Howto repair a retro PC motherboard.

[Music] hello and welcome on this channel I get quite often the request for a how to about the first steps of a retro PC main board repair well every case is of course different and it's hard to make a h two which would fit them all I already made quite a lot of repair videos and hopefully everyone of those was helpful in some ways but in this video I would like to go through a typical order of steps which I usually take during such a repair as an example for the repair today I have this 386 main board I took it from my to-do pile of hardware and all I know about it currently that it is dead however it is a beautiful main board with its dark PCB and golden traces a nice piece for a showcase to be honest this board seems to be in a mint condition as you probably know most of the main board repairs which I do starts with removing of a leaky battery treating the board with white vinegar washing it and restoring corroded traces luckily this board has no battery at all and so it never got a chance to be damaged by that so this repair will be not include those steps however if you're interested why I use white vinegar against leaky batteries I made a video about this topic which you might find interesting sometimes a board is extremely dirty and needs to be washed first but in this case it's not needed to it is just a little bit Dusty so I skip a wash and go for the first step mainboard identification as you can probably see by the CPU and FPU sockets this is a 386 DX main board it is based on Intel chipset here we have a name 386 L revision 1.1 this should already be very helpful I use the Retro web project for identification this is the best resource where you can find thousands of main boards with pictures documentation bios images and more many great people from the community have been working on this project since years and continue to contribute to it every day a big thank you at this point to every one of them so what we have is a main board named 386 L let's just search by name and here it is seems to be elite group computer systems or short ECS 386 L here we have documentation with jumper settings bios image hasn't been yet uploaded this is something what I could contribute after this repair and here is the photo of the board so I can look up something if I need to for example if some parts are missing and I need to find out what exactly nice but what to do if you don't know the name of the main board no problem we just have to look up for some attributes as I said this is a 386 DX board due to the sockets furthermore we have 1 2 3 4 5 6 16 bit ISO slots and furthermore another 2 8 bit ISO slots this should be enough to set up some filters on the Retro web search so we go to the mainboard search type of processor family is 386 DX add expansion slots 6 * 16bit Isa add 2 * 8 bit Isa and select show images and search I selected just a few attributes and of course we see quite a lot of similar main boards but you can add more detailed information like chipset and other options to improve filtering anyway thanks to the images you can visually go through the list very quickly yeah and here is our board so it is an ECS 386 L interesting observation up front the board in this entry has vlsi chipset but the one I have has Intel chipset though vlsi tag is also written on it most probably this is the same chipset made in corporation and released under different names well now where we identified the board let's move on to the next step checking for shorts one one of the most common issues on the old main boards are shorted tantalum capacitors which can put power supply voltages to the ground that's why it is important to check if there are any shorts between the power supply pins for that matter check out the pin out of the power connector this is a standard at Power connector in the middle you have four ground pins and left and right from it there are are plus - 5 Vol plus - 12vt and power good pins none of those should show you less than 10 ohm resistance to the ground usually it goes up to a kiloohms but I saw already boards where the resistance was as low as 20 ohms but as long as it's greater than 10 ohms it's probably okay this tab is also very important to do because not all mainboards use the standard PCU pin out there are many early at main boards which use connectors which look the same but have proprietary Pinal if you connect a standard PCU to such a main board it will probably die instantly so be careful and when checking the power for shorts pay attention that what you measure makes sense four ground pins in the middle must be connected together and of course they are the ground uh three outer pins on one side have to be all connected together and 2 plus 5 volts if you measure something like this you most likely have a main board with the standard at Power connector so our board here has a standard at Power connector and no shorts between voltage lines and the ground but what if you would find a short well I can't predict every possible culprit but in such a case most common issue is one of the tantalum capacitors these yellow decoupling caps which are sitting near the power connector between the iso slots or near ic's these caps are usually used for the voltage stabilization and if you find one of those shorted it is enough for testing to just remove it if the short has gone afterwards just replace it with a capacitor of the same value it is by the way absolutely okay if you replace tantalum caps with electrolytic ones okay so as I said we have no shorts let's connect it to the PSU and move on to the next step check the CPU voltage for this step you need to search for the CPU pin out in this case we have a 386 DX and we want to check if the VCC voltage is plus 5 volts as required and we seem to have steady 5 volts nice and the next step check the clocks every mainboard has multiple crystals which generate various clocks on the main board for this step you will need an oscilloscope first of all let's check the CPU clock on a 36 DX this is the PIN F12 and the CPU expects double clock there let's see and we are getting 50 MHz here so the inserted CPU would run at 25 MHz that looks good this clock is generated by this Crystal oscillator and 50 MHz is what is written on top of it however there are more crystals on this board and they all should deliver what is written on the enclosure this one gives us 14.3 MHz and this is right this one 16 MHz also what it promised this small realtime claw Crystal it should give around 32 khz home um 56 khz is wrong also the waveform looks strange it should be sine wave actually this looks wrong now but this Crystal controls only the time and as long as we get something here it should actually work also 56 khz is not far away from 32 khz so let's pretend for now that this one is okay maybe we will come back later to it but what to do in general if you measure wrong clock or no clock at all well in such a case I would check the small ceramic caps around and try to replace the crystal oscillators so where voltage and the CPU clock look good we can go to the next tab inserting the CPU there is not much to tell about this you should always apply some deoxide to the socket to exclude any contact issues as for the CPU this board expects a 25 MHz 386 DX but I don't have any however I have one which is specified for 40 MHz and it will of course run at 25 MHz just fine and we can move on to the next step power on self test for this step you would need a post analyzer card when trying to start the main board sends some status information to the isabs which can be read with such a post analym I can insert the card vertically into the slot but I use additional angled slot razor so you can better see the display from Top oh and you should also insert a PC speaker since some boards don't show any post codes at all but have only bip codes instead most boards have both however right the reset LED turned off this means that the reset line is not holding the system back but there are no beeps and no post codes on the card so this board is definitely dead and we can move on to the next step analyze address and data lines in this step you can use the oscilloscope to measure the related line directly in the iso slot however I don't like it to do that way first of all the data pin zero is located just opposite from plus 12v on the other side and if you slip from the pin you can send 12vt directly into the data bus which will kill your main board instantly and most probably also burn the chipset that's why I prefer to use my prototyping board for assid measurements I also don't need to search for the isos slot pin out but since all the measure pins are already marked and visible it is much more comfortable to use this card if you find it useful to you for repairs or development you can find the Gerber files in my GitHub repository okay let's see what we have first let's see once again that the oscillator works properly yep it does and we have our 14.3 MHz now the address lines okay they seem to be dead in the background I'm turning the system on and off to see if there is any activity happening on the address bars but it seems to be completely dead once again let's see on the reset line when the system turns on the reset should go high briefly and then low again yes this looks good as well now let's take a look at the data lines they are all dead as well hanging high all the time yeah the address lines are also still dead some seem to be high some are low but I don't see any activity what does this mean well the address and data bus is often connected through additional transceivers to the CPU and the chipset those are usually 74 LS 245 or similar ic's if you measure something strange on those buses it's worth it to search for those ic's near the iso slot and check if what is going in and out there makes sense this would be something what I would do on many boards but on this one there are no additional transceivers and the buses are connected directly with the chipset so either the chipset is faulty what I don't hope because that would be dead end or there is something else wrong what controls the chipset and we are moving on to the next step check the BIOS Ram in this state the step doesn't make a lot of sense since we don't see any activity on the buses whatsoever but this is what I would do at some point anyway so let's see if the BIOS Ram is all right first of all it's worth it to take a look at the sticker if there is any the problem is that I got many main boards which somebody else was already trying to revive and sometimes bios ROMs from other main boards were installed as I got them obviously this one seem to be the original one since we clearly see 36 L written on it just like the name on the board uh that is good and if it works I can contribute the BIOS ROM image to the Retro web project and it looks good I could read this eom on my computer and the data seems to make sense there we see 386 L version 1.0 and this looks right to me so this bios eom can go back into the board of course with some deoxide again but what to do if the BIOS ROM is wrong in such case you can try to find a bios image on the Retro web project if you are lucky someone already uploaded a copy and if not try to find it similar board with the same chipset for which the BIOS image is available if the eom IC is that you will have to find a replacement eoms can be taken from scrap or replaced by a pin compatible e proms the eom with a window on top have to be erased with a UV lamp like this one e proms are easier to handle since they are electrically erasable and they don't need additional equipment as you see I always keep a stack of spare ROMs for that purpose many mainboards which I get from scrap have those ic's removed and so I need them quite often but on this main board the ROM is okay and we can move on to the next step searching full damages this step can be made in the beginning of course but if you know what works and what not you can look more closely at places which make sense and so since we have seen problems with the data and address lines which go into the chips I concentrated more on this area and I found three issues first of all near the chipset there is one damaged capacitor it seems to be on the 5vt rail and is probably for power stabilization for the chipset it's not shorted and is most probably not responsible for the problems it is kind of optional and could only make the system unstable but not totally dead so we can move on down here we have another destroyed tantalum capacitor it was obviously destroyed by some violent Force the um case ceramic enclosure got cracked and this one has to be replaced however also this one is not shorted and is once again probably a decoupling capacitor for power stabilization once again something what is probably not causing the issues which we saw however nearby there is some damage to the PCB which was obviously applied by the same ction which cracked the capacitor nearby and this is very interesting let's take a look under the microscope so here once again the poor tantalum capacitor this here is the CPU and the traces are going from it up to the damaged spot on the PCB here we have a scratch going over the two traces but it looks to be only on the surface the copper seems to be not damaged and up here here we have a big dent in the PCB a whole piece of PCB split away and took a part of one Trace with it these traces are going from the CPU to the chipset and the probability that this is our culprit is very high so let's fix the [Music] traces [Music] [Music] [Music] [Music] [Music] [Music] my [Music] [Music] [Music] oh [Music] [Applause] [Music] e it's done let's check the continuity looks good as you probably noticed uh I also removed already the cracked tantalum capacitor now it's time for another test let's insert the post analyzer card and the speaker once again yay would you look at that we have a Survivor the PC speaker bips periodically and probably asks for some memory here I have some memory modules I'll take four times 1 Megabyte 386dx is a 32bit CPU with 32bit memory bandwidth each memory module is 8 bit so on this system we have to install the memory in groups of four modules I populate the first four slots these are eight bips in a row which means missing graphics card let me add a VGA card and an IDE controller with compact flash adapter as hard disk replacement beautiful we have the post screen here is the 386 L version 1.0 message from 1991 which we saw already in the Bios ROM previously and we have 4 megab Ram which is also right the battery is not connected so the message about Simo settings is okay without battery it's not able to save any settings anyway so let's take a look on the bias settings first of all I have no floppy connected Drive C the compact flash card geometry is 987 cylinders 16 hats and 63 sectors makes 486 m es let's take a brief look at advanced settings here we have some memory Shadow settings nothing exciting and here are memory timing settings I guess I'll leave everything at default this video is not about tweaking anyway we just want to see one happy working mainboard so let's see if it boots into Doss ladies and gentlemen congratulations this revived board has booted the dis operation system let's see some Benchmark results for example Landmark well 25 MHz 36 DX is correct I would say one interesting thing is I don't see any cach ic's on the board I'm curious if the chipet has any integrated cache let's give cash check a go and see what it reports well as expected this machine doesn't seem to have any cash that's unfortunate but at least we should make an ultimate test and start a game this seem seems to work as well very very nice well I would almost call it a day this main world is alive again and you hopefully did find this walk through kind of useful after such a fix I usually cover the repair traces with some nail varnish for protection and I will have to replace the capacitors later one thing I would like to show you though as we tested the address and the data bosses um maybe you remember that those were completely dead there was no movement whatsoever now where the system works I would like to show you what would it look like if they would work as you see the signals are now jumping happily between five and 0 volts also on the data bars there is obviously some um activity this is how it should look like on all of the pins and if you see a signal hanging high or low all the time on one of the pins this should give you a hint where to search next as I said often the bus transceivers are the culprit but some boards like this one don't have any so it depends where to look for oh and one more thing I totally forgot the RTC clock if you remember I measured over 50 khz on this Crystal which controls the realtime counter I don't know if you noticed but the last time we've been in the Bios the seconds were ticking about twice as fast if you look carefully you can also see how some seconds are even swallowed completely this is a result of the too high clock which the crystal generates let's measure it once again this time I just use my multimeter it can measure frequencies up to um 5 MHz or so well now it shows 66 khz not exactly what we had last time but maybe the crystal got even worse and the clock is sliding further away from this speack let's just replace [Music] it here I have a similar Crystal which I took from scrap it should give give us proper 32 [Music] khz and yay we have now the expected 32 khz let's take a look at the bias type well unfortunately this bias type has an issue which ends up in irregular screen updates so it looks like the seconds would jump all the time but let me start a timer and see if it counts properly in average and yes this looks good to me it's just jumpy but in 10 seconds it counted 10 seconds as well and this is it once again I hope you enjoyed this howto and the repair it is of course just the way how to start and some times you have more complicated cases where you have to study a lot of data sheets make a lot of measurements do logical assumptions but all of that is highly specific for each issue and it is hard to predict everything what could be wrong for every possible mainboard however I hope you will find it helpful and this how to will at least give you a good start please don't forget to give me some thumbs up if you liked it or down if not every feedback is appreciated and so far thank you for watching and goodbye

Channel: Necroware

Views: 21,926

Keywords: retro, hardware, soldering, repair, review, nerd

Id: A0OXbKaYMbQ

Length: 28min 4sec (1684 seconds)

Published: Fri Mar 01 2024