together slowly failing PSUs. just my $0,02 on, i've seen switchmode PSU going bad in many ways. some just stop working, some start producing ripple and some start running up their output voltage. also some regulators like to get instable when the caps go bad. i had at least two instances of only the crowbar saving a one of a kind piece of equipment
At least always test all the (easily reachable) caps, and do that first, if you want to keep them original where possible
On your bench, with black thick sharpie write: "Do not trust their lies" and have a drawing of a ground symbol above it. No, but seriously. Trust no ground to be the kind you probe. This mistake or error of judgement has probably taken down more units than any other. Most of the times, people don't even know what they've done.
The Babyface Keeps Gettin ME every Damned tim3! LOL
Recommend "Beast in Black" and "Nightwish" if you like heavy metal and anime. And i am pretty sure you like both.
Three steps back, five steps forward. 👍
Oh an LSI. I have a few of those. Yeah right that's a ticket!
That's not a power supply, it's an arc welder.
5V 120 A - oh wow.
USB C power delivery put to shame 🤣
Yeah PDP time!!!
7:55 "power suppies"??
I would have replaced ALL of the electrolytics as a first step.
And used my battery-powered isolated (floating) Hantek 'scope.
I am not "precious" about keeping failure points, merely for performance art.
I’m of this school also. In something this old, if you have any thought of having be a runner at all versus “I got it to boot - great, off to recycling” then they just need to be replaced. No nostalgia value.
Grzeczna Trusia, chce JEŚ 😎
❤ Hi, thanks for my special spot in the intro for quite some videos but feel free to put me back in the regular list and maybe put somebody else there ❤
I love these old iron videos (PSU, disks, etc), so interesting! Thank you!
BTW: If you shave the beard, you need to get a shorter haircut! :D
Thanks!
And a haircut is on the list of things to do, I just didn't have time to drag myself to the barber just yet.
@@UsagiElectric I can understand why, that PSU monopolized all your time! 😁 I'll be patiently waiting for the part 2, where you beat it into submission & get all the volts. 😶🌫️😶🌫️😶🌫️
I have some original manuals you may be interested in. Don't know how to contact you. I reached out to Vintage Geek Aaron to get yours but he has not gotten back to me yet. Docs are original documentation for Dec equipment I had, specifically VAX and RSTS for the PDP-11. Also docs on the pdp-11 6502 cross assembler that was used to assemble early roms for computers. If you can use these before I recycle let me know!
EEVblog #279 - How NOT To Blow Up Your Oscilloscope
It's worth watching.
I swear every time you shave I suddenly get jumpscared. Completely different look. Not that either is bad, far from it, but my brain always takes a second to register the total visual 180 lmao.
Somebody get this man a "fanspin!" T-shirt!
I said "shorted cap" 5 minutes in. Also you're a dirty reformer. You bad boy, you. I know parts are expensive but just replace 'em. I do agree with your ethos of finding the problem rather than shotgunning everything though.
-10 years on the shave ;)