Super-cool sysadmin fixes PCs with gravity, or his fists Welcome to another Friday and therefore to another edition of On-Call, The Register's regular recycling of readers' recollections! This week, meet “Donald” [Why did that pseudonym come to mind? - Ed] who shared a tale that riffs on last week's dark, magical On-Call. Donald's tale starts with a late eighties gig “doing tech …
I wrote a small TSR (Terminate and Stay Resident) program in X86 machine code to automatically park the heads on all of our hard disks, as they didn't auto-park but stopped where-ever they had last read. One of our workstations had a sticky drive too, so I swapped it for the one from my workstation so I had the problem, and every Monday morning I just opened the back of the machine and swivelled the HD back and forth (it wasn't screwed in) until I heard the disk spin up. Lasted several years like that until we upgraded.
"The heads would *probably* be parked until the disks were spinning..."
Didn't disk drives use to have a parking track on the disk where the heads could land and take off? Any extreme shock could damage the heads - even when powered off.
"Back when I were a youth the approved method of fixing vertical hold problems on a TV was to give the blooming thing a thump on the side."
An early 1970's cartoon depicted a couple of TV repairmen peering under the bonnet of their broken down van, one suggesting to the other that maybe a hammer would fix it.
Everyone I knew got the joke.
Actually had to pay a guy to get my car going again, ROADSIDE support. Car wouldn't start.
I was covered in baby vomit, it was Christmas Eve, just late enough in the afternoon that everyone had gone home. Wife small child, and unrepentant baby in tow. Stuck in the middle of town and the car doesn't start. This is what nightmares are made of.
His solution rubber mallet firm thump to the rear bumper.
Car starts perfectly.
(I'm dumbfounded)
Rationale:
The thing had a submersible fuel pump in the fuel tank. Thump on bumper translated to thump on pump, allowing the thing to start.
Anonymous for personal safety reasons.
Most V8 and some V6 Ford starter motor solenoids are notorious for failing at inopportune times. A good wack with a tire iron will usually unstick it in an emergency. (You often have advance warning that permanent failure is immanent because the car fails to start when hot, but starts fine when it cools down & the tow-driver gets in and casually starts it ... fortunately starters are an easy R&R.)
Come to think of it, Granpa used to call a 5lb ball-peen "Ford tool #1" ...
"It sets a bad precedent/example for the users though."
It also hardly ever fixes the problem permanently. The vast majority of such TV faults were caused by unsoldered joints oxidising (no, not dry ones. Just about every set leaving the factory in the days of manual assembly had at least one spot on the board the solderers had missed. Wave soldering pretty much eliminated this cause in the mid-late 1980s)
As for sticktion drives - replace them. I know it's easier said than done but in an environment where the labour cost of having someone look at it needs to be factored in, repeat calls will outweigh the price of a new drive.
If you're talking about pre-1980-ish sets: they were full of modules, connected to the main boards with the cheapest connectors known to man*, and a thump would move the buggers a little creating fresh contact between the module and its socket. One more decade back you'd find most colour sets still having a few tubes (apart from the CRT itself) because of the rather prohibitive price of semiconductors able to deal with the 25kV required to drive the picture tube. Those would sit in sockets, also prone to oxidising and degraded contact over time. That part of the circuit was also used for horizontal deflection.
* a friend started his career in the early 1980s designing video recorders. He was told not to bother about trying to reduce parts count: that was the job of a specialised department that would take his design and try to trim the BOM while not affecting functionality and quality too much.
* a friend started his career in the early 1980s designing video recorders. He was told not to bother about trying to reduce parts count: that was the job of a specialised department that would take his design and try to trim the BOM while not affecting functionality and quality too much.
The real meaning of "Quality Control".
Consider that a bonus, since it would hopefully mean upgrading from the old MFM based disks to something modern and up-to-data, like PATA
Back when MFM was common, you didn't "just switch" to IDE. You'd need a different controller, of which you could only hope it was compatible with your system's motherboard
IDE didn't need a "controller", per se, just an extension to the system bus ... The name is Integrated Drive Electronics for a reason. If the motherboard couldn't handle a generic IDE card, chances are that motherboard caused all kinds of other problems due to bad design.
They are knackered anyway. A bloody good whack did the top.
The unoffically-approved means of fixing drives from the old Sparc Storage Array (good old 9GB SCSI drives that had a tendancy to not spin up when you switched the array on) was to take the drive out of the array, knock it on the desk (firmly but not *too* hard) and then plug it back in.
Worked abou 80% of the time.
After hearing the frustration of a fellow support colleague who'd forgotten his changed-earlier-that-morning password and couldn't login, I told him that switching his PC off, waiting 30 seconds, and then switching on whilst pressing the embossed ICL badge on the PC would reset his account to a default password of "iforgotit". He sounded sceptical but tried it and it worked. Bless him... his little face lit up with his new-found knowledge, and he couldn't wait to try it out the next time someone called in with a similar problem. As it was lunchtime, he even bought me a pint.
I didn't have the heart to tell him I'd reset his password to "iforgotit " whilst he waited for his PC to reboot...
complained that his screen was so dim he couldn't read it any more - new screen please.
I found it, and brightness already turned up 11. It was, however, filthy. A quick look at my boss, (the BOFH but only an ex-policeman, so not so bad) we thoroughly cleaned exactly half the screen.
Dirt was a collection of cigarette output, dust from adjoining factory, and general scum. Given the entire environment was a bit grimey this could have been quite predictable for the customers.
The clean half was now so bright that it looked like a nuclear detonation.
We left a note suggesting they may give their devices a wipe from time to time...
the proper method of 'repair' was to take the drive in hand and oscillate the case in the the same plane as the disks. Working on the disks alone reduced the chance of percussive un-maintenance of other PC parts.
I did accumulate several of these that were chucked out at work and they ran great for years unless turned off. I must have had nearly 200MB of them at one point!
Aaah, the good old thump.
1) I used to service an own-brand laptop PC, that was built around a Uniwill chassis. Not a bad machine for the time (2004-ish) but used desktop class processors, as most did back then. A desktop P4 in a laptop chassis creates "a smidgen" of heat. This machine also used to use a socketed BIOS chip, fairly close to the processor, and eventually the heat seemed to "creep" the chip out of the socket with hot/cold cycles.
In they'd come for repair, I'd open the box, take the laptop out, drop it from about 4" which shook the BIOS back into the socket, boot it up, test and return.
2) Toshiba laptops, back in the days they were grey and the best you could get. I'm thinking a 2520CDT or something. Anyway, one range used AMD K6-2 processors and Toshiba had just moved to those low profile CPU sockets that slide side to side to lock the processor. You just put a large flat blade screwdriver in a hole and lever it over. Except their sockets were moving back by themselves after a few months, unlocking the processor and causing no POST.
The quick fix was two screws to lift the keyboard, take out the long screw directly in the middle of the chassis, slide the longest flat screwdriver you have under the top case until you feel it touch the side of the processor. Smack the end of the screwdriver which locks the processor again.
(Eventually Toshiba came up with a fix that involved something that looked like a lego brick to drop into the screwdriver hole, stopping it from moving back)
A note on hard drives - not necessarily from "the olden days" but drives covering the last fifteen years or so have a head support when off. A little capacitor stores enough juice to ping the arm into a plastic holder which is away from the platter surface as the power is removed, which means they can put up with a fair amount of G if powered down. On a laptop drive, often the (glass) platters will shatter before the head/arm are damaged.
If it's attached, I'd guess that you'd break the data/power connectors before you caused any damage to the disks, or the parked head.
I can confirm though that you can drop an unpowered harddrive from above head height, onto a solid floor and it will still work fine. Well, by fine I mean that it managed to hold an OS for fifteen minutes while the the PC was QC'd and it got sent to a customer regardless.
"eventually the heat seemed to "creep" the chip out of the socket with hot/cold cycles."
My Apple ][ had several chips that were in sockets. Every couple of years the machine would play up. The cure was to take each of the chips out of its socket and then replace it.
The sockets had tin plated contacts - not gold. The chips were drawing a low current - which apparently allowed the build up of oxide between the chip pins and the socket.
Some of the guys I worked with there started out doing field repairs on home computers - at the time, Amiga 500s and Atari STs were the big players.
Most 500 faults were fixed by just taking out every removable IC and replacing them again, for that same reason I guess or to reseat them. They could confidently go into a customers house with just a chip puller and a philips screwdriver and fix most issues.
"The sockets had tin plated contacts - not gold"
The oxide buildup wasn't due to the low current. That was the dissimilar metals of the socket and the IC pins causing corrosion in the presence of moisture (in the air). the low current just meant that the oxide layer wasn't being punched through.
Gold plated sockets need gold pins. Tin (solder) plate need tin pins. Mix at your peril.
"The sockets had tin plated contacts - not gold. The chips were drawing a low current - which apparently allowed the build up of oxide between the chip pins and the socket."
My old Acorn A3000 had a tin plated header for the RAM expansion (in those days Acorn machines either has sockets for discrete chips or as in this case used their own design of memory modules) and was notorious for the pins oxidising, requiring the module to be removed and reseated every so often, although mine never suffered from that problem.
On a laptop drive, often the (glass) platters will shatter before the head/arm are damaged.
We had an engineer (in the late 90's) who was tasked with measuring mobile phone coverage. Rather than going through the hassle of parking, booting up his laptop, doing the measurements and shutting it down again, decided that he would drive along with his laptop on all the time.
After his third hard-drive[1] in a month he was banned from doing it. Not only could he not do the measurements but it would cost us about a day to recover his laptop, put a new drive it in, re-image it and then put his specialist software back on.
[1] Out of interest, I opened one of the failed hard drives. A nice assorment of magnetically-coated glass crumbles fell out..
Boiler stopped working at our NFP - called the installer who was on the other side of town.
Sounds like impeller stiction. Give the pump three slaps, each 40% of a good punch, give me a call back if it doesn't work.
Now my lovely colleagues assume that I'm also a heating engineer..
DEC once marketed an Unibus serial interface that consisted of the actual Unibus board (might have been two, can't really recall), a 19" 4HE box with three line cards fitted with the D25 connectors, and an interconnect cable. Problem was, the backplane in the distribution box wasn't stiff enough, and inserting the middle card would bend the backplane causing the middle of the edge connector to not make contact. Which caused three or four of the ports on that card to stay dead.
The field change fix was to fit a piece of wood between the backplane and the rear panel of the box. This was informally known as the "2x4 fix", although the actual size of the piece was a fair bit smaller.
many moons ago i used to work PC support, and we had a particular range of PC's in the estate - the Mitsubishi VS550 - there was apricot involvement also - all i know was that someone thought these were a suitable replacement for the aging IBM model 50's that they had.
And they were a massive improvement.
Right until they stopped working. After extensive analysis (not by me i hasten to add - I wasn't that clever to figure out what was happening) it was found that the heat dispersal for the CPU was inefficient and cause the CPU over time to ease itself out of it's socket.
Our solution... switch of PC, lift PC 2-3 inches off desk. Drop. The inertia tended to force the chips back in to the sockets enough to let it carry on for a while.
Thumbs up because.... eyyyyyy
Used to work for ACT/Apricot. The original Xi had a 10Mb drive, made by Rodime in Scotland.
We had endless reports of drives not spinning in the mornings, especially when the weather was cold.
The drives seemed to have been built with a fair amount of moisture inside the casing (it can be damp in Glenrothes) and condensation formed on the platters. The surface tension where the heads were resting was strong enough to stop the drive spinning.
Rodime's "official" fix was to slap the right hand side of the box. The heads bounced into the air, and the platters started moving before they came down again.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
