I know it is old! It has a full height 30MB drive. What it has begun doing is hanging up during boot the first time I turn it on. After the memory test it just hangs. If I then power it off and power it back on, then it will work. I can hear it spin up and make a few noises at the end of the spin up cycle, maybe I should listen closer to the first time I power it up to see if it sounds different. If I run scandisk with s surface test is always passes fine. It just started doing this in the last few months, though the first time i got it, it was having problems spinning up until repeating cycling finally got it to do it and I ran it upside down because of someones tip to try to distribute the bearing oil/grease. I could run it upside down again for 24 hours if that would help.
- VCF East - April 12 - 14, Infoage Museum, Wall NJ
- VCF South West - June 14 - 16, Davidson-Gundy Alumni Center at University of Texas at Dallas
- VCF West - Aug 2 - 3, Computer History Museum, Mountain View, CA
- VCF Midwest - Sept 7 - 8 2024, Schaumburg, IL
- VCF SoCal - Mid February 2025, Location TBD, Sothern CA
-
Please review our updated Terms and Rules here
PC/AT 2170 hard drive starting to have an issue
- Thread starter alank2
- Start date
NeXT
Veteran Member
Is your "fix" only on a hard power cycle or will a warm reset from pressing the reset switching working as well?
ST251
Experienced Member
If it's possible, could you maybe make audio recordings of the drive starting, both when it's making the system hang and when it starts normally?
modem7
10k Member
Refer to [here]. If this is the hard drive's spindle not reaching adequate speed at first power-on, due to some sort of lubrication issue, then I would expect the IBM POST to issue a '1780-Disk 0 Failure' error (because the drive would not be responding as 'ready' at the time of the 1780 test).
Might this be a temperature related problem that has nothing to do with the hard drive? In the CMOS SETUP, temporarily set the hard drive type to 0, then when the 5170 is 'cold', see if the 'at first power-on, system hangs after memory test' happens when you boot from a floppy.
Between those two, are you waiting for the drive's spindle to stop, or is it a relatively quick power off/on cycle that you're doing?
A 1782 error sometimes only appears after a timeout of 1 minute. Are you waiting at least that long after the memory test?
Might this be a temperature related problem that has nothing to do with the hard drive? In the CMOS SETUP, temporarily set the hard drive type to 0, then when the 5170 is 'cold', see if the 'at first power-on, system hangs after memory test' happens when you boot from a floppy.
I probably don't usually wait long enough for the error - part of me is concerned that the drive didn't startup properly and my fear is that it is hurting itself in that condition. Like it isn't spinning and just the motor is just generating heat instead of running. It is running so maybe that isn't a concern.
I do think I've seen the error at least once.
It is a quick power cycle - I turn it off long enough to hear the drive spin down, then I power it back up.
I just tried it again, and while I will say it worked and booted, at the end of the startup cycle is a distinctive two tone sound like "beep-beep". This time it made that sound 3 times while the memory test was going on (it has 8MB memory which takes awhile to count).
I do think I've seen the error at least once.
It is a quick power cycle - I turn it off long enough to hear the drive spin down, then I power it back up.
I just tried it again, and while I will say it worked and booted, at the end of the startup cycle is a distinctive two tone sound like "beep-beep". This time it made that sound 3 times while the memory test was going on (it has 8MB memory which takes awhile to count).
More on this - if I wait it will come up with a 1790 error. Oddly before that it will cycle through the two floppy drives and the hard drive light will blink on and off quickly like it is being checked and everything is fine. But it isn't. Waiting will give the 1790, followed by floppy access, followed by cassette basic. If I CTRL-ALT-DEL in cassette basic, it will do the same thing again, completely with HDD blink and 1790. A power cycle at this point will boot the HDD however.
modem7
10k Member
So, per [here]:
1. The controller internal diagnostic passed; and
2. The controller's 'recalibrate' (take heads to cylinder 0) of the first hard drive worked; and
3. The controller's read of certain sectors FAILED.
Re step 2. At this point, if (due to a problem) the spindle was not up-to-speed, the drive would not report as 'ready' and the POST would issue a 1780 error.
Re step 3. One of the sectors is the last sector ("GET MAX CYL, HEAD, SECTOR") on the drive, and so as part of that, the heads are expected to move to the final/maximum cylinder.
The hard drive light on the front of the IBM 5170 is connected to the controller, and so is expected to be on when the controller is getting the drive do something, or TRYING to get the drive to do something.
Your "A power cycle at this point will boot the HDD however." suggests to me that the problem is the hard drive. I have spare drives, and so if I was in your situation, I would quickly prove/disprove the drive.
As an experiment, have you tried low-level formatting the drive (followed by FDISK, etc.) ?
modem7
10k Member
If this was a spindle 'spin up speed issue', a 1780 is the expected error, not 1790.
But if this is a drive problem of some sort, maybe replacing caps will do something. But before that, I suggest that you first try something simpler, low-level formatting the drive (followed by FDISK, etc.)
modem7
10k Member
I have seen the following behaviour on more than one of my MFM drives:
The drive is low-level formatted, partitioned, then high-level formatted. Running the surface scan in Norton Disk Doctor (NDD) and SprinRite shows no problems at all as all sectors of the drive are read. Then over days, weeks, months, areas of the drive become unreliable. I become aware of that when I go to run a certain program (that loaded off the drive with no problem before) and DOS tells me there was a problem reading. If I run the surface scan in NDD, I hear the drive 'hunting' (back to cylinder 0 then back to the cylinder under test) in certain areas of the drive. Sometimes NDD will mark the corresponding cluster as 'bad'.
Later, if I then again low-level format, partition, then high-level format the drive, NDD and SprinRite show that all sectors on the drive are easily readable again.
The Seagate ST-4026 drive (defect table on drive's label is empty) on my primary IBM 5170 is in this category. Hours ago, there were many clusters marked as bad, and I could hear the surface scan in NDD having problems reading certain areas. I just now low-level formatted, partitioned, then high-level formatted it. NDD and SprinRite now show that all sectors on the drive are easily readable again. I am not surprised by some sectors becoming problematic - the drive is very old. The interesting behaviour about this for me is that low-level formatting a sector, can initially result in that sector being no problem at all, and then within as little as days, the sector starts to become unreliable.
The drive has gotten worse. Unbelievable is that it actually said Starting MS-DOS this time, but returned an error while booting. It sounds like it is spinning up/down. Any thoughts on whether there is anything that can be done to repair it? I can pull it from the PC and put it on a bench power supply to see if it acts the same.
Are you guys able to listen to the audio file above? I am surprised no one here has any ideas. Should I start by pulling the drive and putting it on my bench power supply? What would cause it to not be able to spinup? Old capacitors I might be able to replace?
modem7
10k Member
The spindle speed sounds very much like it is varying. For example, at about the 0:57 mark, the speed picks up, then starting at about 1:10, the speed sounds like it is erratic. It certainly doesn't sound right.
If you had an oscilloscope, you would be able to confirm the varying speed by monitoring the frequency of the index pulses.
Some ideas, assuming varying spindle speed:
1. PSU: Unstable +12V supplied to drive, due to poor/faulty regulation of the +12V from the power supply ?
2. Connection: Unstable +12V supplied to drive, due to poor +12V connection (and/or grounds) on Molex connector - see [here].
3. Drive: Unstable +12V within drive, due to a problem in drive (e.g. a component on the +12V line 'breaking down').
4. Drive spindle: Lubrication or bearing related
5. Drive spindle motor: Windings
6. Drive: Electronics that drives the spindle motor.
Yes.
Well, the 'replace all aluminium electrolytic capacitors' shotgun approach does sometimes work. I remember buying some faulty 8" floppy drives, and thinking something like, "These drives are really old. Will I go straight into diagnosis using my oscilloscope, or will I try replacing all aluminium electrolytic caps first" (having many spares on hand). The latter worked, saving me effort.
FWIW, A few years ago my 5170's PSU failed, The symptoms were, The 5170 would boot successfully and sometimes not, This got worse until it wouldn't boot at all, On power up i could hear the hard drive spinning up and down, I checked the hard drive on the bench with a different PSU and it spun up perfectly fine so i pulled the PSU and checked the Caps, I replaced 2 Caps in the secondary side but left the others because they tested "In spec", The 2 replacement Caps cured the problem but the PSU failed again nigh on a year later, The Caps i didn't replace failed, 1 was a dead short and the Transformer burned. I wish i'd done a full recap the first time, I ended up Recapping a 300W AT PSU i had and transplanted the guts into the old 5170 PSU case.