ICL System 25 restoration during 2011
Post this page to popular social media
Each TNMOC project has either a working group or project team assigned to do the work. Working groups are either managed in association with the CCS (Computer Conservation Society) or solely within the Museum.
Below are updates on the ICL System 25 restoration project by Johan Iversen and Delwyn Holroyd during 2011.
Previous years updates: 2010.
10/12/2011 update from Johan Iversen
Just in case this caused the tape scrunching I cleaned the head and pinch wheel of the CMT drive. Then with the knowledge from last week about the CMT IO channel, I successfully wrote a bootstrap tape that loads and runs the 'Hello World' program. I am now working on a utility that can be loaded into memory after a program has been written, that will create a bootstrap and program load tape.
Also some semi-good news the scrunched tape from last week partially reads, so it can be used to show the initial bootstrap sequence.
The power supply for the other System 25 has not been refitted yet as I first need to get a new set of fans. But a temporary test to check the supply will be done soon.
03/12/2011 update from Johan Iversen
Had a little success. Using a bootstrap command I found in the ‘Programmers Reference Booklet’ I was able to boot of a System CMT. However the bad news is after the second attempt the machine scrunched the tape. I sorted the tape out but alas it would not read any more. However not all is lost, first the tape is part of a set of tapes that we do not have and secondly it made me realise where I was going wrong. The boot instruction accesses the slow IO channel while all alone I thought the CMT was on the FAC channel. Which in the old days the mag. tape was. The clue is also in the assigned device numbers, 8 and 9 the top end of the slow device numbers. Also with this knowledge looking in the maintenance manual it seems to confirm that the R controller driving the CMT is on the slow IO channel. So we will see if I’m right next time.
Phil has repaired the power supply for the other System but it has not yet been installed.
26/11/2011 update from Johan Iversen
The system from Europa Sports has been connected to a VDU and powered up without the system drive (which was initially removed during the move). Initially every thing seemed fine, but once my back was turned the system went up in smoke. Initial sniffing investigation made us think it was the VDU, there was no reason to suspect the System 25 as it had worked for so many years before and up to the move. So the VDU was stripped down, however nothing was found to be at fault. So further sniffing investigation made us conclude that it has to be the System 25 power supply. So it was removed and handed over to our resident power supply expert and fixer, Phil. During the removal parts of a cooling fan were found in the bottom of the cabinet, see photo. Further investigation showed that all the cooling fans for the card cage were faulty and require replacing.
However another and very unfortunate side affect of the of the System 25 going up in smoke, is that the particles of smoke most probably caused the head crash on the 2966 EDS80 drive, oops!. Sorry Delwyn. :(
Attempts are still ongoing to make the CMT work, but all have failed so far. I'm sure it’s a software issue and not hardware, because the controller appears to recognise the CMT drive when a tape is loaded and rewinds. I am currently still trying to understand how the control store is configured and was about to compare the two machines before one of them went up in smoke.
27/10/2011 update from Johan Iversen
Today I went to Leicester to collect the two System 25's from Europa Sports. I arrived early so that I could have a quick play to familiarise myself with the software again. Jo, at Europa Sports had powered the system up the day before and it had worked fine. However, when I tried to boot the system nothing happened. I checked the hard drive and it did not seem to be spinning, I could not hear any thing. I tried power cycling the machine but still nothing. Jo then suggested that I call John the engineer who used to support it. He suggest that I take out the hard disk and give it a gentle shake. Horror this was a 25 year old disk. Still I tried this and when I turned the system on and entered the boot sequence code the system came up. So after this little scare and before dismantling the System I had quick lesson from Jo on how to use the software.
22/10/2011 update from Johan Iversen
Now that the basic system is working I am trying to get the CMT (cartridge tape) drive to work so that more complicated programs can be written and saved. However all attempts to access the tape drive so far have failed. This could be due to the fact that the control store is not configured ? However according to the limited documentation on the system, a cold boot should initialise the control store, I quote - "At power on, a default configuration is supplied by the processor. For all processor types, partition 0 is configured to bring on-line device 0 on IOC 0 (which is a VDU) and device 2 (which is a printer). The cartridge tape drive is brought on-line as device 8 on partition 0, or as devices 8 and 9 if two cartridge tape drives are used. Common store is set up, but the rest of the store is unconfigured". Still it could be the interface board or the drive at fault. Still to be investigated
24/09/2011 update from Johan Iversen
After some email discussion with Delwyn during the week about the interface, which is supposed to be a balanced-pair serial line. It was decided that the next thing to try was to swap the two wires on this interface. After this was done and the System 25 was powered up the first sign of something new happening was the flashing SEL in status line of the terminal. The terminal was put on-line and the instruction 010P010080 was entered, resulting in 80 '0's being displayed, the contents of memory starting from address 1000. WOW! The System 25 had executed its first instruction. The next hour was spent familiarising myself with the terminal and writing some simple programs, in machine code, of course. As there are no means of storing the program, the programs were kept very simple. The final program, seen in the photos below, sits in a loop which displays ‘Hello World’ twice on the terminal, then after a short delay clears the screen.
17/09/2011 update from Johan Iversen
For quite a few Saturday's now the batteries have been left charging for the day. The voltage of each cell is checked before charging starts, and each week the voltage has dropped (to be expected) but overall the cell voltage is higher than the previous week. Only one cell seems to drop a lot more than the others but it is getting better, to.
Delwyn did his magic again found the circuits for the boards and after a quick scan he commented that it could be because there is no standby battery voltage. So this week we decided to install the batteries. We checked that the minimum board configuration looked sensible, however, we think, it looked like the connections to the terminal had been wired to the wrong port, so this was corrected. With the front panel key in the ‘off’ position, the system 25 was connected to the mains and the main PSU switched on, the PSU and fans started OK. Next the key was turned to the ‘ON’ position, LED indicators on the system boards and front panel lit up, then one-by-one they went out. The ‘Fail’ lamp on the front panel went out and the ‘DC On’ lamp lit, the 25 was up, eureka! But there was nothing on the terminal except for its self-check - to be expected if I remember right from the old days. Every time a key was pressed the terminal beeped in error. After routing around the only documentation we have, we found what we think is the key sequence to put the 25 into a Load state. This state causes the CPU to read 10 chars from the terminal (10 chars = 1 instruction) and to then execute the entered instruction. In the old days the instruction entered would cause the boot-strap to be loaded from disc. We entered 010P010080 which should display 80 chars. from address 1000 to the terminal - it did not. subsequent key presses caused the terminal to beep in error. This meant the terminal was waiting for the previous I/O to complete. We then decided to scope the I/O, the picture on the left shows the trace we got.
The trace (right) shows the 25 polling sequence, the first burst is a DLE (0x10), followed by the poll char (0x03), which is a load request on device 0. Each burst is 8 bits, 7 bit data and even parity. So brilliant progress has been made. Now Delwyn is going to modify his interface board so that we can snoop the I/O interface and see what's going on.
14/05/2011 update from Johan Iversen
The power supply has now been soaked tested at 16A for two days, over two Saturdays. The power supply held up well, initially there were some very dusty smells but they soon cleared. The power supply was then refitted into the System 25 cabinet and one of the model 91 terminals has been wired up to the system for testing. I have simplified the configuration of the System 25 down to what I think is a bare minimum. So with the front panel key in the OFF position, the System 25 cabinet was powered up the power supply and fans started ok and on the front panel DCoff status light was ON. The key was now turned to the ON position, LED indicators on the system boards lit up. The +5V was measured and re-adjusted as it was found to be a bit high. However, sadly, that was it. The power supply was running fine, but the LED's that were lit, especially on the processor board where indicating that is was failing it establishment checks. So without any manuals and spare boards this will now leave us with some serious head scratching to do.
07/05/2011 update from Johan Iversen
Had a great visit to the company in Leicester (details to be released later) and saw their working System 25plus (similar to the museum one on display) and also saw their older model System 25 (similar to another system the museum has) that they are using for spares, See photos below.
I saw the machine working and running there purchase order software, I was given notes on how to use it, then played with it for about an hour. Also got the chance to familiarise myself with the VDU, which will help when we connect it to the museums System 25. Another find was a proprietary interface box for converting System 25 serial I/O to Centronics.
I informed them that the museum is very interested in acquiring both systems and that we will arrange collection of the systems in the next 3 months, which they were happy with. I also brought back 3 VDU's out of 9, one known to be working (apart from a sticky Enter key).
All in all a very worthwhile visit... we just need to find the necessary funding to hire a Van, and a few bodies as the systems are up stairs and the lift has been condemned (sounds familiar - see the first project report in collecting and delivery of the museums system 25!).
09/04/2011 update from Johan Iversen
The power supply load has been assembled and is complete enough now for testing the System 25 PSU. The PSU has had two load soak tests:
27/3/2011 for 2hrs +5V at 10A, +12V at 1.2A, -12V at 1.2A
2/4/2011 +5V at 10A for 4hrs then up'd to 15A for another 2.5hrs, +12V at 1.2A, -12V at 1.2A
Initially there were a few of the old dusty smells but they have gone away. The +5V wriggles around a bit, can be up to 0.2V but this is thought to be due to there being no sense wire connected.
Next stage to re-install the PSU back into the System 25 cabinet.
I have also spoken to someone who works at the company in Leicester, who has a complete working System 25, a spare System 25 used as spares and a number of model 91 VDU's, which I have been looking for. I'm hoping to visit them to soon to see exactly what they have.
12/03/2011 update from Johan Iversen
Basic power supply tests and repairs (by Phil H) have been done. It now needs a proper load test.
I'm currently building a power supply load (that can hopefully be used by other projects to), most of the mechanical construction is done, the front panel has just been drilled by Tony at Ceravision, so now the load just needs to be assembled so it can then be used to soak test the System 25 PSU.
Myself and Delwyn are in touch with some people who might be able to help with getting some working Model 91 VDU's to hang of the System 25 - its looking good.