Saturday, 29 August 2020

In memory of the Hilger and Watts X-ray diffractometer.

The following photographs were kindly provided by Prof David Watkin (Chemical Crystallography, Department of Chemistry at the University of Oxford) and are reproduced with permission. They show the earlier linear diffractometer (Y190) which, I think, was manually operated and the later 4-circle instrument (Y290) which was controlled by a PDP-8 minicomputer. 

The following pictures were kindly provided by Prof Jurgen Kopf (Institute of Inorganic and Applied Chemistry, University of Hamburg) and are reproduced with permission. They show the original machine in the early 70's and then again in the early 80's some time after a newer PDP computer was installed. 

The following pictures show an interesting upgrade to the machine in about 1990 when it was fitted with modern motors and a cryostream. 

The new system was originally controlled by an Atari and later a linux PC with a custom Qt GUI. 

There are more pictures at this link and there is a paper describing the modifications made to the machine. However, it eventually went to the great resting place for diffractometers in about 2007. 

The following pictures are photographs by Hilger and Watts which I found on the website of Dr Jiri Hybler (Institute of Physics, Czech Academy of Sciences, Prague) and they are reproduced with his permission. I guess permission ought to be sought from the company but they were shut down a long time ago, although the internet suggests it was dissolved as recently as 2017. I'm not sure how to contact them electronically so I'm assuming that educational use is a good excuse. They are quite high-quality scans, so are well-worth preserving.  

Next, one which I found in the book 'X-ray Crystallography: An Introduction to the Theory and Practice of Single-Crystal Structure Analysis' by G. H. W. Milburn (1973, Butterworths). The book is long out-of-print so I guess copyright is now with the author, although again I don't know how to contact him easily. Also, its a Hilger and Watts promotional photograph which might need their permission but I guess its OK to put here for educational use (hmmm). You can also see this picture in the official brochure further down this page. 

Another one from a book (Methods in Enzymology) shows a machine modified for protein crystallography with 5 detectors, which was at the Laboratory of Molecular Biophysics, University of Oxford. This instrument was developed by Prof D. C. Phillips and could measure up to five X-ray reflections quasi-simultaneously due to the addition of a tiltable linear array of 5 counters to the detector arm. For info, the original machine was one of the early Y230 Ferranti-driven models. 

This is Fig. 6 in the article "On the design of diffractometers to measure a number of reflections simultaneously" by P. J. Artymiuk and D. C. Phillips (1985) in Methods Enzymol. 114, 397-415. Copyright for this image belongs to the publisher, Elsevier, and permission to use it only on this website was obtained on 18/06/20, license No. 4852240484639. 

Finally, I found a photograph of a machine that used to be in the laboratory of Prof J. Kraut at UCSD. The photograph was on the website of Dr Ray Salemme ( whom I tried to contact for permission. However, I then found out that he had sadly passed away in July 2019 so I guess copyright is with his estate. Anyhow, I hope its OK to re-use it here! 

Rick Marshall (former Hilger and Watts sales engineer) has very kindly provided the following two extremely rare documents: 

and some additional recollections!

Here we have a Y290 installation manual

Some gems from a former Hilger and Watts engineer, Derek Coggrave.  

"Regarding Hilger and Watts, as far as I remember I started work there around 1962/63. I spent around 6 months working in Highbury assembling various optical instruments - it was a sort of training to become acquainted with the company and their methods. The reason for employing me was that the production of the linear diffractometer was due to start and I would be assigned to test and install it. 
I haven’t got any official paperwork such as brochures. One of the reasons is that the writing department was always very much behind in producing documentation. On my first visit to Japan to install a four circle there was no instruction manual for the instrument. The customer complained and the agent became very agitated. I was there for about seven weeks and so wrote a manual by hand for the user in the evenings and weekends. The agent had the thing typed and the drawings copied and this was given to the customer before I left. Later this was then used to produce the official version. 
The linear diffractometer didn’t last long because it was superseded by the four-circle. As far as I can remember the first one was produced around 1965. The electronic research department had PDP-8 No 4 so Hilgers was one of the first users. It was a great disappointment to me when on later computers the switches on the front, which one could use for loading a program, disappeared. It is amazing that in just over fifty years we have advanced from the PDP-8 with 4K of memory and costing around £8K (my house only cost £5.7K at the time). There was a book written by U.W Arndt and B.T.M. Willis with the title ‘Single Crystal Diffractometry’ which has photos and other excellent diagrams showing the construction of the diffractometer. 
I remember the long-arm five-circle. That was an idea of the researchers in Oxford and they asked Hilger and Watts to manufacture one for them. I installed it and later when I was working at Birkbeck they asked me to go down and service it - by that time H&W were in Thanet and many of the staff had left. I had to get permission from Crystallography and assumed that they would refuse. However, David Phillips had a lot of influence and I guess that if the department had refused he might have taken offence. So, I was allowed to go with an agreed level of remuneration plus expenses. I can't remember much about it now, but I know the arm was long enough to need supporting on wheels that ran round a track on the base unit. This was to balance the strain on the bearings. The detector housing was moved up and down the arm depending on the unit cell. John Marsh was the technician there I worked with. I used to take John out to the pub for lunch. However, he was always complaining about David because he had ideas for building equipment and David Phillips was very critical on costs and value for money and very often John was refused.
There was an interesting letter in the FT on the 21st July 2019 about the Royal Institution (RI) where David Phillips was mentioned. That struck a note with me because I went there several times to install and service a four-circle for Prof. Phillips. After a couple of years the instrument was moved to Oxford when he moved there. He was working with Dr Arndt in the RI who had had a hand in designing both the linear diffractometer and the 4-circle. It was around 1965/6 I went to the RI with Len Wood for the installation. One difficulty was the hydraulic lifts and getting heavy items to the upper floors. The transformer, which weighed half a ton, made the lift sink as soon as the first two wheels reached the lift floor. A toss of a coin determined who stayed in the lift and with help, the transformer was run into the lift as fast as we could push it. As my colleague and the transformer sank from view, we climbed the stairs and waited. Thankfully, after a few minutes he, the lift and transformer reappeared. 
Later, another colleague servicing the instruments worked until around 10 pm. Turning the lab lights off he found the rest of building in darkness. He groped his way to the front door - locked and bolted. Daunted by a night alone in the RI he climbed upwards looking for help. Seeing light around the edge of a door, he knocked loudly expecting a porter. However, the flat was occupied by Sir Lawrence Bragg who appeared in his dressing gown, produced a set of keys and allowed my colleague’s exit. 
Prior to that, probably around 1962/63 I had quite often visited the department in Oxford run by Dorothy Hodgkin. I installed her linear diffractometer with Len Woods. We were always encouraged to entertain the customer so Len and myself took Dorothy to the Royal Oxford Hotel for lunch a couple of times. We were quite right-wing in our politics and she was very left, although we didn’t know it at the time. However, she was always extremely polite and very gracious. 
In comparison, on one occasion when we were in the Royal Institution the telephone in the lab rang so I answered it. Before I could ask who was calling Dr Arndt rushed in yelling at me for picking the receiver up. Prof. Phillips on the other hand was always very polite but was inclined to ask rather quizzical and penetrating questions, testing whether one was up to the job. But, he was always quite content provided things were done to his satisfaction. 
On one occasion I went to Switzerland to service a linear diffractometer. There were three fellows working in the lab solving structures - it was a pharmaceuticals company although I cannot remember which one now - that was in around 1963. The fellows who worked there kept beer in a tank in a dark room that they used for developing films from an X-ray camera. The water was chilly. Instead of coffee in the mornings they produced the beer. On one occasion one of the managers suddenly appeared and as he entered the lab the other three grabbed their glasses and held them under the table; I followed suit.
Anyway, it is a long time since Hilger and Watts vanished. They were taken over by the Rank Organisation in 1967, and I believe later sold on. At the time Rank wanted to expand into other areas after making a lot of money with Xerox copiers. However, they delved into industries they didn’t understand, lost pertinence and then as far as I know sold everything off and the carcass disintegrated."
Some comments from Prof Lindsay Sawyer (Edinburgh). 
"The pale blue Y290 4-circle was their Mark 2 version. Physics in Edinburgh had a Mark 1 (same colour as the linear diffractometer) which relied on a huge box of Ferranti electronics and too-coarse gratings for positioning the circles. Positioning to 1/10 of a fringe was unreliable at the time and most of the folk in the lab became expert at keeping the thing going to finish data-collection. There was also a peg-legged night watchman who did the night shift. I think only 1.5 crystal structures (by Richard Nelmes) were ever determined using it and it sat for most of my PhD growing nettles. 
I did use the linear diffractometer which was a mechanical analogue of reciprocal space. You set beta* by rotating the two screw drives appropriately. You then dialed-up on the control box the a* and c* increments and then rotated the crystal at a known peak position until you maximised the counts, locked it to the slides and off you went. (Setting a crystal on the 4-circle used a similar if computerised procedure). The Edinburgh (ex-RI) linear diffractometer had 3 counters so if you had an orthogonal space group (like me) you could collect 3 layers at the same time. Eddie Komorowski collected 6A triclinic data on it, 1 layer at a time! The output on the linear diffractometer was 5-hole paper tape, on the 4 circle, 8-hole. I may still have some rolls somewhere. Happy days!"
More from Derek Coggrave.
"Ferranti developed the electronics to drive the 4 circle but this was difficult to use and service, and led to the development of effectively a rival system by Manchester University. I can remember a couple of  people from Manchester turning up when the first electronics was delivered to the research department run by Arthur Long, although I didn't meet them personally. I was working in production. As far as I remember, Ferranti ordered six 4 circles that were delivered to them in Scotland. The colour was grey, and only one was sold I believe, although I cannot remember whom the customer was. I know the Ferranti 4 circle had a different catalogue number and that was probably Y230. 
With the Ferranti electronics, trying to divide a 360 degree sine waveform to get ten evenly space outputs would have been a nightmare, especially with the electronics available at the time - lots of instability. The chaps in Manchester dropped that system and with the finer grating only had to measure when the waveform crossed from + to - which could be done digitally - much greater stability. When I built the electronics for the one in Birkbeck I used op-amps for the reading heads. Once they had been setup they never required adjusting again. 
Probably because Ferranti was having difficulty in selling their kit, Hilger & Watts bought the Ferranti 4 circle unit, which might have made sense if the idea was to buy back the hardware at a discount. However, the Hilger management kept the Ferranti unit running for some time. This seemed a waste of resources - having two units, one in Scotland and one in Camden Town producing two different systems for the same purpose and effectively competing with each other. 
However, as the Ferranti unit never managed to sell any more systems it was eventually closed - probably Hilger had to pay redundancies relieving Ferranti of this obligation and the hardware was brought back to London and repainted blue.  
Subsequently, H&W was taken over by Rank Precision Industries. The problem at Rank was that they had a large cash flow from their copier production. The story I heard was that a couple of Rank executives received information about an engineer who had developed a copying machine. There was just one that the fellow had built in his garage. This is all hearsay but apparently, the Rank executives paid a visit and agreed with the designer to finance the production of the copiers with the market split - the designer took the USA and Rank the rest of the world. You can probably check all this if it is of interest. 
Later, Rank bought up various companies hoping to repeat the success of Xerox. However, selling scientific instruments tends to be a one off process. A university might buy dozens of copiers but only one 4 circle. The other problem was that H&W had growing problems. Many of the managerial positions were occupied by long serving employees who had spent decades engaged in the production of optical instruments and had no knowledge of electronics or the science involved. Secondly, the catalogue of instruments they manufactured was as far as I remember around ninety. Many only sold a few each year: no doubt uneconomic. The main housing of these devices was often a casting and it was always difficult to reach the adjustment mechanisms. I can remember using dental mirrors, torches, dental probes to hook springs and right-angled screwdrivers. For example, the 4 circle would have been much easier to access if a thick steel plate had been used as the basic mount and attached to a frame with removable panels all round. On one occasion a customer asked to purchase the box of jigs that were used to align the 4 circle - they wanted a price. I couldn’t get an answer on the phone so when I got back to the factory I went up to the costing department and inquired. The fellow I spoke to told me that it was not priced because they didn’t expect anyone to buy it, and we’ve been very busy. How do you manage with other instruments, I asked him. Ah, he said, if we haven’t got time we look at other manufacturers catalogues for something similar and get a price from that. So, Rank took over H&W because they thought they could ramp up production not understanding that H&W was struggling to control its rapid growth and had a back catalogue of instruments not suitable for production in large numbers.  
However, Rank did provide a good redundancy scheme and I took advantage of that in 1971. Leaving was quite a wrench, abandoning a job in manufacturing that I had very much enjoyed. 
Regarding the sites such as Camberwell and Camden Town, in both cases much of what went on had started out or expanded into private houses that adjoined the factory. Around that time, local councils were refusing to allow the redevelopment of sites such as Camberwell and Camden Town because they did not want industrial processes mixed up with domestic housing. Also having production split between several sites had disadvantages.  
Thus, Rank chose the move to Thanet and a factory large enough to locate all these functions on one site. The basic idea might have been logical but not the venue. It was in an area where other opportunities for employment in engineering were low. Those who didn’t want to take the risk and there were quite a lot, took redundancy.
The other problem with Thanet was that if one was later made redundant after having moved, this might have meant returning to the London area. However, as house prices were rising rapidly at the time, moving back would have been financially impossible. And, my wife didn’t want to live in an area so remote.  
After that, I lost touch with what happened to the factory, but I guess Rank soon realised that they had bought a company that did not in any way meet their expectations. 
I can remember that before the move to Thanet it was divided into four divisions. Each division had four directors with a central office with four more directors. The names were printed on the official notepaper and I can remember crossing off those who resigned or were sacked. As far as I can remember there were not many left after two years.  
One of the salesmen told me that he was called to the head office and when he got there he was, much to his surprise, given his redundancy papers. He had driven there in a company car. When he got back outside the car had gone. He went back again and asked why. He was told that because he was no longer employed by H&W he couldn’t drive the car. How much do you want for it, he asked. He bought the car on the spot and drove home.  
This perhaps illustrated that Rank did not really understand the business it had acquired and was poor at managing people. The prime purpose at H&W was the design and building of scientific instruments for a limited market, which because of technological developments was subject to sudden change and required a process of constant redesign and upgrades. A lot of investment was required in personnel and equipment, which presumably was not forthcoming, so the company fell apart. Rank lost patience. 
Oxford Instruments, which at the time was a competitor of H&W, has thrived until today. So, I believe the opportunity was there but the management skills and the will to succeed was wanting. As far as I’m aware, Rank Precision Industries no longer exists, which again illustrates that Rank, whose ethos was in the realm of entertainment did not purchase companies that were suitable additions to their portfolio. There was no synergy between entertainment and scientific instrument manufacturing. It was like trying to mix oil and water.
About five years after I joined Birkbeck (1971) a Hilger colleague phoned to say that there was a four-circle left and did we want it at a discount. The department bought it but we had to build our own electronics. The department bought a General Automation computer and it was my job to build the electronics - different because the PDP-8 interface was unusable with a SPC-16 (and a change from 12 bit to 16 bit words). The only operating change was that the motors started and slowed down in 32 steps to stop shaking protein crystals around in their liquid. Peter Lindley did an excellent job of writing the programs for crystal orientation and I wrote the rest. I was taking an MSc at the time and used it as a project in my second year (part time). It was difficult to find the time at work so I finished wiring up the interface boards (wire - wrap) at home over the Easter holiday. There were also other problems such as the SPC-16 didn’t support two word digital arithmetic. Anyway, it was all very time consuming and I wrote a lot of the programs at home. Those were the days. 
In the end the SPC-16 broke down. General Automation wouldn’t let us have the circuit drawings so we couldn’t repair it ourselves and they wanted some enormous sum to do the work. Tracing out the circuits on the PCBs would have been almost impossible because they were multi-layered. This was different from the PDP-8 because we did have the drawings and when something went wrong it was fairly easy to track the fault. Another problem was that the SPC-16 did[n't?] support two word integers and couldn’t output to the teleprinter. "

Frank van Meurs has very kindly sent me some Enraf publicity material from the 1950's confirming that H&W cooperated with them in that era. The Enraf-Nonius CAD4 diffractometer was later to demolish the Y290.

Enraf and Hilger used the same agent in France in 1955. Pictures from a 1958 Enraf leaflet. 

I found advertisement by the US agent for H&W (Engis) in an online journal. 

Note to self: Engis still seems to exist so might drop them a line in case they have any memorabilia! 

Finally, Beirut? Not quite, it is (or was) Hilger&Watts!

~  Now your voice is still  ~

- - New Article - -

A Y290 on the western edge of Europe

By Patrick McArdle (Galway) August 2020

In August 1981 a small ad in Chemistry in Britain placed by Professor George Ferguson advertised a Hilger and Watts Y290 for £5,000. I contacted George in Guelph about buying it and he stressed that a Weissenberg camera was needed to get data collection started on the Y290. I knew of a long forgotten Weissenberg camera that was lying in a cupboard in another University in Dublin. I got permission to use the Weissenberg and I was sure that it could be added to one of the spare X-ray windows on our JEOL PXRD. I went to Professor Frank Coll, the head of the Chemistry Department in Galway, and said to him that we had recently spent more than £5,000 each on IR spectrometers and that I thought I could set up single crystal diffraction in Galway for £5,000. He told me that the University accountant would give me a cheque for £5,000 and he wished me good luck.

With my long-time friend and colleague Des Cunningham I went to Guelph and we stayed for a week with George Ferguson learning how to operate the Y290. We gave George the cheque and he told us that John Ralph had serviced the Y290 in Guelph and that we should get him to come to Galway when we had the machine setup.

We had absolutely no funds and setting up the Y290 in Galway could not have been done without the help of many people especially Professor Philip Walton from the physics department who had experience with X-ray generators, Kevin Carey of Digital Equipment in Galway and John Ralph. The existing X-ray setup in Galway had one generator connected to JEOL PXRD and XRF machines. Philip said he knew where there was a high voltage switch and 20 meters of HV cable which was out of use in one of the hospitals. To illustrate the type of practical help provided I have indicated in the diagram the setup we had and the HV switch labelled S.

I said to Philip “we don’t have any connectors for the ends of the HV cable at the X-ray tube or the switch”. He said “I will get one for the tube made up in the workshop” and then he picked up the cable at the point marked x cut it with his penknife stripped the cut ends pushed the wires into the HV sockets on the switch and filled the sockets with oil. The other great asset we had in Galway was the only Digital Equipment factory outside the U.S. I realized that we had a serious problem with the teletype in that it would not work with 50 Hz ac. I contacted Kevin Carey and asked him if he knew anyone who had a teletype and he told me that the Colaiste Iognaid secondary school in Galway known locally as the “Jes”, had been given a PDP11 and a room full of VT52 terminals by Digital and that they had teletypes for an older system which was now redundant. There can’t have been many schools that had that level of computer facilities in the early 1980s. A few years later Digital gave all Galway second level schools login facilities to a large computer in Galway

When we got the Y290 working any faults the PDP8 developed were repaired by Kevin Carey who would come during lunch hour and quickly find the flip chip where the fault was. John Ralph had told us that we were lucky not to have a straight 8 which had discrete transistors rather than the 7400 series logic DILs which we had. I believe that Hilger & Watts built 25 Y290s and then ceased production. John Ralph bought the rights and built 2 further machines one of which he sold to Max Perutz at MRC Cambridge and the other to the Pasteur Institute in Paris. The Y290 was operated using a teletype which was used to read in the programs from paper tape and punch the output onto paper tape. A full data set required a tea chest full of paper tape. We soaked the paper tape roll before use with Mazola cooking oil to lubricate the punch. We could not afford X-ray diffraction film for the Weissenberg so we used out of date hospital X-ray film. The paper tape was a lot of trouble and it was just possible to read the tapes into the University DEC20 computer. We collected data on about 50 crystals this way but to speed things up we needed to automate the Y290 operations and stop using Weissenberg unit cell data. John Ralph told us that there was an upgrade available based on a PDP11 which cost £45,000. We could not afford this so I learned some machine code programing using an Acorn Atom microcomputer and a PDP12 manual. I asked the University accounts department for a loan of £400 pounds to buy a BBC micro and then set about upgrading the Y290.

Retired Weissenberg camera, Y290, CAD4 and MAR in 2020

Automating the Y290

I wrote what I now know to be a disassembler program for the PDP8 on the Acorn Atom and analysed the PDP8 programs.

The plan devised depended on making a tiny alteration to the BIN LOADER program.

The PDP8 RIM loader program was loaded with the data switches and RIM loader was used to load BIN LOADER. BIN LOADER could then read in the programs needed to operate the Y290 from paper tape BIN LOADER had a go address of 7777. All of the Y290 programs had a go address of 0200.

The last instruction in BIN LOADER was HALT. By great good fortune there was a location on the same page as the HALT instruction which had 0200 in it. I changed HALT to JUMP INDIRECT/ the location containing 0200.

We left the data switches at 7777 and three reed switches were attached to the STOP, LOAD ADDRESS and START switches. The three reed switches were connected to three of the PIA pins on the BBC micro. Thus the BBC micro could by operating STOP LOAD ADDRESS and START run the BIN LOADER to get the PDP8 to read a program the BBC micro sent to it. The program would then auto start when it reached the JUMP INDIRECT (0200). Complete automation !!! and it worked. Seamus Kellehan a member of the chemistry department technical staff built several of these “interfaces” and we sold enough of them at a modest price to repay the University accounts department.

Getting rid of the need for unit cell and reflection data from the Weissenberg

The plan here was to use the Weissenberg to obtain a random orientation rotation photograph. This would be a curved version of the way other diffractometers used a flat Polaroid Camera to take random orientation photographs. In such a photograph each reflection can appear four times and the x and y values of these spots can be used to calculate Theta and Chi for each reflection. Setting these values on the Y290 allowed Phi to be obtained by spinning the Phi axis until the reflection was found. The second thing that was required was to incorporate the Busing and Levy 4-circle geometry calculations1 , least squares on the cell dimensions and unit cell transformations into a FORTRAN program. Rex Dark of the mathematics department in Galway had a look at the Busing and Levy paper and said it should be OK. A few days later Rex returned with 7 pages of neat fountain pen equations and matrix transformations and asked if he could have a set of 12 reflections to try it out. I gave him a reflection set and started to write the FORTRAN version of Rex’s equations. I had not got very far with the programming when a few days later Rex returned to say that it worked and he handed me many pages of tiny writing. I asked him how he did it, did he use a calculator? “No” he said “I used 7 figure log tables”. This was an astonishing feat. The final version of this program was called BRVCEL2 and I know that BRVCEL was in use in the Royal Military College in Shrivenham during the worst of the troubles in Northern Ireland.

The BBC micro driven Y290 increased the number structures to about 200. It was the first 4-circle diffractometer in Ireland north or south and was eventually replaced by a CAD4 which was the second 4-circle diffractometer in Ireland. The CAD4 was replaced by a MAR image plate, the first area detector system in Ireland, which had the Y290 tube shield and monochromator as its X-ray source. The Y290 and the CAD4 were used together for a while and we found that the ESDs on the cell dimensions were better on the Y290. This was a testament to the angle setting accuracy of the Y290 MoirĂ© fringe method which was not affected by gear train wear or slack. The least squares part of BRVCEL was written by Tim Higgins and it showed up some errors in the first version of the unit cell least squares program that came with the CAD4. BRVCEL lives on in the Oscail software package which uses some of its subroutines to check and transform unit cells.3   

(1)  Busing, W. R.; Levy, H. A., Angle calculations for 3- and 4-circle X-ray and neutron diffractometers. Acta Crystallographica 1967, 22, (4), 457-464.

(2)  Higgins, T.; Dark, R.; McArdle, P.; Slmmie, J., BRVCEL—A computer program for cell reduction and Bravais lattice determination. Comput. Chem. 1990, 14, (1), 33-36.

(3)  McArdle, P., Oscail, a program package for small-molecule single-crystal crystallography with crystal morphology prediction and molecular modelling. J. Appl. Crystallogr. 2017, 50, (1), 320 - 326.


Patrick McArdle LĂșnasa 2020