BTW, your final solution is exactly what we would have done in the day, that is, upgrade the BIOS at the same time the modern disk drive was installed. I know that AT is a pain, but it's a magnificent machine and it defined the modern PC world.
The 5170's original IBM BIOS is just total garbage. I think IBM rushed it or something.. The clone manufacturers did a way better job than IBM. Compaq's Deskpro 8086/286/386 BIOSs for instance: These are way more compatible with the original IBM 5150/5160 BIOS than the 5170! The irony :-) The Quadtel BIOS is a great alternative when you're on the verge of attacking your 5170 with a battle axe ;-)
@@pipschannel1222 I know people want to keep the machine "original" because it's IBM, but I'd choose the Quadtel BIOS any day if it meant no disk swapping for trivial things like setting the disk size and changing the clock. Keep the old ROMs, just don't use them!
@@amirpourghoureiyan1637 Yeah I did the same thing with my early 1985 type 1 unit. The original BIOS comes in handy if or rather when the on board memory gets bad. It's got a useful feature where it literally displays the location of the wrong chip(s) during POSTing so when my memory started te become flaky I put the original mask ROMs in, swapped out some of those weird piggyback chips and she lived once again ;-)
@@ericwood3709 Mac were okay until they broke, then they became near impossible to repair for a normal human, a concept Apple has stayed with ever since
I know LGR was having issues with setting up a 1.44M drive back when he added one to his 5170, he actually used GSetup to set the drive in the BIOS but I think he also had a later BIOS and/or different version of GSetup. Otherwise I love the look and sound of the AT, glad you were able to rescue it!
I bought my 5170 from the same Ebay seller that LGR bought his from and went through pretty much all of those same issues (the seller had a bunch of new old stock late model (Type 3 MB) 5170s that he was selling for $500 each). If you don't have the ability to load gsetup on a 5.25 disk, you can also do it manually using GSETUP_BASIC. You run the application on another (even current) computer and set all of the settings you want, which generates a BASIC program that you manually type into the 5170 and run. That allowed me to set my 3.5" 1.44MB floppy as "A" and then boot a bootdisk I made and let me go from there.
Glad you showed IDEDIAG! I have shared that with so many people that are having a hard time using a CF card because they don't know what to use for the geometry. Great little tool.
Adrian's struggles brought back the nostalgic feelings of what it was like trying to troubleshoot PCs back in the day, but much more difficult because the likelyhood a that a replacement component is defective is roughly 50÷ 😄
@@GigsTaggartI spent many hours playing 'jumper bingo' on hardware fished from the trash....the manufacturers page on CompuServe or the local BBS were the only hope for documentation back then.
You know that one Simpsons episode with Homer trying to assemble a grill and he runs at it screaming, impaling it with an umbrella? That would have been me with this computer troubleshooting nightmare.
Re: 601 - The original IBM combo card must have BOTH the hard drive AND floppy base IO jumpers set to default (primary) to satisfy the IBM BIOS. The previous owner perhaps changed the HDC base IO to accommodate the hard card?
@Adrian If I recall correctly, the BIOS settings only apply to on-board ports. Add-on cards didn't rely on BIOS settings to operate. Early in your video, when you first connected the 1.44 stiffy drive, it's LED came on during post, even though it wasn't configured yet. I may be wrong, but it's possible those BIOS settings only apply to motherboards with on-board IDE/FDC interfaces. What I would suggest, is that you completely disable the IDE/FDC ports in the (original IBM) BIOS and check if the ISA cards detects the drives automatically. One interesting thing I noticed, is that the BIOS "swap drives" setting seemed to make a difference. But then again, the first 1.44 stiffy drive didn't work with that setting. It could be that it's actually operating in 1.2MB mode, which in most cases will read/write to the disk, but you'll run into corruption issues. I would love to see the system work with its original BIOS and controller cards 😊
Thanks for the trip down memory lane! When I got hired to my first programming job while finishing my master's degree, I asked the boss/owner if I could come in over the weekend and familiarize myself with the system I would be using. I had never touched an IBM PC! I read over the reference books with awe. That's a great memory :) Good work! Gotta be persistent!!
Just came across your channel. Double beep I seem to recall is saying keyboard not detected properly. I'll try to dig up my old documentation in a few weeks and see if I can find a reference for 601 we error.
I remember from the old days, that to overcome the 1024 cylinders limit you would use the LBA addressing, which was multiplying the heads on expense of the cyls, to keep the overall capacity unchanged.
Yeap. To get past 504MB you have to switch to LBA. And have a BIOS and OS that understands it. (Since DOS doesn't normally load drivers for IDE. Some IDE cards had an extension ROM to fix the issue. And at one point, I was loading a driver to fix that on my 386. But DOS was stupid; I quickly went to Linux. Except on some oddball Compaq's -- EISA??? -- that had to boot linux from DOS to get memory arranged correctly. Novel needed the same trick, btw.)
Back then IBM showed their assembly sources in the Technical Reference Manual for that page. Apparently they also provided the original BIOS source files on a disk.
@@ovalteen4404 Yes. I posted a comment last week with links to the manuals for the AT and noting on what page you could find the code that produces this error. I don't know why that comment has now vanished from here and from my post history. I'm not going to do the research again, but someone else might wish to now that you know the information is out there.
Adrian, many thanks for this video. I enjoyed an hour of retro PC maintenance and deep dive into the struggles of the early PC era. Keep up the good work and again many thanks!
Your struggles reminds me of some of the struggles my father would often run into when he fixed pcs as a side business (his day job being installing and repairing sun computers for Sun) and he would always be in a bad mood if he'd had to deal with any of the following: amstrad computers, Packard Bell computers, compaq computers, IBM computers and problems that turned out to be yet another problem with oracle (at work, oracle had a rather bad reputation amongst sun engineers as they'd always blame it on the Sun hardware when 9/10 times it was oracle).
You seem to be experiencing all the same issues and pitfalls that I fell in to when I was trying to setup my 5170 clone. Glad you were able to figure stuff out! Thanks for the info about that Quadtel BIOS. I ended up using a different Phoenix BIOS than you used and just dealt with the disk error, so now I got more options, awesome.
I just found your channel and saw the three videos for the 5170. What an introduction to your channel. OMG, the amount of frustration you put up with trying to solve that 601 error. I miss the old systems sometimes, but NOT the frustrations of configuring hardware to work properly.
I wouldn't worry so much about the colors of the drive matching. Not even IBM did that. I've got a PS/2 in my basement that has one drive in beige like the case and the other drive in white. Delivered like that from the factory.
What an epic tale. I have had similar problems with a Pentium II based machine that just did not want to cooperate. In fact it's still acting oddly and I know I need to replace the floppy drive on it. Honestly I really admire Adrian's patience on this. I have a bit of a temper and I really do rage at these things, which is absurd I know. I need to remember Adrian's chuckle at the end of this video to recognise that it's just an old computer, be patient with it.
55:58 Wow. It's been decades since I've done a low level format. I'd bet that not many folks would understand what a low level format does; its important for everyone to know that a low level format is not something that you want to do with today's 'modern' hard drives....
And this video demonstrates why I preferred the Compaq DeskPro 286's over the 5170, although I did prefer a Model M over the standard Compaq keyboards. Never had any of these issues with a DeskPro 286 and they ran faster overall than the 5170s
I like going back and wathcing this over and over. Reminds me of the proplems of the olden days. I started working on computers professionally in 1993. We still had lot of customers with the older machines trying to upgrade them and use them. I really bugs me now, because over the years I upgraded whole companies, and I literally sent 1000's of these off to recyclers.
Same could be said of Ford Model T's and A's I suppose. How many cool 1960's vehicles were destroyed because of some engine problem or tranny problem I wonder?
Absolutely compelling viewing. I'm sure we have all gone down a similar rabbit hole and it can drive you mad. The difference is that you actually managed to solve the problems. Fab. Don't worry about the purity thing. I think the purity police allow the swapping of peripherals for operational reasons :-) The motherboard, CPU and case are still there. BTW, you have put me off looking for a 286! :-)
Alas, some comments are complaining about his rubber dome keyboard not being a model M! So the purity police don’t even jive with peripheral changes… but apparently the LCD monitor is fine? Very strange.
A little testing with QEMU and 4 MB RAM: the "Largest EXE Size" in MS-DOS 4.01 was 574992 bytes, but in PC DOS 5.0 (probably because it made use of XMS) was 636816. Confirming that even on 286 you're likely to get more efficient use of RAM from 5 than 4. (With 6.22 it's 607872 - still better than 4 but 5 may be your best choice.)
I'll have to try out QEMU on my 286 systems. I always run MS-DOS 6.22 on anything 386 and up. MS-DOS 5 seem to work better on my 286 and older systems as it also doesn't use as much hard drive space. I've found on a given 286, 1MB of RAM is normally more than enough so DOS can be loaded high. I only up to 2MB or more on a 286 if I want to run Windows 3.0 or Windows 3.1 on it. Most of my 286s just run DOS. I install Windows 3.0 on a TurboXT with an EMS RAM card. OMG it was painfully slow, but was neat to see running and using the EMS RAM. Because I was using PC-DOS 3.30, on that system, I had to partition the ST-251 40MB hard drive into two 20MB drives.
What is still amazing me is your inexhaustible patiente. Perhaps because you don't take what expressions you do with your face with every issue, but your voice sounds so calm. I would be hammering some things! :D
Your video reminds me of the very early 1990's when I switched from Commodore machines to PC clones. BIOS and hardware compatibility nightmares. As I recall, when PCI came in things got way better.
Nice job on this one! I would have been beating my head on the wall by day 2 or 3. But you're an unstoppable force when it comes to rebuilding these machines!
Have you heard of Compaq MS-DOS 3.31? It's a special release of MS-DOS 3.3 with only the big disk support from DOS 4+ added. If you're tight on memory but still need big disks, or if you just want to avoid DOS 4.01 alltogether due to its reputation, this is the version to use.
Regarding the utility to set keyboard repeat&delay values: Isn't that MS-DOS's MODE CON KEYBOARD REPEAT and MODE CON KEYBOARD DELAY or something? And the "primary" or "secondary" floppy disk controller [setting] lets you use the "onboard" controller plus one in an expansion card slot, and set whether the onboard or the card one is used as #0. Basically you can use up to 4 FDDs, "A" and "B" per controller (Note: no drive letters, just the way FDDs are adressed per cable; with those 4 wires flipped between the two drives). EDIT: a word in [] for clarity. Also: What happened here? 45:30: The BIOS "forgets" your "swap floppies" settings on reboot, so you rebooted from your 5 1/4" floppy. This feature is for IT personnel who wants to boot a diagnostic floppy disk ONCE, but without messing with the office worker's PC setup.
I just realized youtube hasn't fed your videos to my feed in a month. I was like dang it's been a while since I've seen Adrian. Hopefully binging your recent vids fixes it lol
Hey Adrian... did you notice how there are 2 sets of holes on AT drive rails? the reason is, that "half height bay" for a second hard drive can actually be used for a full height hard drive on a single floppy system. You attach the rails in the second hole so the drive sits further back, allowing you to put the metal bracket and that blank drive slot cover back in place. Also, AT's never directly supported 1.44 floppies in the bios. You could use them on 5170's, but in order to be able to properly, you 1: had to tell setup it was a 720K drive. Dos would be able to boot from it, read to it, and even write to it. HOWEVER, in order to format it, you needed to use driver.sys, which was a device driver included with dos that would let you tell dos how many sectors, heads and tracks a drive had. (yes, it was literally named driver.sys) optionally, later versions of dos also let you specify the number of heads, sectors and tracks directly in the format command. If you did not do either of the above, when you formatted a disk, it would only format to 720K. (it would assume 80 tracks, 9 sectors instead of 18 sectors) 1.44 meg drives were not introduced for IBM systems until the PS/2s came out. And IBM simultaneously discontinued the 5170s when that happened. So there was never a BIOS for the 5170 that supported 1.44 meg drives. The bios used with the motherboards that have the 256k chips instead of the 128K piggyback chips (the third gen bios, the one you tried using) will support 720K (the 6mhz version was used in 5170-239 ATs, whereas the 8mhz was the 5170-339, both shipped with 512K, a seagate st-4038 full height 30 meg drive (bios type 8) Fun fact - there is a difference later revs between the 6mhz bios and the 8mhz bios. There is a timing check on the 6mhz one to make sure it is not running at 8mhz. This was because people were buying 6mhz systems for several hundred dollars less and then swapping out the 12mhz crystal for a 16mhz one. Most of the time a 6mhz 286 could handle 8mhz. If not, it was still cheaper to swap it out for an 8mhz part than it was to purchase the system as an 8mhz machine. Anyway, however, that 601 error you got was not a bios compatibility issue. Post should just see it as an 80 track drive and assume it is either a 1.2 or 720K. My theory is that Gsetup has options in it that the bios itself does not support, and you were getting the 601 because of some corruption in the bios ram due to using gsetup. Some extended bios ram byte, one AT setup does not set or use, was set to some value by gsetup, and it is confusing the POST because the IBM bios does not know what to make of it. Just a theory. Could be wrong. To test this, I'd suggest wiping bios ram (IIRC, there is actually a jumper to do this on the MB, or you could use that basic program you mention.), putting IBM bios back in, and try using ONLY IBM AT setup, and not even touching GSetup.. edit: also... that stand alone floppy drive card you tried... was that an 8 bit card? if so, bare in mind that these did not have the transfer rate needed to support either 1.2 or 1.44 meg drives. Those will work with 360 and 720k drives only. (although there are special 8 bit floppy controllers with buffering between the drive and the bus, which also have their own bios, that will allow the use of 1.2 or 1.44 meg drives on 8 bit systems... these are specialty cards, though. many of these also supported 4 floppies)
I was heavily in the PC world in the mid- to late-80s and I never remember an AT or any 286 (e.g. HP Vectra) with a 1.44Mb disk - I didn't see that until the first PS/2 came out. It's not period correct in any case to have a 1.44Mb diskette drive. The AT came out in 84, and the 1.44Mb disk in 86. By then the AT was in the rearview mirror of the Deskpro 386 and IBM's own PS/2.
I encountered a lot of IBM 5170s in the early 90s and more often than not they had been retrofitted with a 1.44MB drive. It was a common upgrade to help keep those machines useful. The PS/2 drove people to install 3.5-inch floppies in other PCs so they could read the new disks.
This takes me back. My first impression of troubleshooting PCs felt like this coming from Amiga which "just work". Though in hindsight my opinion was very biased, Amigas can be just as frustrating when you're dealing with obscure problems.
Well, I have to commend you for sticking with this one, I would have probably stopped and used this one for parts long before your time into it. :) Glad I was on the Amiga back then, fjew. :)
I had one of those on my desk at work in the mid 90s for the express purpose of managing the company telephone switch. The PC people said they hated that machine more than any.
When you get around to working on the original 5170 power supply, don't forget the load resistor for the 12V rail, since you don't have a big old full-height MFM hard drive installed to draw enough amps on the 12V rail, and the power supply will shut off.
When you formatted the hard drive did use fdisk /mbr ? Fixes a lot of problems... i had issues with getting a compact flash card to boot. Fixed the problem. Maybe it might work on the drive.
I think this could work as well with the CF card: - Install your favourite BIOS - Reformat the CF card with the 1570 having its new BIOS installed - transfer the contents back on the CF, careful not to use an image for this. The card was working with the AMI BIOS, so that indicates a problem with the geometry match-up. And that could be solved by reformatting the card on the configuration (i.e: disk controller, system BIOS and system motherboard) you're actually going to use it with. This should _always_ be your first step in troubleshooting this issue. Even some 386 boards I encountered with the exact same issue (drive would read small files fine, not larger files; drive would not boot) had it vanish after doing this.
The unknown memory board shown at 4:47 says "MEMOPLUS" on it. The best I was able to do is find a page listing memory expansion board that tells me it could be a 2MB board made by Suntek Information. The A.4 board is for IBM PC, AT, PS/2 Models 25, 30. The X.4 board is for IBM XT. I haven't found any additional information. For reference the information was at found www.atarimagazines.com/compute/issue112/Memory_Expansion_Boards.php which showed page 42 from Issue 112 (Sept. 89) of COMPUTE! magazine.
24:40 You probably need the matching IBM setup program with the newer IBM AT BIOS. From what I remember several versions of the IBM setup program exist.
58:56 I've run into that problem in the past, where drives would not function properly when transplanted between older machines. It seems to have something to do with the way that disk controllers address the disk's layout. That is one of the things that I always appreciate with new machines, the fact that you can create partitions on one machine and move the disk to another and it works perfectly. I have run into problems with newer machines not being able to interpret a disk created in a different machine though it is rare. I always have that odd feeling that it won't work when I transplant disks though, which is something that I have ingrained into my brain from the old days...
I love waching videos like this. all that troubleshooting and trying to make such an old hardware warking.... Just love it I wish I was able to play around with such great machiness as well Great job. Look forward to seing part 3 :)
Given that these bioses use CHS addressing for accessing hard drives (as opposed to LBA that was originally specific to SCSI drives) could it be possible that the compact flash reader received a malformed IO request that it could not complete? Or, to put it differently, would these machines work if you used SCSI hard drive interface for the compact flash hard drive? Super curious. If you ran into the same problem with 386s perhaps the problem is related to how old IDE was trying to access the data. CHS has always been a headache. I'm particularly curious if the bios was setting some "reserved" bit in io that might even have had some meaning, that now caused a headache
OMG this makes me think I truly lucked-out with my IBM 5170 project... not to say that there weren't unexpected issues, but certainly not to this degree! Oh, and I solved the need for IDE support by adding an XT-IDE EEPROM (cheating perhaps) but I get to keep the purity of the original IBM BIOS, use more modern parts, and I can choose from any available FDD or HDD at boot time (like the swap function you were after.)
When you get around to testing the power supply, make sure you meet minimum load requirement for it to regulate properly. It needs 7A on the 5V and 2.5A on the 12V. For AT computers configured with no hard disk, IBM would install a 50W dummy load.
Back in the day I worked for IBM. I was given a 5170 for my desk. I remember adding memory and a 3270 adapter for mainframe access. I also added a 3.5 inch drive as the 1.2 meg drive was compatible with nothing. I certainly did not run into all your problems. But, I had access to all the proper period parts. Ain’t retro great.
Hello Adrian! It CAN boot into DOS with a 1.44MB 3.5 inch drive! You just need a drive where you can set the drive to be drive 0 instead of drive 1. From the 386 era, PC's use them as drive 1 but it does not work with the 5170. Jumpers must be set(or soldered on newer TEAC drives) to drive 0 in your case with the ORIGINAL BIOS.
Remember these times, when I need to mess with new HDD Head and cylinders settings in BIOS.. 😂 Nostalgic, especially, when these settings wasn't on HDD sticker.. Yeah, BIOS issues may take time to figure it out.. Try, just for fun, connecting IDE CD ROM drive.. I think, that BIOS wouldn't recognize it (back in day's, I run in that issue)..
Now you are seeing why Intel dumped the '286 so fast. Here's a fun fact, in order to use all the memory you have to go into protected mode, which is incompatabe with DOS. You can't go back to "DOS mode" once you start Protected Mode without halting the CPU. The 386 can go back and forth so you can use things like use things like DPMI to use all the ram you have for DOS.
Seem to recall that on these early IBM machines you also can't use the 287 in protected mode, but that may have been a BIOS issue. I have however never seen a 286 that had the FPU, likely because they never had enough RAM to run anything that could utilize it
Now that I see the restriction to 3.5" with 720K, I remember to have had the same problem back then with a XT clone. A BIOS without built in setup was not uncommon at the time. Even my first 386 lacked that and came with a DOS setup program.
By the way: It's easy to overclock the AT, just swap the clock crystal for a higher frequency. I used to run mine at 10Mhz and at one point had 12.5 MB memory in it for testing purposes. At the time the memory was so expensive that it was a lot more than the PC At itself, and the AT was not cheap either. All that memory was for customer network servers running Novell NetWare 286, one of the few things that could actually use the 286 and memory to its full capacity.
Adrian had a different video with the oscillating mod. But he had to have a mod done to remove the over locking clock bios routine. And a few other things hidden "features.". If I remember correctly
I have vivid memories of my dad trying to install a 1.44 MB floppy drive on an XT that he had and I definitely remember that he ultimately gave up on it.
Don't know where the dividing line is, but if you go far enough back, beyind just requiring some sort of TSR to override some BIOS behavior, you would also have needed to replace the floppy controller chip with a newer one, because much of the floppy controller circuitry was never on the drive, but instead on ISA card or motherboard.
Funny to know that "All in Wonder" was something different before it was the Graphics Card + Tuner combo that associated with that name when growing up.
I remember those days. It was bad enough when compatible hardware was available, although the hardware did get better for several years even after the 386 came out in 1987. Once the 386 achieved dominance, and especially once the 486 landed, there wasn't much further boundary-pushing on the 286 front. Frankly I wouldn't try to put any IDE solution in a 5170, not even one using spinning rust. (SCSI and RLL are certainly fair game though.) If I can't get hardware appropriate to the time, then I'd just have no hard drive at all, even if that makes the machine essentially useless for more than games. IDE did not exist when this machine came out, so if the machine balks at using IDE, I would prefer to keep it stock and lose the functionality because _none of these machines are particularly useful_ in reality. The same is true of the 1.44 MB drive. If the hardware wasn't amenable to supporting it, I wouldn't install one no matter how convenient it might be, because that's just not how this machine would have ever worked in its useful lifespan. If I want a 5170, I want a 5170. If I want a machine that represents a souped-up 286 rig, then I'll make one out of clone hardware that isn't so buggy. But this machine is now neither. The only thing really original at this point is the motherboard and its 6 MHz clock rate, which is a negative, so basically we've ended up with the worst of both worlds -- a slow 286 that doesn't behave anything like a stock 5170. I admit that stock 5170 behavior is infuriating, but I feel that if that's what you're building, those are the constraints you have to work with. Otherwise it's just another 286 build in an IBM case, at which point you'd enjoy using it a lot more if it had a 12 or 16 MHz clone board in it instead. Once I established that I would never be able to enjoy using a period-correct 5170 because it's buggy and slow, I'd get it minimally working with as much original hardware as possible, and only period-correct replacements when needed. No modern workarounds. Then I'd send it off to a museum or collection, and build myself a clone rig that actually works with the stuff I keep around. Sadly, this build ends up being neither of those things, especially given the amount of work involved.
Hi Adrian, would you mind putting a link to the shiny tool for taking off the chips from tight places? The one seen on minute 33:30. Thank you for another interesting and entertaining video!
I recall my painful experience setting up my first PC which run V20 with 384K on-board, the extra RAM card hold the extra RAM, altogether only 1MB, and 360K floppy only, booting DOS 3.3. Today BIOS is so sweet to setup compare to the old one.
Maybe these EPROMs were just too slow... I once fixed an Asteroids machine, the ROMs had disintegrated and I had to burn new ones. They verified fine, but the machine didn't like two of them. I thought, maybe some bits already flipped back, that can happen, you just have to burn the image twice. But they did verify. Asteroids runs at 1.5MHz and the 6502 is pretty quick when it comes to access times. So when I had to burn ROMs for a slower system, I used these and they worked fine.
@Adrian's Digital Basement Daft question perhaps, but I do wonder if some of those 6xx class errors occur with the older BIOS versions because the code is expecting some hardware to be available that was included on the drive IO card, rather than the Motherboard? A bit like the way they included the Gate A20 controller in the Keyboard?!? I don't know, that just came in a flashback, so I've no idea if I'm making it up or not!
Also, those settings from IDEDIAG seemed very wrong to me. Sector size of 528 bytes? Track size of 16896 bytes? Those both seem off. I have never seen any media with anything other than exact powers of 2 in those categories. You would expect 512 byte sectors and 16384 byte tracks. At least the values it gave are multiples of each other, but I think that program is probably having issues because it's a compact flash card.
This video is the perfect example of why I simply do *not* understand this weird "purity" obsession with some collectors. Who cares if it's all original? The people who actually USED these machines (and that's their purpose, to be used) wouldn't have wanted to keep it stock. They modded these machines, upgraded BIOSes, added new drives, CPUs, etc. If I find machines from this era that are heavily modded, I lean into that and make sure to try and keep the mods functioning, instead of undoing all that work to restore it to a useless, frustrating to use, "virgin" machine. It just makes absolutely zero sense to me. I collect old computers because I find using them fun -- I don't just put them on a shelf and look at them.
Thats why I passed on the IBM XT computers and built generic 386 and 486 computers instead. Too many problems and limitations on the IBM XT computers. My goal was for best DOS game play anyways.
An ide controller card with an external bios will probably solve that issue like you said. I have a k6-2 machine that under DOS the onboard IDE didn’t support LBA under DOS so my 3.2gb hard drive was only visible as 502mb. I bought a promise IDE controller card and all 3.2gb is now usable. Not quite the same issue, but it should fix the problem(although he’d need an ISA one with an external bios such as an XTIDE)
Not sure if either of these things were brought up already but: 1. Another network type that used coax and eventually twisted pair was Arcnet. Arcnet was popular during 80's and early 90's. 2. Problems with booting HD when it otherwise works: I doubt you had a problem with the partition table itself or else you wouldn't have been reading data correctly from the disk. However there is bootloader code on the sector 0 that contains the partition table. That probably was bad or incompatible until FDisk re-wrote it. Not sure if there's a command switch to rewrite the code without wiping the table.
The three greatest computer industrial designs: IBM PC/XT/AT, Apple ][ and MacBook Air. PS/2 was bumped by the PC/XT/AT in my top three simply because IBM tried to close the gate after the horse had bolted.
I think those square (rectangular) batteries were 9 or 12 volts, not 6. That may be why it keeps giving you the “162 - options not set” error. Check the original battery and see if it really was 6.
Adrian was THIS close to return this computer to the dumpster whence it came 😅 Great job fixing this up!
Thank you for using "whence" correctly. I groan when I see people trying to be smart with 'from from where' gibberish.
@@AlTerego816 It's possible to educate people without insulting them.
You were this close to burning 500
BTW, your final solution is exactly what we would have done in the day, that is, upgrade the BIOS at the same time the modern disk drive was installed. I know that AT is a pain, but it's a magnificent machine and it defined the modern PC world.
The 5170's original IBM BIOS is just total garbage. I think IBM rushed it or something.. The clone manufacturers did a way better job than IBM. Compaq's Deskpro 8086/286/386 BIOSs for instance: These are way more compatible with the original IBM 5150/5160 BIOS than the 5170! The irony :-)
The Quadtel BIOS is a great alternative when you're on the verge of attacking your 5170 with a battle axe ;-)
@@pipschannel1222 I know people want to keep the machine "original" because it's IBM, but I'd choose the Quadtel BIOS any day if it meant no disk swapping for trivial things like setting the disk size and changing the clock. Keep the old ROMs, just don't use them!
@@amirpourghoureiyan1637 Yeah I did the same thing with my early 1985 type 1 unit. The original BIOS comes in handy if or rather when the on board memory gets bad. It's got a useful feature where it literally displays the location of the wrong chip(s) during POSTing so when my memory started te become flaky I put the original mask ROMs in, swapped out some of those weird piggyback chips and she lived once again ;-)
It's a clunky piece of junk next to a Macintosh of the era. It's a shame that this platform became dominant.
@@ericwood3709 Mac were okay until they broke, then they became near impossible to repair for a normal human, a concept Apple has stayed with ever since
I would have reprogrammed the 5170 with an axe if I'd faced all of these frustrations!
it might have been more co-operative if he kept the axe nearby...
by the way, this axe is over 100 years old, its had 2 new heads and 5 new handles and still performs great, just like Adrian's 5170...
@@fumthings That line from Only Fools and Horses about Trigger's road sweeping brush also occurred to me as I watched this episode!
I know LGR was having issues with setting up a 1.44M drive back when he added one to his 5170, he actually used GSetup to set the drive in the BIOS but I think he also had a later BIOS and/or different version of GSetup. Otherwise I love the look and sound of the AT, glad you were able to rescue it!
None of the IBM setup versions have a 1.44MB option. The later BIOS supports it, but not the setup. Hence gsetup.
I bought my 5170 from the same Ebay seller that LGR bought his from and went through pretty much all of those same issues (the seller had a bunch of new old stock late model (Type 3 MB) 5170s that he was selling for $500 each). If you don't have the ability to load gsetup on a 5.25 disk, you can also do it manually using GSETUP_BASIC. You run the application on another (even current) computer and set all of the settings you want, which generates a BASIC program that you manually type into the 5170 and run. That allowed me to set my 3.5" 1.44MB floppy as "A" and then boot a bootdisk I made and let me go from there.
Glad you showed IDEDIAG! I have shared that with so many people that are having a hard time using a CF card because they don't know what to use for the geometry. Great little tool.
try FDISK /MBR on CompactFlash drives that do not boot first. Sometimes something goes weird with the MBR on compact flash cards.
I had to do this on mine as well
Exactly what I was thinking.. I bet that would have fixed it
Adrian's struggles brought back the nostalgic feelings of what it was like trying to troubleshoot PCs back in the day, but much more difficult because the likelyhood a that a replacement component is defective is roughly 50÷ 😄
also couldn't pop on the internet to get a new bios or look up jumpers
@@GigsTaggartI spent many hours playing 'jumper bingo' on hardware fished from the trash....the manufacturers page on CompuServe or the local BBS were the only hope for documentation back then.
You know that one Simpsons episode with Homer trying to assemble a grill and he runs at it screaming, impaling it with an umbrella?
That would have been me with this computer troubleshooting nightmare.
I would have went up a tall structure and threw it to the ground
Yes. Let the hate flow. Feel the power it can give you.
Re: 601 - The original IBM combo card must have BOTH the hard drive AND floppy base IO jumpers set to default (primary) to satisfy the IBM BIOS. The previous owner perhaps changed the HDC base IO to accommodate the hard card?
@Adrian
If I recall correctly, the BIOS settings only apply to on-board ports.
Add-on cards didn't rely on BIOS settings to operate.
Early in your video, when you first connected the 1.44 stiffy drive, it's LED came on during post, even though it wasn't configured yet.
I may be wrong, but it's possible those BIOS settings only apply to motherboards with on-board IDE/FDC interfaces.
What I would suggest, is that you completely disable the IDE/FDC ports in the (original IBM) BIOS and check if the ISA cards detects the drives automatically.
One interesting thing I noticed, is that the BIOS "swap drives" setting seemed to make a difference. But then again, the first 1.44 stiffy drive didn't work with that setting. It could be that it's actually operating in 1.2MB mode, which in most cases will read/write to the disk, but you'll run into corruption issues.
I would love to see the system work with its original BIOS and controller cards 😊
Important points for myself to make note of. Thanks.
Thanks for the trip down memory lane! When I got hired to my first programming job while finishing my master's degree, I asked the boss/owner if I could come in over the weekend and familiarize myself with the system I would be using. I had never touched an IBM PC! I read over the reference books with awe. That's a great memory :) Good work! Gotta be persistent!!
"601-Diskette Error" No doubt Adrian was seeing this error in his sleep for days 😕
Ghosts start whispering "ooOOOOooOooo 601 diskette error oOOOooOOoooo"
Yes, it haunts him to this day (which is in fact not a very long period ;-) )
@@aaronjamt _ooooOOOOOooooOoo Press F1 to continue_
Just came across your channel.
Double beep I seem to recall is saying keyboard not detected properly.
I'll try to dig up my old documentation in a few weeks and see if I can find a reference for 601 we error.
@@TomaszWiszkowski OooOOOOoo keyboard not found OOOOOooooooo *_press F1 to continue_*
IBM PCs machines are notoriously non-IBM PC compatible.
To the point they put out the Aptiva line which WERE fully IBM PC compatible... unlike a lot of their other PC's.
if true IBM PC compatibility was important one could always by a PC clone
@sarahts21 which was too little, too late by that point.
Flashbacks from working on PC's from this era. The frustration is real.
I remember from the old days, that to overcome the 1024 cylinders limit you would use the LBA addressing, which was multiplying the heads on expense of the cyls, to keep the overall capacity unchanged.
I remember this as well....
Yeap. To get past 504MB you have to switch to LBA. And have a BIOS and OS that understands it. (Since DOS doesn't normally load drivers for IDE. Some IDE cards had an extension ROM to fix the issue. And at one point, I was loading a driver to fix that on my 386. But DOS was stupid; I quickly went to Linux. Except on some oddball Compaq's -- EISA??? -- that had to boot linux from DOS to get memory arranged correctly. Novel needed the same trick, btw.)
I can feel your frustration! Id love to see a disassembly and analysis of the 601 error. But the RPN calculator made me smile.
Back then IBM showed their assembly sources in the Technical Reference Manual for that page. Apparently they also provided the original BIOS source files on a disk.
@@ovalteen4404 Yes. I posted a comment last week with links to the manuals for the AT and noting on what page you could find the code that produces this error. I don't know why that comment has now vanished from here and from my post history. I'm not going to do the research again, but someone else might wish to now that you know the information is out there.
One of the best videos on the channel (even if you don't think so yourself, Adrian)! Brings back so much memories! Oh!
Adrian, many thanks for this video. I enjoyed an hour of retro PC maintenance and deep dive into the struggles of the early PC era. Keep up the good work and again many thanks!
Your struggles reminds me of some of the struggles my father would often run into when he fixed pcs as a side business (his day job being installing and repairing sun computers for Sun) and he would always be in a bad mood if he'd had to deal with any of the following: amstrad computers, Packard Bell computers, compaq computers, IBM computers and problems that turned out to be yet another problem with oracle (at work, oracle had a rather bad reputation amongst sun engineers as they'd always blame it on the Sun hardware when 9/10 times it was oracle).
You seem to be experiencing all the same issues and pitfalls that I fell in to when I was trying to setup my 5170 clone. Glad you were able to figure stuff out! Thanks for the info about that Quadtel BIOS. I ended up using a different Phoenix BIOS than you used and just dealt with the disk error, so now I got more options, awesome.
I am glad to see someone post their struggles, instead of cutting together a perfect repair :)
33:37 Nice chip extraction tool. I recently bought one because of your usage of it on this channel.
Don’t feel bad for modernizing this PC. This is the exact reason I don’t fool with old computers - but love to see other people do it!
I just found your channel and saw the three videos for the 5170. What an introduction to your channel. OMG, the amount of frustration you put up with trying to solve that 601 error. I miss the old systems sometimes, but NOT the frustrations of configuring hardware to work properly.
27:12 That cut into another cut made me imagine you had some choice words 😀
Yes, something like this:
th-cam.com/video/5vlhDtkWZGg/w-d-xo.html&lc=z22dvrxguwmfxhp4u04t1aokg0zt5t3qcuxxfjvkcb0cbk0h00410
LOL!!
Adrian, we need an Uncensored Digital Basement channel!
I wouldn't worry so much about the colors of the drive matching. Not even IBM did that. I've got a PS/2 in my basement that has one drive in beige like the case and the other drive in white. Delivered like that from the factory.
What an epic tale. I have had similar problems with a Pentium II based machine that just did not want to cooperate. In fact it's still acting oddly and I know I need to replace the floppy drive on it. Honestly I really admire Adrian's patience on this. I have a bit of a temper and I really do rage at these things, which is absurd I know. I need to remember Adrian's chuckle at the end of this video to recognise that it's just an old computer, be patient with it.
55:58 Wow. It's been decades since I've done a low level format. I'd bet that not many folks would understand what a low level format does; its important for everyone to know that a low level format is not something that you want to do with today's 'modern' hard drives....
The problems you encountered are similar to ones I’m having, so this is very helpful to me. Thanks for making this video!
And this video demonstrates why I preferred the Compaq DeskPro 286's over the 5170, although I did prefer a Model M over the standard Compaq keyboards.
Never had any of these issues with a DeskPro 286 and they ran faster overall than the 5170s
I like going back and wathcing this over and over. Reminds me of the proplems of the olden days. I started working on computers professionally in 1993. We still had lot of customers with the older machines trying to upgrade them and use them. I really bugs me now, because over the years I upgraded whole companies, and I literally sent 1000's of these off to recyclers.
Same could be said of Ford Model T's and A's I suppose. How many cool 1960's vehicles were destroyed because of some engine problem or tranny problem I wonder?
Absolutely compelling viewing. I'm sure we have all gone down a similar rabbit hole and it can drive you mad. The difference is that you actually managed to solve the problems. Fab. Don't worry about the purity thing. I think the purity police allow the swapping of peripherals for operational reasons :-) The motherboard, CPU and case are still there. BTW, you have put me off looking for a 286! :-)
Alas, some comments are complaining about his rubber dome keyboard not being a model M! So the purity police don’t even jive with peripheral changes… but apparently the LCD monitor is fine? Very strange.
This was a fine video. Don’t worry about anything. Always enjoy watching and learning from you
A little testing with QEMU and 4 MB RAM: the "Largest EXE Size" in MS-DOS 4.01 was 574992 bytes, but in PC DOS 5.0 (probably because it made use of XMS) was 636816. Confirming that even on 286 you're likely to get more efficient use of RAM from 5 than 4. (With 6.22 it's 607872 - still better than 4 but 5 may be your best choice.)
I'll have to try out QEMU on my 286 systems. I always run MS-DOS 6.22 on anything 386 and up. MS-DOS 5 seem to work better on my 286 and older systems as it also doesn't use as much hard drive space. I've found on a given 286, 1MB of RAM is normally more than enough so DOS can be loaded high. I only up to 2MB or more on a 286 if I want to run Windows 3.0 or Windows 3.1 on it. Most of my 286s just run DOS. I install Windows 3.0 on a TurboXT with an EMS RAM card. OMG it was painfully slow, but was neat to see running and using the EMS RAM. Because I was using PC-DOS 3.30, on that system, I had to partition the ST-251 40MB hard drive into two 20MB drives.
@@ESDI80 QEMU is a modern emulator, not something that runs on old systems.
PS - Besides the 'Dead parts tell no tales' t-shirt, I think you need a '601-Diskette Error' sticker. ;)
What is still amazing me is your inexhaustible patiente. Perhaps because you don't take what expressions you do with your face with every issue, but your voice sounds so calm. I would be hammering some things! :D
Your video reminds me of the very early 1990's when I switched from Commodore machines to PC clones. BIOS and hardware compatibility nightmares. As I recall, when PCI came in things got way better.
For totally unknown cards, taking a picture and reverse image-searching for it may be worth a shot.
Nice job on this one! I would have been beating my head on the wall by day 2 or 3. But you're an unstoppable force when it comes to rebuilding these machines!
Check the access time rating on the EPROMS you are using. One or both of the ones that didn't work might be to slow for your machine.
Have you heard of Compaq MS-DOS 3.31? It's a special release of MS-DOS 3.3 with only the big disk support from DOS 4+ added. If you're tight on memory but still need big disks, or if you just want to avoid DOS 4.01 alltogether due to its reputation, this is the version to use.
Regarding the utility to set keyboard repeat&delay values: Isn't that MS-DOS's MODE CON KEYBOARD REPEAT and MODE CON KEYBOARD DELAY or something?
And the "primary" or "secondary" floppy disk controller [setting] lets you use the "onboard" controller plus one in an expansion card slot, and set whether the onboard or the card one is used as #0. Basically you can use up to 4 FDDs, "A" and "B" per controller (Note: no drive letters, just the way FDDs are adressed per cable; with those 4 wires flipped between the two drives).
EDIT: a word in [] for clarity.
Also: What happened here? 45:30: The BIOS "forgets" your "swap floppies" settings on reboot, so you rebooted from your 5 1/4" floppy. This feature is for IT personnel who wants to boot a diagnostic floppy disk ONCE, but without messing with the office worker's PC setup.
I just realized youtube hasn't fed your videos to my feed in a month. I was like dang it's been a while since I've seen Adrian. Hopefully binging your recent vids fixes it lol
Hey Adrian... did you notice how there are 2 sets of holes on AT drive rails?
the reason is, that "half height bay" for a second hard drive can actually be used for a full height hard drive on a single floppy system. You attach the rails in the second hole so the drive sits further back, allowing you to put the metal bracket and that blank drive slot cover back in place.
Also, AT's never directly supported 1.44 floppies in the bios. You could use them on 5170's, but in order to be able to properly, you 1: had to tell setup it was a 720K drive. Dos would be able to boot from it, read to it, and even write to it. HOWEVER, in order to format it, you needed to use driver.sys, which was a device driver included with dos that would let you tell dos how many sectors, heads and tracks a drive had.
(yes, it was literally named driver.sys)
optionally, later versions of dos also let you specify the number of heads, sectors and tracks directly in the format command.
If you did not do either of the above, when you formatted a disk, it would only format to 720K. (it would assume 80 tracks, 9 sectors instead of 18 sectors)
1.44 meg drives were not introduced for IBM systems until the PS/2s came out. And IBM simultaneously discontinued the 5170s when that happened. So there was never a BIOS for the 5170 that supported 1.44 meg drives.
The bios used with the motherboards that have the 256k chips instead of the 128K piggyback chips (the third gen bios, the one you tried using) will support 720K (the 6mhz version was used in 5170-239 ATs, whereas the 8mhz was the 5170-339, both shipped with 512K, a seagate st-4038 full height 30 meg drive (bios type 8) Fun fact - there is a difference later revs between the 6mhz bios and the 8mhz bios. There is a timing check on the 6mhz one to make sure it is not running at 8mhz. This was because people were buying 6mhz systems for several hundred dollars less and then swapping out the 12mhz crystal for a 16mhz one. Most of the time a 6mhz 286 could handle 8mhz. If not, it was still cheaper to swap it out for an 8mhz part than it was to purchase the system as an 8mhz machine.
Anyway, however, that 601 error you got was not a bios compatibility issue. Post should just see it as an 80 track drive and assume it is either a 1.2 or 720K.
My theory is that Gsetup has options in it that the bios itself does not support, and you were getting the 601 because of some corruption in the bios ram due to using gsetup. Some extended bios ram byte, one AT setup does not set or use, was set to some value by gsetup, and it is confusing the POST because the IBM bios does not know what to make of it. Just a theory. Could be wrong. To test this, I'd suggest wiping bios ram (IIRC, there is actually a jumper to do this on the MB, or you could use that basic program you mention.), putting IBM bios back in, and try using ONLY IBM AT setup, and not even touching GSetup..
edit: also... that stand alone floppy drive card you tried... was that an 8 bit card? if so, bare in mind that these did not have the transfer rate needed to support either 1.2 or 1.44 meg drives. Those will work with 360 and 720k drives only. (although there are special 8 bit floppy controllers with buffering between the drive and the bus, which also have their own bios, that will allow the use of 1.2 or 1.44 meg drives on 8 bit systems... these are specialty cards, though. many of these also supported 4 floppies)
I was heavily in the PC world in the mid- to late-80s and I never remember an AT or any 286 (e.g. HP Vectra) with a 1.44Mb disk - I didn't see that until the first PS/2 came out. It's not period correct in any case to have a 1.44Mb diskette drive. The AT came out in 84, and the 1.44Mb disk in 86. By then the AT was in the rearview mirror of the Deskpro 386 and IBM's own PS/2.
My first PC in 1991 was a Mitac 16Mhz 286 with VGA and a 1.44MB floppy.
I encountered a lot of IBM 5170s in the early 90s and more often than not they had been retrofitted with a 1.44MB drive. It was a common upgrade to help keep those machines useful. The PS/2 drove people to install 3.5-inch floppies in other PCs so they could read the new disks.
@@thematicschematic 80s guys - 80s :) I realize I'm as old as Methusalah :)
Heathkit had an Eazy-PC in 1987 that came with either 1 or 2 720k drives.
Embrace the 5-1/4. :-D Who needs 1.44MB on a 286?
This takes me back. My first impression of troubleshooting PCs felt like this coming from Amiga which "just work". Though in hindsight my opinion was very biased, Amigas can be just as frustrating when you're dealing with obscure problems.
Well, I have to commend you for sticking with this one, I would have probably stopped and used this one for parts long before your time into it. :) Glad I was on the Amiga back then, fjew. :)
I had one of those on my desk at work in the mid 90s for the express purpose of managing the company telephone switch. The PC people said they hated that machine more than any.
When you get around to working on the original 5170 power supply, don't forget the load resistor for the 12V rail, since you don't have a big old full-height MFM hard drive installed to draw enough amps on the 12V rail, and the power supply will shut off.
^^ Truth.
When you formatted the hard drive did use fdisk /mbr ? Fixes a lot of problems... i had issues with getting a compact flash card to boot. Fixed the problem. Maybe it might work on the drive.
I think this could work as well with the CF card:
- Install your favourite BIOS
- Reformat the CF card with the 1570 having its new BIOS installed
- transfer the contents back on the CF, careful not to use an image for this.
The card was working with the AMI BIOS, so that indicates a problem with the geometry match-up. And that could be solved by reformatting the card on the configuration (i.e: disk controller, system BIOS and system motherboard) you're actually going to use it with. This should _always_ be your first step in troubleshooting this issue. Even some 386 boards I encountered with the exact same issue (drive would read small files fine, not larger files; drive would not boot) had it vanish after doing this.
The unknown memory board shown at 4:47 says "MEMOPLUS" on it. The best I was able to do is find a page listing memory expansion board that tells me it could be a 2MB board made by Suntek Information. The A.4 board is for IBM PC, AT, PS/2 Models 25, 30. The X.4 board is for IBM XT. I haven't found any additional information. For reference the information was at found www.atarimagazines.com/compute/issue112/Memory_Expansion_Boards.php which showed page 42 from Issue 112 (Sept. 89) of COMPUTE! magazine.
On your ATI card, resolder the crystals. I have one and one day lost H-Sync on the VGA side. Resoldering the crystals seems to have fixed it.
You my hero I wouldn't dare go that far! the effort and the accomplishments I do salute! :)
Great work Adrian
24:40 You probably need the matching IBM setup program with the newer IBM AT BIOS. From what I remember several versions of the IBM setup program exist.
58:56 I've run into that problem in the past, where drives would not function properly when transplanted between older machines. It seems to have something to do with the way that disk controllers address the disk's layout. That is one of the things that I always appreciate with new machines, the fact that you can create partitions on one machine and move the disk to another and it works perfectly. I have run into problems with newer machines not being able to interpret a disk created in a different machine though it is rare. I always have that odd feeling that it won't work when I transplant disks though, which is something that I have ingrained into my brain from the old days...
I love waching videos like this. all that troubleshooting and trying to make such an old hardware warking.... Just love it I wish I was able to play around with such great machiness as well Great job. Look forward to seing part 3 :)
Given that these bioses use CHS addressing for accessing hard drives (as opposed to LBA that was originally specific to SCSI drives) could it be possible that the compact flash reader received a malformed IO request that it could not complete?
Or, to put it differently, would these machines work if you used SCSI hard drive interface for the compact flash hard drive?
Super curious. If you ran into the same problem with 386s perhaps the problem is related to how old IDE was trying to access the data. CHS has always been a headache.
I'm particularly curious if the bios was setting some "reserved" bit in io that might even have had some meaning, that now caused a headache
OMG this makes me think I truly lucked-out with my IBM 5170 project... not to say that there weren't unexpected issues, but certainly not to this degree! Oh, and I solved the need for IDE support by adding an XT-IDE EEPROM (cheating perhaps) but I get to keep the purity of the original IBM BIOS, use more modern parts, and I can choose from any available FDD or HDD at boot time (like the swap function you were after.)
You have more patience than I do, that’s for sure. Nice job on getting it working.
When you get around to testing the power supply, make sure you meet minimum load requirement for it to regulate properly. It needs 7A on the 5V and 2.5A on the 12V. For AT computers configured with no hard disk, IBM would install a 50W dummy load.
And don't use paperclips or a Dremel.
Many 3.5" fdd's frontplates could be exchanged.
50:14 its funny when I think of the year 2000 I think its recent but then realize that was 21 years ago. That 2000 trips me up.
Back in the day I worked for IBM. I was given a 5170 for my desk. I remember adding memory and a 3270 adapter for mainframe access. I also added a 3.5 inch drive as the 1.2 meg drive was compatible with nothing. I certainly did not run into all your problems. But, I had access to all the proper period parts. Ain’t retro great.
Hello Adrian! It CAN boot into DOS with a 1.44MB 3.5 inch drive! You just need a drive where you can set the drive to be drive 0 instead of drive 1. From the 386 era, PC's use them as drive 1 but it does not work with the 5170. Jumpers must be set(or soldered on newer TEAC drives) to drive 0 in your case with the ORIGINAL BIOS.
Love your videos since the begining, great work Adrian!
Remember these times, when I need to mess with new HDD Head and cylinders settings in BIOS.. 😂 Nostalgic, especially, when these settings wasn't on HDD sticker..
Yeah, BIOS issues may take time to figure it out.. Try, just for fun, connecting IDE CD ROM drive.. I think, that BIOS wouldn't recognize it (back in day's, I run in that issue)..
Now you are seeing why Intel dumped the '286 so fast. Here's a fun fact, in order to use all the memory you have to go into protected mode, which is incompatabe with DOS. You can't go back to "DOS mode" once you start Protected Mode without halting the CPU. The 386 can go back and forth so you can use things like use things like DPMI to use all the ram you have for DOS.
Seem to recall that on these early IBM machines you also can't use the 287 in protected mode, but that may have been a BIOS issue. I have however never seen a 286 that had the FPU, likely because they never had enough RAM to run anything that could utilize it
IBM had a hardware trick to get to get back into real mode: they reset the CPU. I think it was Bill Gates who called the 80286 brain damaged.
12:05 I love the way the 5¼" green diskette teleports over to meet its 3½" cousins.
The screen capture is really nice
Overall cool diagnostic work, love writing my code to these
i have really enjoyed watching this, and your problem solving ! great work and patience !
Now that I see the restriction to 3.5" with 720K, I remember to have had the same problem back then with a XT clone. A BIOS without built in setup was not uncommon at the time. Even my first 386 lacked that and came with a DOS setup program.
By the way: It's easy to overclock the AT, just swap the clock crystal for a higher frequency. I used to run mine at 10Mhz and at one point had 12.5 MB memory in it for testing purposes. At the time the memory was so expensive that it was a lot more than the PC At itself, and the AT was not cheap either. All that memory was for customer network servers running Novell NetWare 286, one of the few things that could actually use the 286 and memory to its full capacity.
Adrian had a different video with the oscillating mod. But he had to have a mod done to remove the over locking clock bios routine. And a few other things hidden "features.". If I remember correctly
Trying to fix a 35 years old IBM at midnight...hmm you are a braver man than me!
been trying to track one of these down for years...finally got one for a small fortune but man, what a good looking machine!
Great follow up video. Would love to see a workshop tour video some day!
I think Dave Just Dave toured Adrian's lab a few years back.
I think this is my favorite cover picture for your videos so far. Like, "what?"
I really loved this one . Just goes to show that even the best get challenged every now and then .
I have vivid memories of my dad trying to install a 1.44 MB floppy drive on an XT that he had and I definitely remember that he ultimately gave up on it.
Don't know where the dividing line is, but if you go far enough back, beyind just requiring some sort of TSR to override some BIOS behavior, you would also have needed to replace the floppy controller chip with a newer one, because much of the floppy controller circuitry was never on the drive, but instead on ISA card or motherboard.
"Who the hell is General Failure, and what is he doing on my diskdrive?" 😆
Jesus Christ. You have the patience of a saint. I would've given up on it and not touched the machine ever again.
Funny to know that "All in Wonder" was something different before it was the Graphics Card + Tuner combo that associated with that name when growing up.
I remember those days. It was bad enough when compatible hardware was available, although the hardware did get better for several years even after the 386 came out in 1987. Once the 386 achieved dominance, and especially once the 486 landed, there wasn't much further boundary-pushing on the 286 front.
Frankly I wouldn't try to put any IDE solution in a 5170, not even one using spinning rust. (SCSI and RLL are certainly fair game though.) If I can't get hardware appropriate to the time, then I'd just have no hard drive at all, even if that makes the machine essentially useless for more than games. IDE did not exist when this machine came out, so if the machine balks at using IDE, I would prefer to keep it stock and lose the functionality because _none of these machines are particularly useful_ in reality. The same is true of the 1.44 MB drive. If the hardware wasn't amenable to supporting it, I wouldn't install one no matter how convenient it might be, because that's just not how this machine would have ever worked in its useful lifespan. If I want a 5170, I want a 5170. If I want a machine that represents a souped-up 286 rig, then I'll make one out of clone hardware that isn't so buggy. But this machine is now neither. The only thing really original at this point is the motherboard and its 6 MHz clock rate, which is a negative, so basically we've ended up with the worst of both worlds -- a slow 286 that doesn't behave anything like a stock 5170.
I admit that stock 5170 behavior is infuriating, but I feel that if that's what you're building, those are the constraints you have to work with. Otherwise it's just another 286 build in an IBM case, at which point you'd enjoy using it a lot more if it had a 12 or 16 MHz clone board in it instead. Once I established that I would never be able to enjoy using a period-correct 5170 because it's buggy and slow, I'd get it minimally working with as much original hardware as possible, and only period-correct replacements when needed. No modern workarounds. Then I'd send it off to a museum or collection, and build myself a clone rig that actually works with the stuff I keep around. Sadly, this build ends up being neither of those things, especially given the amount of work involved.
I agree
Hi Adrian, would you mind putting a link to the shiny tool for taking off the chips from tight places? The one seen on minute 33:30. Thank you for another interesting and entertaining video!
You deserve a medal for perseverance
When I see an RPN calculator, I know I've found a friend. :-)
Lol, this was the point were I went : Ok, he is definitely a (real) engineer ;)
Only the braves use rpn.
RPN Squad Assemble
I recall my painful experience setting up my first PC which run V20 with 384K on-board, the extra RAM card hold the extra RAM, altogether only 1MB, and 360K floppy only, booting DOS 3.3. Today BIOS is so sweet to setup compare to the old one.
Maybe these EPROMs were just too slow...
I once fixed an Asteroids machine, the ROMs had disintegrated and I had to burn new ones. They verified fine, but the machine didn't like two of them. I thought, maybe some bits already flipped back, that can happen, you just have to burn the image twice. But they did verify. Asteroids runs at 1.5MHz and the 6502 is pretty quick when it comes to access times. So when I had to burn ROMs for a slower system, I used these and they worked fine.
VGACopy/386 includes a boot sector with a boot manager that can boot A: B: or C:
It works pretty well and should start on a 286.
@Adrian's Digital Basement Daft question perhaps, but I do wonder if some of those 6xx class errors occur with the older BIOS versions because the code is expecting some hardware to be available that was included on the drive IO card, rather than the Motherboard?
A bit like the way they included the Gate A20 controller in the Keyboard?!?
I don't know, that just came in a flashback, so I've no idea if I'm making it up or not!
The screen shown at 9:30 reminds me to the BIOS of some NEC PC-98 computers.
Also, those settings from IDEDIAG seemed very wrong to me. Sector size of 528 bytes? Track size of 16896 bytes? Those both seem off. I have never seen any media with anything other than exact powers of 2 in those categories. You would expect 512 byte sectors and 16384 byte tracks. At least the values it gave are multiples of each other, but I think that program is probably having issues because it's a compact flash card.
This video is the perfect example of why I simply do *not* understand this weird "purity" obsession with some collectors. Who cares if it's all original? The people who actually USED these machines (and that's their purpose, to be used) wouldn't have wanted to keep it stock. They modded these machines, upgraded BIOSes, added new drives, CPUs, etc. If I find machines from this era that are heavily modded, I lean into that and make sure to try and keep the mods functioning, instead of undoing all that work to restore it to a useless, frustrating to use, "virgin" machine. It just makes absolutely zero sense to me. I collect old computers because I find using them fun -- I don't just put them on a shelf and look at them.
Dont know if this helps Adrian, but that second RAM card says 'memoplus' 'Rev 7' at the bottom of the card
It's actually MEMOPLUS A Rev 7.
You can find them for sale on eBay but no luck finding a manual yet.
Thats why I passed on the IBM XT computers and built generic 386 and 486 computers instead. Too many problems and limitations on the IBM XT computers. My goal was for best DOS game play anyways.
Hi, you could probably use the Phoneix BIOS and XTIDE BOOT ROM on LAN card, which will solve the problem with any modern HDDs ;)
An ide controller card with an external bios will probably solve that issue like you said. I have a k6-2 machine that under DOS the onboard IDE didn’t support LBA under DOS so my 3.2gb hard drive was only visible as 502mb. I bought a promise IDE controller card and all 3.2gb is now usable. Not quite the same issue, but it should fix the problem(although he’d need an ISA one with an external bios such as an XTIDE)
Not sure if either of these things were brought up already but:
1. Another network type that used coax and eventually twisted pair was Arcnet. Arcnet was popular during 80's and early 90's.
2. Problems with booting HD when it otherwise works: I doubt you had a problem with the partition table itself or else you wouldn't have been reading data correctly from the disk. However there is bootloader code on the sector 0 that contains the partition table. That probably was bad or incompatible until FDisk re-wrote it. Not sure if there's a command switch to rewrite the code without wiping the table.
The three greatest computer industrial designs: IBM PC/XT/AT, Apple ][ and MacBook Air. PS/2 was bumped by the PC/XT/AT in my top three simply because IBM tried to close the gate after the horse had bolted.
I think those square (rectangular) batteries were 9 or 12 volts, not 6. That may be why it keeps giving you the “162 - options not set” error. Check the original battery and see if it really was 6.