We use these wireless RS-232 devices with our RF room testers. One goes on the controller unit, and the other on the transmitter inside the RF room. We set up the transmitter inside the RF room and the sweep antenna on the controller outside the RF room. The two devices allow a full RS-232 at around 2 GHz, and the RF room is tested at 100 MHz and one of 2 other frequencies. The 2 GHz used in the transceivers will work fine, because the RF room is tuned for MRI frequencies and doesn’t filter 2 GHz. Obviously, I’m talking about MRI exam rooms which are integrity tested at 100 MHz and 127 or 64 MHz. So those wireless RS-232 transceivers are used with our RF room test units. The dip switches are for matching the two transceivers together, like addressing.
Love overclocking sessions! I used to spend hours squeezing every drop of performance out of my old systems. Endless upgrades until something shiny and new caused me to upgrade.
My recollection was that IBM downgraded the AT to 6 MHz before selling the first units then offered an upgrade to 8 MHz. Worked on one for many years. Love these videos.
More Adrian, this is the sort of thing about your channel I love, random tinkering and testing. Bring on the rest of the boxes mystery contents please.
15:09 Gahh!! more flux!! Flux will stop you from leaving solder blobs everywhere! They're annoying because they're just sticky enough to not come off the solder mask easily but I don't feel like you can just leave them there. Also I think if you warm the board a little it'll stop the solder from solidifying on contact quite so easily. 15:34 Pretty good job! I do this kind of work a lot. What I would do is solder one side, cut, then hold with tweezers and solder the other side. With light wire like that you end up with what you saw quite often which is that when you try to solder the second side you end up undoing the first joint. It can be hard to actually hold wire with tweezers sometimes so what I do is flip the tweezers around and pin the wire down with the butt end (or just hold it down with the side of the tweezers). 15:45 I would always add more flux before reflowing a joint like that. It came out looking okay, but when you melt solder with no flux around it *really* wants to oxidize. 15:52 Looks like you did just fine with the snippers, but a tip for fine wire work like this: #10 x-acto or scalpel blade. You can cut right through the wire with the blade but you might end up gouging the board surface--what you can do instead is push the blade into the wire a little and then use the blade to hold the wire in place while you wiggle it to break it off. 18:46 I usually prefer to use some a sewing needle to do scraping like this but if you don't care about your tweezers that's also fine. 19:38 Copper takes a long time to turn green purely from exposure to air. Oxidation from air also tends to be self-limiting, not penetrating too deep from the surface. Still a good idea with the lacquer though.
One tip for you when dealing with dead boards that have QFP/PQFP chips on them: Inspect the pins with your microscope and tweezers, they like to come off the board either due to degrading solder or flexing of the board and can easily cause a board to appear dead. It's by far the most common fault I encounter while fixing old boards that have no post code activity. And as an aside if you see no post code activity at all on the card it is not the ram that has a problem. Ram init occurs relatively late after rom is executing already, so if there is no acitivity you either have a bad rom, or more commonly bad Voltage rails or lifted/degraded chip contact pads.
Yes, way back in 1984 I overclocked an IBM PC/AT to 8 Mhz! I was an IT techie working for an International Oil company based in London England. We worked closely with IBM and tested and verified all of the early IBM Personal Computers and associated Hardware and Software products for use in our company. I remember getting 4 of the very first PC/ATs in the country for testing. Based on Technical journal articles and chat on BBS's that discussed early overclocking the AT to 8 Mhz and being an electronics enthusiast, I purchased a couple of 16Mhz crystals. I remember they had wire legs so I had to solder proper sized legs to them for the on board sockets. But yes I successfully ran two of the machines at 8 Mhz with no problems at all, apart from running noticeably hotter. I received many visits from other techies, from within the company and even from IBM UK who were based just along the river, to see benchmark utilities and other programs running at "breakneck" speed on my souped-up AT. Yes we were all geeks even back then!!
The 386 motherboards like this one were used in slim form factor desktop PCs made by many off brand manufacturers. I had one and loved it! It was so much smaller then usual PCs at the time
Nice video, looking forward to the next part(s)! Maybe move the little monitor to the right hand side so it's nearer the camera? And recheck that trace on the 486 board - I think it's still broken further along to the left of the repair! Look at 17:40
Hi Adrian, I worked for a Taiwan company in 1990/94. Sunnytech, this company made this motherboard you have here. I remember most motherboards had buffer problems.
22:09, the Firefly Serial to Bluetooth resembles a Serial to Bluetooth device stenographers use to connect their steno machines wirelessly to their computers.
Are we interested in seeing these types of videos? Speaking just for myself; Of Course!! This type of stuff is why I subscribed....and I'm sure I'm not alone in this.... Seeing you revive these old boards and computers is why we're here! Of course, your charming personality, and detailed explanations as to why and how you do the things are all a big part of it, so yeah... please, lets see the other items in the box...don't tease!! :-D You are great, do your thing, and thank you for doing these fascinating and entertaining videos... :-)
I remember reading in a computer repair / upgrade book that people did over clock the 5170 but changing out the crystal like you did. I also remember reading that IBM updated the BIOS to not support over clocking in later BIOS revisions and I'm wondering if this is why your XT/IDE is failing. IBM had mentioned that the type 1 motherboard can't handle proper timings higher than 6 MHz.
IBM equipment - bulletproof! THE best. I remember a PS/2 model 55 I ran across in my consulting days that printed invoices on a dot matrix Proprinter. It has been running flat out for 11 straight years! No power management, no time out, just 24/7 operation all the time, every day, for 11 straight years.
I had no idea you were in Portland. I'm outside of Astoria. You're my current favorite TH-camr. Such an interesting channel. Back when I was a kid my brother and i got like garage sale computers and tinkered with them. In retrospect, I wish I had those old computers now. Lots of 286, 386, and 486 computers. Keep it up man!
A strange mix of EEV, branchus and Phil’s computer lab… love it Learning a lot. Some things I wouldn’t mind you could explain a bit slower for us mac fan boys. Still learning the pc world.
Best video for a while. Random junk could be new series of content. Love idea of overlooked stuff getting attention it shouldn't really be getting. Thumbs up from me
In 1986 I used to run my IBM PC/AT at 10MHz with a 20MHz crystal. For some reason I had to swap the floppy drive for a different brand to get it working at 10MHz. At one point I had 12.5 MB memory in it for testing purposes.
Those BT serial adapters would be handy for us IT folk. Makes it easier to stick one on the back of a network switch, router, server, storage appliance, etc., and configure from a laptop without having a cable stretched out from behind a rack. :-)
I know I'm a bit late to this video. SIP memory is outright awful! I dealt with a bunch of Memorex Telex '386 PCs using SIP memory. The machines would lose memory or fail to post more often than not due to the SIPs creeping out of the sockets. Eventually, the company replaced those machines with something else. On the awesome old IBM AT motherboard, the reason for the slightly bent pins on the ISA slots is that the sockets were installed manually and had to be held in place when put through the wave soldering machine. The bending of the legs prevented the sockets from lifting up. This was a common practice when I worked in PCB assembly in the early 1980s. The terminals I built back then had socketed EPROMs and other support chips as well as the CPUs. We were instructed to bend over the corner pins, alternate corners such as pin 1 and pin 20 and ensure the bent over pins were lined up with the etches. Today, none of this is an issue because these components are robotically inserted or surface mounted and human interaction is taking nearly 99% out of the process. How times have changed in 40+ years!
I have the 407 version of this scope. It can do simultaneous display on the internal screen and HDMI. I also solved the working space issue by mounting the base 180 degrees so it points backwards, flipped the power connection to the lamps around, and clamped the base towards the back at around 2 inches off the bench. Works great but is now more of a semi permanent install. Also protected the plastics exposed at the edge with some Aluminium tape.
I'm so jealous your 5170 board just worked out of the box, so to speak. I picked up a 5170 board a while back and never got it working. It's definitely a later revision than yours, since there's no bodge wires and most of the chips are dated around mid-85. Interestingly, some of the double-stacked RAM chips have a blue line on them, no idea what that means. And the keyboard controller is a ceramic Intel D8742 with a die window, which is different from the one on your board. I haven't exactly tried much troubleshooting, just the basics of checking for shorts and confirming voltages. I really need at least a POST card to see what's going on. Also, thanks for the tip about using a PLCC socket to connect a different 286 chip. Might have to give that a try someday. Anyway, very cool to see some old school overclocking! And I am glad you at least got some parts working.
@@adriansdigitalbasement Just finished watching it, I hope you can get it stable, because apart from that, it's an outstanding effort. I really like the idea of having an arduino crystal where you can change the frequency to anything you want.
Years ago we used very similar Bluetooth serial adapters for educational “scribbler” robots in an intro to computer science class. The robots had a serial port and the dongles allowed them to communicate wirelessly with the lab computers.
Weirdly enough I'm very interested in these early PC mother boards. I remember my brothers talking about working on theirs and occasionally explaining some of it to me.
Regarding the 486-DX50, maybe when fixing the trace you have bridged it to the nearby trace. The added wire looks like it is on top of the nearby wire.
"Linus Gabriel Sebastian (born August 20, 1986) is a Canadian TH-cam personality." So he wasn't born yet. The AT was released on August 14, 1984 so that's slightly over a two year lead.
We use those bluetooth to serial adapters to console into routers in the data center so your laptop isn't tethered to a short cable. Comes in handy in densely populated racks or if your in a tiny closet and there's nowhere good to park your laptop.
The Firefly is a Bluetooth to RS232 serial adapter. It attaches to a RS232 serial port making it a Bluetooth Class 1 wireless connection, capable of transmitting your serial data up to 100 meters (330 feet) away. You can use the Bluetooth Firefly with another Firefly in "cable replacement mode" to make your RS232 cable wireless, or you can pair it with a BluePort XP (battery powered), FirePlug(USB serial adapter) or BluePlug (USB adapter). The Firefly Bluetooth serial adapter may also be paired with any other Bluetooth device that supports Bluetooth SPP (Serial Port Profile). Note that many devices with built-in Bluetooth (e.g. PC's, Laptops, Handhelds and phones) utilize Bluetooth Class 2 radios, which only has a range up to 10 meters, so if you use a Firefly with a Class 2 device, you will only get the effective range of the Class 2 device and not the Firefly. The Firefly supports the latest and greatest Bluetooth specification 2.0 (which provides faster Enhanced Data Rate (EDR) and lower Bit Error Rate (BER)), but is also compatible with devices supporting the older 1.x specifications. The Firefly has dip switches to make set-up of the most common serial baud rates quick and easy. It also supports "AT" commands which allow you to change the firefly settings with messages sent through the serial port.
Also the 386 CPUs have a chipset between the ISA bus and the CPU, which the 286 CPUs didn't have. That chipset could probably negotiate or detect the XTIDE better, whereas the 286 might have been just talking too fast to the XTIDE directly.
55:00 The Harris 80286 manages to stay so cool, because Harris made a CMOS version of the 80286 processor. The Intel processors (EDIT: only some of them) are HMOS processors (which is an NMOS variant). NMOS/HMOS chips have a noticeable static consumption, mostly independent of clock frequency, whereas CMOS gates only get warm when they actively switch something.
@@eDoc2020 For some reason I got the wrong impression that Intel never made a CMOS 80286. This is obviously wrong. The dysfunctional 12MHz part in the video is CMOS. I couldn't get a good view at the original 6MHz processor, though. Possibly that one still was NMOS.
They make metalic Sharpie colours, ie, silver, gold, etc, that is easy to see on black. I had to make a whole new stand with 1ft reach and a higher column for my makshift microscope. I say keep going, because it is entertaining.
I love the video!. I love the overclocking of the IBM 5170 motherboard. I have to check my stash of old MB to see if they have those bad batteries. Thanks for the warning!
i like how you get so excited at overclocking to 8mhz lol. i remember back in the day after this stuff overclocking a pentium 75 to 200 mhz sadly it wouldnt work with my pci 56k modem which required mmx on the chip
17:24 this is where kapton tape helps. you can tape it in place and cut the other side and solder one side. then switch the tape over to the other side while you solder it.
if its 92-52, with 52 weeks in a year, why would it be written that way instead of 93-00? wait.. computers.. I get it. literally when the chip was manufactured, the computer programming for stamping the date on the chip wouldn't have the .14 'days' ... that would drive my OCD nutz
@@dinomagick Week counting starts with the 1st week, though, so it's 1-52. The last day (or 2 in a leap year) would probably just be part of an 8 (or 9)-day 52nd week.
Uh, guys. You are mis-reading those date codes. This chip was made on the 25th week of 1989. For Intel CPUs of this era only one digit after the L indicates the year followed by 2-digit week. So L9 means 1989 and 25 is the week of production.
@@mattj5155 I believe we are referring to different parts of the video. We're talking about the 386SX board shown at 5:25. Adrian read 92-52 in the copper layer and then said it probably meant 1995.
It's a common thing to believe (I did so myself) that the 386SX uses the same bus protocol as the 286. But that's not completely true. In fact the 386SX uses a smaller edition of the 386DX bus interface, limited to 16 data lines and 24 address bits (23 address lines, no A0). The interfaces are very similar, so you can easily build chipsets that support both, but the detailed cycle timing is different. The 80286 did not ensure validness of the address lines during the whole bus cycle. In particular, at least if another bus cycle is pending, the new address is output during the last clock period of the previous cycle. There is no way to prevent that. A board that supports ISA needs to latch the 286 address bits to ensure the address bits staying valid long enough. The idea of this feature is to start decoding the next address (most importantly: detect whether that address is able to take a 0WS 16-bit cycle) as early as possible. On the 386, this feature got modified: By default, the address lines stay valid during the whole cycle, so the new address appears half a clock later than on the 286. This timing would make 0WS cycles harder to implement, as the address appears later, but data is still expected at the same time. On the 386, the board can operate an input to the 386, called "NAB" (next address to bus) anytime during an active bus cylce to tell the processor that is may output the next address. This is called "address pipelining". Some 386 boards have a jumper to turn the pipelining feature on or off (possibly just interrupting the NAB line). A good 286/386SX chipset knows about the address timings of both the 286 and the 386 processor. Interestingly, Intel abandoned address pipelining with the 80486 processor, as they introduced burst cycles. Probably Intel assumed that one-word-per-clock bursting offsets the disadvantage incurred by address pipelining being no longer available, and not having to care about address or decoder output latching simplifies devices on the local bus. With the Pentium, address pipelining and the NAB line reappeared.
I have the 386DX-40 version of that SX motherboard. Mine has only 5 slots and all are 16bit ISA. It has 8 SIMM slots and a spot for a math coprocessor. I bought it in November 1993 and it cost just shy of $100.
That 386 SX-40 motherboard at 2:52 - I have the same board in my "Frankenstein 386" machine. I use it for imaging floppies and running Windows 3.11 for the lulz.
Also that IBM 5170 motherboard might even be a pre-production model, as the 5170 wasn't launched until mid-August 1984 but some of the date codes you have there are from April-May. So you either have one of the very first units sold or possibly even a review or test unit! Would love to know the history of that board.
Pretty sure the firefly modules are basically "Virtual serial cables" using bluetooth for a bridge between 2 PCs.. I had researched similar devices years ago to connect my weather station to a PC collecting data in the basement..
Adrian, I love your videos. Thanks for another great one. Your wire tinning skill on the bodge wire is fantastic. Please share some tips on that. I would recommend that you look into getting some tacky flux in a syringe. It will help with those small soldering jobs.
Hi adrian just to say yes i like those video you do all of them been watching you every time you release something by the way im from canada continue the good work
Personally I would get some of the UV solder mask or at least some of the glorified nail varnish solder mask. The problem with clear stuff is it's hard to tell where it is after you've done repairs. You can't miss the coloured stuff.
If that is a concern, why not use regular colored nail polish? Personally I am a fan of the clear stuff because it makes it easier to inspect repairs. A possible middle ground could be placing a dot of colored nail polish nearby or otherwise marking the board.
31:50 It's 4.77 * 3, because the 8088 needs a 33% duty cycle clock signal, which was obtained on the original PC/XT by dividing the 14.31818MHz by three. Intel supplied a suitable clock generator as companion for the 8086/8088, it is the 8284.
Hi Adrian! Some motherboards may not work at all without a battery. You should try using a battery pack on the external battery pins!
facts
Personally, I love these longer-form videos, I'd love to see parts 2, and 3, and 4, and 5...
We use these wireless RS-232 devices with our RF room testers. One goes on the controller unit, and the other on the transmitter inside the RF room.
We set up the transmitter inside the RF room and the sweep antenna on the controller outside the RF room. The two devices allow a full RS-232 at around 2 GHz, and the RF room is tested at 100 MHz and one of 2 other frequencies. The 2 GHz used in the transceivers will work fine, because the RF room is tuned for MRI frequencies and doesn’t filter 2 GHz.
Obviously, I’m talking about MRI exam rooms which are integrity tested at 100 MHz and 127 or 64 MHz. So those wireless RS-232 transceivers are used with our RF room test units.
The dip switches are for matching the two transceivers together, like addressing.
Nice overclocking, love how excited you get from a whopping 9mhz lol I do enjoy full box treasure hunt and exploration.
Please continue the treasure hunt. Feels like I’m back in time at a used parts computer store digging through their dumpster and finding treasures
Love overclocking sessions! I used to spend hours squeezing every drop of performance out of my old systems. Endless upgrades until something shiny and new caused me to upgrade.
Good job "debugging" that second board! Things were so much more simple back in the 286 / 386 days. Great video!
He removed the tic that marked it as good, no wonder it doesn't work any more xD
My recollection was that IBM downgraded the AT to 6 MHz before selling the first units then offered an upgrade to 8 MHz. Worked on one for many years. Love these videos.
More Adrian, this is the sort of thing about your channel I love, random tinkering and testing. Bring on the rest of the boxes mystery contents please.
We want more, don’t do the rest of the box without us!
15:09 Gahh!! more flux!! Flux will stop you from leaving solder blobs everywhere! They're annoying because they're just sticky enough to not come off the solder mask easily but I don't feel like you can just leave them there. Also I think if you warm the board a little it'll stop the solder from solidifying on contact quite so easily.
15:34 Pretty good job! I do this kind of work a lot. What I would do is solder one side, cut, then hold with tweezers and solder the other side. With light wire like that you end up with what you saw quite often which is that when you try to solder the second side you end up undoing the first joint. It can be hard to actually hold wire with tweezers sometimes so what I do is flip the tweezers around and pin the wire down with the butt end (or just hold it down with the side of the tweezers).
15:45 I would always add more flux before reflowing a joint like that. It came out looking okay, but when you melt solder with no flux around it *really* wants to oxidize.
15:52 Looks like you did just fine with the snippers, but a tip for fine wire work like this: #10 x-acto or scalpel blade. You can cut right through the wire with the blade but you might end up gouging the board surface--what you can do instead is push the blade into the wire a little and then use the blade to hold the wire in place while you wiggle it to break it off.
18:46 I usually prefer to use some a sewing needle to do scraping like this but if you don't care about your tweezers that's also fine.
19:38 Copper takes a long time to turn green purely from exposure to air. Oxidation from air also tends to be self-limiting, not penetrating too deep from the surface. Still a good idea with the lacquer though.
One tip for you when dealing with dead boards that have QFP/PQFP chips on them:
Inspect the pins with your microscope and tweezers, they like to come off the board either due to degrading solder or flexing of the board and can easily cause a board to appear dead.
It's by far the most common fault I encounter while fixing old boards that have no post code activity.
And as an aside if you see no post code activity at all on the card it is not the ram that has a problem.
Ram init occurs relatively late after rom is executing already, so if there is no acitivity you either have a bad rom, or more commonly bad Voltage rails or lifted/degraded chip contact pads.
Why does the 286 motherboard take so long to finish its tea?
Because it only takes SIPs.
[womp womp]
hahahaha I'm stealing that.
I love it.
The DX50 you repaired the trace, then broke it again with your tweezers @17.39
Didn’t notice that the first time, but watching it again I saw it.
Yeah, kinda sad the board got declared dead because he was so concerned at poking at fuzz.
Yes, way back in 1984 I overclocked an IBM PC/AT to 8 Mhz! I was an IT techie working for an International Oil company based in London England. We worked closely with IBM and tested and verified all of the early IBM Personal Computers and associated Hardware and Software products for use in our company. I remember getting 4 of the very first PC/ATs in the country for testing.
Based on Technical journal articles and chat on BBS's that discussed early overclocking the AT to 8 Mhz and being an electronics enthusiast, I purchased a couple of 16Mhz crystals. I remember they had wire legs so I had to solder proper sized legs to them for the on board sockets.
But yes I successfully ran two of the machines at 8 Mhz with no problems at all, apart from running noticeably hotter. I received many visits from other techies, from within the company and even from IBM UK who were based just along the river, to see benchmark utilities and other programs running at "breakneck" speed on my souped-up AT. Yes we were all geeks even back then!!
The 386 motherboards like this one were used in slim form factor desktop PCs made by many off brand manufacturers.
I had one and loved it! It was so much smaller then usual PCs at the time
Always a nice surprise when I reload TH-cam and see a freshly released video!
Specially when you are done with work.
Mr. Microscope 🔬 reveals itself as the new sensation of the channel. 👍 😉
Nice video, looking forward to the next part(s)! Maybe move the little monitor to the right hand side so it's nearer the camera? And recheck that trace on the 486 board - I think it's still broken further along to the left of the repair! Look at 17:40
Hi Adrian, I worked for a Taiwan company in 1990/94. Sunnytech, this company made this motherboard you have here. I remember most motherboards had buffer problems.
22:09, the Firefly Serial to Bluetooth resembles a Serial to Bluetooth device stenographers use to connect their steno machines wirelessly to their computers.
Are we interested in seeing these types of videos?
Speaking just for myself; Of Course!! This type of stuff is why I subscribed....and I'm sure I'm not alone in this....
Seeing you revive these old boards and computers is why we're here!
Of course, your charming personality, and detailed explanations as to why and how you do the things are all a big part of it, so yeah... please, lets see the other items in the box...don't tease!! :-D
You are great, do your thing, and thank you for doing these fascinating and entertaining videos... :-)
I want more! I want more of this kind of videos. I love it.
I remember reading in a computer repair / upgrade book that people did over clock the 5170 but changing out the crystal like you did. I also remember reading that IBM updated the BIOS to not support over clocking in later BIOS revisions and I'm wondering if this is why your XT/IDE is failing. IBM had mentioned that the type 1 motherboard can't handle proper timings higher than 6 MHz.
18:07 the track to C3 looks toast along with another underneath.
Almost an hour!!! Ahhhhhh
Great video!!! Got my recliner, lunch, and some basement!!
That AT overclock was fun. Good job!
IBM equipment - bulletproof! THE best. I remember a PS/2 model 55 I ran across in my consulting days that printed invoices on a dot matrix Proprinter. It has been running flat out for 11 straight years! No power management, no time out, just 24/7 operation all the time, every day, for 11 straight years.
Those overclocks are fantastic!
YES! Great video. Thank you.
I had no idea you were in Portland. I'm outside of Astoria. You're my current favorite TH-camr. Such an interesting channel. Back when I was a kid my brother and i got like garage sale computers and tinkered with them. In retrospect, I wish I had those old computers now. Lots of 286, 386, and 486 computers. Keep it up man!
I miss the time when you could find such a parts box in a garage sale for 3$ :(
486DX BOARD and the 286 board MAY REQUIRE CMOS BATTERY they both have ext batt connectors.
There's also the fact that he broke the trace again while going after those bits of fuzz... That might be the cause of why the board didn't work.
I could watch these debugging videos for hours
A strange mix of EEV, branchus and Phil’s computer lab… love it
Learning a lot. Some things I wouldn’t mind you could explain a bit slower for us mac fan boys. Still learning the pc world.
Intel macs are very similar to pc’s architecturally, and of course you know those macs can run Windows, Linux, Unix, and whatever you like….
Amazing job. More of these videos please.
Ok guys, this video better be popular. Enjoyed it!
Best video for a while. Random junk could be new series of content. Love idea of overlooked stuff getting attention it shouldn't really be getting. Thumbs up from me
They LOVED them some jumpers back in the day!
In 1986 I used to run my IBM PC/AT at 10MHz with a 20MHz crystal. For some reason I had to swap the floppy drive for a different brand to get it working at 10MHz. At one point I had 12.5 MB memory in it for testing purposes.
Those BT serial adapters would be handy for us IT folk. Makes it easier to stick one on the back of a network switch, router, server, storage appliance, etc., and configure from a laptop without having a cable stretched out from behind a rack. :-)
A fun tour of mystery PC boards, loved it. Those off-brand motherboards were everywhere in the UK too in the 90's.
Nothing stops you from inserting a SIMM slot into a SIPP socket =)
That idea's as crazy as inserting a PLCC socket into a PGA socket. Nobody's that crazy.
Oh wait...
Kind of interesting what gems are hidden in that box!
I know I'm a bit late to this video.
SIP memory is outright awful! I dealt with a bunch of Memorex Telex '386 PCs using SIP memory. The machines would lose memory or fail to post more often than not due to the SIPs creeping out of the sockets. Eventually, the company replaced those machines with something else.
On the awesome old IBM AT motherboard, the reason for the slightly bent pins on the ISA slots is that the sockets were installed manually and had to be held in place when put through the wave soldering machine. The bending of the legs prevented the sockets from lifting up. This was a common practice when I worked in PCB assembly in the early 1980s. The terminals I built back then had socketed EPROMs and other support chips as well as the CPUs. We were instructed to bend over the corner pins, alternate corners such as pin 1 and pin 20 and ensure the bent over pins were lined up with the etches. Today, none of this is an issue because these components are robotically inserted or surface mounted and human interaction is taking nearly 99% out of the process. How times have changed in 40+ years!
Good work on the debugging. :)
I'm loving the 'microscope cam' (for lack of a better word).
I have the 407 version of this scope. It can do simultaneous display on the internal screen and HDMI. I also solved the working space issue by mounting the base 180 degrees so it points backwards, flipped the power connection to the lamps around, and clamped the base towards the back at around 2 inches off the bench. Works great but is now more of a semi permanent install. Also protected the plastics exposed at the edge with some Aluminium tape.
The microscope footage is a great addition!
Yes please, I want MOAR of this
Good to see you "de bug" the original way! 😆
I'm so jealous your 5170 board just worked out of the box, so to speak. I picked up a 5170 board a while back and never got it working. It's definitely a later revision than yours, since there's no bodge wires and most of the chips are dated around mid-85. Interestingly, some of the double-stacked RAM chips have a blue line on them, no idea what that means. And the keyboard controller is a ceramic Intel D8742 with a die window, which is different from the one on your board.
I haven't exactly tried much troubleshooting, just the basics of checking for shorts and confirming voltages. I really need at least a POST card to see what's going on. Also, thanks for the tip about using a PLCC socket to connect a different 286 chip. Might have to give that a try someday.
Anyway, very cool to see some old school overclocking! And I am glad you at least got some parts working.
Always check about 5pm UK time figuring that Adrian has just got out of bed 😂
Here's me in the UK still watching at 5am. :D
Can't wait to see what else is in the box. :)
I love how you're sooooooo happy swapping the crystals to get that extra overclocking speeeeeeeeed, honestly, made my day :)
There is a followup to this on my second channel as well.
@@adriansdigitalbasement Just finished watching it, I hope you can get it stable, because apart from that, it's an outstanding effort. I really like the idea of having an arduino crystal where you can change the frequency to anything you want.
Years ago we used very similar Bluetooth serial adapters for educational “scribbler” robots in an intro to computer science class. The robots had a serial port and the dongles allowed them to communicate wirelessly with the lab computers.
Post'n and Roast'n!!! "Pushing the limits!! "
Weirdly enough I'm very interested in these early PC mother boards. I remember my brothers talking about working on theirs and occasionally explaining some of it to me.
The surprise tick at 13:30 was amusing.
50:32 This is why I love your channel.
Regarding the 486-DX50, maybe when fixing the trace you have bridged it to the nearby trace. The added wire looks like it is on top of the nearby wire.
awesome video. really enjoy the trouble shooting process and a mixed box of unknown parts.
I love your enthusiasm, it reminds me the little pleasures in life.
Looking forward to the next video. These are the best video's.
Adrian, get a fibreglass pencil to clean up those tracks. Way better than scraping and much less damaging.
Really enjoying the microscope! Great addition to the video!
Eat your heart out Linus, this is some real overclocking here! (Was Linus even born yet when this board came out)???
"Linus Gabriel Sebastian (born August 20, 1986) is a Canadian TH-cam personality." So he wasn't born yet. The AT was released on August 14, 1984 so that's slightly over a two year lead.
@@eDoc2020 Haha, thanks! I was 5 when this board came out.
We use those bluetooth to serial adapters to console into routers in the data center so your laptop isn't tethered to a short cable. Comes in handy in densely populated racks or if your in a tiny closet and there's nowhere good to park your laptop.
That 286 - 12 brings back memories had one back in the early 90's first computer I built all the parts from tiger direct.
The Firefly is a Bluetooth to RS232 serial adapter. It attaches to a RS232 serial port making it a Bluetooth Class 1 wireless connection, capable of transmitting your serial data up to 100 meters (330 feet) away.
You can use the Bluetooth Firefly with another Firefly in "cable replacement mode" to make your RS232 cable wireless, or you can pair it with a BluePort XP (battery powered), FirePlug(USB serial adapter) or BluePlug (USB adapter). The Firefly Bluetooth serial adapter may also be paired with any other Bluetooth device that supports Bluetooth SPP (Serial Port Profile). Note that many devices with built-in Bluetooth (e.g. PC's, Laptops, Handhelds and phones) utilize Bluetooth Class 2 radios, which only has a range up to 10 meters, so if you use a Firefly with a Class 2 device, you will only get the effective range of the Class 2 device and not the Firefly.
The Firefly supports the latest and greatest Bluetooth specification 2.0 (which provides faster Enhanced Data Rate (EDR) and lower Bit Error Rate (BER)), but is also compatible with devices supporting the older 1.x specifications.
The Firefly has dip switches to make set-up of the most common serial baud rates quick and easy. It also supports "AT" commands which allow you to change the firefly settings with messages sent through the serial port.
13:50 - Great debugging technique.
The microscope really adds a lot to the channel!
Uh oh, Adrian's pulling out a microscope in a video, he's slowly becoming the Louis Rossman of aged computers.
The early IBM AT BIOS was okay with running at 8MHz. The later versions had checks which prevented booting if it was running too fast.
Also the 386 CPUs have a chipset between the ISA bus and the CPU, which the 286 CPUs didn't have. That chipset could probably negotiate or detect the XTIDE better, whereas the 286 might have been just talking too fast to the XTIDE directly.
55:00 The Harris 80286 manages to stay so cool, because Harris made a CMOS version of the 80286 processor. The Intel processors (EDIT: only some of them) are HMOS processors (which is an NMOS variant). NMOS/HMOS chips have a noticeable static consumption, mostly independent of clock frequency, whereas CMOS gates only get warm when they actively switch something.
I thought the Intel chip said 80C286 so wouldn't that also be CMOS?
@@eDoc2020 For some reason I got the wrong impression that Intel never made a CMOS 80286. This is obviously wrong. The dysfunctional 12MHz part in the video is CMOS. I couldn't get a good view at the original 6MHz processor, though. Possibly that one still was NMOS.
This was a pretty cool video, not sure how it found me but I watched all of it!
They make metalic Sharpie colours, ie, silver, gold, etc, that is easy to see on black.
I had to make a whole new stand with 1ft reach and a higher column for my makshift microscope.
I say keep going, because it is entertaining.
I love the video!. I love the overclocking of the IBM 5170 motherboard. I have to check my stash of old MB to see if they have those bad batteries. Thanks for the warning!
Enjoyed these troubleshooting videos. Always informative and entertaining.
Love this type of video.
i like how you get so excited at overclocking to 8mhz lol. i remember back in the day after this stuff overclocking a pentium 75 to 200 mhz sadly it wouldnt work with my pci 56k modem which required mmx on the chip
17:24 this is where kapton tape helps. you can tape it in place and cut the other side and solder one side. then switch the tape over to the other side while you solder it.
Wouldn't the 92-52 indicate the 52nd week of 1992? That also matches with the chip dates you mentioned.
Was going to say the same thing.
if its 92-52, with 52 weeks in a year, why would it be written that way instead of 93-00? wait.. computers.. I get it. literally when the chip was manufactured, the computer programming for stamping the date on the chip wouldn't have the .14 'days' ... that would drive my OCD nutz
@@dinomagick Week counting starts with the 1st week, though, so it's 1-52. The last day (or 2 in a leap year) would probably just be part of an 8 (or 9)-day 52nd week.
Uh, guys. You are mis-reading those date codes. This chip was made on the 25th week of 1989. For Intel CPUs of this era only one digit after the L indicates the year followed by 2-digit week. So L9 means 1989 and 25 is the week of production.
@@mattj5155 I believe we are referring to different parts of the video. We're talking about the 386SX board shown at 5:25. Adrian read 92-52 in the copper layer and then said it probably meant 1995.
It's a common thing to believe (I did so myself) that the 386SX uses the same bus protocol as the 286. But that's not completely true. In fact the 386SX uses a smaller edition of the 386DX bus interface, limited to 16 data lines and 24 address bits (23 address lines, no A0). The interfaces are very similar, so you can easily build chipsets that support both, but the detailed cycle timing is different. The 80286 did not ensure validness of the address lines during the whole bus cycle. In particular, at least if another bus cycle is pending, the new address is output during the last clock period of the previous cycle. There is no way to prevent that. A board that supports ISA needs to latch the 286 address bits to ensure the address bits staying valid long enough. The idea of this feature is to start decoding the next address (most importantly: detect whether that address is able to take a 0WS 16-bit cycle) as early as possible.
On the 386, this feature got modified: By default, the address lines stay valid during the whole cycle, so the new address appears half a clock later than on the 286. This timing would make 0WS cycles harder to implement, as the address appears later, but data is still expected at the same time. On the 386, the board can operate an input to the 386, called "NAB" (next address to bus) anytime during an active bus cylce to tell the processor that is may output the next address. This is called "address pipelining". Some 386 boards have a jumper to turn the pipelining feature on or off (possibly just interrupting the NAB line).
A good 286/386SX chipset knows about the address timings of both the 286 and the 386 processor.
Interestingly, Intel abandoned address pipelining with the 80486 processor, as they introduced burst cycles. Probably Intel assumed that one-word-per-clock bursting offsets the disadvantage incurred by address pipelining being no longer available, and not having to care about address or decoder output latching simplifies devices on the local bus. With the Pentium, address pipelining and the NAB line reappeared.
My first pc had a 486 dx50. Built from used parts from a local computer show.
I have the 386DX-40 version of that SX motherboard. Mine has only 5 slots and all are 16bit ISA. It has 8 SIMM slots and a spot for a math coprocessor. I bought it in November 1993 and it cost just shy of $100.
at 13:50 you really did "de-bug" the board!
i love that eyoyo. i think thats the video you got me hooked.
yes im interested ,you do good content over ][ channels i like this period of technology and you bring it to me thanks Adrian
That 386 SX-40 motherboard at 2:52 - I have the same board in my "Frankenstein 386" machine. I use it for imaging floppies and running Windows 3.11 for the lulz.
Also that IBM 5170 motherboard might even be a pre-production model, as the 5170 wasn't launched until mid-August 1984 but some of the date codes you have there are from April-May. So you either have one of the very first units sold or possibly even a review or test unit! Would love to know the history of that board.
Wonderful Adrian, great vid.
Pretty sure the firefly modules are basically "Virtual serial cables" using bluetooth for a bridge between 2 PCs.. I had researched similar devices years ago to connect my weather station to a PC collecting data in the basement..
I for one found it interesting and would certainly like to see more of the same.
Adrian, I love your videos. Thanks for another great one. Your wire tinning skill on the bodge wire is fantastic. Please share some tips on that. I would recommend that you look into getting some tacky flux in a syringe. It will help with those small soldering jobs.
Microscope based repairs, Rossmann style!
Hi adrian just to say yes i like those video you do all of them been watching you every time you release something by the way im from canada continue the good work
Those Varta batteries are the worst. Good video, liked and subscribed.
Personally I would get some of the UV solder mask or at least some of the glorified nail varnish solder mask. The problem with clear stuff is it's hard to tell where it is after you've done repairs. You can't miss the coloured stuff.
If that is a concern, why not use regular colored nail polish?
Personally I am a fan of the clear stuff because it makes it easier to inspect repairs. A possible middle ground could be placing a dot of colored nail polish nearby or otherwise marking the board.
31:50 It's 4.77 * 3, because the 8088 needs a 33% duty cycle clock signal, which was obtained on the original PC/XT by dividing the 14.31818MHz by three. Intel supplied a suitable clock generator as companion for the 8086/8088, it is the 8284.