I finally got the hard drive installed in an IBM 390E I bought in the spring. It came with a 20G drive with a broken pin and a caddy pin to slot adapter with a matching missing contact (at least after I removed the broken pin from it). The final solution came in flipping the adapter 180 degrees, after I'd made a contact from copper and slid it into place after the adapter was on the hdd.
This machine now has a PII 366, upgraded from a 333, 256MB ram (thanks, jimbo ), and a Samsung 40G 5400rpm 8Mb cache drive. Just put 2K in it.
It has some suspend type functions in its bios, which I'm not used to yet. I've been using an old Acer which lacks any of those for anything but 95. (That's the machine that had the display corrupting. That problem went away and it's been in daily use for months now.)
Here's the problem: it's locked up solid about three times, once after an extended on time, once defragging. When locked up nothing responds, keyboard, cursor, C/A/D won't bring up Task Manager. One further time it just shut off totally. The battery is completely gone, but the psu adapter connection is solid -- moving it around will not kill it.
So I'm suspecting a heat issue. I've had the keyboard off three times now, and noticed that the contact for the cover/heatsink over the cpu has a pad. I could clean the surface of it and the cpu and apply fresh grease, but if I removed the pad, which doesn't inspire confidence, I'd have too much thickness missing. I suppose I could try and find some copper of the appropriate thickness and use an adhesive thermal product, but that's getting pretty complicated. I don't think a PII of this vintage will have any temp reporting, nor would I expect the IBM bios to retrieve it either. I could probably check the hdd temp, but the Samsungs run cool.
Any suggestions greatly appreciated, especially anyone who's experienced this type of lockup.
I dunno, if you're using a doctored cable for the HD, you might be having issues with the connection. A lot of times when a hard disk is disconnected while the system is running, the system will just lock up. Perhaps as the drive gets warmer, it expands enough to affect the cable connection. Maybe that one pin getting disconnected is enough that windows loses the drive. Just an outside guess.
That could be a possiblilty. There's no cable, its just an adapter that fits onto the hdd pins and slides onto a slot connector when the hdd caddy is slid in. I could make another contact to try and address that. I tried to find the adapter, both at local electronics stores and online, without any luck.
Caddies are crap. The machines should be designed so a hdd just plugs in directly, underneath an access cover. I think some of the new ones do just that.
Inspiration: conductive silicone grease of the type used for alum/copper connections!
Last edited by DanceMan; December 8th, 2005 at 06:19 PM.
I'm going to run memtest86+ from the Ultimate Boot CD (floppy on this doesn't work) and that should eliminate heat as a problem and keep the hdd out of the equation. Then I'll remake the copper contact and try off the hdd again.
Thanks for pointing me towards the drive.
I'm only running it to load the system. I think I ran memtest many months ago, but that might have been with the 333 cpu in it. Too long ago to remember correctly.
Memtest appears to have locked up at 2 hrs 5 min. With 0 errors.
How much does memtest load the cpu? I'm going to run one of the cpu load utilities from the cd and see what happens. The hdd is still in the system. After a cpu test I'll remove it and try some of the tests again. That should tell whether the hdd is causing the lockups.
Ran Cpu Burn-in for 2hr 40min with no problems, so I think I can discount any cpu heat issues. Ran Everest about 15 min after that and cpu was 57C, HDD was 43C. Computer locked up in less than an hour running on the hdd. I'm running memtest86+ again, this time with the hdd removed. At the moment it's at 1hr 20min. If I get any lockups, I'll remove one of the sodimms and test one at a time.
Memtest86+ locked up at 2hr 9min, very close to the previous 2:05. It was soon after memtest began a new mode. I think I had already removed the hdd. So I removed one of the ram sticks, the Kingston, and started memtest on the other, one with Hyundai (Hynix ?) chips. Just got home from a long day on a rock show. It's run for 25hr 22min at this point with no errors, no lockup. The other stick, the Kingston, is going into the same slot now.
I still wonder if heat could be a factor, for the ram if not the cpu. The ram slots sit right under the cpu, and IBM put a paper thin copper sheet on the inside of the ram cover, likely as a bit of cooling. With the ram maxed, there will be more heat. But sitting under the cpu means it's already in a warm spot.
While swapping ram I noticed that the slot I'm using for the test has an insulating sheet under the ram, between it and a circuit board immediately underneath. The other slot, the one the Kingston was in originally, is bare.
Last edited by DanceMan; December 11th, 2005 at 05:11 AM.
The Kingston stick ran for 10hrs 15min with no errors or lockup.
I made an insulating sheet from a piece of plastic to fit under the second ram slot, and I'm now testing one stick at a time using the second slot. If both sticks pass this, I'll fill both slots and test again.
Tested both sticks alone in the second slot with a piece of clear, stiff but flexible plastic to insulate the stick from the circuit board underneath. The Kingston stick ran 6hrs 44min without problems. The Hynix stick ran 5 and 10 hours without errors or lockup, but locked when I pressed Esc to exit (first test) and locked when the comp was bumped as it was picked up (restest).
With both sticks installed and the second insulator in place, memtest ran 3hr 20min without lockup (would only do just over 2 hr before) but the comp locked up when the comp was picked up.
So I can conclude that neither stick is bad, but the Hynix stick is prone to lockup the comp when used in the second slot, or if both slots are filled, if the comp is moved after an extended period of operation. I think that heat might be the cause, and that one of the sticks is more susceptible to the heat.
It just repeated the same behaviour. Ran for 3hr 19min without a problem, but locked up when I picked up the laptop. No bumps, but I moved it around, sort of medium gently. It's running off an Ultimate Boot CD, with no hard drive installed. But the memtest program should be running completely from ram or cache once it starts. Before I put the insulator in place, it would lock up on its own just over 2 hours on memtest. So that's helped. Now I have to find a way to help it more.
I have 64Mb sticks. I'm going to try them along with the Kingston stick. If they prove resistant to movement after it's run for 3 hours, then maybe replacing the Hynix stick with another 128Mb stick will work. And I'm going to see if I can do anything to improve heatsinking on the cpu and the ram.
Ran one more test with both 128Mb sticks, ram cover off and a fan blowing directly on the ram. It ran memtest for 4hr 15min without locking up and wouldn't lock up even when picked up and moved around. So it's definitely temp, and specifically one stick that's temp sensitive.
Ran it for 12hr with the 128Mb Kingston stick and a stick of 64Mb PC66. No errors, no lockup, and it wouldn't lock even when picked up and moved. Just for interest, the chips on the 64 are MT, which must be Micron.