Huge IBM 1800 find (and need some help)
cctech at ljw.me.uk
Sat Mar 21 14:58:10 CDT 2015
On 21/03/15 19:39, Jon Elson wrote:
> On 03/21/2015 01:47 PM, Johannes Thelen wrote:
>> Thanks Jon for tips!
>> No, that is not option, I mean 2311 is have to get back to life ;) I
>> think use ultrasonic or carbon ice cleaning for 2311 and disk packs.
>> Both are quire gentle cleaning methods.
> You have to be very careful with anything that requires temperature
> extremes. Can anybody verify the 2311 uses the hydraulic head
> positioner? If the oxide coating of the packs has deteriorated, which
> I think is quite possible, then the disks will not be usable. Anybody
> else have experience with 14" disk packs from 40" years ago?
Yes, hydraulic actuator. Assuming it hasn't run for a long time then
flushing and replacing the hydraulic fluid would be necessary.
> The other problem is this machine uses SLT. That was never a
> super-reliable technology. They start with 1/2" square ceramic
> substrates, and fire thin film resistors onto them, then print and
> fire thick film wiring. Then they use bump-bonded discrete
> transistors and diodes, and solder these by bump bonds to the wiring,
> and add leads. Then, because the transistors and diodes are not
> well-passivated, they smear silicone goop over it and epoxy on the
> die-drawn aluminum covers. These are not totally hermetic, and oxygen
> gets inside and slowly deteriorates the semiconductor devices. They
> also had some problems with the solder joints to them developing cracks.
The plus side is that you can pretty well replace a module with a few
transistors/diodes/resistors, if you have the manual that shows you what
is inside each one. Also, the same SLT card will be used in many places
so it's easy to swap them around. For the Model 30 there was a manual
that showed where each card was used, which facilitated this swapping.
Of course there were odd cards that only appeared once or twice.
> So, the likelihood of the CPU actually working after FORTY years of
> storage is pretty small. There are, of course, no replacements for
> SLT modules, as IBM stopped making them in the late 1960's! It was a
> form of diode-transistor logic, and not terribly dense. A single
> flip-flop would not fit in one SLT module, it took at least 2, and the
> usual scheme was to integrate some steering gates with it and ended up
> with about 4 modules to implement the typical register bit. They
> didn't use what we think of today as a FF, they used D latches, as it
> was simpler to implement, but required several clock phases. One
> clock phase to latch input data to the ALU, wait for the ALU to
> settle, then a pulse to latch the result out of the ALU.
The power supplies would be the first thing to work on - they'll be
switchers and have the potential to make a very large bang.
>> I like to dump core and all disc packs for future uses ( = make
>> original configuration).
> I wouldn't worry about what is left in core, it would be interesting,
> but you might have to do a lot of testing and repair before enough of
> the machine is working to read it out.
> Be very careful with the disk drives. If you even get a scratch on
> the heads, you will not be able to replace them. Also, if anything at
> all happens to the heads, you will need an alignment pack to realign
> any replaced heads. Do you have ANY experience with old-school
> mainframe equipment? If you think you will just dust this thing off,
> mount the disk packs and fire it up, you are quite crazy!
I got away with it, but the Model 30 was only 10 (or fewer) years out of
service. By defeating interlocks it's probably possible to get a 2311 to
spin up without a pack and have it load the heads (with something soft
between them!) to check head loading and cylinder actuation. I'm not
sure what diagnostic facilities exist for the 1800, but the 2841
controller had microcode routines to manually test drives.
Lawrence Wilkinson lawrence at ljw.me.uk
The IBM 360/30 page http://www.ljw.me.uk/ibm360
More information about the cctalk