Harwell Dekatron / WITCH Restoration - 2009
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 you will find the progress made on the restoration during 2009.
Updates for 2010.
For more detailed information on this system please visit the CCS Harwell Dekatron / WITCH page.
23/12/2009 update from Tony Frazer
Prototyped and tested a relief +375VDC power supply for Rack 5 (Dekatron Stores 50-99).
It needs a better filament supply to properly heat the two EL360 pentodes to provide the required output current under load. I have an alternative transformer which may solve this problem.
Apart from that, it works just fine and I can start preparing a suitable chassis so it can be properly assembled and mounted on the rear of Rack 4 (Rack 5 is not pre-drilled for panel mounting so we can't use that one).
15/12/2009 update from Tony Frazer
Eddie completed adjustment/replacement work on the Relay Sets on racks 1 and 2, and is now working on the shift unit from the arithmetic rack, half of which has now been cleaned and adjusted.
Due to the frequency with which we have been blowing fuses, Johan has fitted a fuse holder board and bracket discretely inside the 50v power supply (using the existing protruding studding) so that we can use standard cartridge fuses for the mains and 50v output. This will minimise future disturbance of the original fuse holders and ease routine maintenance.
I have completed functional testing of the Rectifier Unit under documented current loads without incident and refitted it to the power supply rack ready for soak testing.
My next step is a detailed examination of the last unit on the power supply rack - the stabiliser unit.
I have gathered parts together to build a separate +375VDC 200mA stabilised power supply for the Dekatrons in rack 5 (due to the overloading previously noted when repairing the rectifier unit) and I have started assembly of a temporary prototype. I will report more on this when I get it working - suffice to say it is of contemporary design!
08/12/2009 update from Tony Frazer
Testing of the recently rewired Rectifier Unit continues. The picture on the left shows dummy loads comprising lots of 7K5 10W resistors soldered together to draw current equal to that noted I made a 3mm aluminium bracket on which the 4x47R current limiting resistors have been mounted.
This was fitted in place using existing mounting hardware and holes. This can be seen immediately below the load in the picture.on the schematics at the expected voltage.
The first HT secondary circuit I chose to test is the one that had previously 'cooked' and required a replacement choke and current limiting resistors. The initial test revealed a fault in the rectifier (over 800vAC in, only 50v out!) so a tagboard was annexed to my aluminium bracket on which 4 rectifier arms were mounted. Each arm comprises 2x1N4007 diodes with voltage equalizing resistors (possibly unnecessary with modern silicon rectifier diodes), which should be good for 1A at over 1kV PIV. This fixed the problem of low output voltage which is now around 600 VDC and the circuit was successfully placed under load (200mA).
The initial test of a second HT secondary circuit indicated a blown resistor (one of a parallel pair), so a replacement will be sought before testing of this circuit continues.
Eddie cleaned and adjusted the relays in the Power Distribution Unit, which has now been returned to the rack ready for testing. Also, cleaned and adjusted half of the relays in the Shift Relay Unit on the arithmetic rack.
Johan continued testing the 50v supply and did some insulation tests on the mains cable and transformers - no problems found.
01/12/2009 update from Tony Frazer
Reassembly and rewiring of the HT Rectifier Unit is almost complete.
During the course of reassembly, a number of tests (without loading) were performed - first I wired and tested the primary circuits, which include the -10v supply for the relay operated by the thermal 55-second delay device, then the remaining LT secondary circuits. OK so far. Next step will be to connect and test the 3 HT secondary circuits.
Eddie has replaced some heavily worn relay contact sets - there is one particular circuit path through several of the digit selector relay sets that has experienced heavy wear. The cause is a mystery at the moment as based on the schematic diagrams it doesn't appear to be heavily loaded or switching a high voltage.
After replacing the blown supply fuse, Johan continued testing the 50v relay supply. It is incredibly inefficient! It draws 1 amp from the mains with only the internal rheostat as a load. By connecting a chain of wire wound-resistors to the 50v output, it was supplying 4 amps to the external load - then the supply fuse blew again! This is likely because we did not take into account the inefficiency of the unit in choosing the fuse wire.
25/11/2009 update from Tony Frazer
This week I concentrated my efforts on the HT Rectifier Unit...
The bubbling vat has been set up again to continue electrolytic removal of rust from the front grille and chrome surround. My last attempt was encouraging but I don't think I left it running long enough to achieve the desired results. After one day of operation, half of the nails dropped off the support wires which were below the water line! The support wires had rusted away! I must remember to ensure the top of the nail is out of the water when I set it running again.
As mentioned in an earlier report, a choke was found to have gone open circuit and had oozed pitch from the casing seams. It was also evident that the choke had 'cooked' over a period of time, resulting in the interior of the chassis being coated with a sticky brown residue of condensed pitch. I re-potted a replacement choke of the same type (from our stock of old spares) in the original casing using pitch salvaged from the original. This was done over a camping stove with a tin can in the alcove of Hut 24. The pitch caught fire a couple of times but was easily extinguished.
The only way to clean the HT Rectifier unit was to dismantle the entire unit, clean the chassis plates, frames and individual components, then reassemble and re-wire the unit. This is well under way, I just need to re-fit the repaired choke on the chassis, replace four blown wire-wound resistors (discretely, leaving the originals in place), replace a damaged Plessey connector on the rear chassis plate and do the wiring (for which I have obtained 1kV 1.8A rated plate wiring wire in the correct colours.)
Eddie has made a start on replacing damaged and worn contacts on the relays and there is plenty more of this work to do as we will then move on to the relay/valve units of the arithmetic unit and stores.
17/11/2009 update from Tony Frazer
This week I adjusted the M22 tape reader to read 5-hole paper tapes and I wrote a simple visual basic program to interface with it and capture the data.
Some problems reading tapes were encountered, causing the reader to halt. I first tried adjusting the intensity of the infra-red light, but eventually found that the problem was due to a rapid build up of dust on the reader which was causing the read errors.
Cleaning the reader with isopropanol cleared the problem and enabled the vast majority of tapes to be read.
Around 170 tapes have now been read, covering problems such as Sin x, Fibonacci series, square and cube roots, arithmetic progressions, clear stores, test stores and arithmetic.
Thanks to John P, we now have a supply of low current nickel/silver fuse wire (1A, 1.75A and 3A) which will enable us to replace many blown fuses in the machine.
11/11/2009 update from Tony Frazer
Johan continued work on the 50v power supply and successfully powered it up with a variac and dummy load. All looked well, but shortly after the excitement and photographs, one of the mains supply fuses blew, so we will be re-examining the unit at the weekend.
Johan found some replacement fuseholders in D-Block which are identical to those on the Power Distribution Unit. I replaced one which had cracked. I also inspected all chassis power sockets and replaced all perished rubber sleeving on thru-chassis wires and fuseholder terminals with shrink sleeving. Similarly, perished grommets were replaced. The wiring appears in generally good order, although some joints need to be reworked as some wires have been weakened at the joints due to mechanical disturbance possibly during previous maintenance activities.
I replaced as many of the spring-loaded fuses on the front panel of the Power Distribution Unit as we had suitable intact spares for. However, I now need to investigate sources of low fusing current wire (1, 1.5 and 2 amps) to repair the box full of dead ones. Strands of braid and hifi loudspeaker cable have been suggested as commonly available sizes of Cu wire are too thick. I will assemble a test rig and test a few samples to destruction!
Eddie continues to clean and adjust the relay sets. This is almost complete - Eddie will then revisit each relay set found to have damaged contacts and replace them where necessary.
I had a pleasant chat with Prof. Sir Maurice Wilkes this morning - he was impressed with the design of the computer and the economy of materials used in its construction.
02/11/2009 update by Tony Frazer
The order of electrolytic capacitors etc. arrived last week, so there is a capacitor theme to this week's report...
Johan fitted a replacement capacitor alongside the original capacitors in the 50v power supply.
I cleaned the Distribution Unit and replaced a capacitor, also replaced 6 capacitors in the Rectifier Unit and the capacitor in one of the 110v power supplies.
Eddie continues to clean relay sets and straighten contacts. Some of this work is to correct poor maintenance at some point in the past.
27/10/2009 update by Tony Frazer
I re-wired the two 110v power supplies due to perished rubber insulation and repaired a fuseholder. Insulation checked and powered up with an anglepoise lamp as a load.
Johan found an identical Mains Switch to replace the faulty one on the 50v power supply in D-block, which is now fitted. The assembly of four electrolytic capacitors dated 1947 was found to be short circuit and we decided to replace them with modern equivalent components. We discounted the idea of drilling out the electrolyte and using the cans to pot up replacement components (which is common practice in antique radio restoration) on the grounds that we would be disguising the repair, which is unsafe and pointless and that the original capacitors would be physically damaged. We decided that the original capacitors will remain in position, intact, but disconnected. A single new component will replace all four of them, adjacent to the assembly.> John Pether has removed the dead choke from its can by melting out the pitch with a hot air gun and blowtorch. John also re-attached the paxolin connector plate with rivets, cleaned and under-coated the can ready for painting in due course.
Eddie continues to methodically clean and adjust the relay sets...
19/10/2009 update by Tony Frazer
The "dodgy contact due to broken plastic" I mentioned last week was found in the Plessey output socket on the Rectifier Unit - it is fairly minor and wouldn't account for the blown components.
I completed re-routing the wiring under the Input Table and have started inspection and cleaning of the Tape Readers. The only markings on the interior frame of the first unit I inspected are "A.W.E.3.". Eddie suggested the possibility that these may be Post Office units and may appear in the PO Engineering Instructions.
I drilled out the rivets and removed the bottom plate from the defective choke from the Rectifier Unit - it is full of pitch and the Black & Decker hot air gun isn't powerful enough to remove it - any ideas?
I removed the Voltage Test Unit and Power Distribution Unit from the PSU Rack. Initial inspection of the PDU revealed perished/missing grommets and some defective panel switches (which may be repairable.)
Eddie continues to inspect and adjust relay sets, finding some heavily worn* contacts which require replacement and evidence of poor maintenance at some point in the past where relay spring contacts were 'bent' rather than making the proper adjustments (we have spare spring sets which can be fitted later.)
*same contact position in a series of relay sets D1,2,3,4 & 5 which are interconnected - suggests a heavy load or back EMF causing arcing. We will revisit this later.
Johan and I spent some time at the weekend trying to organise the layout of the Tape Edit Desk - it looks cluttered so it would be helpful if we found another desk similar to the two we already have - does anyone know of a spare one on the park?
Next steps - Test rig for tape readers (50v pulses to solonoids, 5 LEDs), continue cleaning tape readers.
13/10/2009 update by Tony Frazer
Eddie continues to clean and adjust relay sets and has traced the +50v wiring on the relay racks.
I am routing the wiring loom under the input table.
I have traced the schematic for the Rectifier Unit - it appears that the failed components (choke and 3 resistors) are all on the +370v supply circuit. Looking at the latest schematics with the pencil annotations we found at Birmingham, there is reference to a known "dodgy contact due to broken plastic" in one of the connectors - I will investigate this next!
Thanks to a visit by a Dekatron enthusiast on Saturday, I now have enough information to design a simple Dekatron tester/reconditioner with variable guide voltage and pulse width/frequency which should enable us to find the threshold/pulse width at which a tube "sticks" on a cathode and allow manual bidirectional stepping to get the discharge onto the guide electrode adjacent to the sticky cathode on which sputtering may have occurred. It will then be possible to increase the anode current and leave it "cooking" for a while to reverse the effect. Repeat testing should show an improved threshold before sticking occurs.
Apparently, severely damaged tubes can sometimes be spotted visually due to cathodes burning off and appearing shorter in length!
05/10/2009 update by Tony Frazer
I examined the Rectifier Unit this week... One of the HT chokes is open circuit and pitch had oozed out of cracks in the casing and through the mounting holes at the base. There also appears to be a thin layer of oily residue on the underside of the chassis, wiring and components. This suggests a prolonged overload condition on the HT causing the pitch to liquify and the choke to fail. A thorough clean and some rewiring is required.
All thru-chassis grommets have deteriorated and will require replacement. 3 of 4 wirewound resistors in the front of the unit have burned out. The cause of this failure is not yet known but is probably external to the rectifier unit as there are no apparent shorts or other failures on the unit which may account for its present condition.
The good news is that all other components in circuit measure OK under no-load conditions (multimeter and tracker/scope). Also, Johan and I found an almost identical replacement choke in D-block - but thanks to Tony S and Phil H for helping in the search for a replacement. Also, the big Copper Oxide rectifier stacks on the front of the unit have been replaced (possibly at Wolverhampton) by a solid state rectifier - fortunately they left the original components in situ to preserve its original appearance. Curiously, all secondary fuses are intact!
The next step is to drill out the rivets holding the failed choke casing together. Then it should be possible to remove the innards using the hot air gun. The intention is to re-pot the replacement choke in the original casing. Then we can build suitable dummy loads for each supply in turn and bring them up slowly using a variac whilst monitoring the current.
I started mounting the 6 input paper tape readers and their control box on one of the tables I varnished last week. I need to feed the wiring loom underneath and attach the socket box to the underside of the table.
There are now 2 Dekatron spinners in operation for visitors to see. The dim one on the left is a GC10A (the original type the machine was built with) and the brighter one on the right is a GC10B (which was used to replace failed GC10A in various positions in the computer.)
Johan and I found a small wooden table in D-block which I have sanded down and started varnishing. This is destined to be for the large printer as there will not be enough room on the other table for the offline tape preparation equipment AND the printer.
Eddie continues to methodically remove, clean and adjust the relay sets.
Johan started examining and cleaning the 50V Relay Supply - the mains and output cables will require replacement due to perishing and cracking. There are some electrolytic capacitors dated 1947 which will need to be electrically re-formed, or possibly replaced.
On Monday, we were visited by Gurney Thomas and his daughter Caroline, and Dick and Doreen Barnes. Gurney and Dick are two of the original designers of the computer. Gurney (whom neither Kevin M nor I had previously met) designed the Dekatron store units having previously worked on Dekatron counting equipment at Harwell.
28/09/2009 update from Tony Frazer
Eddie started cleaning and adjusting relay sets. A broken socket on a print layout patch panel was found - so we will need to find a replacement. A quick examination with the magnifying glass (I must get my eyes tested) revealed "Radiospares" - so they must have been reasonably common. A bent "V" connector on the back of a relay set was repaired.
I used Graham's electrolytic technique to remove rust from the chromed grille of the Rectifier Unit. The chrome has come up really well, but I think the grille (which appears to be steel) needs a little more treatment. I have spent much of the week sanding down and varnishing two wooden tables, which I thought we could mount the tape readers, perforator, printer and Teletype on. I started cleaning the tape reader mounting plates ready for mounting on a table.
John Pether and Peter Merryman have provided numerous boxes of valves for sorting. It won't mean much listing types, but the highlights so far are a box of GC10A's (the original Dekatrons the HDC was fitted with). Many other useful types found. This evening I will start testing a box full of EF91's.
I finished one of the Dekatron Spinners I have been building, so that can be put on display - I will do some "words" explaining in simple terms what it is doing and how that is relevant to the HDC. Next step - continue cleaning tape readers and their mounting plates; drawing schematic of Rectifier Unit and preliminary examination/testing of components (no power up yet!)
20/09/2009 update from Tony Frazer
Our new volunteer Eddie Washington has been examining the relay units to assess what work is required to get them back into working order. First impressions are that they all need cleaning and adjustment which will commence shortly. This is Eddies area of expertise and he has all the proper tools for this time-consuming and fiddly task. In the meantime, Eddie has made a start cleaning the exterior of the relay cans which are suffering from a build up of grime and from flaking paint. It will be necessary to re-paint the can covers - fortunately we have spares, so we can keep one in its present condition to illustrate their state before/after restoration.
Double agent Peter Onion (also on the Elliott 803 working group) has read some 5-hole paper tapes which were found with the machine at Birmingham - of particular interest was a box labelled "...key problem" which we believed may contain the tapes for the work that was done at Wolverhampton for Chubb locks. At present it looks like the box contains a Pythagoras program :( but no positive identification of the Chubb tapes. However, there are lots of tapes still to read - we have only scratched the surface and there is a lot more work to do on this before we know what we actually have! Peter's tape reading gizmo is optical and some problems were encountered reading tapes which had become translucent (this would not be a problem for the HDC's tape readers which use mechanical feelers rather than optics - but this would damage the old fragile tapes.)
I (Tony F) have prepared an inventory of the valves/tubes actually in situ on the machine. This differs from the list I obtained from the schematics for a number of reasons: The 12E1 tubes (beam tetrodes) specified for the power supply are actually EL360's (pentodes) - this appears to have been done in all cases where a 12E1 was expected. The ECC91's specified for general logic use appear to have been replaced on a haphazard basis with E90CC and E92CC - both of these types were designed for computer use and also have a greater life expectancy. I have contacted a valve enthusiast who is active in the valve collecting community and he has offered to coordinate collections of spare valves in the UK. He has contacted his counterpart in the USA who is keen to do the same over there. We will also pass the revised valve list to John Pether and Peter Merryman so we get any types of use to us (fortunately little or no conflict with types sought for Colossus!). A few useful valves have already been donated by visitors. It may be some time before we have the Dekatron stores on the HDC working, so in the meantime I am building some Dekatron Spinners to show visitors what various types of Dekatrons look like in operation.
I started cleaning up the Rectifier Unit from the PSU rack - apart from some pitch running down from the chokes it looks in reasonable cosmetic order. However, it is evident that our schematic does not match the unit - so the first step will be to draw up a new one. Looking forward to trying Graham Wallace's electrolysis technique to restore the chrome front panel and grille.
Johan and I did some rearrangement in the HDC restoration area. Firstly, we moved the racks to the left to make room for the input/output and tape preparation desks. We also removed the tape readers from the Dexion stand. There is now a much better view "around the back" where much of the restoration work will be carried out. Johan has started looking at the 50v power supply for the relays.
Kevin M has ordered the barrier and arranged for two excellent pull-up display stands which enhance the display and attract visitors to the vicinity of the money jar.