You made the day of a 51 years old boy! It brought me back to my endless nights of fiddling, programming and trying to muffle the sound of my dot matrix to avoid awaking my dad... lol! 8 mbs on a 286.. crazy!
I had a 386dx20 with sipp sockets. When I had the money to upgrade to 2 MB, I went to a small local computer shop and I was told to wait a couple of minutes, because they had to solder the pins to their simm modules. I guess that was the reason for the 5$
@@daybyterI imagine that the might had some kind of rig so they can place the SIMM and solder the pins quickly and reliably... Or maybe I'm overthinking and they were really good at soldering 😅
@@RetroTinkerer Nice, good find! So $11 more per module (in today's money.) Not huge by any means, but seemingly money that didn't need to be spent. I can't imagine the 286 board would have cost $40 more just to add SIMM sockets at the factory.... but who knows. And also interesting they had the pins and likely a jig to help them solder the pins on.
I seem to recall something happening with DRAM / silicon prices around 1989 ? I remember the Kobe earthquake in 1995 had an impact on DRAM prices in the late 90s, but I can't remember what the 1989 thing was? Maybe it was the Loma Prieta quake in San Fran in October 1989? What I do remember more clearly, is around 1993, when Dad bought 8MB of SIMMs for my PC. It cost £200 (GBP) back then, so around £25 per Megabyte. :o
The oscillator next to the FPU socket is for separate clocking the FPU if desired. The jumper should choose between synchronous mode for the FPU which would be the half of the CPU clock (24 MHz /4) or asynchronous mode which gives you the possibility to clock the FPU higher if you install a 287XL or ITT up to 20 MHz. I recently pushed one of my 286 boards to the limit with this experiments. Cheers, Peter
This is great! I was a PC repair and clone build technician from 1990-1995 in Denver, at the start of my career. I'm 47 now, and this brings back great memories!
This is an exciting video. Changing to more modern memory slots. Very cool. Very impressive. I have never seen any of those old type memory with the pins installed. I started building computers in 1995. And where I worked, I purchased some of their computers. I know when I first started, I had to install a card with the IDE pins. And the floppy as well. I built computers for neighbors, and I could warranty them for 3 years, that is what they gave me on every computer board I purchased. Watching you convert the sim sockets Rocks! I am so happy it worked!
Fantastic conversion! I converted simm modules to SIP by using "turned pin" style male-male pin headers. They're pretty easy to solder on, they're the correct pin pitch, and they mate perfectly with the turned pin style female sockets on the motherboard.. What's easier though is you can just stuff the raw simm sockets in to the sipp headers. No need to solder or anything. If the spacing is right just stick 'em right in and they work. Granted de-soldering the old sockets and putting the headers in proper is better but the easy way works in a pinch. Also nice this motherboard with those "3 chip" modules. Older 286 boards often only post with the "9 chip" sort.
Indeed I think this is the best method to make SIPP modules, I have also done this on a 386 mainboard. Though if I had the SIMM sockets Adrian has at that moment, I would have used those in the same way he did. And the board really looks cool like he said, and it looks very professional.
Nice, I have some of those male to male strips, I think that's a much better idea than using pins from old chips. I'll just stick a strip of 30 pins into one of those SIPP strips I removed from this board, and I can solder up some more as tests. I have a bunch of other SIPP motherboards for testing.
When I used to work as a PCB assember and later as a tech, I was shown to bend a pin or two over carefully to hold a socket or IC in place prior to soldering. This made the assembly and soldering process go pretty quickly without risking having something loosen up and end up partially soldered on the board. This was really important prior to putting PCBs through the wave soldering machine.
SIPs soldered directly to the board are hell. I've pulled a bunch of SIPs from industrial controller boards. The solder used to mount the pins must have a much higher melting point than the solder used to mount them on the board. I suspect there's some silver in there. Excellent episode, full of technical tips. 👍
Is really about the only thing that would make good use of a hot rod 286 with that much RAM - well, there is the 286 16-bit version of OS/2, if can find a copy of that. Win 3.x would sort of too, but it’s such a kludge compared to actual protected mode operating systems that have been designed for that from the ground up.
I started my computer life in the 80's and think I remember SIPs came before SIMMs. SIMMs came about to make it easer to fit modules can cheeper to make them.
It might be my bad recollection but I seem to remember in your second channel, in a video about a box of random junk, was an industrial equipment board containing a "Make-it 486" CPU upgrade for 286. Assuming it's not my memory playing tricks, you need to test it out on this motherboard!
Somewhere in the early 2000s was the best time to get this era of equipment. Even though my first computer was actually a 386SX, I got to buy up older stuff severely cheap in thrift stores and the like. I once got an original IBM XT with a working hard drive for $5! Imagine that today. Anyway, I did at one point have some generic 286 machine and found the Make-It 486 part, and sure enough, it did come up as a "486" as far as any diagnostics were concerned. However, an attempt to install NT 4.0 on it failed, although I don't remember precisely why.
Sadly I don't have a 286 -> 486 upgrade. I have a few 386SX -> 486 upgrades though. I also have a 286->386SX upgrade, so that would be another way to push the limit on this board.
@@curtissimmons1085 I imagine anything is possible with enough glue logic. I have an "Intel Inboard" 386 upgrade designed for the original IBM PC/XT, which sure enough can boot that machine as a 386. But the reality is, it's just an SBC jammed in an ISA slot and basically takes over the former 8088's bus. There's very little of the original PC doing anything but being reduced to a backplane. One fun thing though is the Inboard's 386DX is socketed, so any upgrade part in theory should work in it. I found a 386 -> 486DX2-50 upgrade part, and that's how that XT boots now. (Amusing to see Microsoft Diagnostics say "486 CPU, XT bus") That said, depending on your exact definition, there's no reason that you couldn't do something like that to make an early motherboard have more capability. But it's not quite the same as simply inserting or snapping-over an upgrade module that takes over as CPU but leaves the basic "spirit" of the original computer intact. Part of the problem will be clock timing with all the other guts... You'd have a CPU craving high speed buses and yet being bottlenecked to some relatively incredibly low speed bus.
My favorite old-computer overclocking adventure was getting an Am5x86-133 running with a 50MHz base clock. I needed to use a giant heatsink, but it was stable.
A 286 had a faster clock rate, didn't have the 8/16 bus bottleneck, didn't have multiplexing, had a 6 byte PFQ like the 8086 (unlike 4 on the 8088), had a separate address unit, and a hardware multiplier. And there were additional instructions for better control.
I think the oscillator Adrian has connected is originally for the FPU so you can run one at a clock different from the CPU. But since there is no FPU in the system it can work like that.
We had this board in my dad’s pc! We upgraded from a turbo XT. 1MBx4 SIPP RAM, 16 mHz Harris 80286, and i287, too. We were able to skip a generation from 1992 to 1994 and later would put an i486sx in that case.
I only worked the PC repair shop from Aug'95 to Sep'97. Only saw one mobo with SIPP. Never saw a 286 with that much memory. Never saw simm slower than 80ns (that I know of). You show all the coolest different stuff.
One of my earliest vintage PC purchases was a unknown motherboard for £10. Hit the jackpot, it was a 286 board with 4mb on the board in DIP chips and 4 SIPP sockets. Even better, it has a 286-486 upgrade chip in it. Really nice board, but I did have to repair a few Varta ravaged traces.
You should totally do a video going through writing a simple TSR in assembler or C to make a keyboard combination change the clock speed on the fly. Since it's just changing memory registers, it would be super simple to do either hooking onto int 9 or int 16. If you're interested and you see this, I'd be willing to write one for you with the relevant information (though it would be fun to watch you do it if you have the knowledge or you just want to learn). I used to make a lot of drivers for DOS back in the day.
I have a big spool of some tiny solder I’ve been trying to work through for years. I take a length of it and fold it it half, then twist the whole length up like a rope and it’s almost big enough to use 😅
Hi Adrian, many years ago (1980 - 1990's) i worked at a PC manufactures, and we use to also add the SIPP pin to SIMM Modules, the pins came on a roll of 10,000 or so and came in various pitch spacing, so it was easy to cut a batch of pins for modules which were held in the pitch spacing by a metal ribbon which you broke off after soldering the pins, with a bit of flux and a little skill you could convert a SIMM to a SIPP in about 10 Minutes. Regards Martin
What a nice upgrade for this board, it's lucky that he had these new old stock SIMM connectors. I board an old 386 board which also featured these sipp slots. The seller forcibly cramped normal square pinheaders soldered to the modules into the SIPP turned sockets which was rude and destructive. Yes, in case of the 5170 there is full speed on the ISA slot cards, no advanced chipset mechanisms there, only the wait state decoders in U87. The zero wait state can be defined per memory decoded region with the 0WS input designed by IBM in the AT, so variable speed can be used transparently per memory region which is a very cool design. The board being used in the video is years further developed and may even operate separate buses for the RAM and the ISA slot connectors. This would allow for much easier speed separation in accessing the different buses. I am developing a 286 AT and made a thread online and published in GitHub, in my design all the RAM is SRAM and running on a ISA slot connector. I haven't tested zero wait states on the SRAM yet but I will try this out at some point. For 16 bit access, for example the ROM chips will definitely need a wait state at above 10Mhz or the system won't POST. Anyway I love this video since I am working on a 286 system for more than a year and have seen a lot of stuff, so it's great to see someone else tinkering with this type of system and seeing what happens.
I really enjoyed this video. Being you were able to upgrade the Ram slots. Very impressive. I have plenty of those old boards in storage. I will do the same upgrade. Thank you for up loading this video. My experience with computers started with the 8088 models. I lived close to a Fries Electronics. That is where i purchased my computer boards, ram, Processor chip and Ram. I really do not like to purchase computers, I rather build my own.
I replaced my Hakko 301 with a Metcal desoldering gun. The difference is night and day. The Metcal can suck up an enormous amount of solder without needing cleaning. And it has better heat up time, stability and quick change tips in a bunch of sizes. They’re expensive new but you can find good deals used if you’re patient.
new yes, but used I’ve managed to find Metcal desoldering stations for similar or less money than a new Hakko FR-301. Just need to keep an eye on eBay, Craigslist and FB marketplace.
Metcal's are definitely a great upgrade. I just bit the bullet and replaced my crappy old desoldering station with a Pace. I wish I had saved up for one years ago. Its quality and speed is absolutely amazing. If you value your time, it's totally worth it. Also, it helps to add a nice big dab of fresh solder to old joints before trying to suck them up.
Not to mention the variety of quick change tips available for it. Just this morning I desoldered both the 1mm leads and 3mm mounts of a small stepper motor in about 2 minutes.
This is so nostalgic for me. I only encountered SIPP once from garage sale computers when I was very young. I proceeded to throw away the motherboard and sipp ram since it barely had any and I had a box full of simms. Was neat to find all this out about this. Thanks for the post.
For the really thin solder, what I do, it pull out a long piece and fold it in half and twist it. If it is still to long then fold it in half again. My silver bearing solder is really thin like this.
For converting the SIM to SIP you could just use a row of male-male header pins, spacing will be easy, alignment will be easy, just lay it on the SIM and solder it on.
I have been having nightmares about changing the memory sockets on my Amiga 4000, and although I have also watched others, like Chris, make it look easy, I still didn't feel confident enough to give it a try. Watching your successful effort, has now encouraged me enough to give it a try. Nice work
My first computer was a 1mb 286 with a 40mb harddrive EGA monitor in 1989 (I was in middle school). cost my parents $1350. I was so happy, read the DOS 3.3 book from cover to cover. Played games on it. First upgrade was a 3.5 HD floppy drive and later a 2400 baud modem when I discovered BBSes. Later ran my own BBS with a friend.
I have another idea for the SIPP/SIMM conversion dilemma. Take some single SIMM sockets and put SIPP legs on those! Then you can swap in any SIMMs you want, and move the "format converters" to any motherboard you want. Of course if you find the upgrading process to be reliable and not too much trouble, then by all means stick with it! Making every motherboard in your collection not suck in this particular way is certainly one way to solve the problem. It's good that you have three rolls of the "hair thin" solder, as you can put all three spools on a stick and twist the three wires together to get something that more closely matches the solder diameter you're accustomed to. When you only have one roll, you have to do it in segments which is much more annoying, although in either case you can use a drill or power screwdriver to twist the wires together, and collect the result on a new spool. It's not like any twisting stress is going to matter once they're melted.
Back in the day I got a pile of proto boards that had 8x1mb SIPs per board soldered to the board. I would strip them and re-pin them with sip strips I got from a local shop and use them for various upgrades.
To make a SIPP, put the pins in a strip of female header, butt it up to the simm and drag solder them down. Takes about 40 seconds a SIPP when you get practiced. Alignment is perfect. Works when SIMM conversion of the mobo is hard or impossible.
I would put a piece of kapton under the crystal. On those old motherboards, on corner of the crystal has a positive voltage on the pin. If the pin on the socket is proud of the plastic surrounding it, it can touch the grounded body of the oscillator module it will short out and usually will burn an internal trace on the motherboard. I lost a few motherboards like that. In retrospect, I should have tried adding a bodge wire from the 5v rail to that pin. But I learned to put a piece of some sort of insulting plastic under the oscillating to protect it.
You can buy those pins, they're called lead frame pins. They come on a strip (or frame) for easy installation and then you cut the strip off after they're soldered. I don't know how easy it is to get them in small quantities though.
Back in those days, I used to remove the pin and convert SIP to SIMM, they're same thing just SIMM socket is more expensive, price of SIP and SIMM just a few USD, when you make a lot of these those metal pins just cost human labor, at those days is not that expensive.
Some very early produced SIPs were narrower lacking sides that lock in SIMM slots. You could still use those as SIMMs by using one proper SIMM to hold su h converted SIP in place. They also worked on motherboards where simms were just pushed downward instead of inserted in angle and then "click".
I had one of those 32-bit VLB cache hard drive controller cards that used these same memory modules. It was a Tekram and I could swear it doubled performance of my 486dx2
The 286 was a real step up in capabilities but they were quickly eclipsed by the 386. I had a 286 system that I eventually jacked up to 2 MB and 12 MHz. It was an awesome DOS/Windoze machine. And I justified trying dual boot Linux on my 386 machine having the 286 as a backup in case I screwed things up on the 386. I did have to heat sink the 286 and 287 for the overclock. They ran HOT at 12 MHz.
I had a 16 MHz 286 from toward the end of the run, when 286 boards at any speed were becoming hard to obtain because production had all re-focused on the 386sx (which uses pretty much the same motherboards, with socket and BIOS changes). Clock for clock, it was slightly faster than a 386DX -- but of course, only for the things it could do at all. That was also the machine where I accidentally set my CGA card on fire when I set the CPU/ISA bus ratio to 1:1. (It was repairable, but I wasn't gonna mess with a 16 MHz bus after that.) From a reliability standpoint, and for ease of finding matching contemporary parts, I think the "best" retro 286 system is a slow 386sx! Or skip the 286 and 386sx entirely and de-Turbo a 386DX when you want that 286 experience.
I swapped the 8 SIMM sockets on a Macintosh IIcx recently. They were the crummy plastic ones and all the tabs had snapped. It was quite tricky getting the sockets, I also used a combination of hot air with a desoldering pump. Definitely something I don't want to be doing on a regular basis 🤣
I once soldered wires from SIMMs straight into an STFM board (not even SIPP, sockets IIRC) and that was still less hassle than upgrading my original ST
Another great video! You can still get the pins to make SIPP sticks from 30-pin SIMMs. Look for y1 leadframe. I think the TE 1544210-2 leadframe would work.
I had a 286 board with 4mb on the mobo in CHIPS, not memory modules! And it had a Harris 286 CPU running at 20 mhz. It was awesome. I have another 20mhz Harris chip in a drawer, just bought it from eBay a few months ago.
Should note that this board seems to have a development BIOS - says so right on the splash screen. Explains the wacky super-fine-grained control and the direct register manipulation screen.
Frequency divider is needed because some software didn't work correctly (was too fast for example) when run on higher frequency clock. It's mostly code that doesn't respect timer but instead use instructions in sync with their execution time. It's in similar way done on older 8bit machines.
I've had good luck soldering a row pin headers to things, and then heating/removing the plastic spacer. (for other random projects) That might work for installing sips pins on to a chip.
I've only been soldering for a few years now, but I'll just say what I've observed/works best for me. When you soldered in the slots, you went along a row working your way to the other side. Myself, I like to tack either side first so that the slot cannot move while I'm soldering it. Once I tacked either side with two spots, I verify the slot is flat. Its more so a issue with dupont headers like on a rpi or pico. Plus if you're concerned about melting the connectors plastic, you can alternate between either side. Also flux. The thicker kester stuff you had was likely rosin core. The smaller stuff probably wasn't. Either way. I always add extra flux, it makes your joints nice and shiny. MG chemicals make a alright one. The go to seems to be Amtech stuff. However they recently rebranded to Stirri and you can buy DIRECTLY from them unlike Amtech. Hot air is kinda eh. I warped a xbox 360 board using it to clear holes when I recapped it. However as long as you don't put too much heat into it too fast its fine. Just be careful, I'm sure you know. Give a board the case of the bends.
Reminds me of the Kaypro II that I had way back when. It had a Z80 with a clock speed of 2.5 MHz. A popular mod back then was to double the clock speed to 5 MHz. The recommended way to do this was to put a toggle switch on the front panel so that you could switch back and forth in case of problems.
In 1994 I had a 386SX PC that used those SIMM memory modules. On that era, internet wasn't available on the entire island, only the capital city. I had to buy those modules locally. 1 megabyte of RAM cost around a $100.00 each. I put a Math Coprocessor, another $100.00, and added 6 megabytes of RAM to run Lotus 1-2-3 for Windows, for Windows 3.11.
I never had to worry about the price of SIPP memory because I never dealt with it personally. At a company I worked for in the early 90s, we had a bunch of truly awful Memorex-Telex '386s that had them. The machines were weird to begin with and the issue we had was memory working its way out of the SIPP sockets over time causing all kinds of problems, as you can imagine made them even worse. When they were finally scrapped, we couldn't wait to dump them into the scrap pile.
To convert SIMMs to SIPPs you need leadframe pins that will clamp around the PCB and can be easily soldered. Digikey has A122965TR-ND (Tape and reel) and A122965CT-ND (cut tape.)
The previous episode with the board that had the damaed RAM catches comes back to mind, remembering you had that vast supply of NOS sockets, that board could have the sockets replaced now... :D
15:52 it almost looks like the through hole plating came out on the 16th pin. However, if that happens, the pad most likely is toast too, and it seems fine there. Obviously, since it worked.
I remember back in the 90's buying a bunch of SIP modules at a local car boot sale. I had no idea about them at the time byt was super excited of the prospect of potentially upgrading my 386 back home as all I had was 1MB in simm format. I tried to look up as much information back then on them but I found nothing anyway I remember tracing out the memory modules with a multi-meter to make sure they where electrically the same. In the end I de-soldered the pins and tested it in the system unfortunately They never worked so may have been damaged somewhere along there life. But I always wondered about these if they would have worked.
I'd use those generic pin sockets that you can buy for cheap to build a sacrificial jig to hold the pins in alignment while you solder them to the board, rather than having to solder them one-by-one.
It wouldn't be sacrificial if your soldering's halfway tidy. I always try to have connectors plugged into something while soldering, to sink heat away and hold the pins in line when the plastic gets soft.
I love this SCAT design because the chip is so elaborate that it includes so many options and flexibility. The AT system is very complex which is why certain speed configurations may not work. In case of running the system bus at 12Mhz as he has tested, this should easily work. I designed a 286 system based on the 5170 core concept using 3 CPLDs, and this is easily able to run at 16Mhz, with no restrictions on the ISA slots and system bus. I was surprised to hear that he was not able to boot at 12Mhz, I suspect the CF card is the reason and should be replaced with a fast laptop harddisk. The disk controllers on those are much better than the controllers inside CF cards. About the oscillators, I believe the SCAT has various clock configurations, where different input pins are used to base the CPU clock on. The datasheet will explain this in more detail, which will explain the PCB layout. I think it's possible to use the other pin on the SCAT if configured correctly. The keyboard controller should run at 8Mhz to be sure it won't cause problems, a little faster does seem to work also in my tests but it depends on the keyboard controller chip.
First time I overclocked I didn't even know it. Had a 486 board that had been upgraded from an 80mhz to a 120, then later I got an AMD 133. There was no jumper setting to equal 133 so I kept trying different jumper combs until it booted, later I noticed the benchmarks all showed it overperforming and it was running at 160. Back in those days you could easily get 20-50% better performance by overclocking I was one of those people who had a dual celeron 300A board running at 2X450 and easily beat the P450 and that was king of the hill for desktop computers at that time.
I have a 386sx that I've used the bios clock divider to closely match the 286 12mhz computer years ago. Other than giving me better memory management, it was very close to what I wanted. The divider could also let you slow down a fast processor for older games when needed.
That's why I often say that the ultimate 286 is a 386sx. You won't notice the difference unless you specifically invoke the 386 part of the instruction set. The 286 is ever so slightly faster clock for clock, but considering the 386sx clocks a lot higher (if you want it to), that is pretty much lost in the noise. You also get more sophisticated BIOS and better support for large (by the standards of the time) hard drives. Make it a 33 or 40 MHz 386sx and you can kinda skip the 386DX generation and move on to the 486DX. This is a practical solution to the fact that most of us don't have digital basements to keep one of everything.
Local shop used to sell both for same price, both for two much, so bought memory in CompUSA, Fry's Electronics, somewhere on sale, or ordered mail order for cheaper if I could wait. Either way never bought SIPPs by themselves, only in used systems.
There are some SIMM to SIPP adapters ready made, though for the cost and fuss at this point they're probably not worth it. Definitely have a soft spot for 286s, so nice to see one taken to a level most of us could have only dreamed of.
Don't take the pins off the 40 pin dip. Just position the sim against the ice and solder the pins to the simm. The break the pins off by flexing it back and forth. You could file the end of the pins off carefully if you wanted too, but id just leave them as is. It won't hurt anything.
Not in post communist countries, I have some 286 boards with '93 datecodes. And even some XTs from early 90s, compared to local production 8bit computers from the 80s or something like C64 or Atari 8bits, XTs/286s were way more useful for office work.
That's because most manufacturers had made the slight modifications required to support 386sx chips since the whole point of the 386sx is that it essentially was a "386 overdrive" that fit nicely into existing 16-bit circuit designs. Then manufacturers didn't need to go to Harris and AMD for high clocked chips. That was pretty much the end of the line for Harris as an x86 chip manufacturer. AMD, on the other hand... you've probably heard of them. 🤣
The 286 survived into the early 90s in some markets because of super cost reduced designs like this with a very low component count and 286 second source CPUs being available at very low cost. This was by no means the most obsolete tech still being produced as new then: Commodore was selling new C64's - a 1982 8-bit computer - into European markets until the day they went bankrupt in 1994.
In the late 80s/early 90s, those SIPPs were being phased out and so I'll bet the motherboard manufacturer was able to get those sockets for next to nothing because the suppliers couldn't get them out of their warehouses fast enough. On the other hand, they were able to sell the memory modules for a slight premium because supply was dwindling and if you had no choice but to buy SIPPs, they would make an extra buck off of you if they could.
The keyboard controller is almost definitely running down clocked from the system clock. Very often I see it tied directly to the 32khz oscillator for the real time clock. Early XTs and even 286s hadn't standardized the CMOS / system clock (as it's not always present) it's not surprising that it is using a fractional clock to the system clock. So when you put the 4 MHz oscillator in there the resulting clock may have been below the refresh limit on the dynamic registers in the keyboard controller. The advantage to using a slower clock on a keyboard controller means fewer counts for delays such as de-bounce thus requiring less space to store those counters.
If I recall SIPP's were more expensive at the time, but it was not by as much as the pin socket vs the SIMM socket. I can't recall the exact pricing though. SIPPs did not have a long life in the retail space. They are just a variant of the SIMMs and you can make SIMMs into SIPPS by soldering pins on, which is likely less dangerous to the board than removing the Pin sockets from the mainboard, but if you can manage to remove the Pin Sockets, then all power to you.
The bios offers the slower CPU speeds because old XT programs were reliant on the actual CPU speed for their timing. Having that option in the bios is better than having to run a slowdown program and more convenient
A very delicate de-soldering job. I was wondering if anyone have experience using a heated (regulated) plate laying the board on, to remove the connector strips. Not tried it myself, but seen old household irons used as tthe heater plate, with strips of laminate to fence off the heated area. Otherwise I'd go with Adrian's method to avoid too much grief.
You made the day of a 51 years old boy! It brought me back to my endless nights of fiddling, programming and trying to muffle the sound of my dot matrix to avoid awaking my dad... lol! 8 mbs on a 286.. crazy!
I'm a tad older but it took me back to those good old days.
Looked in an old computer shopper 1992, several sellers have SIMM memory listings with a line that says: ” SIPP $5 more per module”
I had a 386dx20 with sipp sockets. When I had the money to upgrade to 2 MB, I went to a small local computer shop and I was told to wait a couple of minutes, because they had to solder the pins to their simm modules. I guess that was the reason for the 5$
@@daybyterI imagine that the might had some kind of rig so they can place the SIMM and solder the pins quickly and reliably... Or maybe I'm overthinking and they were really good at soldering 😅
@@RetroTinkerer Nice, good find! So $11 more per module (in today's money.) Not huge by any means, but seemingly money that didn't need to be spent. I can't imagine the 286 board would have cost $40 more just to add SIMM sockets at the factory.... but who knows. And also interesting they had the pins and likely a jig to help them solder the pins on.
I seem to recall something happening with DRAM / silicon prices around 1989 ?
I remember the Kobe earthquake in 1995 had an impact on DRAM prices in the late 90s, but I can't remember what the 1989 thing was?
Maybe it was the Loma Prieta quake in San Fran in October 1989?
What I do remember more clearly, is around 1993, when Dad bought 8MB of SIMMs for my PC.
It cost £200 (GBP) back then, so around £25 per Megabyte. :o
The oscillator next to the FPU socket is for separate clocking the FPU if desired. The jumper should choose between synchronous mode for the FPU which would be the half of the CPU clock (24 MHz /4) or asynchronous mode which gives you the possibility to clock the FPU higher if you install a 287XL or ITT up to 20 MHz. I recently pushed one of my 286 boards to the limit with this experiments. Cheers, Peter
Exactly. Good info.
This is great!
I was a PC repair and clone build technician from 1990-1995 in Denver, at the start of my career.
I'm 47 now, and this brings back great memories!
This is an exciting video. Changing to more modern memory slots. Very cool. Very impressive. I have never seen any of those old type memory with the pins installed. I started building computers in 1995. And where I worked, I purchased some of their computers. I know when I first started, I had to install a card with the IDE pins. And the floppy as well. I built computers for neighbors, and I could warranty them for 3 years, that is what they gave me on every computer board I purchased. Watching you convert the sim sockets Rocks! I am so happy it worked!
I wish I could have watched you on youtube in the 90’s. I’ve learned so much from you. Thank you.
Was more fun back in the 90s not having UA-cam.
You can certainly make use of the thin solder in a pinch, just twist together several strands
And he has three spools to draw from at once so he doesn't have to do (so much of) the loopty-loop action to build up the thickness.
Luckily I have my old 0.8mm stuff again -- I was shocked how the price of US made solder has seemingly double since I last ordered some in 2019!
Fantastic conversion!
I converted simm modules to SIP by using "turned pin" style male-male pin headers. They're pretty easy to solder on, they're the correct pin pitch, and they mate perfectly with the turned pin style female sockets on the motherboard.. What's easier though is you can just stuff the raw simm sockets in to the sipp headers. No need to solder or anything. If the spacing is right just stick 'em right in and they work. Granted de-soldering the old sockets and putting the headers in proper is better but the easy way works in a pinch.
Also nice this motherboard with those "3 chip" modules. Older 286 boards often only post with the "9 chip" sort.
Indeed I think this is the best method to make SIPP modules, I have also done this on a 386 mainboard. Though if I had the SIMM sockets Adrian has at that moment, I would have used those in the same way he did. And the board really looks cool like he said, and it looks very professional.
Nice, I have some of those male to male strips, I think that's a much better idea than using pins from old chips. I'll just stick a strip of 30 pins into one of those SIPP strips I removed from this board, and I can solder up some more as tests. I have a bunch of other SIPP motherboards for testing.
ahhh i can remember back in the late 80s we had a 286 at home .... great memories with monkey island and lemmings and stuff :D
When I used to work as a PCB assember and later as a tech, I was shown to bend a pin or two over carefully to hold a socket or IC in place prior to soldering. This made the assembly and soldering process go pretty quickly without risking having something loosen up and end up partially soldered on the board. This was really important prior to putting PCBs through the wave soldering machine.
Hi Adrian, Great job you did with this motherboard. Nice to see you working and testing on. Greetings from Steven from the Netherlands
I've added just round pin strips and soldered them on simm. The one you can use on breadboard or wherever. Worked beautifully.
@17:01 Love the enthusiasm!
Nice to see CMOS which isn't focussed on changing fan LED colours!
SIPs soldered directly to the board are hell.
I've pulled a bunch of SIPs from industrial controller boards. The solder used to mount the pins must have a much higher melting point than the solder used to mount them on the board. I suspect there's some silver in there.
Excellent episode, full of technical tips. 👍
Great for resisting vibration, but hell if you need to upgrade.
That much RAM in a 286 cries out for Xenix or OS/2!
I think I had a copy of SCO UNIX back then... old memories are coming back!
Is really about the only thing that would make good use of a hot rod 286 with that much RAM - well, there is the 286 16-bit version of OS/2, if can find a copy of that. Win 3.x would sort of too, but it’s such a kludge compared to actual protected mode operating systems that have been designed for that from the ground up.
I started my computer life in the 80's and think I remember SIPs came before SIMMs. SIMMs came about to make it easer to fit modules can cheeper to make them.
simms are steamlined sipps
Unexpected Adrian and Michael MJD crossover! :D
It might be my bad recollection but I seem to remember in your second channel, in a video about a box of random junk, was an industrial equipment board containing a "Make-it 486" CPU upgrade for 286. Assuming it's not my memory playing tricks, you need to test it out on this motherboard!
Somewhere in the early 2000s was the best time to get this era of equipment. Even though my first computer was actually a 386SX, I got to buy up older stuff severely cheap in thrift stores and the like. I once got an original IBM XT with a working hard drive for $5! Imagine that today. Anyway, I did at one point have some generic 286 machine and found the Make-It 486 part, and sure enough, it did come up as a "486" as far as any diagnostics were concerned. However, an attempt to install NT 4.0 on it failed, although I don't remember precisely why.
Sadly I don't have a 286 -> 486 upgrade. I have a few 386SX -> 486 upgrades though. I also have a 286->386SX upgrade, so that would be another way to push the limit on this board.
wonder if with enough interposers you could get a pentium running on a 286 mb. its been done on a 386 mb
@@curtissimmons1085 I imagine anything is possible with enough glue logic. I have an "Intel Inboard" 386 upgrade designed for the original IBM PC/XT, which sure enough can boot that machine as a 386. But the reality is, it's just an SBC jammed in an ISA slot and basically takes over the former 8088's bus. There's very little of the original PC doing anything but being reduced to a backplane.
One fun thing though is the Inboard's 386DX is socketed, so any upgrade part in theory should work in it. I found a 386 -> 486DX2-50 upgrade part, and that's how that XT boots now. (Amusing to see Microsoft Diagnostics say "486 CPU, XT bus")
That said, depending on your exact definition, there's no reason that you couldn't do something like that to make an early motherboard have more capability. But it's not quite the same as simply inserting or snapping-over an upgrade module that takes over as CPU but leaves the basic "spirit" of the original computer intact.
Part of the problem will be clock timing with all the other guts... You'd have a CPU craving high speed buses and yet being bottlenecked to some relatively incredibly low speed bus.
@@curtissimmons1085 I'm imagining a stack of interposers that would dwarf a modern CPU cooler :D
Great video! Love seeing this board upgraded and it's very neat to see the chip(set) options!
Best ADB video I have ever seen,
My favorite old-computer overclocking adventure was getting an Am5x86-133 running with a 50MHz base clock. I needed to use a giant heatsink, but it was stable.
with the super thin solder, you can always double/triple etc it up, twist it together and it will work just like a thicker solder wire
A 286 had a faster clock rate, didn't have the 8/16 bus bottleneck, didn't have multiplexing, had a 6 byte PFQ like the 8086 (unlike 4 on the 8088), had a separate address unit, and a hardware multiplier. And there were additional instructions for better control.
My guess would be that that first oscillator location is for a separate clock for the FPU.
I think the oscillator Adrian has connected is originally for the FPU so you can run one at a clock different from the CPU. But since there is no FPU in the system it can work like that.
We had this board in my dad’s pc! We upgraded from a turbo XT. 1MBx4 SIPP RAM, 16 mHz Harris 80286, and i287, too. We were able to skip a generation from 1992 to 1994 and later would put an i486sx in that case.
I only worked the PC repair shop from Aug'95 to Sep'97. Only saw one mobo with SIPP. Never saw a 286 with that much memory. Never saw simm slower than 80ns (that I know of). You show all the coolest different stuff.
One of my earliest vintage PC purchases was a unknown motherboard for £10.
Hit the jackpot, it was a 286 board with 4mb on the board in DIP chips and 4 SIPP sockets. Even better, it has a 286-486 upgrade chip in it. Really nice board, but I did have to repair a few Varta ravaged traces.
You used to get simm sockets with thicker pins, that would just push into the sipp slots. They were a bit wiggly, but they worked.
286 Motherboards are so nostalgic!
You should totally do a video going through writing a simple TSR in assembler or C to make a keyboard combination change the clock speed on the fly. Since it's just changing memory registers, it would be super simple to do either hooking onto int 9 or int 16.
If you're interested and you see this, I'd be willing to write one for you with the relevant information (though it would be fun to watch you do it if you have the knowledge or you just want to learn). I used to make a lot of drivers for DOS back in the day.
I have a big spool of some tiny solder I’ve been trying to work through for years. I take a length of it and fold it it half, then twist the whole length up like a rope and it’s almost big enough to use 😅
Hi Adrian, many years ago (1980 - 1990's) i worked at a PC manufactures, and we use to also add the SIPP pin to SIMM Modules, the pins came on a roll of 10,000 or so and came in various pitch spacing, so it was easy to cut a batch of pins for modules which were held in the pitch spacing by a metal ribbon which you broke off after soldering the pins, with a bit of flux and a little skill you could convert a SIMM to a SIPP in about 10 Minutes.
Regards Martin
LOL you got that bad boy running at Commodore 64 clockspeeds! Impressive, Adrian!
I had a 80386SX board that had SIPPs. (Probably a modified '286 motherboard!) What a pain. Those pins were so fragile!
What a nice upgrade for this board, it's lucky that he had these new old stock SIMM connectors. I board an old 386 board which also featured these sipp slots. The seller forcibly cramped normal square pinheaders soldered to the modules into the SIPP turned sockets which was rude and destructive. Yes, in case of the 5170 there is full speed on the ISA slot cards, no advanced chipset mechanisms there, only the wait state decoders in U87. The zero wait state can be defined per memory decoded region with the 0WS input designed by IBM in the AT, so variable speed can be used transparently per memory region which is a very cool design. The board being used in the video is years further developed and may even operate separate buses for the RAM and the ISA slot connectors. This would allow for much easier speed separation in accessing the different buses. I am developing a 286 AT and made a thread online and published in GitHub, in my design all the RAM is SRAM and running on a ISA slot connector. I haven't tested zero wait states on the SRAM yet but I will try this out at some point. For 16 bit access, for example the ROM chips will definitely need a wait state at above 10Mhz or the system won't POST. Anyway I love this video since I am working on a 286 system for more than a year and have seen a lot of stuff, so it's great to see someone else tinkering with this type of system and seeing what happens.
I really enjoyed this video. Being you were able to upgrade the Ram slots. Very impressive. I have plenty of those old boards in storage. I will do the same upgrade. Thank you for up loading this video. My experience with computers started with the 8088 models. I lived close to a Fries Electronics. That is where i purchased my computer boards, ram, Processor chip and Ram. I really do not like to purchase computers, I rather build my own.
I replaced my Hakko 301 with a Metcal desoldering gun. The difference is night and day. The Metcal can suck up an enormous amount of solder without needing cleaning. And it has better heat up time, stability and quick change tips in a bunch of sizes. They’re expensive new but you can find good deals used if you’re patient.
new yes, but used I’ve managed to find Metcal desoldering stations for similar or less money than a new Hakko FR-301. Just need to keep an eye on eBay, Craigslist and FB marketplace.
Metcal's are definitely a great upgrade. I just bit the bullet and replaced my crappy old desoldering station with a Pace. I wish I had saved up for one years ago. Its quality and speed is absolutely amazing. If you value your time, it's totally worth it. Also, it helps to add a nice big dab of fresh solder to old joints before trying to suck them up.
Not to mention the variety of quick change tips available for it. Just this morning I desoldered both the 1mm leads and 3mm mounts of a small stepper motor in about 2 minutes.
Desoldered stuff for years with a cheapo Chinese gun, upgraded to a Pace SX-90 + MBT-250 a year ago, never looked back
This is so nostalgic for me. I only encountered SIPP once from garage sale computers when I was very young. I proceeded to throw away the motherboard and sipp ram since it barely had any and I had a box full of simms. Was neat to find all this out about this. Thanks for the post.
For the really thin solder, what I do, it pull out a long piece and fold it in half and twist it. If it is still to long then fold it in half again. My silver bearing solder is really thin like this.
This board offers nice opportunity to get perfect system for early 8088 games. Nice!
For converting the SIM to SIP you could just use a row of male-male header pins, spacing will be easy, alignment will be easy, just lay it on the SIM and solder it on.
I have been having nightmares about changing the memory sockets on my Amiga 4000, and although I have also watched others, like Chris, make it look easy, I still didn't feel confident enough to give it a try. Watching your successful effort, has now encouraged me enough to give it a try. Nice work
That's one of those, back in the day, who would EVER have 8MB?!?!?
Nice work Adrian.
My first computer was a 1mb 286 with a 40mb harddrive EGA monitor in 1989 (I was in middle school). cost my parents $1350. I was so happy, read the DOS 3.3 book from cover to cover.
Played games on it. First upgrade was a 3.5 HD floppy drive and later a 2400 baud modem when I discovered BBSes. Later ran my own BBS with a friend.
I have another idea for the SIPP/SIMM conversion dilemma. Take some single SIMM sockets and put SIPP legs on those! Then you can swap in any SIMMs you want, and move the "format converters" to any motherboard you want. Of course if you find the upgrading process to be reliable and not too much trouble, then by all means stick with it! Making every motherboard in your collection not suck in this particular way is certainly one way to solve the problem.
It's good that you have three rolls of the "hair thin" solder, as you can put all three spools on a stick and twist the three wires together to get something that more closely matches the solder diameter you're accustomed to. When you only have one roll, you have to do it in segments which is much more annoying, although in either case you can use a drill or power screwdriver to twist the wires together, and collect the result on a new spool. It's not like any twisting stress is going to matter once they're melted.
That is exactly what I do with my tiny angel hair solder roll. The drill trick makes quick work of it.
Outstanding episode!!! Keep up the awesome work!
Great video Adrian, lots of useful information. Cheers.
Back in the day I got a pile of proto boards that had 8x1mb SIPs per board soldered to the board. I would strip them and re-pin them with sip strips I got from a local shop and use them for various upgrades.
To make a SIPP, put the pins in a strip of female header, butt it up to the simm and drag solder them down.
Takes about 40 seconds a SIPP when you get practiced. Alignment is perfect.
Works when SIMM conversion of the mobo is hard or impossible.
Or just take pin header with correct pitch...the additional hight won´t matter....
I would put a piece of kapton under the crystal. On those old motherboards, on corner of the crystal has a positive voltage on the pin. If the pin on the socket is proud of the plastic surrounding it, it can touch the grounded body of the oscillator module it will short out and usually will burn an internal trace on the motherboard. I lost a few motherboards like that. In retrospect, I should have tried adding a bodge wire from the 5v rail to that pin. But I learned to put a piece of some sort of insulting plastic under the oscillating to protect it.
You can buy those pins, they're called lead frame pins. They come on a strip (or frame) for easy installation and then you cut the strip off after they're soldered. I don't know how easy it is to get them in small quantities though.
Back in those days, I used to remove the pin and convert SIP to SIMM, they're same thing just SIMM socket is more expensive, price of SIP and SIMM just a few USD, when you make a lot of these those metal pins just cost human labor, at those days is not that expensive.
Some very early produced SIPs were narrower lacking sides that lock in SIMM slots. You could still use those as SIMMs by using one proper SIMM to hold su h converted SIP in place. They also worked on motherboards where simms were just pushed downward instead of inserted in angle and then "click".
I had one of those 32-bit VLB cache hard drive controller cards that used these same memory modules. It was a Tekram and I could swear it doubled performance of my 486dx2
Fold the solder over and twest it and it will be usable. No since in wasting it.
The 286 was a real step up in capabilities but they were quickly eclipsed by the 386. I had a 286 system that I eventually jacked up to 2 MB and 12 MHz. It was an awesome DOS/Windoze machine. And I justified trying dual boot Linux on my 386 machine having the 286 as a backup in case I screwed things up on the 386. I did have to heat sink the 286 and 287 for the overclock. They ran HOT at 12 MHz.
I had a 16 MHz 286 from toward the end of the run, when 286 boards at any speed were becoming hard to obtain because production had all re-focused on the 386sx (which uses pretty much the same motherboards, with socket and BIOS changes). Clock for clock, it was slightly faster than a 386DX -- but of course, only for the things it could do at all. That was also the machine where I accidentally set my CGA card on fire when I set the CPU/ISA bus ratio to 1:1. (It was repairable, but I wasn't gonna mess with a 16 MHz bus after that.) From a reliability standpoint, and for ease of finding matching contemporary parts, I think the "best" retro 286 system is a slow 386sx! Or skip the 286 and 386sx entirely and de-Turbo a 386DX when you want that 286 experience.
I swapped the 8 SIMM sockets on a Macintosh IIcx recently. They were the crummy plastic ones and all the tabs had snapped. It was quite tricky getting the sockets, I also used a combination of hot air with a desoldering pump. Definitely something I don't want to be doing on a regular basis 🤣
I still have all my old Computer Shoppers back at my parents' house. When I get back there, I can look up the SIMM vs SIPP pricing.
I once soldered wires from SIMMs straight into an STFM board (not even SIPP, sockets IIRC) and that was still less hassle than upgrading my original ST
Happy to see the plexus im in the background
You are a GENIUS. I only know 1% of what you know. I understand what you are doing, but I cant do the same! 😂😅
To convert SIMMs into SIPPs you can use the "precision" type (round pins) goldpin male connectors.
Another great video! You can still get the pins to make SIPP sticks from 30-pin SIMMs. Look for y1 leadframe. I think the TE 1544210-2 leadframe would work.
I had a 286 board with 4mb on the mobo in CHIPS, not memory modules! And it had a Harris 286 CPU running at 20 mhz. It was awesome. I have another 20mhz Harris chip in a drawer, just bought it from eBay a few months ago.
Should note that this board seems to have a development BIOS - says so right on the splash screen. Explains the wacky super-fine-grained control and the direct register manipulation screen.
Just what I needed after a long day!! Thank you, Adrian!!
Frequency divider is needed because some software didn't work correctly (was too fast for example) when run on higher frequency clock. It's mostly code that doesn't respect timer but instead use instructions in sync with their execution time. It's in similar way done on older 8bit machines.
You can buy the pins on a strip that fit on a PCB.
I've had good luck soldering a row pin headers to things, and then heating/removing the plastic spacer. (for other random projects) That might work for installing sips pins on to a chip.
I've only been soldering for a few years now, but I'll just say what I've observed/works best for me. When you soldered in the slots, you went along a row working your way to the other side. Myself, I like to tack either side first so that the slot cannot move while I'm soldering it. Once I tacked either side with two spots, I verify the slot is flat. Its more so a issue with dupont headers like on a rpi or pico. Plus if you're concerned about melting the connectors plastic, you can alternate between either side.
Also flux. The thicker kester stuff you had was likely rosin core. The smaller stuff probably wasn't. Either way. I always add extra flux, it makes your joints nice and shiny. MG chemicals make a alright one. The go to seems to be Amtech stuff. However they recently rebranded to Stirri and you can buy DIRECTLY from them unlike Amtech.
Hot air is kinda eh. I warped a xbox 360 board using it to clear holes when I recapped it. However as long as you don't put too much heat into it too fast its fine. Just be careful, I'm sure you know. Give a board the case of the bends.
Reminds me of the Kaypro II that I had way back when. It had a Z80 with a clock speed of 2.5 MHz. A popular mod back then was to double the clock speed to 5 MHz. The recommended way to do this was to put a toggle switch on the front panel so that you could switch back and forth in case of problems.
In 1994 I had a 386SX PC that used those SIMM memory modules. On that era, internet wasn't available on the entire island, only the capital city. I had to buy those modules locally. 1 megabyte of RAM cost around a $100.00 each. I put a Math Coprocessor, another $100.00, and added 6 megabytes of RAM to run Lotus 1-2-3 for Windows, for Windows 3.11.
I never had to worry about the price of SIPP memory because I never dealt with it personally. At a company I worked for in the early 90s, we had a bunch of truly awful Memorex-Telex '386s that had them. The machines were weird to begin with and the issue we had was memory working its way out of the SIPP sockets over time causing all kinds of problems, as you can imagine made them even worse. When they were finally scrapped, we couldn't wait to dump them into the scrap pile.
To convert SIMMs to SIPPs you need leadframe pins that will clamp around the PCB and can be easily soldered. Digikey has A122965TR-ND (Tape and reel) and A122965CT-ND (cut tape.)
The previous episode with the board that had the damaed RAM catches comes back to mind, remembering you had that vast supply of NOS sockets, that board could have the sockets replaced now... :D
15:52 it almost looks like the through hole plating came out on the 16th pin. However, if that happens, the pad most likely is toast too, and it seems fine there. Obviously, since it worked.
I remember back in the 90's buying a bunch of SIP modules at a local car boot sale. I had no idea about them at the time byt was super excited of the prospect of potentially upgrading my 386 back home as all I had was 1MB in simm format. I tried to look up as much information back then on them but I found nothing anyway I remember tracing out the memory modules with a multi-meter to make sure they where electrically the same. In the end I de-soldered the pins and tested it in the system unfortunately They never worked so may have been damaged somewhere along there life. But I always wondered about these if they would have worked.
I'd use those generic pin sockets that you can buy for cheap to build a sacrificial jig to hold the pins in alignment while you solder them to the board, rather than having to solder them one-by-one.
It wouldn't be sacrificial if your soldering's halfway tidy. I always try to have connectors plugged into something while soldering, to sink heat away and hold the pins in line when the plastic gets soft.
Oh C'Mon! I Want One!!
I love this SCAT design because the chip is so elaborate that it includes so many options and flexibility. The AT system is very complex which is why certain speed configurations may not work. In case of running the system bus at 12Mhz as he has tested, this should easily work. I designed a 286 system based on the 5170 core concept using 3 CPLDs, and this is easily able to run at 16Mhz, with no restrictions on the ISA slots and system bus. I was surprised to hear that he was not able to boot at 12Mhz, I suspect the CF card is the reason and should be replaced with a fast laptop harddisk. The disk controllers on those are much better than the controllers inside CF cards. About the oscillators, I believe the SCAT has various clock configurations, where different input pins are used to base the CPU clock on. The datasheet will explain this in more detail, which will explain the PCB layout. I think it's possible to use the other pin on the SCAT if configured correctly. The keyboard controller should run at 8Mhz to be sure it won't cause problems, a little faster does seem to work also in my tests but it depends on the keyboard controller chip.
Pimp My 286
Just add a 60" flat screen between the ISA slots, put the crystal oscillator on hydraulic lifts (for easy removal), and gold-plate the SIMM pads.
First time I overclocked I didn't even know it. Had a 486 board that had been upgraded from an 80mhz to a 120, then later I got an AMD 133. There was no jumper setting to equal 133 so I kept trying different jumper combs until it booted, later I noticed the benchmarks all showed it overperforming and it was running at 160. Back in those days you could easily get 20-50% better performance by overclocking I was one of those people who had a dual celeron 300A board running at 2X450 and easily beat the P450 and that was king of the hill for desktop computers at that time.
Absolute Genius!!
I have a 386sx that I've used the bios clock divider to closely match the 286 12mhz computer years ago. Other than giving me better memory management, it was very close to what I wanted. The divider could also let you slow down a fast processor for older games when needed.
That's why I often say that the ultimate 286 is a 386sx. You won't notice the difference unless you specifically invoke the 386 part of the instruction set. The 286 is ever so slightly faster clock for clock, but considering the 386sx clocks a lot higher (if you want it to), that is pretty much lost in the noise. You also get more sophisticated BIOS and better support for large (by the standards of the time) hard drives. Make it a 33 or 40 MHz 386sx and you can kinda skip the 386DX generation and move on to the 486DX. This is a practical solution to the fact that most of us don't have digital basements to keep one of everything.
There's the explanation for such dividers range - some custom programs might provide errors while running at higher speeds.
It might be signal integrity over your bodge wire. Some coax might help.
Local shop used to sell both for same price, both for two much, so bought memory in CompUSA, Fry's Electronics, somewhere on sale, or ordered mail order for cheaper if I could wait. Either way never bought SIPPs by themselves, only in used systems.
There are some SIMM to SIPP adapters ready made, though for the cost and fuss at this point they're probably not worth it. Definitely have a soft spot for 286s, so nice to see one taken to a level most of us could have only dreamed of.
32:45 Nice EEVBlog cameo! 😀
Don't take the pins off the 40 pin dip. Just position the sim against the ice and solder the pins to the simm. The break the pins off by flexing it back and forth. You could file the end of the pins off carefully if you wanted too, but id just leave them as is. It won't hurt anything.
1990 sounds really late for a 286 based board. Nice job on the simm slot install 👍
ive seen a few over the years with 1992 dates on the components
Not in post communist countries, I have some 286 boards with '93 datecodes. And even some XTs from early 90s, compared to local production 8bit computers from the 80s or something like C64 or Atari 8bits, XTs/286s were way more useful for office work.
@@snap_oversteer a fast 286 could run windows 3.1 and do some thick wire networking, more than enough for remote access to servers applications.
That's because most manufacturers had made the slight modifications required to support 386sx chips since the whole point of the 386sx is that it essentially was a "386 overdrive" that fit nicely into existing 16-bit circuit designs. Then manufacturers didn't need to go to Harris and AMD for high clocked chips. That was pretty much the end of the line for Harris as an x86 chip manufacturer. AMD, on the other hand... you've probably heard of them. 🤣
The 286 survived into the early 90s in some markets because of super cost reduced designs like this with a very low component count and 286 second source CPUs being available at very low cost. This was by no means the most obsolete tech still being produced as new then: Commodore was selling new C64's - a 1982 8-bit computer - into European markets until the day they went bankrupt in 1994.
In the late 80s/early 90s, those SIPPs were being phased out and so I'll bet the motherboard manufacturer was able to get those sockets for next to nothing because the suppliers couldn't get them out of their warehouses fast enough. On the other hand, they were able to sell the memory modules for a slight premium because supply was dwindling and if you had no choice but to buy SIPPs, they would make an extra buck off of you if they could.
Just use dedicated pins if you need to make SIPP. For example 1544210-2 by TE-Connectivity. Available at Mouser for example.
The keyboard controller is almost definitely running down clocked from the system clock. Very often I see it tied directly to the 32khz oscillator for the real time clock. Early XTs and even 286s hadn't standardized the CMOS / system clock (as it's not always present) it's not surprising that it is using a fractional clock to the system clock. So when you put the 4 MHz oscillator in there the resulting clock may have been below the refresh limit on the dynamic registers in the keyboard controller. The advantage to using a slower clock on a keyboard controller means fewer counts for delays such as de-bounce thus requiring less space to store those counters.
If I recall SIPP's were more expensive at the time, but it was not by as much as the pin socket vs the SIMM socket. I can't recall the exact pricing though. SIPPs did not have a long life in the retail space. They are just a variant of the SIMMs and you can make SIMMs into SIPPS by soldering pins on, which is likely less dangerous to the board than removing the Pin sockets from the mainboard, but if you can manage to remove the Pin Sockets, then all power to you.
really awesome! If you don't have the time to desolder the SIPP sockets, just plug in the 30pin SIMM sockets :) Works like a charm :)
Well done
The bios offers the slower CPU speeds because old XT programs were reliant on the actual CPU speed for their timing. Having that option in the bios is better than having to run a slowdown program and more convenient
I'd stuck the SIMM sockets into the headers and jbweld them down :D
I think the oscillator socket was meant to asynchronously clock the 80287...
Good work ❤
A very delicate de-soldering job.
I was wondering if anyone have experience using a heated (regulated) plate laying the board on, to remove the connector strips.
Not tried it myself, but seen old household irons used as tthe heater plate, with strips of laminate to fence off the heated area.
Otherwise I'd go with Adrian's method to avoid too much grief.