Fixing a PC that won't turn on (Bonus magic smoke!)

[@pvc988]

5.150V What a fitting voltage for that machine.

[@zoomosis]

I was wondering if anyone else noticed that! :D

[@Kundalini12]

Maybe that's why it's known as the 5150

[@gyrgrls]

In California, 5150 has another meaning, according to the WIC (Welfare and Institutions Code). It means, literally, a danger! LOL

[@sonicunleashedfan124]

lmao

[@bdhale34]

5150 is also the name of Eddie Van Halen's studio and the model of an amazing tube amp by Peavey

[@50shadesofbeige88]

Classic ADB. its crazy to think I've been watching for 4 years!

[@rymixxx]

Oh man, you just seem like the nicest person. Keep doing what you do. I thoroughly enjoy going on this retro tech adventure with you.

[@crowbarviking3890]

My dad who was an Engineer in the 80s used to say "If you are looking for a short, apply direct voltage to the line and see where it burns" Now considering he had a basic infinite supply of new parts and pcbs that approach was good. Now that the parts are old and rare.. well... I always watch Louis and when he looks for shorts he also applies voltage (lower though) and adds a lot of alcohol.. so he can see where it dissipates. Or uses a thermal camera.

[@jaykay18]

The way I always learned was "Red to red, you're dead. Black to black, no flak." Excellent technique for diagnosing. Sometimes you just have to play your hunch, which worked out quite nicely. Thank you for the work that you do, keeping machines like this, like the ones I (we) grew up with running. In 13 more years we'll all be so old we'll have trouble reading that 80-column text, but that machine, at 50 years old, will still do everything it originally could, and is still the basis for computers today. IBM really had it together back in 81.

[@KatTrapable]

IBM is still around today somehow. Thanks Adrian for showing this. I don't think I had one growing up since they were expensive :3

[@jaykay18]

I know all about IBM and their exit from the PC market, and their presence today. My father worked for IBM for over 30 years.

[@thecaptain2281]

The way I learned it, " Black next to Black is a friend to Jack, red next yellow will kill a fellow. "

[@mushroomsamba82]

red touches black, friend of jack. red touches yellow, kills a fellow.

[@andy16666]

IBM is today and really always was a service company. Their PC business and their server business were intended as platforms for broader service offerings. The same is true today. IBM still offers mainframes and POWER servers for big data and big compute applications. But they are still only there because they form a core part of solutions that they couldn't offer otherwise. PCs and x86 servers are commodities now, there's really no need for IBM to make these themselves anymore,

[@robertoneill1979]

I used to work at In-Circuit Test debugging IBM AT motherboards. For tracing shorts we had a gadget we called a "toneohm"; basically a very sensitive ohmmeter which emitted a tone, the pitch of which would get higher as you closed in on the shorting component. Really helped with tracking down those pesky shorts.

[@indylmc]

I can't believe you figured that out. Back in the day (40 years ago) when I worked on electronics I never had any luck at component troubleshooting. Replacing the board was the way I went. Good for you by the way, that is not by any means, easy to figure out.

[@andrewobrien3973]

Funny to think my grandfather was part of the design team behind that power supply. Glad to see its still holding up :)

[@Ice_Karma]

I believe the original IBM PC floppy format was actually 160 kB, single-sided, 40 tracks, 8 sectors per track, until someone, I can't remember who-maybe Tim Paterson?-was looking at the floppy controller specs and the way IBM was configuring it, and realized it would be easy to squeeze a ninth sector in, increasing the single-sided capacity to 180 kB, and the double-sided capacity from 320 kB to 360 kB. (For what it's worth, it was actually easy to squeeze a _tenth_ sector in, even, leading to capacities ranging from, e.g. on a 3½" DD disk, nominal 720 kB formatted, 800 kB with 80 tracks to as much 843 kB with 85, at the risk of compatibility issues.)

[@michaelhawthorne8696]

Tants are notorious.....Great when they work but a bastard when they go bad.... they are a very stable cap but have a bad rep over time in use. Good find Adrian

[@ObiWanBillKenobi]

This is a textbook example video of how to track down and bullseye the fault of a vintage computer not turning on. Thank you! I had the exact same symptom and solution to a vintage computer last year.

[@edward_grabczewski]

Thanks for taking the time to explain your troubleshooting logic. Most people gloss over that and leave viewers confused! I have just bought a UK 5150 (I live there!) but it hasn't been turned on in 30 years, so I'm wondering how to test the computer.

[@FloppydriveMaestro]

Capacitor smoke stinks. Its a good job these old pcs have weak power supplies. It often saves them from certain death.

[@OfflineSetup]

You stink

[@neilaldenarmstrong9806]

Tantalum smoke smells a lot worse than electrolytic. :)

[@georgechambers3197]

And to think 30 years ago I desecrated one of those by putting in a 486 motherboard. It still had the black power supply and worked great. I worked at BPA there in Portland and they had a plethora of them that went to scrap so I salvaged a couple for use at work. Great stuff Adrian, I hadn't seen the guts of a 5150 for a very long time!

[@TheCondoInRedondo]

George, what you claim is fascinating. Truly! That said... I'm really struggling to see how this was physically possible. How exactly did you overcome the card slot spacing incompatibility? The slots of the 5150 chassis were spaced one inch on center. The aftermarket 486 motherboard's card sockets were spaced 3/4-inch center-to-center. The openings in the back of the 5150 chassis would not have lined up with the card sockets of the motherboard. And how did you close the cover of the 5150 chassis once populated with 5170 height cards of 1990? Again, I'm struggling to visualize how this was done. PC (5150) and XT (5160) chassis were not tall enough for the AT (5170) era cards, like (for example) the ATI 8514/Ultra. Finally, do you recall which hard drive you used such that the original black 5150 power supply didn't shut down? I was never able to make that power supply work with 386 or 486 systems plus an inboard hard drive. The current draw was too much for the 12 volt line when the disk drive would first power up. Could never get past the BIOS when I tried this with AMI's first 386 motherboard (Serial Number 7). I even thought this was a flaw with the BIOS and I worked on this with AMI directly only to determine that the problem was the 5150's power supply, not AMI's motherboard. That's why the 5160 (XT) came with a beefier power supply... and retrofit drives for the 5150 came with an expansion chassis fitted with their own power supplies.

[@ultrametric9317]

I was looking back through your videos trying to determine the beginning of ADB - there are several criteria you can use - 1) First appearance of the ADB logo 2) First shout of "IT FREAKIN' WORKS!" 3) First kick-ass fix. I chose 3. That's this video :) Good job!

[@bluerizlagirl]

Tantalum capacitors _used to be_ preferred in certain applications because they offered more capacitance in the same amount of space as regular electrolytic capacitors, and the leakage current was lower for the same capacitance. But they have always had a bad reputation among people who replace capacitors for a living. Meanwhile, capacitor technology has not stood still. Modern electrolytic capacitors are better (even less leakage, and more µF/mm³) than 35-year-old tantalum ones were when they were new. Just fit some 105 degree Panasonic electrolytic capacitors and it'll probably outlast you. For working on multi-layer boards with plated-through holes, you probably will need some sort of vacuum desoldering apparatus, whether this is a fancy temperature-controled electronic device with a heated hollow tip or just one of the old-fashioned spring-loaded plunger ones (slow because they are only good for one joint, then you have to reset the spring and eject the old solder).

[@misterhat5823]

I'd use a newer ceramic. Lower ESR than an electrolytic and closer to the original. Also nearly infinite lifespan.

[@PileOfEmptyTapes]

@@misterhat5823 Infinite lifespan - you wish. Through-hole ones tend to be very reliable, yes. Surface-mount ceramics, by contrast, are notoriously prone to developing stress fractures (especially when mishandled) and eventually shorting out as silver migrates through those. You could say SMD and RoHS have ruined what used to be a reliable technology. Thankfully you can get automotive grade ceramics designed to prevent these issues, bigger but also a lot more dependable. Did I mention that the dielectric in high-capacitance ceramics is so crummy that they can lose over 50% of capacitance at rated voltage? They're also piezoelectric and can make quite the racket when subjected to AC at times.

[@danmackintosh6325]

In this case I'd imagine it would be a through hole cap that was fitted. I learned a little here today, not used to old computer equipment (more CRT TV and radio) and never really saw tantalums as a major issue before now. Certainly surprised to see one fail short, I guess the memes are right after all and it *will* be the caps,,,

[@bluerizlagirl]

@@danmackintosh6325 Capacitors are very often to blame for problems with electronic circuits, but it's a serious mistake to try a "scattergun" replacement. If you test after replacing every one or two capacitors, then at least any new fault you may have introduced by mis-wiring a replacement capacitor will show up sooner.

[@misterhat5823]

@@PileOfEmptyTapes I must disagree. SMD parts are perfectly reliable. You just have to know how to use (handle) them.

[@therealjammit]

Another idea for looking for shorted parts is to use your variable power supply on the mainboard in place of the computer power supply. Set the voltage on the power supply to the correct voltage and the current to minimum. Turn on the power and start cranking up the current until you get to something "reasonable" (500ma to 1A). Use the voltmeter and poke around suspected parts and look for the lowest voltage. By injecting a constant current you make the power supply and the voltmeter into a crude ohm meter capable of measuring really low ohms accurately. Since you're only looking for the lowest resistance in the circuit, absolute resistance measurements aren't relevant and the lowest measured voltage will suffice.

[@jaycee1980]

As soon as I saw those dipped tantalums i thought "One of those buggers". They were used mainly because they had low ESR compared to electrolytics of the time. Nowadays we have low ESR electrolytics and you can just substitute them. MLCC ceramics are very low ESR as well but are also highly voltage dependant, so its usually not a good idea to substitute them in

[@gymnasiast90]

Hadn't seen your videos before, but I like your narration: clear pronounciation and coherent, without uhms and ers. That's not something that every KZbinr has mastered.

[@conundrum112]

Those were the very first PC’s I worked on as an IT Analyst. Previous to that I used to be a Mainframe Computer Operator. 40 years working with computers, retired now, lol I still have an original with a cassette interface and matching monitor and keyboard. I had eight of them at one stage, keyboards, monitors, but alas all gone except one now.

[@axonis2306]

Can you do a video about it? I'd love to see the BASIC tape commands in action

[@mikededmon]

That was my first computer and I had a 30MB hard drive. I was Big Man on Campus at that time. All my friends had 2 floppy drives. but, that machine informed my career. I loved that computer.

[@samuellourenco1050]

Nice methodology! Finding a short is always difficult, but you've nailed it.

[@MrKillswitch88]

This is why always checking those tantalum caps when something of this vintage quits working as they love to short and sometimes pop explosively.

[@BEdmonson85]

Well, according to Louis Rossmann, you can't have too much wrong... you got fan spin! 🤣🤣🤣

[@RainerTrunk]

Oh yes that's true and it is not an Apple. What an lucky Man.

[@tompepper4789]

Great video. And we didn't have to hear P P bus over and over.

[@schnuuuu]

Your "overload" is my "open line". ;)

[@FalconFour]

Same exact issue on the 5150 I played with, and with 8-Bit Guy's Compaq Portable. Shorted tantalum cap dragging the whole rail down. I ended up finding the cap on mine by clipping a leg of every tantalum cap on the board until there was no more short, then soldered the good ones back together. Might be an easier way to find it ;)

[@PileOfEmptyTapes]

Tantalums often tended to be effectively undersized back then. They're low ESR but also very small and seemingly struggle to handle the inrush current at turn-on in the long run. A 16 V type on +12V? That's the minimum even for an Al 'lytic, for a tantalum it's borderline suicidal. I'd put in more like a 35 V.

[@glynwatkins9968]

Tantalum is used in a variety of alloys to add high strength, ductility and a high melting point. When drawn into a fine wire, it's used as a filament for evaporating metals such as aluminum. More than half of tantalum's use is for electrolytic capacitors and vacuum furnace parts.

[@only257]

Glyn Watkins ☺️

[@henryatkinson1479]

My 5150 revA (serial in the fist 1k units, ceramic chips, 1981 rev1 bios and all) did the same thing recently. Glad this vid popped up as it helped me troubleshoot it.

[@philiphoeffer7442]

OL stands for OPEN LINE. Not OVERLOAD :)

[@utfigyii5987]

Philip Hoeffer i thought that was over limit

[@taldmd]

or Open Loop

[@Jordyperson]

@Great Value Bleach breaking news man finds god in multi-meter

[@sadmac356]

I always wondered!

[@grizzli211]

Over Limit not open line :)

[@glynwatkins9968]

Tantalum is a chemical element with the symbol Ta and atomic number 73. Previously known as tantalium, it is named after Tantalus, a villain from Greek mythology. Tantalum is a rare, hard, blue-gray, lustrous transition metal that is highly corrosion-resistant.

[@ericromano8078]

I got scared when you had it narrowed down to that one capacitor, but I knew from the description that something was going to blow still. Nice fix and it really makes me want to find some old junk computers to dig into and fix up.

[@LarryDeSilva64]

Great diagnosing job Adrian. That little cap took 5 amps at 12 volts before it blew wow who knew!

[@misterhat5823]

5A at a fraction of a volt. Rewatch that part. There was only about a volt and most of it dropped by the test leads.

[@andy16666]

Yeah, it never hit 12 volts until the cap went open. If you work it out with ohm's law based on the resistance he measured (~0.05 ohms), you'd expect the cap to drop 0.25 volts at 5 amps. Obviously the leads drop some and the resistance changes as the cap heats up. But that's still enough to deliver 1.25 watts of heat to the cap, more than enough to cook it.

[@AndruRomin]

Way to throw a temper tantalum...🤔

[@gusbert]

Good diagnosis techniques, better than removing random components to see if the short was fixed. What a coincidence that the first cap you tested was the bad one. BTW, even at 5A the cap was only dissipating 1.5W assuming your measurement of 0.06 ohms was correct.

[@kevincozens6837]

Years ago when i was tracking down a short between PCB traces on a board that just came back from the board assembly company I found it very helpful to have three digits of precision on the ohms scale and not just two. A minimum of two decimal places really helps to narrow down the location of a short.

[@johnpossum556]

caps basically block DC which is why this didn't blow for a long time. Not much current passes through it. You showed that even though you don't have a thorough parts-level understanding of electronics you can still fix 'em. Good job!

[@300maze]

the cap was shorted so 5A went through it

[@johnpossum556]

@@MrHBSoftware sleep deprivation most likely. I'm extremely suspect of the DMM because I have seen them charge up a cap and give wrong readings and I thought that might be the case here. Tantalums are generally superb at holding a charge. Besides, do the math....It was 12 volts, right? Are you so naive as to believe your meter readings when they show 60 watts (5v * 12vdc) going through a tiny part that small? That's about the size of a TO-92 package and most of them are not even good for 1 watt.

[@ibazulic]

Capacitors typically burn out because of too high voltage, it's the electric field inside the capacitor that ultimately destroys it. That's why they usually have a rated voltage written down along with the capacity. For instance, 10 uF 120 V means it can handle up to 120 V across its electrodes. The smoke in the end is probably not due to it blowing up but because of the fact we basically threw 60 W of power on it (12 V * 5 amps) which the casing most likely cannot handle. It might also be due to overvoltage, I am not sure if you mentioned exactly to which rail this capacitor was hooked exactly. Another thing to note is that capacitors don't really work in DC environments and they function more or less as a switch. This is because they act as a short circuit at the moment of power on, but as they charge, the current through them fades. Once the capacitor is fully charged, there is no more current flowing through the capacitor and hence the circuit is broken. The time it takes for the capacitor to charge is proportional to the capacitance. Other than that, yet another great video! I find myself going through your library and just watching old stuff as i'm fascinated with the old computers and how advance they were for their age. Hope to see more in the future!

[@dank1837]

Adrian. Great video but could you one day do a TANDY 2000 VIDEO!! Please!!! They’re pretty much all broken and I need some direction. Like no one has done an in depth on one of the few computers to use the INTEL 80186 PROCESSOR! Thanks. Keep up the great work!

[@jjock3239]

I know this is a very old video, but there was still something to learn from it. I enjoyed and learned from your technique of checking for minute differences on resistance, to determine a possible bad capacitor.

[@TheLOD2010]

The orange cable is power good. It is a simple 5V signal that tells the system the status of the PSU. If power good goes to ground, than the system has to stop. These function is up to day as the grey cabel on ATX2.xx is power good too.

[@thedungeondelver]

Excellent job, Adrian. Amazing to think that the base configuration on these old beasts was barely more than that of the C64. In fact I've heard of some small businesses that used the C64 versus the IBM PC back in the day. But from a technical/EE stand point this is just pure voodoo. If I was confronted with that problem I wouldn't even have the faintest idea where to begin looking.

[@thulinp]

For bypass caps, you don't typically need all of them. Their effect is more like a team effort. With just one missing it should work just fine.

[@buenopower1]

This channel is so fun to watch man. Gawd. And I don't even know what's going on half the time lol

[@sleora]

Just 3 days ago I was trying to repair my 5150 that wouldn't start as well, I tested the caps around the power supply connectors on the motherboard on the -12V rail and I had indeed a shorted cap. Thing is, it's a 64k-256k model and caps on it don't have the + marking them, so the cap I replaced was soldered backwards and instantly and spectacularly blow up the moment I powered the machine on. Thankfully nothing was damaged (besides my heart) and with the next cap the machine powered on flawlessly. I just had to solder the cap under the motherboard because when removing the previous blown up cap I ripped the pad of the ground pin, so I had to use a nearby one.

[@televisionandcheese]

Wow 5 amps at 12v before it blew!! Really enjoyed this video, it was actually very entertaining and interesting to watch, even though I don't understand most of it. :)

[@SAerror1]

The lab power supply reduces the voltage significantly to limit the current through up to the max of 5A. If you look at the voltage readout it was less than 1V most of the time, so you were only putting a couple watts of actual power through it. It’s a bit confusing at first that this is what happens when you current limit a power supply but it makes sense when you think about ohms law and have a component that’s shorted out.

[@andy16666]

@@SAerror1 Yes! According to ohm's law, if the resistance stayed a constant 0.05 ohms with 5 amps across it, the voltage drop would be 0.25 volts. Obviously the leads would drop probably half a volt or so as well, and also the resistance probably varies with temperature as it heats up and nears failure. Still, that means at 5 amps, and 0.25 volts it was dissipating 1.25 watts. That's more than enough to cook something that size.

[@CODMarioWarfare]

It was holding steady around 7 or so watts for a minute there. Once it started failing the voltage went up a bit, peaking around 10 watts by my eye before basically failing open.

[@andy16666]

​@@CODMarioWarfare Exactly! I was calculating the minimum you'd expect to see, according to the theory. As the cap started to fail (at being a resistor), and as its temperature rose, its resistance increased quickly, thereby increasing the voltage drop and the dissipated power. Remember also that the leads probably have more resistance than the cap initially.

[@gyrgrls]

Methinks most of the wattage was dissipated in the leads and power supply. 5 amps at .012 volts ~ about 300 mw in the shorted cap. Unless the cap acted as a PTC thermistor... but that's a new test for the next dead MB, I guess.

[@jmpattillo]

What a great video. I love it when the problem isn't so obvious and it takes a while to figure out. Of course, it is almost always a capacitor, but I like that you so clearly identified which one.

[@TheCondoInRedondo]

The single-sided diskette drive could indeed hold 180Kbytes. However, that was with nine sectors per track, which was not the standard configuration for DOS. DOS defaulted to eight sectors per track, or 160Kbytes. Byte Magazine (and PC Magazine) published the command line tweak you could employ to get that extra sector. I think it was as simple as adding slash nine "/9" to the format command. But IBM was very conservative about performance and squeezing in the ninth sector was something IBM was not comfortable doing. In those days, the diskette drives were using stepper motors with stainless steel ribbons linking the stepper spindle to the head actuator and head alignments sometimes got out of calibration due to temperature changes. For example: A diskette recorded in the heat of summer might misread on a chilly day because of that stainless steel band changing dimensions... if you packed-in more than X number of tracks.. So IBM chose to err on the conservative side and limit the "official" capacity of the single-sided diskette drives to "160 Kbytes" and the number of tracks to a conservative value where reasonable temperature variations would not render a diskette unreadable. BTW - It wasn't long before you could purchase a "notch punch" to convert a single-sided diskette into a double-sided "flippy floppy" as we called it. The single-sided diskette drive would not be capable of reading both surfaces of the rotating disk simultaneously, of course. The diskettes were sold as "single-sided", but were (in reality) double-sided specimens that often had a flaw on one surface. So that disk was sold as a "single-sided" inside a sleeve with just one notch to prevent "flippy" usage. But if you manually cut a notch on the opposite side of the sleeve and flipped the diskette over, when you performed a format of the disk, any flaws in a sector on the unofficial surface would be detected during the format and the sector would be locked out by DOS, leaving the vast majority of sectors on the "flippy" side quite usable. BTW - I am the engineer who taught IBM's Boca engineers how to diagnose boot issues in the 5150 using a simple AM radio, tuned between stations. All 5150s were shipped to retail outlets in pieces. I would have to assemble each PC with the configuration of cards and options ordered by the customer. Then (of course) I would run diagnostics using the special diagnostics diskette before sliding the shell back on. Well, I accidentally had the office radio mis-tuned and discovered that the open PC would "sing" through the AM radio as soon as the PC was turned-on...a very repeatable sequence of chirps and pings and other sounds corresponding to which phase of the BIOS sequence was running. You could even tell how much RAM was installed (even without a monitor connected) because that portion of the boot sequence would take longer for a machine equipped with 128K, 192K or 256K. You could tell if the monochrome adapter was installed or the color one by the "detour" in the chirps as that phase of the BIOS would run. I am also the engineer who found (the hard way) the grounding flaw in the original color video adapter. I got shocked when assembling a PC for a customer and quickly identified the source of the shock as that card and reported it to Boca Raton because any substitute color adapter would deliver the identical shock. It wasn't just a glitch with one copy. They sent a "ready team" to North White Plains Computerland to meet me and see a demonstration of the problem. They verified the flaw and that triggered an EC plus a partial recall. That's when I also demonstrated the "singing" BIOS to them and showed them one motherboard that would fail before completing the BIOS phase. I illustrated how you could tell what was wrong by following the "song" until it stopped. That told you what component was being tested when the boot failed.

[@BoHolbo]

NERD ALERT!!! My favorite type of nerd of the most useful variety.. 😁

[@FireballXL55]

Hi Adrian yes I use a bench PSU and a thermal camera, and set the voltage lower than the rail I am testing. I set the current limit at 1A and usually this is enough to see the shorted component, it may have worked in your system because the faulty tant was dissipating 0.4W at that current and in such a small package I am sure you would see the difference.

[@Daniel009007]

its looks beautiful , Absolute beauty ... and best part is there is no dust on boards, So clean

[@RunnerBalz]

Wow, this 5150 is in great condition!

[@denshi-oji494]

Well, there have been so many changes to it, it is more like an XT at this point. I am actually surprised the stacked DRAM mod was not done on it however.

[@chico305SIGMA]

The IBM 5150s were built like tanks. They used metal for their basic construction of the case vs plastic like Apple. IBM definitely had more experience with packaging. The 5150 looks like it's from the 90's but it's actually from 81. Lol

[@Mirandorl]

25:12 for the magic smoke : )

[@mgzukows]

Why didn't you just inject say 1V on the 12V line and just see what heats up? You can just put alcohol in the suspected area and it will just evaporate instantly so you know what's up. It wouldn't blow anything up. Just heat up over time.

[@misterhat5823]

That's exactly what I was going to comment. Being that it's only RAM and bypass caps, I would have just seen what got hot with a couple of amps flowing.

[@Marfprojects]

Problem is he did wanted to do this. But even on 2 amps the cap want even warm. Around 5 amps it failed. You don't want to let so many current on the board. Even with this vintage.

[@ThisRandomUsername]

@@Marfprojects Yes, at 1,05V it was at 4,1A. Something you really don't want to happen through a board you care too much about. Not a lot of heat being output in the component (~4,3W), but a lot of current for the supply traces to handle.

[@therealjammit]

Just looking for a hot part would work in some cases (I call it the burning finger test), but on a part that has a really low ohms short the part would have a lower voltage drop across it than the copper traces leading to it. This would actually cause the bad part to be cooler than the traces leading to it.

[@johnrickard8512]

Why WOULDN'T you want to blow it up? :P

[@theoisle]

My 5150 had the same problem plus. The sockets for the end chip on Banks 2 and 3 had corrosion from a memory battery on the expansion board above them. It literally dripped down. After fixing that I found that C6 and C7 on the end of bank 3 were both shorted.

[@VintageTechFan]

The power connectors are supposed to be coded (some of the projections are a little longer) .. but in reality, that only makes the difference between going in quite stiff and a little more stiff in my experience.

[@theryanwitski]

Subscribed! Love the videos. Perfect balance of vintage computers and restoring. Looking forward to more repairs!

[@BadSector]

man, I used to repair these back in the 80's, good times. when I saw all this add in card, that is where we would find the most problems.

[@Renville80]

Although not necessarily a recommended technique, what I do when changing a through-hole component, I will add excess solder to the joint then gently tap the board on the bench. 9 times out of 10, the lead stub and most of the solder drips right out, leaving a small remainder to sip out with a plunger type solder extractor. Then it’s just to clean off the flux etc. and prep the new part for installation.

[@rapidfiringneurons]

6:42 +12v and -12v are very common and are part of the modern ATX PSU specification, as is +5V. The -5V rail is the one needed for those DRAM chips, and most modern computers do not have this voltage. On later IBM offerings like the XT 5160, the -5V rail was already not really needed (some expansion cards used it, but that was the only thing).

[@winfr34k]

Videos like this are why KZbin is still great. Keep it up. :) Retro extravaganzaaaa....

[@TheOcculus]

Old tantalum capacitors. Not cheap parts when the 5150 was made. They were expensive so most computer manufacturers didn't use them. I swap out the tantalums along with the electrolytics in my old test equpment. TEK and HP, among others, used em liberally. Kemet has failure mitigation built into their modern tantalum caps to limit damage to circuit boards from smoking tantalums.

[@TanjoGalbi]

In all my years working with the old PC power connectors I never once put them in wrong. My boss however did one day! He was in his office and I heard a bang. I went in to see what happened and saw the magic smoke coming out of the PC case. Thankfully he was a cool boss and did not mind me laughing at his stupidity lol

[@edsirett3860]

As someone whose job was getting PCs to talks to Machine tools in the early 1980s I am fairly sure the -12V supply was used for the serial port (on the DB25 connector at the back Called COM1 in DOS).

[@MarkALong64]

Good fault finding technique but one thing to consider is that a multimeter in resistance mode is applying a small voltage to the system and tiny caps will start to charge just a little and so the reading will change over time.

[@uliseslay24]

Nice video, thanks!. The tantalum is used because of low ESR or something, the electrolytic does nothing there, you could have just done without with all the other capacitors around it.

[@_Thrackerzod]

As I recall the original 5150 power supply is rated too low (only 65 watts) to provide sufficient power to a hard drive. That's one reason the XT got a beefier 130 watt power supply. I would replace it with an XT power supply personally.

[@AntonioGarcia-ti9wc]

It's always a great pleasure to watch your videos. Informative and captivating. I wish you a great new year 2024.

[@andygozzo72]

i had a multi IO card in a 286 that kept popping its multilayer ceramic caps (the small ones you mentioned) it didnt trip the psu out but every so often you'd hear a crack and see a puff of smoke come out....the psu also had a mains suppressor cap blow its guts out....yes those dreaded RIFAs !

[@andygozzo72]

i've never yet come across a shorted tant, but had a few small electrolytics go low resistance (not total short) in a small cd player 'hifi' system...,and in a late '70s Roberts Radio (unusual as the type used were normally very reliable)

[@pentiummmx2294]

had a RIFA go out on my Apple IIe and IIgs. it was stinky when they failed. fortunately, they're easy to replace if you know soldering

[@trainingtheworld5093]

Very methodical troubleshooting. Great video as usual. If I were you I would recap the whole board with electrolytic caps. You just never know with those ancient tantalums.

[@MoosesValley]

Nice job Adrian. That's was one tought little component. I'd be inclined to leave the others as is ... if they are that tough, they could have decades of life still in 'em.

[@osgrov]

I enjoyed the diagnostic you did here, quite educational for a beginner like myself. :) Glad you found it! That's a beautiful 5150.

[@projectartichoke]

Clever idea using the trace resistance to find a shorted component! I have to remember this one.

[@PJBonoVox]

Saw 'smoke' in the title and clicked. Can't help myself.

[@zarkeh3013]

@ 25:14

[@dieSpinnt]

That was a good guess, Adrian. You should try a lottery ticket soon:) ...and a pity, unfortunately. I would have liked to see a deductive error search: Checking the current on the rails with an adapter or current clamp. Then when you _know_ there is something wrong with the current, you can follow the traces with a hall-effect based probe or build your own and measure the current through the individual capacitors. Imagine the dozens of paths of parallel loads in your resistance measuring method, all with unknown values. And you found the short! Godlike...:) Another method is the thermal one. Testing what gets hot via thermal imaging or the "holy finger". But it can be inappropriate because you can destroy your repair object when running a long time in fault condition while powered. But I'm wrong anyway. All capacitors should be swapped anyways. But where is the fun then? It is always a pleasure to find the error by logical action. Thank you for the video and keep on hacking:)

[@zakmartin6394]

This EXACTLY what happened to me. Thanks for the help

[@BastetFurry]

Why use tantalum, what is the benefit against normal caps? Especially if i look how tantalum is produced, shouldn't we avoid it if we can?

[@hippynurd]

O.L. Stands for Out of Limit, It simply means that the value is not readable by that device. We like to think that and open or short are absolutes, but they are not, and the manufacturers know that too. I knew it was going to be a cap, but I was a bit surprised that it was a tantalum. Its amazing how one part can totally wreck a complex machine.

[@andchip.s]

Nice fault finding, and Bonus magic smoke! What where you look at towards the end of the video, was it the mini screen on the camera?

[@saturn5tony]

Priceless cap blow out 😅 also pricess troubleshooting as well, thanks for sharing!

[@thecaptain2281]

+Adrian Black You should do a video showing the re-cap job. It'll be interesting to see what you use as replacements and if you can manage to keep it authentic or semi-authentic.

[@orangeActiondotcom]

Great technique! A video title like "diagnosing bad tantalum capacitors" might help people who have a similar hunch on other equipment but aren't sure how to proceed.

[@Tim3ru]

Least invasive maneuver straight to it', great work! Plus escaping captured magic smoke... (:

[@denshi-oji494]

another caution about applying 12 volts only to the motherboard... I am glad you did not do it, as you talked about on the video, I am not sure about the 4116, but earlier DRAM had a very sensitive need about voltage rails needing to be applied, within ranges from each other. 12 Volts, without the 5 volt supplies would very quickly destroy the older DRAM chips.

[@michaelwirtzfeld7847]

Sweet diagnosis. Reminds me of my high-school electronics class!

[@littleloner1159]

Haha that old little thing didn't wanted to die

[@Jenny_Digital]

Just a very important point regarding 4116 DRAM IC's but if you don't have all three rails powered you _can_ damage them, so pull 'em out next time you do that!!

[@andygozzo72]

especially the -5v line, if that goes off while the +12v is present, they will quickly fry...

[@Californiansurfer]

I built a clone in 1989 , I still remember the dip switch’s. ATA architecture is still used today. Wow, I am old.

[@clarkss12]

Thanks for this video, it brings back old memories..... :) I remember my first IBM, just like this one, died on me. Had to take it to a special shop (as I recall), and they had to replace one of the memory chips..... Don't even remember what I did with that computer. Must have be in the 70's maybe??

[@juniorbcm5375]

Great way of detecting which cap was shorted. Although the Louis Rossmann method of injecting current would have been interesting as well :)

[@lol4fun]

Likely been pointed out already but your measuring caps on the same rail and charging them with the meter which is what's giving the changing readings

[@Doom2pro]

As Dave would say, ram some high current into the board and blow the ass out of the bad component.

[@Ballgolong]

Hey Adrian. Which GoTek model is that? It looks like the 1.44MB model. To my knowledge the 5150 did not support 1.44MB drives. Could you possibly proved the model of the floppy controller you are using as well as the GoTek model? Thanks

[@stevec00ps]

Your multimeter battery is low - that can affect the voltage readings too

[@aaronbrandenburg2441]

By the way if anyone at all is curious out there I've heard the term for using a beef your power supply to blow a component deliberately to find where it is. Is called Power blasting.

[@PaulKostrzewa]

I have two 5150s and one of them is refusing to power on... thanks for the repair video, I'm hoping it helps get my wayward machine back in line.

[@charlesdorval394]

lmao to this day I still remember the black go together, let's just say I was quite lucky in my younger days hehehe Great video on troubleshooting, tantalums are real fire hazard in my book Oh, for the record, since this was a question in a test I had last week, O.L. when on the Ohms ranges... it's Open Line. That time too I got lucky, overload was also in the answer list, I'll admit I've almost been caught, my courses aren't in English, thankfully the teacher did put "Open Line" in parentheses for one of the choices, else I was screwed lol

[@kaylaandjimbryant8258]

cool deal man! nice find on the revision A. too bad a 3 in wonder card won't do much.

[@adam850]

Do you think multi-layer ceramic capacitors may be a better replacement option?

[@danielmantione]

I think I would have inserted a ceramic capacitor. Experience with other old computers such as Commodore shows they never fail and it also has become the modern standard; for bypass they are the right capacitor. Dirt cheap and effective. Your multimeter looks very similar to my Voltcraft VC830. Always interresting to see that completely different brands are based on the same designs.

[@MontegaB]

Adrian, what's the little chip spudger you have there with the red handle? That looks handy...

[@MontegaB]

Thanks! I thought it might be Wiha. I'm a big fan of theirs.