I received an email from Jochen telling m that he owns since 1986 a complete System/3 Model 10,
including two IBM 5444 disk drives, a IBM 5424 MFCU and the IBM 5203 printer.
The CPU is equipped with 16kilo (not mega or giga) bytes of core memory.
The system is completed with an IBM 5496
It was fully operational in 1986 before it was disassembled and moved to a storage location.
In the late 1970's the system belonged to a company where Jochen's father worked as an operator.
They named it "Kasimir
". It comes from a German poem that start with: 'Kasimir der Weise spricht....'.
Someone in the company suggested to name the computer "Kasimir der Weise", they short it to "Kasimir",
put this name to the front panel and from then on the System/3 had its own individual name.
When the company replace this system, it was donated to the Jochen's school.
Jochen operated the system, but when left the schole no one remembered how to operate the
machine and they decided to remove it.
Fortunately they asked Jochen he wanted it. He accepted it the system with alot of material
like (empty) punching cards, disk cartridges, all CE manuals and RPG training documents.
In 2014 the system was moved from storage to a summer time cottage near Trier Germany.
Jochen goal is to make this system operational again and I offered him to help.
A date was set on 27 June 2015
Below you will find the restoration project blog.
Positioning and Assembly
First hurdle we had to taken was how to position the system in the room and not to block any doors.
After moving the CPU, printer and MFCU around a bit we found the right location.
We bolted the 5203 printer to the CPU and installed the rear modesty skirt between the CPU and the MFCU.
Now we had to unravel all the cables coming out of the CPU.
After some sorting out and with some still readable markings (30 y ago) from Jochen we where
able to route every cable to its right location.
Routing signal and power cables from the CPU to the MFCU.Reformating the Power Supplies
Just powering up an old system that has been in storage for 30 years is not such a good idea.
First all power suppliers where rewired from 220V to 240V.
Rewiring the +24V power supply inside the MFCU and inside the CPU.
reformating the capacitors the load was disconnected from the supply
and a variac was connected at the primairy side of the transformer.
Reformating the CPU bulk power supply and the +60V power supply inside the 5203 printer.
Slowly the primairy AC voltage was increased until the nominal DC voltage was reached at the secondair side.
This was done for:
up the CPU
- the +24V power supply for the power up sequence
- the CPU bulk power supply (-4V, +6V, -30V etc)
- the +24V inside the MFCU
- the +60V inside the printer.
The CPU was connected to the 230/400V 3 phase mains.
Nothing happend. The Emergency Power Off knob was pulled.
After restore the EPO knob we had immediately "TH CHK" and "PWR CHK" lights.
Sounds a bit strange, but this is as expected. The TH CHK is turned off after pressing the Check Reset button.
I turned on the system while Jochen was holding the EPO knob just in case.
Some smoke came out of the CPU. Poweroff !
After some searching we found the smoker, a resistor on the -4V regulator card.
Please contact me if you have an spare -4V regulator card.
The IBM part number is: 5808959 -or- 5860519.
Here stops our project.
I probably have this card spare at home, but at that moment not with me.
Back at home I found the schematics of this card
. This is pure luck!
The chance of finding IBM schematics at component level outside IBM labs is almost zero.
A repair at component level of this card is now feasible.
Will be continued. Core memory
The core memory might be a real problem.
This picture is from another module. (32kb and also broken)
These modules where build by IBM Mainz and
they used some foam between the core planes to route the cooling
air throught it.
Unfortunatly this black foam turns into some sort of tar over the years and starts eating away the X/Y wires.
A quick inspection with an Ohm meter confirmed my suspicion and that Kasimir's memory module is unrepairable broken.
The only solution to this problem is replacing it with modern solid state chips. This is a project on its own.
To get this system up & running again several problems need to be resolved:
- Working regulator card and see what happens when the power up sequence continuous.
- Replacing the core memory with a more modern solid state one.
- Some small parts are missing, but not essentail to get the system working again
Jochen 'operating' this system again after 30 years !