Hi Adrian, that was a nice journey. I know very well how a 5 minutes task ends up in a week of work, so you are not alone sometimes asking WTF ;) Anyhow, my module has all data pins wired to the chip and you have to cut non required ones. If the mainboard has some of the pins connected, which should be floating you potentially could put the chip into a wrong state. You can find which pins have to be cut on the project site. Especially pin 21 (~RCL) can be problematic. Some boards require it, others don't work properly if you leave it uncut. In regards of Motorola/Intel selection, I already made a new version with a jumper, just have to make it public.
Thanks for the info Scorp👌I like your Dallas replacements a lot and I'm going to try and make mine work with the IBM PS/2 Model 30-286, the 55sx and the PS/1 using the fixes you described. Oh and there's a guy out there using your design and selling really badly assembled Dallas replacements on Ebay, using the wrong pins and really bad solder joints.. Maybe there isn't much you can do about it (open source design?) I just thought you should know..That's all.
@@pipschannel1222: Remember, for any PS/2s (and microchannel systems from other brands) have moved the "Century Byte" offset in the NVRAM configuration - so the Dallas RTC replacements with a 'C' (for 'Century', meaning Y2K-fixed) in the part number won't work. Saying it for the people less familiar with PS/2s.
Adrian, you have been in IT for most of your career. You have to know by now the MOST dangerous words in IT are "This will just take a minute or two.." Goes double for any hardware issues.
Adrian I can't thank you enough, I happen to have the exact machine with the exact rom and memory expansion card. Also happen yo do same mistake you did lol you made my day. The damn 162 memory error is gone once I did the same math and gsetup. Much appreciated 🙏
Hey Adrian, I love that you were honest enough to not edit out your mistakes here, which would have been quite east to do. In not doing so, I believe that I've learned more from this. I'm a new subscriber and based off of this video, I'm sure I'll become a regular visitor. Thank you for all that you do.
This was my 2nd PC I owned as a kid, mine was modified with an IDE controller and 40GB HDD before I even got it, I later added Dos 5.0/Windows 3.0, 4MB RAM, and a VGA card. but mine was an 8MHz 286, this was circa 1994 before I later upgraded to a 386DX-40, and AMD 486/50 @ 66MHz machines. (The first two machines I built entirely myself circa 1996, Ironically using some components sans the HDD from the 286 and garage sale parts.)
The VGA card is a ATI VGA Wonder-16 that support 800x600 and 1024x768 SVGA modes, not the original 8-bit VGA-only version. And yes, the middle port is a Microsoft bus mouse port. Apparently it still works in 8-bit slots (autoconfigured), not all 16-bit cards does that. Not sure if the timer chip is actually broken, it could also be residual BIOS code designed for 6MHz, even if the specific overclock check has been removed. So it might be worth trying the alternative non-IBM BIOS at 10 or even 12/12.5 MHz...
@@GrandpasPlace A S-Video port has 4 pins plus a plastic bar at the bottom, this connector isn't even close (18:53). It has WAY more pins and is missing the wide slot. And if we look at the Wilkipedia entry for Bus Mouse it's an exact for the slightly irregular 9-pin layout of that. Moreover, the manul for this exact card says the port is "compatible with Microsoft Bus Mouse" so, yeah, I'm 100% sure. Since it was an ATI card my first though when seeing it was actually to wonder if it was one of those proprietary connector ATI put on many of their All-In-Wonder cards, that came with a break-out cable splits it into audio L/R, composite and S-Video connectors (later models sometimes could also do component). They had a few different models of this connector so it's kind of hard to rule out. But that was much later, the first AIW came out the mid 90's (Wikipedia says 1996) and this card came out 1988 and besides there's no TV Tuner on the card. So Adrian's guess of Bus Mouse seemed the most likely port, Wikipedia agreed and the manual made it 100% sure.
Bought one of these at a computer store for $120 used in 1995, it was 8 Mhz 286 and had the 4MB RAM card for ISA slot. Wish I kept it original, but a friend gave me a 386SX 40Mhz motherboard and I yanked the 286 board out and made it fit... power supply didnt like the 386SX 40Mhz and I had to flip power on and then flip power off and on fast to warm boot it to get it to post and run. But 386 SX 40Mhz was so much better than the 286 8Mhz for Windows 3.11 and DOS Games. I had a Trident 1MB VGA card for it, but it originally was a VGA monitor and card system. I added a 14.4 modem to it to get onto AOL 3.0 around 1996 and learned the hard way that COM1 and COM3 share same interrupt as serial mouse was COM1 and 14.4 model was jumpered for COM3. Had to keep moving mouse to keep the AOL internet connection until I moved to COM4 for different interrupt as COM2 wasnt in use to IRQ share conflict.
I LOVE that Adrian is not afraid to show the absolute frustration that can come with this hobby as much as he shows the absolute joyous triumph of getting something to work! Ive certainly felt the pain and frustration of chasing bugs into the wee hours of the morning only to have a major component burn out or discover that the answer was right in front of me the whole time. Its the lows and highs that makes it so thrilling.
Agreed 100% on how handy it is for regular VGA monitors to switch video modes nearly instantaneously. I always keep a plain VGA monitor around just in case as a backup. The downside of course is the high flicker 60 Hz refresh rate, though.
@@herrbonk3635 I've grown sensitive to the point where I can tell a refresh rate within ~ 5 Hz just by looking at it, until it reaches 75 Hz or so. Anything from 85 onward is mostly unnoticeable for me. And yes it could be worse. My original 512k XGA card had that awful 43 Hz interlaced signal in its highest resolution.
I had the VGA Wonder back in the day as an upgrade to my CGA system, and it was absolutely awesome. With my CGA monitor, it would "display" EGA and VGA... kind of. It was a jittery mess in some modes, but the 320x200 16 color mode worked like a champ. And when I could finally afford a VGA monitor, my system was ready.
I love when he talks about people probably using VGA for too long, I work in IT for a school district, and the entire district is still VGA. They are flat panel led monitors, but VGA, not a single digital video signal to be seen!
Battery idea, before I hear what you did: lithium-ion 18650 constantly on charge. Charge from 5V or 12V, discharge through a voltage regulator that gives the appropriate voltage for the BIOS memory. Bonus points if you add an I/O bracket with a button and LEDs that show the BIOS battery state of charge, and allow you to inject voltage only into the battery charger.
This reminds me of my first year of vintage computing 2 or 3 years ago with my 386 tall tower I got for $35. Forgetting to plug something in, or plugging it in in the wrong place, or not realizing I had the jumpers set wrong, or not knowing how do diagnose anything, or not knowing that I had to use FDISK on a hard drive before I could format it, or not knowing that capacitors sometimes go bad, or not knowing what a capacitor is, or not knowing how to properly measure a capacitor, and on and on and on. Even the greats are fallible. 😅
Those old vga wonder cards are great. I have one in my XT and bring able to run the original monochrome display, or a vga monitor off the same card with any video mode is incredible
Yeah The VGA Wonder has a BUS mouse port. I used a 5170 6 mhz when I worked for a GIS company in 1990-96. Used to compile digitizer table programs. The PC had 640K, a 30MB HD, IBM EGA card and two 5.25" floppies. The receipt was still in the case, $18,000 CDN. They made a lot of money with that PC.
I did own a VGA Wonder at some time... yes, the round plug is for a mouse. The card had quite some modes available, what I used the most was a 132 character text mode. And if memory serves, with the right driver it had hardware mouse cursor support. That meant zero CPU cylcles lost for moving the mouse over any display mode. But the last info is a bit hazy as I didn't use the mouse myself. It was also the last card I recall that came with a simple "how to program" manual.
132-column mode is also why I like IBM XGA adaptors. The PC-DOS E editor and the Lynx browser are among some applications that make use of it. I just wish the video mode would've taken off a bit more than it did.
NEC was the gold standard for multi-frequency / multi-mode displays. That model is a little earlier than you might think - my Dell Precision PentiumPro 200 workstation came with that display (I had the 17inch version) as an optional upgrade in 1995 (running NT 3.51 - alongside my Alpha box). It was the immediate successor to the NEC FG line (4FG, 5FG, 6FG). Very, very good monitors (IMHO) from the company that coined the term “multisync”.
IBM was probably thinking that the processor on the board would be the only one ever installed, and designed the board accordingly. None of the manufacturers/designers had to worry about clock frequency ranges beyond a couple megahertz, until the end of the 286's run and into the 386 processors... I wonder if the ISA bus is directly coupled on the original AT, and when the first design was drawn up that gave ISA it's own timing and fully buffered IO to the processor side. Also, if there are cards that are generating internal timing and "expecting" ISA bus signals within what were the normal timings of the day, you might get intermittent/hard to reproduce 'unicorn' issues from timing misalignments. Between the 16 and 17.7Mhz timing... the period is 62.5ns versus 56.37ns, so if something was counting cycles for a given task you may end up on one side or another of a given external clock timing with the old designs never expecting that and not providing any checks or graceful failure states. TLDR on the latter... better make sure the full system is totally asynchronous before throwing an indivisible clock frequency at it.
I waited for my CRT screen to die before upgrading to something more modern. It took 15 years of heavy use for the power button to become sufficiently worn out that percussive maintenance was a requirement.
I used to get boxes of PC motherboards from computer auctions in the 90's and and try to get them going. Anything below a 386SX-16 was a hard nope from me. I'm not that much of a glutton for punishment. They usually just fed my TTL chip stash. My first job in 1989 was configuring 8088 and 80286 machines for commercial customers. There is not enough nostalgia in the world to make me want to relive that.
Loved your moment of geeking out over the camera and shutter speed fine adjustment and great CRT recording capability! And that really does excel over other cameras when it comes to recording CRTs; great footage! Thanks for another great video!
You know, after that ordeal maybe Adrian's good bye should be changed to "stay healthy, stay safe, and stay sane". Good to see you figured it out in the end. :)
Well, for an industrial computer, which would forget very important settings when the battery went flat, I replaced the CR2032 with a 3V Lithium AA battery, and a holder that allowed access from the outside.
@@zoomosis Didn't Adrian tell us about 4 AA alkaline batteries? That would still be 6V, though. Anyway, one option would now be to have 2x 3Vlithium batteries in a battery holder. That's 2000mAh, or 12 Wh. That should be enough to last more than a year... AFAIK the problem comes from these systems being turned off for long times.
My mother bought herself an NEC PC in '95 or '96, and it came bundled with this MultiSync E500 display; I distinctly remember being in awe as I dialed in the image with those control buttons and its OSD and comparing it to my Acer's fiddly knobs that got dirty and scratchy in its annoyingly short lifetime! I think it's usually CTRL+X or CTRL+Q to get out of those Origin games.
Hi Adrian. I used to remember for some of the NEC multisync monitors, they have an RGB drive/or manual color balance setting in a different sub-menu. This usually allows you to crank up the individual RGB contrast levels to help improve your picture contrast. Also, within another sub-menu, there may be a video level selection between 0.7v and 1.0v. If the setting is in the 1.0v option, the contrast will be only 70% maximum. If that Nec multi-sync also supports 15Khz, it makes it really valuable for the Amiga home computers, especially the newer AGA Amigas which can use 31Khz in desktop modes and 15Khz modes for video games.
I can second the part about being great for AGA Amigas. I have an A1200 and back in the day I had to have dual monitors with a VGA screen hooked up to the RGB out for Workbench and productivity programs in 480p with a TV hooked up via RF for playing games. I now use a modern Phillips TV that allows me to hook up via VGA and display both modes, but if you're insistent on using CRTs(unfortunately the only CRT TV I have is an old thing with built in VCR due to budgetary restraints) then that monitor is the best solution.
@@fattomandeibu A utility called 'FlickFix' would patch the Amiga OS for the A4000/A1200 which made the amiga reboot in 31KHz doublescan mode. Most games would then run in the VGA 31khz mode.
@@BrianHG.Ocean.Fitness Most of my games were booter floppies that would automatically switch to PAL. Even later tried using WHDLoad, and it would do the same thing even when running from inside Workbench. I assume this is to stop the games being cropped if run on a NTSC Amiga.
@@fattomandeibu Once this startup has first been run at least once: maprom devs:kick.a4000 flickfix TOTCLKS=0x073 VGAONLY Everytime you hit CTL-[A] [A] to reboot, any AGA Amiga will reboot in 31Khz mode forced and any game will also assume PAL 50hz 31khz or NTSC 60hz 31khz double scan mode which most VGA screens will do. So long as you do not power off the Amiga, or it doesn't hard crash, if those 2 lines at my startup-sequence has been run, I can boot a floppy game and it will also run in 31Khz. Only a few games, really un-did the 'flick-fix', usually 256 color AGA games.
@@BrianHG.Ocean.Fitness Might've been useful 20 years ago, but the only CRT I have left is 15kHz only anyways. I now use a Phillips LCD screen that can handle either.
Whiskey, Tango, Foxtrot, eh? Real-time clocks can be dicey, as you have uncovered here. Tying to push a 5170 beyond the state of its art is a gutsy move, anyway, so kudos to you for that. While I'm passing out the kudos, you did very well to give that shout out to Necroware - that dude is a board-repairin' maniac! Keep at it, and I have every confidence you will figure this out. Question: what speed is the RAM? Back then they were specified in nanoseconds, and typical speeds for an AT ranged from 450 ns to about 120 ns. I think... the 5170 running at 8 MHz would need 150 ns RAM to be stable. You might try replacing this with some faster RAM when you next try to overclock it? For 12 MHz, you would want something running about 80 ns.
That's why all the manufacturer are trying to make the overclocking better compatibility. What I love is the INFORMTECH board which can take the 80286 up to 20MHz Lanmark test, before 386 take over.
Interesting. I don't recall ever seeing 20 MHz 286 PCs here in Australia. I suspect by the time they were available it made more sense to buy a 386SX-16 instead.
Got to love a BIOS that makes you define memory sizes, then figures them out itself anyway, and fails if they don't match. It's like, hey, BIOS... look... if you want to do it, then you do it. If you already have the answers, then why am I still even involved???
The mouse port being on the video card (if indeed that's what it is) makes sense: a hardware accelerated sprite custom just for the mouse cursor was a thing I saw bandied about in computer mags in the early 1990s, so it may well be the trend started prior to then.
12:18 I’d wager interior lighting is also much brighter relative to screens these days. LCD and OLED brightnesses are nuts compared to CRTs I’d say too. Have you ever thought that might also contribute to dim looking displays?
Awesome trip to the original IBM pc pains of the era, btw your ATI round connector it is I mouse, a damn good one, but it is propietary, I had it in like, another life?, Bought it brand new and I love that card and mouse, died long ago and I didn't knew how to repair it back then so end up in the bin most likely :(
There is a youtuber called shango066 who restore the most broken TV's ever. He has an interesting tool that can restore CRT tubes. I've seen it work very well indeed given the initial state of the tube. So those VGA CRT's are probably just in need of some minor restore work.
Reminds me of programming some the Xilex IMTS Mobile Telephones, with talk around channels. It wouldn't work unless an IMTS System was also programmed, even if the IMTS wasn't even active! Crazy the hoops some equipment has us jump through!
Your speed increase is actually quite good, if you think about. From 6 MHz to 9 MHz would have been a 50% increase. (Three is half of six.) Your 8.8MHz clock speed is roughly a 47% increase. I would call that quite a substantial and worthwhile gain. It's almost the same percentage increase as the difference between a 486 DX2-66 and a 486 DX4-100. (Just on a much smaller scale.) So that's a win! 😀
They're much easier to install (i.e. cheap.) HDMI and mini-HDMI are growing, but VGA is still common. (unless you're into non-PC's... those are RS232 serial console)
Ahh, the power of the Pee CAT. Not quite enought to rule the world, but close to it when that bad boy was first released. 10:30 - Little known fact: The B in IBM stands for BODGE.
I need to get the setup disk for my 5170 and see if I can get the hard drive to boot again. It might have some old stuff on it. Also should see you at VCFMW, I'm bringing a few curious machines from my collection.
Perhaps the overclock puts the read/write timing to the RTC out of spec. It would have been interesting to see if you could write to the Necroware RTC module with the slower oscillator and pull up resistor in place.
perhaps but i am kind of surprised it breaks with such a small overclock. Now Adrian did say the bios may be checking for overclock but then mentioned timer2 so i think that is a different device. however maybe the ibm overclock/bios checking is also running foul of the RTC output. maybe a further bios patch could fix the incompatibility. i cant remember if the old motorola chip worked, and all that needed was to relocate the aa batteries somewhere safe. i guess the moral is, you can overclock until some peripheral chip fails to work properly.
You should get some 3.6V Lithium Thionyl Chloride batteries, they can last for a decade or more under some circumstances, I have test equipment here which was made in the 80's and it still had the original working battery in it !
Well the IBM AT actually requires a 6V battery, small wonder that a CR2032 just barely holds the settings and goes marginal FAST. I just bought a bigger 6V battery pack with 2200 mAh in it for mine, and it's fine.
Suggestion with the new camera, and I know this sounds counter-intuitive, but don't go below about f/5.6. This is because with a FF camera you have half the depth of field as the Lumix, at the same f-stop. This causes a bit of awkwardness with the focus and it makes parts in your hand blurry. The old phrase "f/8 and be there" could apply here. It will require you to increase the ISO or lighting to compensate, but no more so than your old cameras.
I have a feeling that this was smooth sailing compared to what is waiting for my to get my Czechoslovak XT clone working :) (including it playing nicely with an XT IDE)
The issues you were having at 10MHz may actually be issues with the ISA bus. The ISA bus was rated at 8MHz all the way until they started disappearing from motherboard designs. Someone on another channel, who is trying to build a 286-based system from scratch (as in starting with breadboards) was trying to run the chip that controls the ISA bus (cannot remember offhand what part number it is) at 14MHz because another chip he was using was rated for that speed. He had stability issues until he reduced the speed for that chip.
I remember fiddling with the ISA bus clock speeds in the BIOS on some 386 machines back in the day just to eek out some extra performance when playing Microprose Grand Prix. Video cards of that era typically couldn't handle going much past 10 MHz.
@@zoomosis Makes sense as a video card is timing critical whereas most other expansion cards are not (except serial cards of course, but that just gives funky Baud rates).
WTF!!! I had an IBM AT in 1985 when working for Boeing. It had a 10 or 20M full height HD. I changed the crystal to run at 8MHz and never had to change the ROMs. Never had any issues. Maybe they found out what nerds like me were doing, and put a stop to it? I don't remember if the technical reference manual listed the ROM source like they did in the original PC or not. Did they?
If memory serves, in the original video where the limitation was discovered and the ROM altered to bypass it, the underclock detection was set at sub-6MHz while the overclock detection was set at above-8MHz. It seems they wanted to prevent people from running it any faster than they sold them. So you should've been fine even under that scheme, since it was still a spec IBM sold!
Great video and this is definitely something I am going to eventually do to my 5170. I have the REV 3 board running at 8 Mhz although I did replace the stock bios with an AMI one. Even using AA batteries, it eats through 4 of them in about 2 years so this mod would be amazing. Also, I was definitely in the group yelling at the screen about you putting in 512k of base memory instead of 640k although it was because I have done the same thing before.
I think this video showcases why IBM lost the clone wars of the 80s. Their hardware just wasn't user friendly. By the time it was (PS/2 era), they had lost the market already. Even then they still didn't get that people wanted a BIOS menu for configuring the system.
You can still buy Motherboards with VGA Connectors. And I don't mean Server Boards. ASUS Prime B650M-A Wifi for example, which is a brand new AMD Socket AM5 Board.
Adrian, thank for such a detailed video regarding the IBM 5170. You’re right, IBM built a beast of a machine and it’s really heavy. I am restoring a used IBM 5170 but the components are mostly a hodgepodge of parts. No IBM motherboard or parts with even a Chinese power supply that is housed in the traditional style metal case using the indented space for the IBM motherboard; bummer but it’s close to the 5170 that I can afford. I wanted to ask about the ISA IDE controller with the SD card flash drive you used. What size is your flash SD card? How did you configure the setup to address the heads, cylinders, sectors, and etc for the SD card that you used? When you were completed with the setup, did you FDISK and format to configure your SD drive for the MSDOS 6.22 that you used? Thank you for any time you devote to the inquiry.
Think about how the BIOS probably tests that timer - they probably set the timer to fire in x milliseconds, then run a delay loop on the CPU for x+y milliseconds and if the timer hasn't fired by the time the CPU loop finishes, it flags an error. There's probably a loop counter you could increase a bit in the ROM to make this part work again.
This journey is the epitome of working on computer projects, lol. Something unexpected and time consuming to figure out, even thought it might be super simple, almost always seems to happen.
One bad thing about the older CRTs with analog circuits is that there is usually nothing to prevent the PC from trying to use timings that are out of range, risking damage to the display. There also usually isn't EDID so the attached PC can't tell what timings are supported.
Absolutely correct, there's nothing worse than hearing the screech from the line stage as it tries to run at 32kHz when it isn't designed to, sometimes followed by a squeak, then nothing!
I had an "NCL" IDE controller that worked with an IBM AT (and no aftermarket BIOS) by setting the HDD configuration to 'Type 1' - there weren't any timing issues that I saw with it.
@@zoomosis: This was able to auto-configure somehow - I don't know the production timeframe, but it seemed to be designed for the AT and other systems where the original controller was replaced.
If IBM and microsoft would not have managed to establish this one architecture as THE one standard for personal and business computers, we would still be stuck with every manufacturer doing their own thing and all computers being incompatible to each other ... we are kind of getting back to that state, unfortunately .... apple, atari and amiga all used the same motorola 86k architecture back then but were otherwise incompatible to each other ... just like ARM devices nowadays ...
![Adrian's Digital Basement ][](http://yt3.ggpht.com/ytc/AIdro_k7CrQRe71qN6c0Ua7A1P_I8KkxjqLa70u0zYeCvdvnKw=s48-c-k-c0x00ffffff-no-rj)
Hi Adrian, that was a nice journey. I know very well how a 5 minutes task ends up in a week of work, so you are not alone sometimes asking WTF ;) Anyhow, my module has all data pins wired to the chip and you have to cut non required ones. If the mainboard has some of the pins connected, which should be floating you potentially could put the chip into a wrong state. You can find which pins have to be cut on the project site. Especially pin 21 (~RCL) can be problematic. Some boards require it, others don't work properly if you leave it uncut. In regards of Motorola/Intel selection, I already made a new version with a jumper, just have to make it public.
Thanks for the info Scorp👌I like your Dallas replacements a lot and I'm going to try and make mine work with the IBM PS/2 Model 30-286, the 55sx and the PS/1 using the fixes you described.
Oh and there's a guy out there using your design and selling really badly assembled Dallas replacements on Ebay, using the wrong pins and really bad solder joints.. Maybe there isn't much you can do about it (open source design?) I just thought you should know..That's all.
@@pipschannel1222: Remember, for any PS/2s (and microchannel systems from other brands) have moved the "Century Byte" offset in the NVRAM configuration - so the Dallas RTC replacements with a 'C' (for 'Century', meaning Y2K-fixed) in the part number won't work. Saying it for the people less familiar with PS/2s.
I have to say i didnt expect a comment of you here but its nice to see hope to see some videos of you soon again ^^
Lol WTF could have been avoided by some RTFM😮
@@michaelfuller34 or by just buying a proper 6V battery like the thing requires. you can tape it to the back of the case if worried about leaking.
This may be cruel of me but my favorite videos are when things go wrong. It reminds me of my real life experiences.
It always seems to be when I learn the most... 🤔
Adrian, you have been in IT for most of your career. You have to know by now the MOST dangerous words in IT are "This will just take a minute or two.." Goes double for any hardware issues.
Adrian I can't thank you enough, I happen to have the exact machine with the exact rom and memory expansion card. Also happen yo do same mistake you did lol you made my day. The damn 162 memory error is gone once I did the same math and gsetup. Much appreciated 🙏
Hey Adrian,
I love that you were honest enough to not edit out your mistakes here, which would have been quite east to do. In not doing so, I believe that I've learned more from this.
I'm a new subscriber and based off of this video, I'm sure I'll become a regular visitor.
Thank you for all that you do.
This whole video is so completely the period experience for PC computing, it really, really is. xD
This was my 2nd PC I owned as a kid, mine was modified with an IDE controller and 40GB HDD before I even got it, I later added Dos 5.0/Windows 3.0, 4MB RAM, and a VGA card. but mine was an 8MHz 286, this was circa 1994 before I later upgraded to a 386DX-40, and AMD 486/50 @ 66MHz machines. (The first two machines I built entirely myself circa 1996, Ironically using some components sans the HDD from the 286 and garage sale parts.)
I have 2 of the NEC Multisync monitors, and I use them with my Amiga machines. I really like them.
The VGA card is a ATI VGA Wonder-16 that support 800x600 and 1024x768 SVGA modes, not the original 8-bit VGA-only version. And yes, the middle port is a Microsoft bus mouse port. Apparently it still works in 8-bit slots (autoconfigured), not all 16-bit cards does that. Not sure if the timer chip is actually broken, it could also be residual BIOS code designed for 6MHz, even if the specific overclock check has been removed. So it might be worth trying the alternative non-IBM BIOS at 10 or even 12/12.5 MHz...
are you sure that is a mouse port? It looks like an S-Video port
@@GrandpasPlace A S-Video port has 4 pins plus a plastic bar at the bottom, this connector isn't even close (18:53). It has WAY more pins and is missing the wide slot. And if we look at the Wilkipedia entry for Bus Mouse it's an exact for the slightly irregular 9-pin layout of that. Moreover, the manul for this exact card says the port is "compatible with Microsoft Bus Mouse" so, yeah, I'm 100% sure.
Since it was an ATI card my first though when seeing it was actually to wonder if it was one of those proprietary connector ATI put on many of their All-In-Wonder cards, that came with a break-out cable splits it into audio L/R, composite and S-Video connectors (later models sometimes could also do component). They had a few different models of this connector so it's kind of hard to rule out.
But that was much later, the first AIW came out the mid 90's (Wikipedia says 1996) and this card came out 1988 and besides there's no TV Tuner on the card. So Adrian's guess of Bus Mouse seemed the most likely port, Wikipedia agreed and the manual made it 100% sure.
Bought one of these at a computer store for $120 used in 1995, it was 8 Mhz 286 and had the 4MB RAM card for ISA slot. Wish I kept it original, but a friend gave me a 386SX 40Mhz motherboard and I yanked the 286 board out and made it fit... power supply didnt like the 386SX 40Mhz and I had to flip power on and then flip power off and on fast to warm boot it to get it to post and run. But 386 SX 40Mhz was so much better than the 286 8Mhz for Windows 3.11 and DOS Games. I had a Trident 1MB VGA card for it, but it originally was a VGA monitor and card system. I added a 14.4 modem to it to get onto AOL 3.0 around 1996 and learned the hard way that COM1 and COM3 share same interrupt as serial mouse was COM1 and 14.4 model was jumpered for COM3. Had to keep moving mouse to keep the AOL internet connection until I moved to COM4 for different interrupt as COM2 wasnt in use to IRQ share conflict.
Eyoyo monitor upside down...adds to the flavor of this video. I worked on AT systems for a computer company many a moon ago. I feel your pain.
I LOVE that Adrian is not afraid to show the absolute frustration that can come with this hobby as much as he shows the absolute joyous triumph of getting something to work! Ive certainly felt the pain and frustration of chasing bugs into the wee hours of the morning only to have a major component burn out or discover that the answer was right in front of me the whole time. Its the lows and highs that makes it so thrilling.
Your tangents about the specific details are the best part! Things we may not know!
Agreed 100% on how handy it is for regular VGA monitors to switch video modes nearly instantaneously. I always keep a plain VGA monitor around just in case as a backup. The downside of course is the high flicker 60 Hz refresh rate, though.
Nah, some 60Hz is nothing :D I grew up in 50Hz flicker (and not only from my screen).
@@herrbonk3635 I've grown sensitive to the point where I can tell a refresh rate within ~ 5 Hz just by looking at it, until it reaches 75 Hz or so. Anything from 85 onward is mostly unnoticeable for me. And yes it could be worse. My original 512k XGA card had that awful 43 Hz interlaced signal in its highest resolution.
The camera shutter speed tangent with the tangent monitor was great, a tangent with the tangent ;)
I had the VGA Wonder back in the day as an upgrade to my CGA system, and it was absolutely awesome. With my CGA monitor, it would "display" EGA and VGA... kind of. It was a jittery mess in some modes, but the 320x200 16 color mode worked like a champ. And when I could finally afford a VGA monitor, my system was ready.
I love when he talks about people probably using VGA for too long, I work in IT for a school district, and the entire district is still VGA. They are flat panel led monitors, but VGA, not a single digital video signal to be seen!
Battery idea, before I hear what you did: lithium-ion 18650 constantly on charge. Charge from 5V or 12V, discharge through a voltage regulator that gives the appropriate voltage for the BIOS memory. Bonus points if you add an I/O bracket with a button and LEDs that show the BIOS battery state of charge, and allow you to inject voltage only into the battery charger.
This reminds me of my first year of vintage computing 2 or 3 years ago with my 386 tall tower I got for $35. Forgetting to plug something in, or plugging it in in the wrong place, or not realizing I had the jumpers set wrong, or not knowing how do diagnose anything, or not knowing that I had to use FDISK on a hard drive before I could format it, or not knowing that capacitors sometimes go bad, or not knowing what a capacitor is, or not knowing how to properly measure a capacitor, and on and on and on. Even the greats are fallible. 😅
Those old vga wonder cards are great. I have one in my XT and bring able to run the original monochrome display, or a vga monitor off the same card with any video mode is incredible
Loving the camera Adrian. Beautiful CRT images.
The full 10MHz overclock was working perfectly with the original clock chip! Even after running setup and saving changes
The IBM AT is a huge pain whenever you have to lift it!
Back in the day, I used an XT Clone at 12Mhz (turbo on) / 8Mhz (turbo off) which had 1MB of ram and a 30MB hdd and EGA.
Thanks for reminding me why I preffer clones over original IBM PCs 😁
Yeah The VGA Wonder has a BUS mouse port. I used a 5170 6 mhz when I worked for a GIS company in 1990-96. Used to compile digitizer table programs. The PC had 640K, a 30MB HD, IBM EGA card and two 5.25" floppies. The receipt was still in the case, $18,000 CDN. They made a lot of money with that PC.
I knew you would crack it in the end Adrian ... fascinating stuff
I did own a VGA Wonder at some time... yes, the round plug is for a mouse. The card had quite some modes available, what I used the most was a 132 character text mode. And if memory serves, with the right driver it had hardware mouse cursor support. That meant zero CPU cylcles lost for moving the mouse over any display mode. But the last info is a bit hazy as I didn't use the mouse myself. It was also the last card I recall that came with a simple "how to program" manual.
132-column mode is also why I like IBM XGA adaptors. The PC-DOS E editor and the Lynx browser are among some applications that make use of it. I just wish the video mode would've taken off a bit more than it did.
@@DavidWonn back in the day, I did mod the Borland Pascal Turbo Vision library to support that mode. Was quite handy for these text mode "GUI"s.
How fun! Thanks for sharing. I always love these videos!
NEC was the gold standard for multi-frequency / multi-mode displays. That model is a little earlier than you might think - my Dell Precision PentiumPro 200 workstation came with that display (I had the 17inch version) as an optional upgrade in 1995 (running NT 3.51 - alongside my Alpha box). It was the immediate successor to the NEC FG line (4FG, 5FG, 6FG). Very, very good monitors (IMHO) from the company that coined the term “multisync”.
IBM was probably thinking that the processor on the board would be the only one ever installed, and designed the board accordingly. None of the manufacturers/designers had to worry about clock frequency ranges beyond a couple megahertz, until the end of the 286's run and into the 386 processors... I wonder if the ISA bus is directly coupled on the original AT, and when the first design was drawn up that gave ISA it's own timing and fully buffered IO to the processor side.
Also, if there are cards that are generating internal timing and "expecting" ISA bus signals within what were the normal timings of the day, you might get intermittent/hard to reproduce 'unicorn' issues from timing misalignments. Between the 16 and 17.7Mhz timing... the period is 62.5ns versus 56.37ns, so if something was counting cycles for a given task you may end up on one side or another of a given external clock timing with the old designs never expecting that and not providing any checks or graceful failure states.
TLDR on the latter... better make sure the full system is totally asynchronous before throwing an indivisible clock frequency at it.
Don't be disheartened. This was entertaining. 😊
I’ve learned to never expect a project will be quick and easy, because then it won’t be.
I waited for my CRT screen to die before upgrading to something more modern. It took 15 years of heavy use for the power button to become sufficiently worn out that percussive maintenance was a requirement.
I used to get boxes of PC motherboards from computer auctions in the 90's and and try to get them going. Anything below a 386SX-16 was a hard nope from me. I'm not that much of a glutton for punishment. They usually just fed my TTL chip stash. My first job in 1989 was configuring 8088 and 80286 machines for commercial customers. There is not enough nostalgia in the world to make me want to relive that.
Loved your moment of geeking out over the camera and shutter speed fine adjustment and great CRT recording capability! And that really does excel over other cameras when it comes to recording CRTs; great footage! Thanks for another great video!
I guess he missed that it focused on his palm instead of the coil-cell pack directly in front of it - lol -.
Sometimes it's out of sync with the modern camera's refresh rate, causing weird artifacts.
The tangent about the tangent ;)
You know, after that ordeal maybe Adrian's good bye should be changed to "stay healthy, stay safe, and stay sane". Good to see you figured it out in the end. :)
Well, for an industrial computer, which would forget very important settings when the battery went flat, I replaced the CR2032 with a 3V Lithium AA battery, and a holder that allowed access from the outside.
From memory the 5170 systems originally used a 6 volt lithium battery.
@@zoomosis Didn't Adrian tell us about 4 AA alkaline batteries? That would still be 6V, though. Anyway, one option would now be to have 2x 3Vlithium batteries in a battery holder. That's 2000mAh, or 12 Wh. That should be enough to last more than a year...
AFAIK the problem comes from these systems being turned off for long times.
That upside down Eyoyo is killing me. Put some black tape over it!😂
It’s not Eyoyo, it’s the lesser known Australian brand ohoh3!
That's worse than doormats at a crooked angle! Or my monitor not being parallel to the wall ... Welcome to OCD hell 🙃
54:29 would love to see you tune up some of these later vga monitors! You have to connect via serial for many functions
Woah, seeing video of a CRT with no rolling is absolutely wild to me! Thats awesome.
My mother bought herself an NEC PC in '95 or '96, and it came bundled with this MultiSync E500 display; I distinctly remember being in awe as I dialed in the image with those control buttons and its OSD and comparing it to my Acer's fiddly knobs that got dirty and scratchy in its annoyingly short lifetime! I think it's usually CTRL+X or CTRL+Q to get out of those Origin games.
The WOnder circle port on my old one took a RCA jack, lefr and right audio. It was same oin the All in Wonder Pro that you could watch TV on
Ha! I use the midis from ultima VI as my ring tones. I love the character creation song with the gypsy woman.
Hi Adrian. I used to remember for some of the NEC multisync monitors, they have an RGB drive/or manual color balance setting in a different sub-menu. This usually allows you to crank up the individual RGB contrast levels to help improve your picture contrast. Also, within another sub-menu, there may be a video level selection between 0.7v and 1.0v. If the setting is in the 1.0v option, the contrast will be only 70% maximum. If that Nec multi-sync also supports 15Khz, it makes it really valuable for the Amiga home computers, especially the newer AGA Amigas which can use 31Khz in desktop modes and 15Khz modes for video games.
I can second the part about being great for AGA Amigas.
I have an A1200 and back in the day I had to have dual monitors with a VGA screen hooked up to the RGB out for Workbench and productivity programs in 480p with a TV hooked up via RF for playing games.
I now use a modern Phillips TV that allows me to hook up via VGA and display both modes, but if you're insistent on using CRTs(unfortunately the only CRT TV I have is an old thing with built in VCR due to budgetary restraints) then that monitor is the best solution.
@@fattomandeibu A utility called 'FlickFix' would patch the Amiga OS for the A4000/A1200 which made the amiga reboot in 31KHz doublescan mode. Most games would then run in the VGA 31khz mode.
@@BrianHG.Ocean.Fitness Most of my games were booter floppies that would automatically switch to PAL. Even later tried using WHDLoad, and it would do the same thing even when running from inside Workbench.
I assume this is to stop the games being cropped if run on a NTSC Amiga.
@@fattomandeibu Once this startup has first been run at least once:
maprom devs:kick.a4000
flickfix TOTCLKS=0x073 VGAONLY
Everytime you hit CTL-[A] [A] to reboot, any AGA Amiga will reboot in 31Khz mode forced and any game will also assume PAL 50hz 31khz or NTSC 60hz 31khz double scan mode which most VGA screens will do. So long as you do not power off the Amiga, or it doesn't hard crash, if those 2 lines at my startup-sequence has been run, I can boot a floppy game and it will also run in 31Khz. Only a few games, really un-did the 'flick-fix', usually 256 color AGA games.
@@BrianHG.Ocean.Fitness Might've been useful 20 years ago, but the only CRT I have left is 15kHz only anyways. I now use a Phillips LCD screen that can handle either.
IIRC, that middle port in the ATI card was for a proprietary video capture cable
The (not so) good old memories of the IBM PC (and compatibles) ;-)
Whiskey, Tango, Foxtrot, eh? Real-time clocks can be dicey, as you have uncovered here. Tying to push a 5170 beyond the state of its art is a gutsy move, anyway, so kudos to you for that. While I'm passing out the kudos, you did very well to give that shout out to Necroware - that dude is a board-repairin' maniac!
Keep at it, and I have every confidence you will figure this out. Question: what speed is the RAM? Back then they were specified in nanoseconds, and typical speeds for an AT ranged from 450 ns to about 120 ns. I think... the 5170 running at 8 MHz would need 150 ns RAM to be stable. You might try replacing this with some faster RAM when you next try to overclock it? For 12 MHz, you would want something running about 80 ns.
Put the original clock chip back!!! The overclock was working fine with it
He said himself that he never tested the overclock with a battery installed so that may not be the case.
I used to play CounterStrike on one of those NEC monitors. Holy nostaliga.
Welcome to Sony. I hope the E1 does not overheat. Should it, the FX30 is also a very VERY competent studio cam with active cooling.
"first I want to replace this monitor" - Finally something that I could do, too! :D
WTF, indeed. "This will be an easy upgrade..." famous last words. Been there, done that more times than I can remember. :-)
That's why all the manufacturer are trying to make the overclocking better compatibility. What I love is the INFORMTECH board which can take the 80286 up to 20MHz Lanmark test, before 386 take over.
Interesting. I don't recall ever seeing 20 MHz 286 PCs here in Australia. I suspect by the time they were available it made more sense to buy a 386SX-16 instead.
It's an easy thing to overlook if you're not paying attention. I did something similar when I was setting the DIP switches on my AST RAM card.
Got to love a BIOS that makes you define memory sizes, then figures them out itself anyway, and fails if they don't match. It's like, hey, BIOS... look... if you want to do it, then you do it. If you already have the answers, then why am I still even involved???
Those little tape dots holding down the bodge wires are oddly satisfying.
The mouse port being on the video card (if indeed that's what it is) makes sense: a hardware accelerated sprite custom just for the mouse cursor was a thing I saw bandied about in computer mags in the early 1990s, so it may well be the trend started prior to then.
DIP switches and jumpers were the bane of my existence
12:18 I’d wager interior lighting is also much brighter relative to screens these days. LCD and OLED brightnesses are nuts compared to CRTs I’d say too. Have you ever thought that might also contribute to dim looking displays?
Good point... probably causes them to get run brighter too...I hope you have a great day or night!
That answers and explains some issues I have had with older Desktops with the CR32 Batteries going dead.
Awesome trip to the original IBM pc pains of the era, btw your ATI round connector it is I mouse, a damn good one, but it is propietary, I had it in like, another life?, Bought it brand new and I love that card and mouse, died long ago and I didn't knew how to repair it back then so end up in the bin most likely :(
There is a youtuber called shango066 who restore the most broken TV's ever. He has an interesting tool that can restore CRT tubes. I've seen it work very well indeed given the initial state of the tube. So those VGA CRT's are probably just in need of some minor restore work.
Reminds me of programming some the Xilex IMTS Mobile Telephones, with talk around channels. It wouldn't work unless an IMTS System was also programmed, even if the IMTS wasn't even active! Crazy the hoops some equipment has us jump through!
I had one of those NEC MultiSync monitors. It worked well. Used it for years!
That's a bit of a tangent. Going off at a tangent while talking about a tangent monitor. Tell me that was deliberate as you said it totally dead pan 😂
I would be running Razzle Dazzle screensaver on that and every other old PC I can. I just love watching it.
Your speed increase is actually quite good, if you think about.
From 6 MHz to 9 MHz would have been a 50% increase. (Three is half of six.)
Your 8.8MHz clock speed is roughly a 47% increase.
I would call that quite a substantial and worthwhile gain.
It's almost the same percentage increase as the difference between a 486 DX2-66 and a 486 DX4-100. (Just on a much smaller scale.)
So that's a win! 😀
An industry standard built around such an arcane BIOS. What a relief it is that IBM's BIOS was cloned and improved.
Brand new rack mount systems *still* come with vga ports. Crash carts are all vga I suppose.
They're much easier to install (i.e. cheap.) HDMI and mini-HDMI are growing, but VGA is still common. (unless you're into non-PC's... those are RS232 serial console)
I admire your composure, Adrian. Most of my dialogue would've consisted of four-letter words.
Ahh, the power of the Pee CAT. Not quite enought to rule the world, but close to it when that bad boy was first released. 10:30 - Little known fact: The B in IBM stands for BODGE.
I need to get the setup disk for my 5170 and see if I can get the hard drive to boot again. It might have some old stuff on it. Also should see you at VCFMW, I'm bringing a few curious machines from my collection.
Perhaps the overclock puts the read/write timing to the RTC out of spec.
It would have been interesting to see if you could write to the Necroware RTC module with the slower oscillator and pull up resistor in place.
perhaps but i am kind of surprised it breaks with such a small overclock. Now Adrian did say the bios may be checking for overclock but then mentioned timer2 so i think that is a different device. however maybe the ibm overclock/bios checking is also running foul of the RTC output. maybe a further bios patch could fix the incompatibility. i cant remember if the old motorola chip worked, and all that needed was to relocate the aa batteries somewhere safe. i guess the moral is, you can overclock until some peripheral chip fails to work properly.
You should get some 3.6V Lithium Thionyl Chloride batteries, they can last for a decade or more under some circumstances, I have test equipment here which was made in the 80's and it still had the original working battery in it !
I'm pretty sure I saw IBM ATs with Tadiran batteries from the factory and they usually outlasted the useful life of the computer.
Well the IBM AT actually requires a 6V battery, small wonder that a CR2032 just barely holds the settings and goes marginal FAST. I just bought a bigger 6V battery pack with 2200 mAh in it for mine, and it's fine.
that's a soundblaster 1.5 with CMS (SAA1099) installed. IIRC, later models had sockets for CMS chips but a needed PAL was also socketed as well.
the CMS soundtrack for U6 really shines at the end... (no spoilers?) I think the underworld theme is much creepier on CMS than Adlib as well.
This brings back memories of building my first Frankenstein IBM Clone.
Suggestion with the new camera, and I know this sounds counter-intuitive, but don't go below about f/5.6. This is because with a FF camera you have half the depth of field as the Lumix, at the same f-stop. This causes a bit of awkwardness with the focus and it makes parts in your hand blurry. The old phrase "f/8 and be there" could apply here. It will require you to increase the ISO or lighting to compensate, but no more so than your old cameras.
I have a feeling that this was smooth sailing compared to what is waiting for my to get my Czechoslovak XT clone working :) (including it playing nicely with an XT IDE)
Thumbs up for your persistence with this one - I was frustrated with you the whole time!!!
The issues you were having at 10MHz may actually be issues with the ISA bus. The ISA bus was rated at 8MHz all the way until they started disappearing from motherboard designs. Someone on another channel, who is trying to build a 286-based system from scratch (as in starting with breadboards) was trying to run the chip that controls the ISA bus (cannot remember offhand what part number it is) at 14MHz because another chip he was using was rated for that speed. He had stability issues until he reduced the speed for that chip.
I remember fiddling with the ISA bus clock speeds in the BIOS on some 386 machines back in the day just to eek out some extra performance when playing Microprose Grand Prix. Video cards of that era typically couldn't handle going much past 10 MHz.
@@zoomosis Makes sense as a video card is timing critical whereas most other expansion cards are not (except serial cards of course, but that just gives funky Baud rates).
I love seeing thee original sound blaster card! love that font. SOUND BLASTER
WTF!!! I had an IBM AT in 1985 when working for Boeing. It had a 10 or 20M full height HD.
I changed the crystal to run at 8MHz and never had to change the ROMs. Never had any issues.
Maybe they found out what nerds like me were doing, and put a stop to it?
I don't remember if the technical reference manual listed the ROM source like they did in the original PC or not. Did they?
If memory serves, in the original video where the limitation was discovered and the ROM altered to bypass it, the underclock detection was set at sub-6MHz while the overclock detection was set at above-8MHz. It seems they wanted to prevent people from running it any faster than they sold them. So you should've been fine even under that scheme, since it was still a spec IBM sold!
Thanks for confirming I should never own an original XT/AT system (at least with the IBM bios) as I probably don't have the patience 😄
Thanks Adrian you are excellent at explaining the out comes of your to us. Cheers
Great video and this is definitely something I am going to eventually do to my 5170. I have the REV 3 board running at 8 Mhz although I did replace the stock bios with an AMI one. Even using AA batteries, it eats through 4 of them in about 2 years so this mod would be amazing.
Also, I was definitely in the group yelling at the screen about you putting in 512k of base memory instead of 640k although it was because I have done the same thing before.
I think this video showcases why IBM lost the clone wars of the 80s. Their hardware just wasn't user friendly. By the time it was (PS/2 era), they had lost the market already. Even then they still didn't get that people wanted a BIOS menu for configuring the system.
You can still buy Motherboards with VGA Connectors. And I don't mean Server Boards.
ASUS Prime B650M-A Wifi for example, which is a brand new AMD Socket AM5 Board.
Now you know why I went for Commodore and Amiga computers.
Adrian, thank for such a detailed video regarding the IBM 5170. You’re right, IBM built a beast of a machine and it’s really heavy. I am restoring a used IBM 5170 but the components are mostly a hodgepodge of parts. No IBM motherboard or parts with even a Chinese power supply that is housed in the traditional style metal case using the indented space for the IBM motherboard; bummer but it’s close to the 5170 that I can afford. I wanted to ask about the ISA IDE controller with the SD card flash drive you used. What size is your flash SD card? How did you configure the setup to address the heads, cylinders, sectors, and etc for the SD card that you used? When you were completed with the setup, did you FDISK and format to configure your SD drive for the MSDOS 6.22 that you used? Thank you for any time you devote to the inquiry.
Think about how the BIOS probably tests that timer - they probably set the timer to fire in x milliseconds, then run a delay loop on the CPU for x+y milliseconds and if the timer hasn't fired by the time the CPU loop finishes, it flags an error.
There's probably a loop counter you could increase a bit in the ROM to make this part work again.
This journey is the epitome of working on computer projects, lol. Something unexpected and time consuming to figure out, even thought it might be super simple, almost always seems to happen.
16:31 every Dankpods viewer pointed at the screen just now.
OH MY PKCELLS!
Pkcell!
One bad thing about the older CRTs with analog circuits is that there is usually nothing to prevent the PC from trying to use timings that are out of range, risking damage to the display. There also usually isn't EDID so the attached PC can't tell what timings are supported.
Absolutely correct, there's nothing worse than hearing the screech from the line stage as it tries to run at 32kHz when it isn't designed to, sometimes followed by a squeak, then nothing!
@@cambridgemart2075were you trying to setup XFree86 ModeLines?
CRT looks really nice on the new camera!
I had an "NCL" IDE controller that worked with an IBM AT (and no aftermarket BIOS) by setting the HDD configuration to 'Type 1' - there weren't any timing issues that I saw with it.
How was the drive capacity configured? DIP switches? I don't think auto-detection was a thing yet.
@@zoomosis: This was able to auto-configure somehow - I don't know the production timeframe, but it seemed to be designed for the AT and other systems where the original controller was replaced.
What a beautiful AT.
The screen looks rock solid with that new camera! Sweet! Looked at the price on Amazon and nearly choked. :)
I know the feeling. When you get pissed enough, you go back to resolve a failure with double the power.
dude this is like... exactly my experiences every time I try to work on an IBM
Why did we pick this architecture to morph into the modern PC
The clone PC manufacturers fixed a lot of IBM's mistakes over the years. Or bandaged over them. 😀
If IBM and microsoft would not have managed to establish this one architecture as THE one standard for personal and business computers,
we would still be stuck with every manufacturer doing their own thing and all computers being incompatible to each other ... we are kind of getting back to that state, unfortunately .... apple, atari and amiga all used the same motorola 86k architecture back then but were otherwise incompatible to each other ... just like ARM devices nowadays ...