The motherboard had vias for through hole capacitors under each SMT cap, they were at 90 degrees to the SMT pads. They just needed the solder sucking out of the vias. I normally just heat up one leg of the SMT caps with the iron, while gently tilting it to one side, once that leg is clear just heat the other leg to remove the cap.
@@DxDeksor might be but from experience dry joints are way more common. By dry joints people usually mean either a joint where not enough solder had been used initially or if joint wasn't heated up to a proper temperature leading to a weak joint that's essentially being held by stinction instead of solder adhesion to the copper. Or a joint that had developed a micro-crack over time which became a typical illness since the move to the RoHS
@@lexatwo yeah I know, but with hardware you may never know, just look at the commodore 64 ! (Though you're probably going to say it's a specific case and you'd be right, but hey it may happen to other chips, especially ram I think).
Deksor yep, proper investigation is needed to figure out what's wrong here. My approach would be to use thermal imaging cam if available to determine warm spots around area where memory chips are in and then run some stress test in cycle and use compressed air cans to locally cool joints and chips. It would allow to localize the problematic joint or chip and fix it either by replacement or by reflowing - whatever appropriate.
The reboot happens at specific moments, that should make troubleshooting 'easier'. First step - it happens during the POST, while testing the memory - more specifically the extended (above 1MB) / Expanded Memory Second step - it happens after trying to load the mouse driver. In next parts of the video, this seems to be SmartDRV - which also uses the extended/expanded memory. This, for me, narrows it down to the memory controller or the soldered memory. As it seems to getting to work when 'getting warm' - check for dry solder joints in the memory. For some reason pin 29 attracts my eye, but that might be misleading (not seeing it in person) Hope you get it :)
Will try a memtest. Pre-heated the area with a heat gun but results were inconclusive. Also reflowed all the pins of the ram stick. I did notice the board was kinda bent around that Memory stick (before I started work on it). The area of the board isn’t 100% flush.
@@RetroSpector78 Had it once with a 386. The board was squeezed in so tight it was probably a bit warped and caused issues. Loosening the screws a bit helped. So maybe stress on a chip or joint?
I had the boot loop in the machine I'm currently posting on, which is a lot newer than a 386 (running an i7 here), and after a couple of days of despair and hair loss, I decided to simply take the memory sticks and put one at a time and that did the trick. Two of the memory sticks were bad. The computer worked flawlessly after that. Entirely different architectures, but maybe there's some relation. I hope you find out what's wrong. :)
@@thomasandrews9355 Proper way is to use hot air reflow station with kapton tape shielding or special dual-tipped soldering irons or two irons - if space allows to reach contact pads with two irons at the same time. Any twist or cut techniques are not a good behavior, even if being done by experienced person and really carefully. This is just due to a fact that torque forces you create on cap when twisting are huge due to the force amplification you get through your tool handles. Yes cap's pads are usually the weakest points but for heavily leaked cases it might not be the case. Proper desoldering coupled with some vinegar applied to neutralize electrolyte and followed by immersion ultrasonic cleaning are best for cases of heavy leakages.
I know it seems like a weird technique but in this case it’s OK if you get it right. Check out ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-X8N9O3a9jiM.html
i noticed at initial boot it resets while checking the RAM and in dos it resets when attempting to run Smartdrive, which is also RAM-related program. Might there be some bad traces near the memory, or bad soldering contacts? After booting when enough heat has been generated the bad joint makes contact again?
That RAM solder job looks dodgy. Also puts a ton of stress on those legs when the machine is moved around. I would desolder the installed SIMM (possibly could just desolder the power leads?) and use a regular SIMM in one of the socketed slots as a first test.
Another nice video. There's a good chance that pad would have lifted regardless of the method of removal. Twisting caps would be my last resort but I have done it before were absolutely stuck. Sometimes there so much corrision no amount of heat will melt the old solder. If you are doing it again I'd suggest a pair of cutters and snip or crush the old cap to almost release the legs from within it. Maybe even a small screw driver to hold the base down. Looking forward to the next part and hopefully a full working machine. The fact it seems to work after warming would point towards caps but maybe also a cold joint somewhere as other have suggestted. Maybe even reflow all points on the psu? Doing the entire motherboard would take a while but might be worth it.
Attention now is on the memory module (cold joint). Also expecting brand new caps. Opinions seem to be divided 50/50 between caps and cold joints. Ram might also still be an issue.
if you have some "freezer" spray you could wait till the PC boots up, then cool down various component and try to boot up again. I'd start with the memory chips. Had a similar issue with a modern PC where ONE RAM module would fail any RAM test only within the first 5 minutes of use. Then it was fine. (Unfortunately that was enough to corrupt the OS after several BSOD). And no, I am Italian and that weird power button configuration doesn't tell me anything! :)
The caps in the power supply makes sense. The fact that it gets better after being on for awhile suggests that the caps are taking too long to charge up. It also might just get better in general if you leave it on for awhile because older caps can sometimes reform after being off for a long time. (they don't have to be leaky or dead)
Will have all brand new ones in a couple of days. Will do a proper re-cap at that point. Would like to try it with another working PSU but don’t really feel like making cables / adapters / soldering wires to the thing :)
I am pretty sure this computer is affected by “thermal expansion”. After thousands of power cycles from cold to hot certain components can shift their position and this can cause issues when starting from cold. Put flux on the SMD chips and go over them with a hot air rework station..I’m not saying to remove any chips but just to re-solder the components back onto the pads on the board.
What is the solder flux "pen" you're using? It looks quite confortable to use and I'd want to buy one. Thanks. To test the RAM you can use the Memtest86 program, so you haven't to remove the chips from the motherboard. It's free, the installer will just create a bootable floppy disk that will launch automatically the utility when booting the machine from that disk. Then let it run for a couple of days, and after that if you won't see any memory address hightlighted in red on the screen, this will mean that the memory is OK.
Don’t know if you noticed but all the caps an their positions on the riser board not only had PDS for smd caps but also throughholes for standard radial caps. That was sometimes done in the late 80s when moving more and more towards smd but still having lots of tht components. I had seen boards up into the mid 90s that had provisions for smd as well as tht parts.
Replacing the rest of the caps is good, and as already been said, it could be cracked solder joints and stuff, but I'd also look at the the optocoupler in the power supply. I think it's U2 there. Not a primary suspect usually, but they can degrade over time, and their lifetime will depend on how they're driven. We're talking about old hardware, so worth taking a look, or even just replace while you're at it. Also, you can connect a few dummy power resistors to the rails, and perhaps even a current shunt for scope probing. If the power supply stalls, you'll likely see it better in the current than voltage, and you need a predictable dummy load for that.
May be more capacitors - it is common issue with "doesn't work well before heating up". And one of the tests is to cool capacitors one by one - to check which one will cause problems.
So, next stop: Memory! I’m wondering if a cracked joint might be the cause of this issue. Just put some flux on to the joints and some fresh solder. It might be fixed by just doing that! The PSU doesn’t look like it’s faulty any more. You’ll get there 💪 I have again been watching with great pleasure!!! 👍
I had a similar issue with the Compaq Deskpro EXD and 2 PSU (one was brand new). With both PSU the PC shuts down after a few seconds and everytime I powered on it did the same thing, with other PSUs the PC works fine. For some reason I let the power cord plugged in for a while and now works every time, even if it's unplugged for a few days. Thanks for the video!
Nice to see that you made progress with that machine. I think you omitted caps on riser card - try replacing them as well. This cold start boot loop still indicates bad capacitors somewhere. I would also recommend thorough cleaning of all the motherboards chips/soldered RAM with IPA soaked toothbrush. It is also possible that come solder on the mobo cracked and some point - you my also try to correct as many soldering points as you can. Too bad that PSU is not AT compatible, so there is no easy way of swapping it (or maybe just soldering new ATX PSU wires to right spots on motherboard). Good luck with that can - people are watching :)
Did replace 3 caps on the riser (all except for the axial ones) after I created this video. No difference. Started soldering wires to the connector but turned into a horrible spaghetti mess on my bench ;) best thing would be to create an adapter cable. Or need to find compatible connectors I could crimp on my cables to properly insert them in an atx connector. Mine kept falling out :)
@@RetroSpector78 There are ATX -> AT adapters available. Maybe it would be good idea to sacrifice adapter to use with modern ATX power supply (or at least rip out ATX power supply to accomodate old PSU case).
I saw what looks like a potentially bad trace when you were taking the caps off near the keybaord controller. It was on the very edge of the board and looked a bit black. Do you have a POST card? That could tell you what it is crashing on. Could be an issue with the power on reset circuit timing cap as well.
Are the via's near the electrolytic caps corroded? Did you try contact spray on your connectors and sockets? It looks like a bad connection, like a crack in the solder. You can try to reflow some things (be aware of overheating).
Just general stuff, clean the contacts of anything socketed, look around the processor for missing or damaged bypass capacitors, Does it go through the reset loop when the HD is unplugged? Do you have another HD (or better yet CF to IDE) to test in it just to make sure the HD is not doing something strange (noise on power, sending some garbage over the IDE line).
Aand it's a little spaniard, cool! Always find to find out whre one for these Ebay orphans came from. Btw, my Olivetti PCS 33 has similiar issues, I need to dig it out and look after the caps. So much to do...
Does it POST like it should if you "Pause/Break" machine immediately after first cold boot and wait for couple of minutes and then resume? Just curious..Btw I think you possibly could "reverse" the issue by using cold spray at fault location when machine has gotten warm? Not sure, but I'd try that.
I have had an Olivetti PCS 11 for many years, which has a similar design but it is even smaller. Unfortunately it has never worked while I've had it, at the very least the power supply is dead. I want to get it running some day, but I don't have the skills unfortunately. It doesn't help that everything is non-standard, and documentation is almost impossible to find. Anyway, nice to see video of an Olivetti very similar to mine!
I'd be looking for grey solder joints on the power connector on the motherboard, rather than focusing on the PSU, You should look out for tiny "rings" in the solder joints around the soldered pin which would indicate a fractured joint. Also, try gently moving connectors during boot one at a time to see if it causes a re-boot. Nice work, nice machine.
I remember back in the day, a problem with the state of the A20 gate handler. I had to boot the computer, then do another warm boot with the reset button to make the thing work. I don't know if your issue is related, but it might be worth looking into.
Since it needs to "warm up" I wonder if it's more related to temperature than to voltage. What if you pre-heated the board with a hair dryer or something and try "cold" booting it then? would it come up first time? if so you might be looking at broken solder joints or something that might make better contact when the temperature inside the case is warm.
For this PSU one way to grossly check the ouput filter caps is making an AC measurement. YOU ARE NOT SUPPOSED to get any AC reading at all. This will work on true-rms meters (nothing rare these days). For PSU using a stby voltage line, FIRST CHECK that voltage. This stby is the supply for the chips of the PSU and since it is the one that never shuts off, tend to get bad first. This PSU is probably AT. Still the filtering for the control IC needs to be checked. Try to change the main one when you have a replacement, those can go BOOM even if they test fine. I ve replaced them in the AT PSUs I ve done. I took the replacements from modern ATX PSUs, super inexpensive. The AC TEST works nice on laptop chargers also. Also during testing, away from the computer try to LOAD the PSU with a couple a hard drives. Some PSUs start to trash on load, or on the other hand, they need the load in order to regulate. The 3.3V rail is the perfect example.
the large capacitors will not be faulty, i've never known one to fail. did you change the small capacitor on the primary side? that could cause this problem
Is your soldering iron temperature controlled? When soldering using my cheap iron (basic plugs straight into mains type), I find that the solder joints end up bad (‘ball’ and/or ‘burnt’) and I’m try to isolate whether it’s just my technique, my iron or my solder.
Probably. If you don't have much money, get a Yihua 936 it will cost just a tiny bit more than your current iron and you'll get an iron with temp control and standard type of tip. If you have a little more money get a TS100. The heat will transfer much better to the solder point and the tip will last much longer.
I would advise against twisting the caps. If you rip off the pad you have to get creative, and in some more dense boards with many layers you might not even be able to run a wire to replace the pad. Not a problem here, but if you work on a more recent piece of electronics you might not have a possibility to scrape off contact space or even find where the trace goes. Had a thick multilayer board (more than 10 layers) and one trace basically disappeared right under a BGA chip and the via was inacessible from both sides. Major PITA.
To be honest I dobut that this is caused by electrolytics. At least not by the output filter caps in power supply, especially that this started to work after few minutes (assuming after they warmed up a little bit). IMHO this is something related to the memory or it's power supply. Does the motherboard have separate voltage regulator or dc-dc converter for supplying the DRAMs? Could You just run a memtest on this PC?
I have a 386 board with a similar issue - from a cold start, it detects different amounts of memory at POST for a while and Windows fails to boot, but eventually it starts to run perfectly. I have tried different SIMMs and the issue is all the same, so I suspect something is wrong with the board. I also have another 386 machine with the opposite issue - boots up fine initially, but starts crashing earlier and earlier the longer it's turned on.
I would suggest taking a good look at the memory module and the contacts on the chips seated on that module. Due to the point of resetting the memory is a likely suspect. I don't think the chips themselves are bad but i'ts most likely a bad solder joint which makes contact after warming up for some time. Good luck with the project and hopefully this can help you out.
MIght check for missing or damaged resistors or SMD bypass caps on the memory module. I had a DIMM once that one of the resistors or capacitors got knocked off, it would work about 90% of the time, but suddenly would just do strange things.
RetroSpector78 at 7:40, I can see there are 2 vias or through hole pads for each of the capacitor you replaced. Why didn't you just use them ? You've had a better contact and a far better mechanical holding
If you must do the twisting method to remove the caps you should be pressing down at the same time as twisting back and forth, absolutely never pulling.
And indeed they are :) It worked for me but I can imagine not everyone being too happy about it. I think that if the board is relatively new / clean / high quality, and you apply the right amount of force and don't try to pull it upwards there is a decent chance that the pads survive. And it is also a lot quicker. But I'm far from an expert as you can already imagine from looking at my video's :)
4 года назад
the proper way is to use the hot air station, another way to remove those type of smd cap is to cut it in the middle and remove the base, after that you can simply de-solder the leg, i advised against the way of twisting the cap when you work on the old board, you may damage the pad
Crystal oscillators I suspect.... The board was designed to take both smd and through hole caps! Where did you get an ISA tv-card, or better yet, why are you sabotaging the card?
Yeah should have noticed that about the through holes ... was in tunnel-vision mode I guess. Regarding the tv tuner card. Have zero use for it, and was the first card I found in my useless expansion card bin with some smd caps... didn’t even notice it was ISA to be honest. I put those in the same category as modems. Never going to watch TV on a retro computer I think.
My guess would be an issue with the memory. One of the caps was next to the soldered-on memory module, so the electrolyte might have affected it somehow. I'd try reflowing the pins connecting it to the motherboard with fresh solder to rule out a dry solder joint. Or you could remove it entirely and try using SIMMs instead. Although that would probably be pretty difficult without a desoldering gun. You might risk damaging the board trying to desolder it with a standard soldering iron and solder sucker or braid.
It's also not a 30pin module so not sure if I can just remove it and use standard simm modules in the simm sockets and omit the bank 0. And I would also like to get it working while keeping that soldered memory stick. We'll get there :)
Yeah, now that I think about it, bank 0 would have to be present before the SIMMs in bank 1 were populated. But it's definitely worth at least checking the solder joints on the memory module, if not reflowing them.
4 года назад
I probably need to do this with my old Olivetti M24. It's working but it is very unstable, sometimes it hangs on boot. I did only replaced battery, but it was like ten years ago. It can be leaking again.
Twisting is always really a bad idea, no exception IMHO. Way better to use hot air reworking station and a pair of tweezers. However, liked your video as usual. Cheers, M
This is a classic symptom of a cold solder joint somewhere. Or a broken joint. You'll have to reflow a solder joint somewhere, if you can find it. Sometimes they can be a real pain. What happens is, once the machine warms up, the solder joint expands and starts making better connection.
So I thought about this a little more while I was rearranging my workshop. It's kind of interesting that it's actually resetting instead of just crashing or locking up. Makes me wonder if the power-on-reset circuit is acting hinky. There will be a /reset line on the bus (if I am recalling correctly) that will be pulled high through a resistor. When the machine powers up, there's a timer that pulls it low for a split second, to give all the ICs time to get powered up and initialized before execution begins. But if there's something weird going on where /reset is floating sometimes due to a cold solder joint, it could be asserting itself, or maybe if it's losing the 5v supply on upstream somewhere, that might cause it to float too. I'm really just kind of guessing, though, and I don't really know what I am talking about most of the time, so....... >_> I just think it's a little strange that it's cleanly resetting over and over rather than just wigging out and locking up or something.
Yep, as commented, i had a hard time recovering ripped pads using the twist method to take these SMT caps off ... now i'm using tons of flux and hot air station, i've being more successful .. .also , there is another method, swinging it along the pads, but also it's possible to take the pads off, my friend, also expert in retro-computer recovery made a video about it ( it is portuguese, but YT captioned it ) > ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-u3qSlONxFaM.html ... watch from 7 to 10 min .... About the reboot cicle while cold... it seems to be related to the memory , maybe some broken solder , on that daughter board with the ram chips.
Nice video. Have you tried to boot it from floppy with ms-dos? Maybe some application intalled in the hard drive is making some kind of wierd verification, and it causes the reboot.
Think its all a matter of personal preference ... everything has its pro’s and con’s I guess. Worked for me. The one pad that was lifted could have easily been damaged by a heat gun / soldering iron also as it was already far gone.
You’re * probably * right that ripping them is * probably * not going to damage a pad that isn’t already a goner. But it just seems so needlessly violent. >
My advice for this stability problem : heat and add a little bit of solder to each through-hole on the motherboard. If this doesn't fix the issue : reheat all the pads of the SMDs with hot air and flux.
I would look at the +5vdc rail under load with an oscilloscope, trigger set to catch a glitch, and look at what happens when it does the re-boot, that will tell you what direction to head.
DOS 5's himem.sys didn't really test the extended memory. You can just replace it with himem.sys from a DOS 6.x machine and see if that finds an error. Or try CheckIt. Next step: Have Fun! Upgrade the video RAM and install drivers for 800*600 at 256 colors. And add a sound card. I'd probably replace the speaker with a piezo, put a Sound Blaster 2.0 in and solder the old speaker to the sound card.
If you desoldered with that odd onboard SIMM, I suspect that the board would still boot with replaceable SIMMs. Something of a last resort measure for sure.
Maybe there's a bad chip on there that starts working after it's heated up ? Since it reboots everytime when checking the ram, maybe something's wrong with the ram ? To check this you could try to heat up a chip using a hot air station or a simple hair dryer. Or also try the opposite by cooling a chip down using a cool air spray ?
1:48 - Those are not in fact polymer caps. Polymer means solid state, so they can't leak. Below is some info on caps for everyone interested. Those "metallic" capacitors with black paint are most often regular electrolytic capacitors, just in different package (surface mount). Polymers are usually colorful (pink or red mostly), and the stripe on polymers means Positive. This is unlike regular el. caps, where the stripe means Negative. If you are replacing regular el.caps with polymer, remember that polymers most often cause short cirtuit when they fail (in case of reverse voltage or over-voltage, even for very short period of time). If possible, I tend to replace electrolytic caps with ceramic surface-mount caps. Nowadays you can find values like 10u 25v pretty easily.
Careful swapping electrolytics with ceramic. Ceramic caps have issues with derating capacitance vs voltage, and also may cause problems with resonant circuits that depend on higher ESR. (Not an issue with most bulk capacitance / decoupling applications, but some of the 0.47 - 10u electrolytics may have other purposes, for e.g.)
I can see your keyboard is getting reset each time the computer resets (all three lights flicker). Since the keyboard does not rely on anything but 5v and ground from the power supply, I would suspect the power supply is at fault. If there were dry solder joints in the ram you would immediately get parity errors.
Yeah ... some debate going on right now if it's the power supply or dry solder joint. Not that easy to hookup another psu. Would need to solder wires to the board and create a special connector. Don't have those special connectors lying around here.
I have noticed that on the "second" reboot, it only show 1 floppy drive instead of the 2 floppys on the boot that gives another reboot It may mean nothing but...
It did that once ... not really sure what that was about ... but ever since it correctly shows 1 floppy and recognised the floppy drive fine.... also added a 5.25 inch drive today and it detected it just fine. .... think it was just a glitch.
@@RetroSpector78 ok, but visuals and measured voltage seemed to suggest it. (And, old computers used NiCd, but three cells) - But, yes - zoomed in and it's a Sanyo CR-2450@3V
It is on my TODO-list and it will definitely come . It's a nice 486 system and have lots of upgrade goodies for it to make it my ultimate mid nineties multimedia machine.
Do a power rms / load test on your 12 and 5 volt. You might have transistor / mosfet issues.. it doesn’t take much to damage these and you say after a few minutes of being on they start working.. classic sign of a bad transistor. Hot air is how I fix them. Temporarily of course..
Twisting those surface mount caps to remove them is bad workmanship. You're far more likely to lift/tear traces doing that. A better way is to heat one pad and lever the cap up side by side. Also, there was no need to kludge the thruhole caps to fit the surface mount pads. Olivetti kindly provided you with thruhole AND surface mount pads under the component, likely to allow them to use whichever one they could buy more cheaply at the time
Twizers + smd = crack :) Have you seen a capacitor to explode using hot air ? You don't need to heat it that much :D. And I'd rather have a cap to explode instead of breaking a trace ^^
@@RetroSpector78 Definately, use goggles if you are going to use a heat gun. Even at 300 degrees SMD capacitors can explode and take out your eye. Cut them in half while gently pressing down on the board, no issues whatsoever, its faster and cleaner. Clean the board while the legs are still attached to limit that fishy smell. I don't use vinegar as it only damages the board more, baking soda is ok.
Don't breathe the fishy smell. I've desoldered several caps like this in a session and even though the room was well ventilated, after the session, I felt a little pain in my lungs for 2-3 days which is definitely not good I believe. I wonder what's in these fumes but it can't be good. You should wear something to protect yourself from these fumes (I'm going to do this too the next time I do this) As some people have said already, that twist technique is not recommended at all because there's a risk of breaking the pads (kind of like you showed how to not desolder a through hole chip before). It may not always happen, but you get the idea. Use hot air instead.
You DON'T twist off bloody capacitors you remove them by removing the old solder you will rip a pad off or a trace only a mater of time before you do more damage with this method
Saw the technique first on Mr Carlson’s Lab and seemed to work for him. This board was pretty decent quality so decided to try it out. And turned out ok.
