Wow at that explosion!! I suspect you need the DIP RAM installed, to give it base memory. Chances are the SIMMs are for extended memory. Some of the older 286's were funny like that! I realise some of the manuals that support that board show either the DIP RAM, or SIMMs... It does show x 9 type though, so needs parity by the looks of it.
I hope you are right! DIP memory is on the way. So far, I wasn't able to find anything else suspicious about the board and why it wouldn't continue to boot. The memory should arrive next week and I hope it will fix that issue with the memory. Fingers crossed!
I have another 286 board with DIP+SIMM configuration and it runs fine on 4 x 1 MB SIMMs without any DIPs. But my board can have a different implementation.
Great work with the restoration! Octek makes some really nice boards. I had a tantalum cap explode on a sound card a couple years ago. It was totally unexpected and was very startling. The worst part though was the smell. It lingered for hours even with the window open and fan going! .. also so awesome to see two of my favourite TH-camrs troubleshooting together 😁
Thanks! Surprisingly, the tantalum from this board didn't leave a lot of bad smell behind - it was gone in seconds. Maybe it was because it exploded so violently (it was the 12v line after all). I exchange thoughts with Tony often - I learn a lot from his analytics and debugging skills. His videos are always entertaining and educational!
@@alisharifian535actually it stems from „Sakrament nochmal“ (a swear with catholic origin) and was considered rude / blasphemy. Hence, „Sackl Zement nochmal“ is used instead because it sounds very similar.
@@aracon22 himmi herrgott saggrament (heaven, god, sacrament).. usually followed up by greiz gruzifix (cross crucifix), to pile some more blasphemy on the unfortunate subject. We Bavarians really love our religion-based insults and curses. 🙂
the 20MHz Harris made 286 processors are the best 286 processors. They are very reliable and run cooler than the other 12MHz 286 cpu-s. You have a jackpot on that board, my friend.
I had a 20MHz 286 as my first PC and it was giving my friend's low end 386 a run for its money. They felt nearly equal in power, except his could run Doom and mine couldn't. I miss that machine so much. Father sold it off and I was too young to memorize the specs. I would've rebuilt it, but I don't even remember what I had (in the 80s').
Must just be a good day for bad caps on 286 boards...I popped one this morning! After replacing a shorted one that I caught with the multimeter, I plugged it in, and the test supply found one I'd missed on the +5V rail.
I've heard so many bad things about those tantalum beads from the 80s and earlier that I'd be tempted to just replace all of them. They were mainly used because they were very compact and had tight tolerances but modern high-quality electrolytics have much tighter tolerances than the ones back in the day had, so often enough they're a decent replacement for tantalums from that era, or so I've been told.
@@Ragnar8504 Tantalums have different performance characteristics than aluminum electrolytics, so aluminum electrolytics are not always a good substitute. Low ESR and especially polymer electrolytics are often suitable replacements; however, not all vintage tantalums are problems. It seems to be mostly an issue of storage conditions: damp conditions seem to accelerate tantalum crystal growth, which is what causes shorts. Boards stored in damp basements, or worse, from scrapheaps, have higher tantalum failure rates than boards store in climate controlled shops and living spaces.
Harris Semiconductor were a pretty decent chip producer in the 80s and 90s. They had a strong presence with specialist parts for avionics and similar high-end applications. Coincidentally, my first PC also had a Harris 80286, but only at 10MHz. I also had a 10MHz Intel 80287 in it, which was a big help for the PCB CAD software that I used at the time. As for the tantalum beads... I designed lots of boards that used them at that time. Back then, they weren't old, so they mostly only went bang if soldered in backwards, but still with fireworks 😂. Also, you're right about the RAM chips being for the system RAM. That board is marked for 256k x 4 chips (44256) and is laid out for either dual-in-line or Zig-zag packages. It maybe that it always needs populating regardless of any expansion SIMM modules. Edit: I see that the user guide says that using only SIMMs is a perfectly valid option, so that's not the issue.
I'm afraid that the manual is correct and nothing changes once the DIP memory is installed... I still hope it's a mistake or oversight in the manual. However, I also tested all the traces to the DIP and simm sockets. Bank0 connects to both sockets and all connections seem fine - so, yes - it might not be the memory after all - that would be sad and I need to find out when else could be the issue.
I wish I had even a single scrap yard near me that allowed hardware to be bought and taken away. We don't even have a scrap yard near me, let alone one that allows hardware to be purchased, let alone one that has a trove of retro goodies such as this.
Try adding some RAM to the sockets, a lot of 8086 and 80286 will expect RAM to be present on the board itself, the slots will use space further up the memory map. This is a remnant of the IBM AT/XT era and the fact most systems cloned that spec, those systems expected at least 64K at a certain address in the map to even post.
The manual specifies that DIP and simm sockets share the memory banks. I got the DIP memory today. Unfortunately, I have no way of verifying that the memory chips work. Nevertheless, there is no change with DIP memory installed. The board still stops at POST code 10 - same as in the video 😟
Wow! I usually see the tantalums crack, but never a full jetison like that. I'm glad you caught it on camera! How does one go about finding these scrapyard finds? I live in Berlin but my skill with the German language is pretty bad and I never know what to look for.
I can't be much of a help for Germany. I live in the United Arab Emirates. The scrapyard I visit here is very lenient with people going through old hardware and buying them. Initially it's hard, but once you build a friendship, it gets a lot easier.
As someone who used pc's since the early 80's, Im looking forward to the DOS memory management videos. Sometimes it was a pain in the butt to get your new game or software to run due conventional memory, ems memory, himemsys, drivers, etc. But, when you got it finally working, it felt great!
About the 2 BIOS chips: 8086 and 80286 systems require 16bit EPROM but at that time only 8bit EPROMs were available, so they used 2 identical EPROMs (one filled with odd and the other with even bytes) from the BIOS program. Have in mind that 8088 based system such as the original PC and later PC XT have 8bit data bus and required only one 8 bit EPROM. Later in 80386 and later era there were actually 16bit EPROMs available however they were so expensive that chipset manufacturers decided to "emulate" 16bit EPROM using memory and the chipset and store the BIOS program in an ordinary but larger 8bit EPROM.
PS: The block of TTL chips you showed are 244/245 (tristate buffers/transcievers) and 373 (8-bit latches). While those aren't known to be fragile, one could have a stuck bit or something like that. Probably worth a try to take them off and test them separately. Another possibility is: The SIMM sockets are really close to where the battery was. Maybe some of the battery leakage got underneath the sockets and messed something up. Have you already tested continuity between the SIMM sockets, and from the SIMM sockets to the CPU/Chipset?
Yes, I did check continuity between DIP, SIMM, and Chipset. Unfortunately, all checks out. I'll try again if all other suggestions don't resolve the issue.
The reason it has two BIOS chips is that each BIOS chip is 8-bit wide, therefore dual chips make full 16-bits of this architecture. Before the times of large memory capacities it was important to mainain enough bandwidth of BIOS to reduce it's access time. Later, in 386 era, it was changed to cheaper solution - BIOS shadowing in RAM memory - where the BIOS contents was copied to RAM memory and its addresses remapped using 386's MMU. The 286 did not have MMU therefore it was not possible at all in such setups.
The 286 does have an MMU, as in, it is definitely capable of virtual addressing, but it doesn't have V86 mode. Which means its MMU can be only used in protected mode, which is not compatible with DOS applications. And to make it worse, once it enters protected mode, there is no way to leave this mode w/o making a full reset. All of these aren't a problem on 386 and above. Anyway, this is all irrelevant, because I'm fairly sure ROM shadowing is achieved by a hardware feature on the mainboard chipset, not by the CPUs MMU, by simply changing the address decoding, and setting if ROM or RAM is visible at that certain address. This is also nothing rocket science, and has been done many times even on 8 bit computers...
You *can* shadow BIOS using the 386 paging feature in virtual 8086 (V86) mode. This is what EMM386 does if you use the "ROM=F000-FFFF" switch. But this is not how it usually was done on 386/486 boards. Instead, the *chipset* can be switched into a mode in which it presents RAM instead of the ROM at the BIOS address, and it can optionally write-protect the RAM area that contains the ROM copy. It's impossible for the BIOS to use the processor MMU for BIOS shadowing, because there is no protocol for the BIOS to hand off control to a memory manager in an operating system. The operating system expects to be able to use the MMU anyway it likes. So if you find an option to enable shadowing inside the BIOS, you can be sure it's a chipset feature. In case the chipset does not support BIOS shadowing, you can use the MMU for memory remapping as well, but in that case, it's solely an operating system feature unrelated to any BIOS setting, e.g. using the EMM386 command line in MS-DOS.
@@Insomniator1 I don't lie. Please don't tell me I'm lying. Instead I just tell things that are not entirely true. I did not know it is done via chipset on x86 platform. On my native Amiga we use MMU to do this.
Unfortunately, this is a huge problem with old PC mainboards.. the isolating layer in the tantalum capacitors will degrade with years in storage, and at powerup the contacts will fuse together into a short. On the plus side, this makes it rather obvious which one of the capacitors is broken ;-) Also, don't bother retrofitting another old Tantalum on those boards, you'll just get the same problem again some time later. Pull the other tantalums off and replace them with some modern low-ESR electrolytics, those are safer.
Before I clicked on the video I thought to myself: I bet the tants weren't replaced before turning it on. I was not disappointed. Every time with this generation of boards.
Thank you 😊 - I hope we can fix this board - it seems to be a very interesting 286 with a lot of potential. Happy to hear you subbed and will be there along the ride!
I've heard tantalum caps could blow like that, but I'd only ever seen it when someone intentionally overvolted one. Interesting; glad none of those flying red hot bits hit you in the face!
38:48 after foreshadowing "These tantalums like to explode so here goes nothin'" and "I'd better wear my safety specs just in case" Then: "Well that was unexpected". Narrator: "It was not unexpected."
@@danpedersen55 Like ordering things from Aliexpress. You forget what you ordered by the time it arrives, then it's like a nice unexpected present from someone who knows what kind of ultra niche little thing you're in to.
Given how close the corrosion was to those memory sockets, I would look there, to start. Additionally, the chip that was affected near the beginning by the corrosion might have gone bad from the battery leak wicking its way into the package through the legs, so I'd see if there was some way to test that.
Great point regarding that once chip. I tested the memory sockets so much that I don't believe the issue to be there. But let me see if I can figure something out regarding that chip.
I got the DIP memory today, unfortunately, the board does exactly the same thing. Identical behavior with either DIP memory, SIMM modules, or no memory at all.
1:55 The chip next to the power connectors should be a Motorola 16818 iirc, it is the original DALLAS chip without battery and any external components like clock quartz...but combined with CMOS memory for the BIOS settings
It's a very nice motherboard, well worth the time to fix. Good job so far. BTW I got a little explosion of my own yesterday. I was attempting to repair a power supply that was dead (not from a computer, but from an embedded device) and unintentionally shorted 2 terminals that had voltage on them. Oops! :) It was only for a split second though, didn't even trip a fuse. And I don't think it caused any further damage.
Thanks... Yeah, I also blew up a few caps unintentionally. One was on a bread board power supply. I somehow messed up the wiring and blew up one of the caps. Electronics tend to teach you in an interesting way if you make a mistake 😅
@@bitsundbolts The only tantalum cap that ever blew on me was on an IBM 5160 motherboard - you guessed it, on the 12V rail. I replaced all of the 12V rail caps on that board with MLCC ceramic caps rated for 50V. Those never explode.
I don't know why I love those Harris CPUs so much, I even bought one off eBay even though all my (working) 286 machines already have CPUs. Maybe it has something to do with my job, since I work in the broadcast industry and Harris makes broadcast equipment (or at least used to, I think their broadcast division was carved out and sold of, and possibly rebranded out of existence since then). If I see corrosion near the keyboard connector, I usually just replace it. Usually when I'm buying parts off a seller from China, I take a peek in their eBay store and see if they've got other interesting stuff. At one stage I came across a listing for AT keyboard connectors, so I just threw a bunch of them in with my order, so now I've got plenty in stock. I also bought a bunch of PS/2 connectors, but that was specifically because I needed to replace one. For your board, my guess is there's pull-ups on the memory address lines, and the chipset has open drain outputs, so it would the lines would normally float high until the chipset actively pulled them low. That's why you're seeing them high most of the time, the chipset is only pulling them low when it's trying to access the memory. One trap for young players is checking the memory address and data lines on the pins of the connector, rather than the contact pads of the SIMM. There could be corrosion in or under the memory socket that could cause discontinuity between the SIMM and the board. In this case I'd suspect the problem could be elsewhere, since the corrosion looked pretty limited, but I'd still check it just in case.
Good point regarding the pull-ups and high lines. if the chipset can't access the memory correctly due to some other factors (e.g. faulty buffer/logic chips that are right next to the 40 MHz crystal), this could make the lines to remain high. I am testing those logic ICs now. Maybe I'll find something.
I think the two BIOS chips are there for faster access. 16-bit instead of 8-bit, which is why they are often labelled ODD/EVEN (data is interleaved - even and odd bits are there - respectively). However, a solution called Bios Shadow has marginalised this (and one chip is cheaper than two)...
I added an external battery, but that didn't help. I think I tested that the battery reaches the real-time clock, but now I think I should check that again. Thanks for reminding me!
@@bitsundboltsMaybe this board isn't so picky. By post card it just tested DMA, keyboard controller command byte, next will be pin 23,24 blocking unblocking and then it halts on keyboard input buffer free test and issuing BAT command to the keyboard controller.. Did You hit reset with battery inserted? Anyway, this board is very interesting. 😉
I'll try again very soon. Since I didn't get far, I got my hands on two more boards, similar to this one. I'll show them in an upcoming video very soon!
Interesting board, I hope you can make it work and look forward to the EMS video! I will never forget my first (and, I think, only) tantalum explosion. It was an Asus Media Bus card and I was unaware of the explosive nature of these capacitors, so my face was closer to the card that would have been otherwise and I was completely unprepared for the ensuing fireworks xD.
By the way, I think most 286s I have personally seen were Harris and Siemens. I wonder if it was because they were "modern" (1989-91, with the Harris...es? being usually of high frequencies that I think Intel didn't make) and/or a regional thing (I'm from Argentina).
Good that nothing has happened to you when that cap exploded. They are quite dangerous if you are too close and one goes off. Unpredictable! I'll replace them if I get the board to POST...
That's a pretty orderly tantalum blowout. Couldn't have gone any better. I had a PSU on a Dell switch blow up once, where the tantalum cap was. I tore the freakin' traces. Since they fuse and form shorts, they almost always take something else with them. I have a running early build 486 full of tantalums and I an very weary of using it. The moment it blows up, I'm not even going to fix it, but will just go a newer blue socket motherboard, that only has electrolytics.
My favourite component is that crystal in the middle. It seems like they forgot about the crystal, so they deleted a slot in the middle and slapped the crystal on the board. Many motherboards are missing the upper half of the ISA slot, but they usually do not place components over it, just simple don't solder the header.
There is a lack of flux when de-soldering old joints.. try adding a drop of new flux or adding a touch of fresh resin core solder first to aid the flow into the desoldering wick
The memory you showed has a 7 in the end of it's type numbers, suggesting it is 70ns. You would need 60ns according to the manual. The memory that was in there is likely 20 years old, which may make it not as fast as it was when new. It might have worked fine back then, running out of spec, but it will be too slow to run now? You could swap the Xtal for a 32MHz to test if the board will boot at 16MHz before going for 20MHz again. Did you check how stable the voltages are? Those tantalums might not explode on you anymore, but how good are they? I have had my fair share of boards being finicky about what memory modules worked when DIP and 8bit SIMMs were normal and it would not be uncommon to get a machine in under warranty that had memory fail that would work fine in another less critical machine.
That's good old Schrödinger's tantalums. They are simultaneously fine and not-fine until you power them on. The failure is initially not electrical, is not detectable by any measurements known to mankind.
Harris Semiconductor manufactured the HD-6120 12-bit microprocessor (PDP-8 on-a-chip) used in DEC's DECmate series of microcomputers (that followed the PDP-8 series of 12-bit minicomputers and the VT78 [which used the Intersil 12-bit IM6100 PDP-8 on-a-chip]. I've got several examples of all of these, but not the complete series of PDP-8s & DECmates. Also, I had to replace a power supply in an x86 (maybe a 486?) PC, but since the "new" power switch had the same colored wires, I got lazy and didn't perform the more difficult task of changing the power switch too. When I "finished," I plugged the PC in & turned it on. I blew a tantalum cap off of the board and it looked like a small bottle rocket leaving a smoke trail as it shot towards the bookshelf! I immediately turned the PC off and and changed the power switch too (the one in the face of the desktop PC). I went ahead and turned the PC on again and had no problems, even without replacing the blown cap. Just lucky I guess! That PC is still out in the garage and may get used again in the future.
One of the reasons I think these tantalum capacitors pop is that they weren't spec'd correctly back then. After a similar failure of a new tantalum, I found a number of empirical evidence posts stating that tantalums should have double the voltage they're expected to see. In your case, those are rated for 16V but are attached to the 12V rail. This is *way too close* to their max rating and any fluctuation could pop them again. The damage is cumulative and chances of failure increase with age. The voltage rating doesn't have to match the original spec, only the capacitance is important. I'd look for 24V+ tantalums as suitable replacements or you're rolling the dice with every power up.
You are absolutely correct! The tantalum cap that exploded is indeed connected to the 12V rail! That might have been the reason, in combination with age, it exploded. Maybe it would be a good idea to preventively replace the cap around the power connector - at least the ones connecting to 12V. I have seen many times that those are the caps that blow up.
I remember that Necroware had a video some time ago with a motherboard that had a problem of being always in turbo mode (slow speed). He had to replace the chip, responsible for generating a turbo signal. The video is called "Turbo in a retro PC - explanation and repair" on his channel, maybe it will help!
Many years ago. Must be 15 yrs by now. I was sitting at my desk. I suddenly heard a loud BANG! It scared the crap out of me at the time. The noise was from a video card I had laying on a table beside my desk. One of the capacitors had spontaneously burst. I think it was an electrolytic though not tantalum. I didn't have the ability to replace it and I'm sure the card went to e-waste when I chucked out a bunch of stuff a while ago so I don't know exactly the model. At the time I was messing with Geforce2 cards though and I also had a Goodwill Computer Works nearby (they resold donated computers) which had a bevy of cheap computer parts that I often would grab if I needed a part. So it might have been a card I bought from them. Either way your fireworks in this video were great ! ;)
@@bitsundbolts Yes. I never knew they could burst w/o power either. That was the strange thing. I am not even sure it had been plugged in for months before it blew out either.
@@bitsundbolts It's been a long time but I want to say it was an electrolytic and it burst at the top. No rhythm nor reason that I can remember either. Just poof and as I said it was a loud rapport like a firecracker at close range. I've messed with large, old multi-section can capacitors from tube circuits. I used to purchase old gear to modifiy into guitar amps or salvage the parts. So I've worked with some bad caps, but never had one burst without power. I did blow up a can once. I'd changed a power supply from a choke>cap (choke input) filter to cap>choke> cap (capacitor input) unaware that the power transformer was not able to supply enough current w/o the choke. The result was the can burst violently while under full power. Not to mention it was a New-Old Stock part of questionable quality. Even that didn't sound like a firecracker though. It was more like frying bacon since the electolyte boiled and exceeded the pressure rating of the can. There was no safety vent on this type from the 1960s.
Base 64K is on the high addresses for bios shadowing. So, the stable address high may not be a problem. Chck the connections, most likely there is corroded trace/via.
No, "base 64K" is not used for BIOS shadowing, which is optional (if implemented at all) in this kind of board. They used two BIOS chips to provide full 16 bit bus width, reducing the need to shadow the ROM for performance, although the ROM is surely slower than DRAM at 20MHz, so shadowing still would help. Base 64K indeed refers to the lowest 64K of memory, which needs to be operational for the POST to start. The first kilobyte is used for interrupt vectors, the subsequent 256 bytes are used for BIOS variables, and some other part of the base 64K is used as stack memory. Lastly, the memory between 31K and 32K is used to load the boot sector or MBR to start a disk-based operating system.
@@tw11tube Ok, my fault. Bios itself is located on 0xF0000 - 0xFFFFF. So, if you try to observe bus in this state, you can see only the top addresses. CPU and memory address / data bus is directly connected in this platform, right? So, it make sense to just measure pin to pin continuity.
@@petr56321313 As the system is beeping, we can assume that accessing the BIOS works fine. Address lines are most likely not connected straight from the CPU to the bus, because the 286 stops driving the address lines for the current cycle already before the cycle is complete, so there needs to be address latches on the board. Data is likely connected straight through. It makes sense that you see A16..A19 high most of the time, as these pins need to be high when addressing the BIOS. You will see the lines low for a short moment before beeping starts, becasue that's when the BIOS tests the first 64K. When the 286 starts, it starts execution at FFFFF0 (16M - 16), so the BIOS must be visible there (at least the last 16 bytes), which contains a far jump, which in real mode will cause continuation at 0Fxxxx, so just below 1M. This means we expect A20..A23 to be high for a very short time after reset, and then stay low until the memory test reaches extended memory.
I have a 286 too that would not work at all without proper FPM installed. regular sim would not work. I actually got a second 286 board which seem to have on board ram but it errors out. I can somewhat run it with only base memory. 286 in particular seems to be very iffy sometimes
I have an idea. Since keyboard connectors seem to get the brunt of battery corrosion, it might be helpful to wire up a new connector to flying leads and solder the leads to the board for testing. If the board doesn't work you don't have to try and desolder the connector too many times and damage it.
Harris is actually a US military contractor and a lot of their parts are "milspec", with higher thermal tolerances and lower power consumption. They're usually manufactured later years and a process node or two later than their Intel counterparts.
capacitors sure are fun, aren’t they? honestly, after learning how tantalum caps are made I’m kind of impressed they work at all. ditto electrolytic caps in general, I suppose, but tantalum in particular just seems bonkers with that whole lump o’ microscopically spongey metal with a whiff of oxide around whatever gnarly shape the tantalum happened to sinter into thing they’ve got going on
@@bitsundbolts Understandable! I think it might be wise to simply replace the tantalums on the voltage lines before powering up an old motherboard like that. Of course, the video might be a bit less exciting ;)
If you did not find out already on how to check the operation of the octal buffer: find the LE pin (high is latch in) inputs and outputs on 74xx373 parts and use an logic analyser or at minimum two channel scope (direct/continuous sampling as in: without averaging or other software filters) to verify the data is latches in properly.
I'll have a look. Using a two channel scope is still above my abilities, but I'll have a look the datasheets of those chips and try to understand how they work. Maybe my programmer has the ability to test those logic chips - I haven't checked that yet.
I did measure before powering on. No shorts at that time. The cap must have failed with the inrush of 12v after such a long time. I guess from now on, I'll pay attention to the cap on the 12v line. This seems to be the one that most likely will explode.
I got the same Board some years ago. It is not working and i hope i have it restaurated one day 😅 I think it is a good idea to make some videos about dos memory management😊 I like all of your videos, thank you for that interesting stuff.
Hey i remember watching the talk about capacitors by James the Bald Engineer... himself a classic computer restorer and enthusiast, who showed among other things why tantalums explode, or why new polymer ones don't. So what i took home is that after thermal stress or prolonged storage, one should cure tantalums by gradually bringing them once up to voltage, gradually raising the voltage over minutes, then they'll no longer explode, even crappy old ones. So yeah what if one did that type of voltage ramp rig, while holding the whole mainboard artificially in reset in order for it not to go all weird?
I feel your pain :( I went through the same thing a few months ago, a socket 7 motherboard exploded and took with it everything connected to it, sound card, video card and ISA capture card, network card, memories, processor and power supply. There was a short circuit on the motherboard and the AT power supply I was using had no protection circuit. Since that day, I have never tested a computer in the dark again, I created an adapter that sits between the power supply and the motherboard, any surge of voltage or current will trip it in milliseconds. There are a thousand different projects on how to do this, and it's very simple, in fact, you can order everything from PCB Way already assembled.
I'm using a modern power supply from EVGA (supernova G3). It should have all the protection mechanisms, but I did get the occasional surprises like this one. One time however, the power supply did shut off. That was when I built one of the -5v ATX Adapters which had a bad cap.
Short from where to where? Short any rail to ground wouldn't actually harm any attached hardware; except via crossload and the PSU being a piece of garbage that would have killed the PC anyway. This is why i really don't like most old PSUs except a handful trusted manufacturers.
@@SianaGearz I believe it was an IC regulator that completely shorted out, everything on the 5v line simply exploded. Perhaps the battery or a capacitor has leaked, causing the short. The computer was an Itautec RTV A96, and used an Intel AT motherboard with some onboard stuff, like an ATI Mach 64 video, a somewhat unusual motherboard.
@@RicardoRamosRetrocomputacao Yeah nah if you have explosions along the 5V line, it means the 5V line drifted well above nominal. The regulators off the 5V line would be supplying low voltage rails, whatever CPU and specific IC supplies, so a regulator has 5V input, low voltage output, and feedback, correspondingly all regulators along 5V line would worst case, if failing, pass the input through, and there really aren't a lot of them. So i'm thinking your PSU killed the hardware, not the mainboard. If the PSU doesn't have crossload protection (individual rail under/over voltage sensing), then a near-short of 12V line to ground will easily cause 5V line to drift up, while the 12V line dips. Also quite common for shitty ones is a magnetic failure causing the 5V line to be internally shorted to 12V line in the PSU.
Fun with Tantalums! Happened several times on my Amigas that the old buggers simply blew up and left marks on the PCB. Happened most of the times after switching to another power supply. My theory is, that the aging Tantalums were so in tune with the aging caps inside the old power supply that they simply were not able to withstand the power of a new or more recent power supply. Best would be to replace all of them with modern MLCC ceramic caps. But I shouldn´t be the person to say that. MLCC replacements for all my old hardware is on my list since many years but I am just too lazy to start. So I only replace Tantalums in case one turns into fire. ^^
Tants need to be significantly over-rated on voltage, machines of this era didn't do that. 16v caps on a 12v line, which can easily over-volt as it is typically not as well regulated. They age out and fail short circuit, just like MLCCs (in fact I've had way more MLCCs fail short in Amigas I repair for people, but they are used a lot more). It is better to replace with modern electrolytics for most of the use cases, although MLCC probably won't do any harm.
Hunt down an Intel 287XL coprocessor for that thing. It's basically a 387 core in a DIP 40 package compatible with the 286 bus. Its one of the fastest coprocessors out there (not the fastest, that would be a 20 mhz Cyrix FasMath or one of the Harris parts) while still being really common and not expensive. Since it's a 387 at the core, it can usually be overclocked to a really great degree, so you might be able to run it synchronously with the CPU, it might require some tweaking of the coprocessor speed jumpers, though. The 287XL has a pin that toggles an internal clock multiplier meant to adjust for the 2/3 divider that a lot of boards used for the original 287, so experimentation is necessary. It's also a lovely looking gold topped ceramic package, so there's also that. Regarding the Bios, there's a very nice MR-BIOS available in the VOGONS MR-BIOS archive. It supports things like larger HDDs and more than two floppy drives.
I hope you're right! It says differently in the manual, but it would be the first time there is a mistake in one or the wording is chosen poorly and ends up in confusion.
@@bitsundbolts Wonder if it may not be working because some of the other caps are also shorted? Maybe go through and check each of them for shorts until you replace them?
my first introduction was a Zenith EZ PC. According to wiki it's a 286 at 7 mhz made by NEC. My first computer that I actually had permanent access to was a compaq 386. Then I swapped the two (I was not happy about this) and eventually I ended up with both.
wow intentionally designed to prevent the instalation of a hard drive unless you had the EZ PC 3. And I do remember that it had no ISA slots. They were jerks!
Thank you for the great video. I would prior test if the resistance between the right contact (+5) and ground (black) of AT power switch on the mobo is appropriate. Think maybe this system board will be hard to repair
3-2-1 you're going to fix it soon! I can see that! I hope it's not the chipset, that 16MHz to the CPU is indeed suspicious. If I remember right we checked on my 286 (different board) and the speed seems to go to 20MHz straight away. Fingers crossed!
Yes, you're right! I'm looking at the logic chips now. I also socketed the 40 MHz crystal and tried one with 33 MHz - unfortunately, there was no change. Tomorrow, I'll continue with the logic ICs.
6:37 This is what i often see in my own barrel battery boards, if I continue to brush, I continue to see bubbles and it seems to never stop. Does this mean there is still hidden battery leakage? Even though the PCB part looks fine like yours, it's strange.
The manual may be correct for 20mhz cpu. I'd try putting a matching crystal for the DRAM speed you have. To see if the dram is just too slow (given you got memory errors).
I tried a 33 MHz crystal, but no change. I'm in the process of debugging further right now. I may want to try an even slower crystal if I find one. But with 33 MHz, the board behaves exactly the same. I also found 60ns memory - but no change.
I got the memory about an hour ago. Unfortunately, there's no change - same behavior as with the SIMM sockets - or no memory at all. The board behaves identically no matter what. I also got new BIOS chips which work properly - still no change.
@@bitsundbolts Thats unfortunate. You mentioned in the video that there were data/address lines acting weird? Maybe try desoldering the CPU and see if the issue persists. If it does then chipset may be bad.
I have a BIOSTAR , MB-1220VE Harris 286 20 MHz the board had minor battery damage the current memory installed is 16 Meg ram most of it is used as a ram drive
I'll wait for the memory first and see if it makes a difference. Next step would be probing around the board and see if I can find something suspicious.
You discovered the light emitting capacitor :D I do love it when the faulty part self-identifies though!
Haha, well, it identified itself. It didn't hold back even a little bit.
There’s something incredibly satisfying about watching solder braid up close as it absorbs the solder. It’s just so mesmerizing to see
Funny, I thought I was the only one mesmerized by that. Especially in slow motion.
Wow at that explosion!! I suspect you need the DIP RAM installed, to give it base memory. Chances are the SIMMs are for extended memory. Some of the older 286's were funny like that! I realise some of the manuals that support that board show either the DIP RAM, or SIMMs... It does show x 9 type though, so needs parity by the looks of it.
I hope you are right! DIP memory is on the way. So far, I wasn't able to find anything else suspicious about the board and why it wouldn't continue to boot. The memory should arrive next week and I hope it will fix that issue with the memory. Fingers crossed!
Ja 286 er sind wählerisch was RAM angeht
I have the same motherboard and I'm using only 2 SIMM modules, no DIP chips installed, and it's working fine.
@@BaguetesGarage Hmmm, maybe he has a problem with SIMM connectivity then, or it's because it lacks parity or the SIMMs are somehow incompatible?
I have another 286 board with DIP+SIMM configuration and it runs fine on 4 x 1 MB SIMMs without any DIPs. But my board can have a different implementation.
'sacki zement!' wow, I haven't heard that swear since my grandmother (born in austro-hungaria!) passed in 1998.
I had never heard that before... but I think I want to start using it! 🤣
Well, ... instant detour to memory lane. In my case to my grandpa in rural Swabia, God Bless his soul
It's a pretty mild swear.
@@ninjamaster3453 Yeah, it's designed to sound like "Sakrament", which in itself would be considered very mild today.
Great work with the restoration! Octek makes some really nice boards. I had a tantalum cap explode on a sound card a couple years ago. It was totally unexpected and was very startling. The worst part though was the smell. It lingered for hours even with the window open and fan going! .. also so awesome to see two of my favourite TH-camrs troubleshooting together 😁
Thanks! Surprisingly, the tantalum from this board didn't leave a lot of bad smell behind - it was gone in seconds. Maybe it was because it exploded so violently (it was the 12v line after all).
I exchange thoughts with Tony often - I learn a lot from his analytics and debugging skills. His videos are always entertaining and educational!
"Sackl Zement" 😆
What a nice expression (which i havent heard for a long time, although i live in Bavaria...).
Nice video and work! Thank You!
I am learning German, and I had no idea a sack of cement could be used as an expression of contempt.
@@alisharifian535actually it stems from „Sakrament nochmal“ (a swear with catholic origin) and was considered rude / blasphemy. Hence, „Sackl Zement nochmal“ is used instead because it sounds very similar.
💯
Bavarian language has more curses than thlIngan Hol.
@@aracon22 himmi herrgott saggrament (heaven, god, sacrament).. usually followed up by greiz gruzifix (cross crucifix), to pile some more blasphemy on the unfortunate subject. We Bavarians really love our religion-based insults and curses. 🙂
the 20MHz Harris made 286 processors are the best 286 processors. They are very reliable and run cooler than the other 12MHz 286 cpu-s. You have a jackpot on that board, my friend.
Well, let's hope it is repairable... Or just a configuration that isn't right.
Considering that I get POST codes, I guess at least the CPU is ok.
@@bitsundbolts
I have two working 286 motherboards with cpus & ram if you need anything. Adding ram from that era is definitely PITA 😂
I'm working on the board again. I am checking all those ICs near the 40 MHz crystal. Maybe I can find something odd or faulty.
I had a 20MHz 286 as my first PC and it was giving my friend's low end 386 a run for its money. They felt nearly equal in power, except his could run Doom and mine couldn't. I miss that machine so much. Father sold it off and I was too young to memorize the specs. I would've rebuilt it, but I don't even remember what I had (in the 80s').
@@enilenis clock for clock, in 16bit software, the 286 was a hair faster than 386 cpus.
Must just be a good day for bad caps on 286 boards...I popped one this morning! After replacing a shorted one that I caught with the multimeter, I plugged it in, and the test supply found one I'd missed on the +5V rail.
I've heard so many bad things about those tantalum beads from the 80s and earlier that I'd be tempted to just replace all of them. They were mainly used because they were very compact and had tight tolerances but modern high-quality electrolytics have much tighter tolerances than the ones back in the day had, so often enough they're a decent replacement for tantalums from that era, or so I've been told.
@@Ragnar8504 Tantalums have different performance characteristics than aluminum electrolytics, so aluminum electrolytics are not always a good substitute. Low ESR and especially polymer electrolytics are often suitable replacements; however, not all vintage tantalums are problems. It seems to be mostly an issue of storage conditions: damp conditions seem to accelerate tantalum crystal growth, which is what causes shorts. Boards stored in damp basements, or worse, from scrapheaps, have higher tantalum failure rates than boards store in climate controlled shops and living spaces.
Harris Semiconductor were a pretty decent chip producer in the 80s and 90s. They had a strong presence with specialist parts for avionics and similar high-end applications. Coincidentally, my first PC also had a Harris 80286, but only at 10MHz. I also had a 10MHz Intel 80287 in it, which was a big help for the PCB CAD software that I used at the time. As for the tantalum beads... I designed lots of boards that used them at that time. Back then, they weren't old, so they mostly only went bang if soldered in backwards, but still with fireworks 😂. Also, you're right about the RAM chips being for the system RAM. That board is marked for 256k x 4 chips (44256) and is laid out for either dual-in-line or Zig-zag packages. It maybe that it always needs populating regardless of any expansion SIMM modules. Edit: I see that the user guide says that using only SIMMs is a perfectly valid option, so that's not the issue.
I'm afraid that the manual is correct and nothing changes once the DIP memory is installed... I still hope it's a mistake or oversight in the manual.
However, I also tested all the traces to the DIP and simm sockets. Bank0 connects to both sockets and all connections seem fine - so, yes - it might not be the memory after all - that would be sad and I need to find out when else could be the issue.
I wish I had even a single scrap yard near me that allowed hardware to be bought and taken away. We don't even have a scrap yard near me, let alone one that allows hardware to be purchased, let alone one that has a trove of retro goodies such as this.
Man, *yes*
Ultimate bummer. At least we don't have a scrapyard that disallows purchase, forcing you to see the goodies go to absolute waste
This sounds like a perfect video for a YT collab! I'm sure Adrian Black would be keen to get involved too
I wouldn't mind some extra help. Maybe he finds this video and watches it...
Try adding some RAM to the sockets, a lot of 8086 and 80286 will expect RAM to be present on the board itself, the slots will use space further up the memory map. This is a remnant of the IBM AT/XT era and the fact most systems cloned that spec, those systems expected at least 64K at a certain address in the map to even post.
I am also quite sure the dip sockets have to be populated to form the base memory. The simm modules are higher up on the map.
The manual specifies that DIP and simm sockets share the memory banks.
I got the DIP memory today. Unfortunately, I have no way of verifying that the memory chips work. Nevertheless, there is no change with DIP memory installed. The board still stops at POST code 10 - same as in the video 😟
@@bitsundbolts Try to check the manual from the post analiser card itself, it should give you some clue, somewhere to aim...
Good point, I'll check if there are more hints.
@@bitsundbolts Fair enough, of course you can check this. If the DIPs and SIPs are shared then you should be able to tone out from a socket to a slot.
Cute little motherboard. The fireworks display was great 👍. I hope you get it sorted out.
I hope so too! Thanks
Wow! I usually see the tantalums crack, but never a full jetison like that. I'm glad you caught it on camera!
How does one go about finding these scrapyard finds? I live in Berlin but my skill with the German language is pretty bad and I never know what to look for.
I can't be much of a help for Germany. I live in the United Arab Emirates. The scrapyard I visit here is very lenient with people going through old hardware and buying them. Initially it's hard, but once you build a friendship, it gets a lot easier.
As someone who used pc's since the early 80's, Im looking forward to the DOS memory management videos. Sometimes it was a pain in the butt to get your new game or software to run due conventional memory, ems memory, himemsys, drivers, etc. But, when you got it finally working, it felt great!
im living that struggle right now trying to get enough ems memory to play dungeon hack with sound effects
Some went so far as to set up boot disks with the requisite configurations for certain games in order to make the most out of available RAM.
@@blakegriplingph
Secret Weapons of the Luftwaff went even further. My DOS boot disk had to be ver5.0. Setver was ignored somehow.
Watching these, I wish there was a local scrap dealer here to go looking.
Scrap hunting is very exciting! You never know what you'll find! Maybe you can find some place close to your area.
Tony's channel is how I found out my audio interface has really bad crosstalk 😭
Uhh, don't tell me that Tony ruined your Audio Interface for you after you found out...
About the 2 BIOS chips: 8086 and 80286 systems require 16bit EPROM but at that time only 8bit EPROMs were available, so they used 2 identical EPROMs (one filled with odd and the other with even bytes) from the BIOS program. Have in mind that 8088 based system such as the original PC and later PC XT have 8bit data bus and required only one 8 bit EPROM. Later in 80386 and later era there were actually 16bit EPROMs available however they were so expensive that chipset manufacturers decided to "emulate" 16bit EPROM using memory and the chipset and store the BIOS program in an ordinary but larger 8bit EPROM.
"Go to 38:15 if you want to see the explosion."
Bro, for that alone, you've earned my sub, Like, and interest, heh.
I miss videos from CPU Galaxy, hope he is allright!
Thank you very much for your videos, I watch every one of them, have a great day!
I must say, your soldering is almost perfect. Really beautiful work, love seeing it.
PS: The block of TTL chips you showed are 244/245 (tristate buffers/transcievers) and 373 (8-bit latches). While those aren't known to be fragile, one could have a stuck bit or something like that.
Probably worth a try to take them off and test them separately.
Another possibility is:
The SIMM sockets are really close to where the battery was. Maybe some of the battery leakage got underneath the sockets and messed something up. Have you already tested continuity between the SIMM sockets, and from the SIMM sockets to the CPU/Chipset?
Yes, I did check continuity between DIP, SIMM, and Chipset. Unfortunately, all checks out. I'll try again if all other suggestions don't resolve the issue.
The reason it has two BIOS chips is that each BIOS chip is 8-bit wide, therefore dual chips make full 16-bits of this architecture. Before the times of large memory capacities it was important to mainain enough bandwidth of BIOS to reduce it's access time. Later, in 386 era, it was changed to cheaper solution - BIOS shadowing in RAM memory - where the BIOS contents was copied to RAM memory and its addresses remapped using 386's MMU. The 286 did not have MMU therefore it was not possible at all in such setups.
The 286 does have an MMU, as in, it is definitely capable of virtual addressing, but it doesn't have V86 mode. Which means its MMU can be only used in protected mode, which is not compatible with DOS applications. And to make it worse, once it enters protected mode, there is no way to leave this mode w/o making a full reset. All of these aren't a problem on 386 and above. Anyway, this is all irrelevant, because I'm fairly sure ROM shadowing is achieved by a hardware feature on the mainboard chipset, not by the CPUs MMU, by simply changing the address decoding, and setting if ROM or RAM is visible at that certain address. This is also nothing rocket science, and has been done many times even on 8 bit computers...
@@chainq68k Thank you for correcting me on that, I wasn't aware. Anyway it still means basically the same.
You *can* shadow BIOS using the 386 paging feature in virtual 8086 (V86) mode. This is what EMM386 does if you use the "ROM=F000-FFFF" switch. But this is not how it usually was done on 386/486 boards. Instead, the *chipset* can be switched into a mode in which it presents RAM instead of the ROM at the BIOS address, and it can optionally write-protect the RAM area that contains the ROM copy.
It's impossible for the BIOS to use the processor MMU for BIOS shadowing, because there is no protocol for the BIOS to hand off control to a memory manager in an operating system. The operating system expects to be able to use the MMU anyway it likes. So if you find an option to enable shadowing inside the BIOS, you can be sure it's a chipset feature. In case the chipset does not support BIOS shadowing, you can use the MMU for memory remapping as well, but in that case, it's solely an operating system feature unrelated to any BIOS setting, e.g. using the EMM386 command line in MS-DOS.
Don't lie, shadowing is chipset feature, not CPU MMU !
@@Insomniator1 I don't lie. Please don't tell me I'm lying. Instead I just tell things that are not entirely true. I did not know it is done via chipset on x86 platform. On my native Amiga we use MMU to do this.
I have an Octek Fox II and all I can say it's a very nice board, it won't be a letdown if you can repair it. So, fingers very much crossed!
Oh, I hope I'll be able to figure out what is wrong with that board. I'm a bit jealous that you have a working Fox II 😅
Unfortunately, this is a huge problem with old PC mainboards.. the isolating layer in the tantalum capacitors will degrade with years in storage, and at powerup the contacts will fuse together into a short. On the plus side, this makes it rather obvious which one of the capacitors is broken ;-)
Also, don't bother retrofitting another old Tantalum on those boards, you'll just get the same problem again some time later. Pull the other tantalums off and replace them with some modern low-ESR electrolytics, those are safer.
If I get the board to work, I'll replace those caps and put electrolytic capacitors on it.
exactly thats what i do
last 286 board i restored was about 6 years ago been trying to get a pdp-11 for myself to restore
Before I clicked on the video I thought to myself: I bet the tants weren't replaced before turning it on. I was not disappointed. Every time with this generation of boards.
At least we got a nice frame from that video 😅
That is so cool how you see the cooper wicking strip turn silver, and the silver colored solder expands as it is sucked up.
It's worth checking tge chipset IC for corroded/dry solder joints, particularly in corners.
Des "Sackl Zement" kam unerwoated 😂👀
What happened to Harris? Harris 286s were everywhere.
Best wishes.
so.. the explosion in the thumbnail did grab my attention.. but the mystery of why this is not working made me subscribe.. :D
Thank you 😊 - I hope we can fix this board - it seems to be a very interesting 286 with a lot of potential. Happy to hear you subbed and will be there along the ride!
probably the most explosive one of those I've seen yet wo what a bang
So glad they reattached your fingers after that explosive cap
It's a motherboard, not a pager. 💥
I took all the precautions necessary to not be dismembered or harmed. Safety first!!
286 chipsets typically require the bios to be 16bit. Since (E)(E)PROM chips are typically 8bit wide, you needed two of them.
I've heard tantalum caps could blow like that, but I'd only ever seen it when someone intentionally overvolted one. Interesting; glad none of those flying red hot bits hit you in the face!
I was prepared and expected this - just not that it would explode this violently! I was also wearing safety goggles to be sure nothing hits my eyes.
38:48 after foreshadowing "These tantalums like to explode so here goes nothin'" and "I'd better wear my safety specs just in case"
Then: "Well that was unexpected".
Narrator: "It was not unexpected."
I was surprised by the violence of the explosion - not that one did. But yes, I guess it sounded like I was surprised that it did explode.
I like to set myself up to a surprise, if something does not explode, but still get a shock if something does 🫨
@@danpedersen55 Like ordering things from Aliexpress. You forget what you ordered by the time it arrives, then it's like a nice unexpected present from someone who knows what kind of ultra niche little thing you're in to.
Given how close the corrosion was to those memory sockets, I would look there, to start. Additionally, the chip that was affected near the beginning by the corrosion might have gone bad from the battery leak wicking its way into the package through the legs, so I'd see if there was some way to test that.
Great point regarding that once chip. I tested the memory sockets so much that I don't believe the issue to be there. But let me see if I can figure something out regarding that chip.
I think I recall that base DIP mem needs to be populated on 286 boards. SIM slots may be only memory expansion.
I got the DIP memory today, unfortunately, the board does exactly the same thing. Identical behavior with either DIP memory, SIMM modules, or no memory at all.
I have the same board with the same fault. Looks like I'll be getting it out tonight!
Wow! Interesting... The fact that it may be the same fault could mean that there is a solution... Hopefully 🤞
Nice Firework! :) I never boot up my IBM AT again. Thank you :)
Rapid unscheduled disassembly. Nice.
I love watching your solder, very talented
Thanks 👍🙏
Came for the explosion, stayed for the explosion
Hah, that happened to an old 386 board I had - turned it on, and POP! One of the caps blew. Scared the shit out of me. Edit - 6:27 haha EXACTLY that!
1:55 The chip next to the power connectors should be a Motorola 16818 iirc, it is the original DALLAS chip without battery and any external components like clock quartz...but combined with CMOS memory for the BIOS settings
It's a very nice motherboard, well worth the time to fix. Good job so far.
BTW I got a little explosion of my own yesterday. I was attempting to repair a power supply that was dead (not from a computer, but from an embedded device) and unintentionally shorted 2 terminals that had voltage on them. Oops! :) It was only for a split second though, didn't even trip a fuse. And I don't think it caused any further damage.
Thanks... Yeah, I also blew up a few caps unintentionally. One was on a bread board power supply. I somehow messed up the wiring and blew up one of the caps. Electronics tend to teach you in an interesting way if you make a mistake 😅
@@bitsundbolts The only tantalum cap that ever blew on me was on an IBM 5160 motherboard - you guessed it, on the 12V rail. I replaced all of the 12V rail caps on that board with MLCC ceramic caps rated for 50V. Those never explode.
I don't know why I love those Harris CPUs so much, I even bought one off eBay even though all my (working) 286 machines already have CPUs. Maybe it has something to do with my job, since I work in the broadcast industry and Harris makes broadcast equipment (or at least used to, I think their broadcast division was carved out and sold of, and possibly rebranded out of existence since then).
If I see corrosion near the keyboard connector, I usually just replace it. Usually when I'm buying parts off a seller from China, I take a peek in their eBay store and see if they've got other interesting stuff. At one stage I came across a listing for AT keyboard connectors, so I just threw a bunch of them in with my order, so now I've got plenty in stock. I also bought a bunch of PS/2 connectors, but that was specifically because I needed to replace one.
For your board, my guess is there's pull-ups on the memory address lines, and the chipset has open drain outputs, so it would the lines would normally float high until the chipset actively pulled them low. That's why you're seeing them high most of the time, the chipset is only pulling them low when it's trying to access the memory. One trap for young players is checking the memory address and data lines on the pins of the connector, rather than the contact pads of the SIMM. There could be corrosion in or under the memory socket that could cause discontinuity between the SIMM and the board. In this case I'd suspect the problem could be elsewhere, since the corrosion looked pretty limited, but I'd still check it just in case.
Good point regarding the pull-ups and high lines. if the chipset can't access the memory correctly due to some other factors (e.g. faulty buffer/logic chips that are right next to the 40 MHz crystal), this could make the lines to remain high. I am testing those logic ICs now. Maybe I'll find something.
I think the two BIOS chips are there for faster access. 16-bit instead of 8-bit, which is why they are often labelled ODD/EVEN (data is interleaved - even and odd bits are there - respectively). However, a solution called Bios Shadow has marginalised this (and one chip is cheaper than two)...
Great video and excellent job!
For posting issue fixing, try to add battery.. One time I got 386 board, that wouldn't post without it.. 😉
I added an external battery, but that didn't help. I think I tested that the battery reaches the real-time clock, but now I think I should check that again. Thanks for reminding me!
@@bitsundboltsMaybe this board isn't so picky. By post card it just tested DMA, keyboard controller command byte, next will be pin 23,24 blocking unblocking and then it halts on keyboard input buffer free test and issuing BAT command to the keyboard controller..
Did You hit reset with battery inserted?
Anyway, this board is very interesting. 😉
I'll try again very soon. Since I didn't get far, I got my hands on two more boards, similar to this one. I'll show them in an upcoming video very soon!
Exploding Tantalums are like a rite of passage for Retro PC TH-camrs.😉
Interesting board, I hope you can make it work and look forward to the EMS video!
I will never forget my first (and, I think, only) tantalum explosion. It was an Asus Media Bus card and I was unaware of the explosive nature of these capacitors, so my face was closer to the card that would have been otherwise and I was completely unprepared for the ensuing fireworks xD.
By the way, I think most 286s I have personally seen were Harris and Siemens. I wonder if it was because they were "modern" (1989-91, with the Harris...es? being usually of high frequencies that I think Intel didn't make) and/or a regional thing (I'm from Argentina).
Good that nothing has happened to you when that cap exploded. They are quite dangerous if you are too close and one goes off. Unpredictable! I'll replace them if I get the board to POST...
That's a pretty orderly tantalum blowout. Couldn't have gone any better. I had a PSU on a Dell switch blow up once, where the tantalum cap was. I tore the freakin' traces. Since they fuse and form shorts, they almost always take something else with them. I have a running early build 486 full of tantalums and I an very weary of using it. The moment it blows up, I'm not even going to fix it, but will just go a newer blue socket motherboard, that only has electrolytics.
Yes, those tantalum capacitors are a constant threat. I'll replace the caps once I get the board working. Hopefully I can repair it.
My favourite component is that crystal in the middle. It seems like they forgot about the crystal, so they deleted a slot in the middle and slapped the crystal on the board. Many motherboards are missing the upper half of the ISA slot, but they usually do not place components over it, just simple don't solder the header.
Yes, it is a rather odd place for that crystal. A 16 MHz 286 wouldn't need it because the CPU and ISA bus could be derived from the 32 MHz crystal.
There is a lack of flux when de-soldering old joints.. try adding a drop of new flux or adding a touch of fresh resin core solder first to aid the flow into the desoldering wick
The memory you showed has a 7 in the end of it's type numbers, suggesting it is 70ns. You would need 60ns according to the manual. The memory that was in there is likely 20 years old, which may make it not as fast as it was when new. It might have worked fine back then, running out of spec, but it will be too slow to run now? You could swap the Xtal for a 32MHz to test if the board will boot at 16MHz before going for 20MHz again. Did you check how stable the voltages are? Those tantalums might not explode on you anymore, but how good are they? I have had my fair share of boards being finicky about what memory modules worked when DIP and 8bit SIMMs were normal and it would not be uncommon to get a machine in under warranty that had memory fail that would work fine in another less critical machine.
Good point with changing the crystal. I'll consider it. Also, the memory might be too slow. I'll try more options and see if I get somewhere. 👍
Might be worth checking for shorts on the power rails before firing up unknown boards.
I did, there weren't any when I checked. It must have happened when I powered that system on after all that time.
@@bitsundbolts ohh. I don't think that was in the video, which threw me as I'm sure you usually check.
That's good old Schrödinger's tantalums. They are simultaneously fine and not-fine until you power them on. The failure is initially not electrical, is not detectable by any measurements known to mankind.
As usual, very interesting. I wish you good luck and waiting for next part!
Harris Semiconductor manufactured the HD-6120 12-bit microprocessor (PDP-8 on-a-chip) used in DEC's DECmate series of microcomputers (that followed the PDP-8 series of 12-bit minicomputers and the VT78 [which used the Intersil 12-bit IM6100 PDP-8 on-a-chip]. I've got several examples of all of these, but not the complete series of PDP-8s & DECmates.
Also, I had to replace a power supply in an x86 (maybe a 486?) PC, but since the "new" power switch had the same colored wires, I got lazy and didn't perform the more difficult task of changing the power switch too. When I "finished," I plugged the PC in & turned it on. I blew a tantalum cap off of the board and it looked like a small bottle rocket leaving a smoke trail as it shot towards the bookshelf! I immediately turned the PC off and and changed the power switch too (the one in the face of the desktop PC). I went ahead and turned the PC on again and had no problems, even without replacing the blown cap. Just lucky I guess! That PC is still out in the garage and may get used again in the future.
When they said that you would get a 286 system that would blow you away. I thought that they ment in performance.. :D
Haha, oh well...
One of the reasons I think these tantalum capacitors pop is that they weren't spec'd correctly back then. After a similar failure of a new tantalum, I found a number of empirical evidence posts stating that tantalums should have double the voltage they're expected to see. In your case, those are rated for 16V but are attached to the 12V rail. This is *way too close* to their max rating and any fluctuation could pop them again. The damage is cumulative and chances of failure increase with age.
The voltage rating doesn't have to match the original spec, only the capacitance is important. I'd look for 24V+ tantalums as suitable replacements or you're rolling the dice with every power up.
You are absolutely correct! The tantalum cap that exploded is indeed connected to the 12V rail! That might have been the reason, in combination with age, it exploded. Maybe it would be a good idea to preventively replace the cap around the power connector - at least the ones connecting to 12V. I have seen many times that those are the caps that blow up.
Very interesting! I will uprate the voltage on these in future!
@@GadgetUK164Tantalum polymer caps don’t have that disadvantage
@@_irdc I replace the tantals with electrolytic capacitors.
@@Nelson68k that also works, although I’d go for OS-CONs
I remember that Necroware had a video some time ago with a motherboard that had a problem of being always in turbo mode (slow speed). He had to replace the chip, responsible for generating a turbo signal. The video is called "Turbo in a retro PC - explanation and repair" on his channel, maybe it will help!
Boy that had a lot of magic smoke !!!
Many years ago. Must be 15 yrs by now. I was sitting at my desk. I suddenly heard a loud BANG! It scared the crap out of me at the time.
The noise was from a video card I had laying on a table beside my desk. One of the capacitors had spontaneously burst. I think it was an electrolytic though not tantalum. I didn't have the ability to replace it and I'm sure the card went to e-waste when I chucked out a bunch of stuff a while ago so I don't know exactly the model. At the time I was messing with Geforce2 cards though and I also had a Goodwill Computer Works nearby (they resold donated computers) which had a bevy of cheap computer parts that I often would grab if I needed a part. So it might have been a card I bought from them.
Either way your fireworks in this video were great ! ;)
It exploded without being plugged into a system that was powered on? I never heard that this can happen 😮
@@bitsundbolts Yes. I never knew they could burst w/o power either.
That was the strange thing. I am not even sure it had been plugged in
for months before it blew out either.
That is very odd. That must have been a chemical reaction then that built up pressure inside the can. Very weird!
@@bitsundbolts It's been a long time but I want to say it was an electrolytic and it burst at the top.
No rhythm nor reason that I can remember either. Just poof and as I said it was a loud rapport
like a firecracker at close range.
I've messed with large, old multi-section can capacitors from tube circuits.
I used to purchase old gear to modifiy into guitar amps or salvage the parts. So I've worked with
some bad caps, but never had one burst without power.
I did blow up a can once. I'd changed a power supply from a choke>cap (choke input) filter to
cap>choke> cap (capacitor input) unaware that the power transformer was not able to supply enough
current w/o the choke. The result was the can burst violently while under full power.
Not to mention it was a New-Old Stock part of questionable quality. Even that didn't sound like a
firecracker though. It was more like frying bacon since the electolyte boiled and exceeded the
pressure rating of the can. There was no safety vent on this type from the 1960s.
Base 64K is on the high addresses for bios shadowing. So, the stable address high may not be a problem. Chck the connections, most likely there is corroded trace/via.
No, "base 64K" is not used for BIOS shadowing, which is optional (if implemented at all) in this kind of board. They used two BIOS chips to provide full 16 bit bus width, reducing the need to shadow the ROM for performance, although the ROM is surely slower than DRAM at 20MHz, so shadowing still would help.
Base 64K indeed refers to the lowest 64K of memory, which needs to be operational for the POST to start. The first kilobyte is used for interrupt vectors, the subsequent 256 bytes are used for BIOS variables, and some other part of the base 64K is used as stack memory. Lastly, the memory between 31K and 32K is used to load the boot sector or MBR to start a disk-based operating system.
@@tw11tube Ok, my fault. Bios itself is located on 0xF0000 - 0xFFFFF. So, if you try to observe bus in this state, you can see only the top addresses. CPU and memory address / data bus is directly connected in this platform, right? So, it make sense to just measure pin to pin continuity.
@@petr56321313 As the system is beeping, we can assume that accessing the BIOS works fine. Address lines are most likely not connected straight from the CPU to the bus, because the 286 stops driving the address lines for the current cycle already before the cycle is complete, so there needs to be address latches on the board. Data is likely connected straight through.
It makes sense that you see A16..A19 high most of the time, as these pins need to be high when addressing the BIOS. You will see the lines low for a short moment before beeping starts, becasue that's when the BIOS tests the first 64K.
When the 286 starts, it starts execution at FFFFF0 (16M - 16), so the BIOS must be visible there (at least the last 16 bytes), which contains a far jump, which in real mode will cause continuation at 0Fxxxx, so just below 1M. This means we expect A20..A23 to be high for a very short time after reset, and then stay low until the memory test reaches extended memory.
I have a 286 too that would not work at all without proper FPM installed. regular sim would not work. I actually got a second 286 board which seem to have on board ram but it errors out. I can somewhat run it with only base memory. 286 in particular seems to be very iffy sometimes
Let's hope there is a difference with the DIP memory - once it arrives... At least it would eliminate the Chipset as the cause.
I have an idea. Since keyboard connectors seem to get the brunt of battery corrosion, it might be helpful to wire up a new connector to flying leads and solder the leads to the board for testing. If the board doesn't work you don't have to try and desolder the connector too many times and damage it.
When you're done with sorting out the board keep in mind that Headland HT12 chipset needs special drivers to handle memory properly, I use TLBMM.
Yes, I read about it. I never had such a board with that chipset. It would be fun learning about all this - I just need to get it to work.
Harris is actually a US military contractor and a lot of their parts are "milspec", with higher thermal tolerances and lower power consumption. They're usually manufactured later years and a process node or two later than their Intel counterparts.
capacitors sure are fun, aren’t they? honestly, after learning how tantalum caps are made I’m kind of impressed they work at all. ditto electrolytic caps in general, I suppose, but tantalum in particular just seems bonkers with that whole lump o’ microscopically spongey metal with a whiff of oxide around whatever gnarly shape the tantalum happened to sinter into thing they’ve got going on
I think you're off to a good start.
😅 Next time, I wouldn't mind a little bit less fire...
@@bitsundbolts Understandable! I think it might be wise to simply replace the tantalums on the voltage lines before powering up an old motherboard like that. Of course, the video might be a bit less exciting ;)
01:38 that's a Realtime Clock/CMOS RAM chip.
If you did not find out already on how to check the operation of the octal buffer: find the LE pin (high is latch in) inputs and outputs on 74xx373 parts and use an logic analyser or at minimum two channel scope (direct/continuous sampling as in: without averaging or other software filters) to verify the data is latches in properly.
I'll have a look. Using a two channel scope is still above my abilities, but I'll have a look the datasheets of those chips and try to understand how they work. Maybe my programmer has the ability to test those logic chips - I haven't checked that yet.
I'd check the LS245 first, getting stuck high can definitely happen if it fails to go High-Z.
great fireworks! next time measure for shorts in the power connector :)
I did measure before powering on. No shorts at that time. The cap must have failed with the inrush of 12v after such a long time. I guess from now on, I'll pay attention to the cap on the 12v line. This seems to be the one that most likely will explode.
I got the same Board some years ago. It is not working and i hope i have it restaurated one day 😅
I think it is a good idea to make some videos about dos memory management😊 I like all of your videos, thank you for that interesting stuff.
Good luck with your board! I hope you'll be able to fix it.
Thanks for watching my videos! Glad to hear you enjoy them.
Hey i remember watching the talk about capacitors by James the Bald Engineer... himself a classic computer restorer and enthusiast, who showed among other things why tantalums explode, or why new polymer ones don't. So what i took home is that after thermal stress or prolonged storage, one should cure tantalums by gradually bringing them once up to voltage, gradually raising the voltage over minutes, then they'll no longer explode, even crappy old ones.
So yeah what if one did that type of voltage ramp rig, while holding the whole mainboard artificially in reset in order for it not to go all weird?
I feel your pain :( I went through the same thing a few months ago, a socket 7 motherboard exploded and took with it everything connected to it, sound card, video card and ISA capture card, network card, memories, processor and power supply. There was a short circuit on the motherboard and the AT power supply I was using had no protection circuit. Since that day, I have never tested a computer in the dark again, I created an adapter that sits between the power supply and the motherboard, any surge of voltage or current will trip it in milliseconds.
There are a thousand different projects on how to do this, and it's very simple, in fact, you can order everything from PCB Way already assembled.
I'm using a modern power supply from EVGA (supernova G3). It should have all the protection mechanisms, but I did get the occasional surprises like this one.
One time however, the power supply did shut off. That was when I built one of the -5v ATX Adapters which had a bad cap.
Short from where to where? Short any rail to ground wouldn't actually harm any attached hardware; except via crossload and the PSU being a piece of garbage that would have killed the PC anyway. This is why i really don't like most old PSUs except a handful trusted manufacturers.
@@SianaGearz I believe it was an IC regulator that completely shorted out, everything on the 5v line simply exploded. Perhaps the battery or a capacitor has leaked, causing the short. The computer was an Itautec RTV A96, and used an Intel AT motherboard with some onboard stuff, like an ATI Mach 64 video, a somewhat unusual motherboard.
@@RicardoRamosRetrocomputacao Yeah nah if you have explosions along the 5V line, it means the 5V line drifted well above nominal. The regulators off the 5V line would be supplying low voltage rails, whatever CPU and specific IC supplies, so a regulator has 5V input, low voltage output, and feedback, correspondingly all regulators along 5V line would worst case, if failing, pass the input through, and there really aren't a lot of them. So i'm thinking your PSU killed the hardware, not the mainboard.
If the PSU doesn't have crossload protection (individual rail under/over voltage sensing), then a near-short of 12V line to ground will easily cause 5V line to drift up, while the 12V line dips. Also quite common for shitty ones is a magnetic failure causing the 5V line to be internally shorted to 12V line in the PSU.
Fun with Tantalums! Happened several times on my Amigas that the old buggers simply blew up and left marks on the PCB. Happened most of the times after switching to another power supply. My theory is, that the aging Tantalums were so in tune with the aging caps inside the old power supply that they simply were not able to withstand the power of a new or more recent power supply.
Best would be to replace all of them with modern MLCC ceramic caps.
But I shouldn´t be the person to say that. MLCC replacements for all my old hardware is on my list since many years but I am just too lazy to start. So I only replace Tantalums in case one turns into fire. ^^
Tants need to be significantly over-rated on voltage, machines of this era didn't do that. 16v caps on a 12v line, which can easily over-volt as it is typically not as well regulated. They age out and fail short circuit, just like MLCCs (in fact I've had way more MLCCs fail short in Amigas I repair for people, but they are used a lot more). It is better to replace with modern electrolytics for most of the use cases, although MLCC probably won't do any harm.
Hunt down an Intel 287XL coprocessor for that thing. It's basically a 387 core in a DIP 40 package compatible with the 286 bus. Its one of the fastest coprocessors out there (not the fastest, that would be a 20 mhz Cyrix FasMath or one of the Harris parts) while still being really common and not expensive. Since it's a 387 at the core, it can usually be overclocked to a really great degree, so you might be able to run it synchronously with the CPU, it might require some tweaking of the coprocessor speed jumpers, though. The 287XL has a pin that toggles an internal clock multiplier meant to adjust for the 2/3 divider that a lot of boards used for the original 287, so experimentation is necessary. It's also a lovely looking gold topped ceramic package, so there's also that. Regarding the Bios, there's a very nice MR-BIOS available in the VOGONS MR-BIOS archive. It supports things like larger HDDs and more than two floppy drives.
Thanks for mentioning the MR BIOS details - I'll look for it. If I get this board to work, the 287 is going to be a must!
If you were wanting to test with less than 4 Simms remember a 286 has a 16 bit bus, Simms must be installed in pairs.
the DIP sockets need to be populated with the base memory. The SIMMs are purely for expansion.
I hope you're right! It says differently in the manual, but it would be the first time there is a mistake in one or the wording is chosen poorly and ends up in confusion.
Should always replace these old ceramic capacitors. They are known to fail short. Replace them with something other than another disk, like a polymer.
I will replace the caps once I get the board to work. You're right, leaving those caps on the board will just create more issues in the future.
@@bitsundbolts Wonder if it may not be working because some of the other caps are also shorted? Maybe go through and check each of them for shorts until you replace them?
Lunar New Year has come early!
The exploded capacitor looked like an exploded chicken from some kind of looney tunes cartoon XD
All of the Minions look nearly identical lol
my first introduction was a Zenith EZ PC. According to wiki it's a 286 at 7 mhz made by NEC. My first computer that I actually had permanent access to was a compaq 386. Then I swapped the two (I was not happy about this) and eventually I ended up with both.
EZ PC 2 apparently. 2 drives no HDD
wow intentionally designed to prevent the instalation of a hard drive unless you had the EZ PC 3. And I do remember that it had no ISA slots. They were jerks!
and the criticism on wiki is exactly why my sister didn't want it. It wasn't good enough to be useful for what she needed it for.
Pretty sure the letter "C" in the CPU type number indicates a part made with CMOS transistors. This alone permitted a clock-speed boost.
Thank you for the great video. I would prior test if the resistance between the right contact (+5) and ground (black) of AT power switch on the mobo is appropriate. Think maybe this system board will be hard to repair
I definitely will do this from now on. The capacitor that exploded was on the 12v line btw.
I'd be tempted to unsolder the address buffer chips, replace them with sockets, and test the chips off board in a TL866.
That is exactly what I'm doing right now. The next video about this board will be all about that.
3-2-1 you're going to fix it soon! I can see that! I hope it's not the chipset, that 16MHz to the CPU is indeed suspicious. If I remember right we checked on my 286 (different board) and the speed seems to go to 20MHz straight away. Fingers crossed!
Yes, you're right! I'm looking at the logic chips now. I also socketed the 40 MHz crystal and tried one with 33 MHz - unfortunately, there was no change. Tomorrow, I'll continue with the logic ICs.
Great mainboard!!
It is! I hope we will get it fixed!
The chip memory is your base cache memory. You need it installed. The sim memory is system memory.
6:37 This is what i often see in my own barrel battery boards, if I continue to brush, I continue to see bubbles and it seems to never stop. Does this mean there is still hidden battery leakage? Even though the PCB part looks fine like yours, it's strange.
those caps are very amongus
It fits so well because non-poymer type tantalums are generally sus.
Sackzement is such a awesome thing to say instead of shit lmfao
JP2 was probably for selecting FPU speed. Typically 287 run at 2/3 cpu MHz. Possibly you could run at 1/2 or 1/1 with jumper.
There are frequencies present on those pins. I'll check where they go and maybe add the jumper back.
The manual may be correct for 20mhz cpu.
I'd try putting a matching crystal for the DRAM speed you have. To see if the dram is just too slow (given you got memory errors).
I tried a 33 MHz crystal, but no change. I'm in the process of debugging further right now. I may want to try an even slower crystal if I find one. But with 33 MHz, the board behaves exactly the same. I also found 60ns memory - but no change.
@@bitsundbolts fair enough. worth a shot; sorry it didn't work out.
I bet you need to populate the socketed RAM.
Cant wait for you to get yours and see what happens.
I got the memory about an hour ago. Unfortunately, there's no change - same behavior as with the SIMM sockets - or no memory at all. The board behaves identically no matter what.
I also got new BIOS chips which work properly - still no change.
@@bitsundbolts Thats unfortunate. You mentioned in the video that there were data/address lines acting weird? Maybe try desoldering the CPU and see if the issue persists. If it does then chipset may be bad.
I'll check the logic chips over the weekend. There are about 10 on the board. My programmer can test most of them.
I have a BIOSTAR , MB-1220VE Harris 286 20 MHz the board had minor battery damage the current memory installed is 16 Meg ram most of it is used as a ram drive
Great! An ocean of RAM for a 286! Very nice. I hope I can get this board to work some day. Would be nice to also see it boot up with 16 MBs.
45:27 You hovered over a bus-driver as you said it 😂74x245. Other common ones are 241' 615' and 62x' series
..and 244' and 373'. Man they're using everything on that thing :D
that was.... tantalizing
Agree with GadgetUK. But also, look at caps and resisters around the clock crystal between it and the chipset.
I'll wait for the memory first and see if it makes a difference. Next step would be probing around the board and see if I can find something suspicious.