Temperature controllers
Temperature controllers
Page produced with the help of Dominique Brochier, Paul Dagleish, Pierre Andant (retired staff), Eddy Lelièvre-Berna (ILL), Jacques Bossy (CNRS) and Jean-Louis Bret (retired from CNRS/CRTBT).
Thermometry techniques and temperature controllers are indispensable to cryogenics and furnaces. It is vital for our experiments to be able to measure temperatures constantly and with great precision, whilst keeping them accurately regulated and stable over long periods of time. And it's not that easy!
The ILL Annual Reports show that the performance of the industrial devices available in the seventies was disappointing. Once again it was up to the ILL to come up with an in-house measurement and control system; its achievements in this respect made a major contribution to the success of the ILL's sample environment facilities and, therefore, to the quality of the experimental programme.
Early days
For cryogenics the 1972 annual report mentions the purchase and production of three AsGa diode controllers (1.5 T 300). They had been developed at the CNRS/CRTBT by Jean-Louis Bret, Jean-Paul Faure and Gilles Pelletier in 1972 and then commercialised by MERIC [1]. They were based on analogue electronics with AsGa diode thermometry.
It was the best material available at the time and the only competitor was Lake Shore, le ILL's current supplier. The CNRS controllers turned out to be difficult to use for non-specialists and Dominique Brochier now believes that the choice of this diode thermometry was a mistake.
Alain Filhol confirms that the controllers were indeed capricious. Ah that famous "gain" button which no one knew exactly how to use, but which was clearly essential! Everyone had their own ideas.
For the furnaces, we've not been able to come up with much information, except for a message from Pierre Andant:
I remember the "Drush" controllers with manual PID control producing from 0 to 5 volts DC and running a DC amplifier set up to saturate the magnetic circuit of the saturable inductor connected in series to the power transformer and supplying the furnace. The principle still applies today and is preferred to the triac systems which produced parasites at the time and could interfere with the neutron detectors.
The ILL gets to work!
The 1977 annual report shows that ILL had realised that its temperature controllers were not up to the mark.
A precise study has shown that a considerable time was devoted by the experimenters to changes of temperature. It seems more and more necessary to be able to control the temperature by means of the computer operating the instrument, so that in the end temperature will be a parameter like the others, for the sample can be "positioned" in the same way.
ILL decided to produce its own temperature controllers and completely review its thermometry approach. The 1980 annual report speaks for the first time of a microprocessor regulation project, a major innovation. The initial aim was just to copy into digital what was being done in analogue. Dominique Brochier explains:
I developed the controller with one of the computer engineers [Paul Dagleish]. I said "Listen, we can simulate an analogue controller with digital techniques; it'll have to tick over at... I don't know... a tenth of a second." This was just becoming possible, but I didn't personally know any of the techniques. I was just suggesting the direction to be taken.
PTC, the first microprocessor-based controller
It was therefore Paul Dagleish who was given the daunting task of designing and programming the very first prototype microprocessor controller.
At the time the control devices used in industry were all analogue and non-programmable. This meant that a scientist had to come back on site at weekends or during the night simply to change the temperature. It was burdensome and incompatible with the rapid turnover of experiments at ILL.
The project was therefore setting its sights very high:
- an independent device which could be piloted by the instrument computer
- with automatic temperature programming
- and temperature measurement by diode as already in use at ILL, but also by platinum or carbon resistance (more efficient), and by all types of thermocouple,
but it succeeded in all respects. In 1984 the new controllers - baptised as PTCs (Precision Temperature Controller) by Paul Dagleish - were already being tested. In 1985 ILL could claim:
“It is now possible for an experimentalist to verify and change temperature from home.”, and it was all thanks to the famous Minitel which had been brought on to the market in 1982. ILL was indeed at the forefront of progress!
The PTCs were a great success and were soon being requested by other laboratoires. In 1988 ASL (Automatic Systems Laboratories Ltd) in the UK started production under an ILL licence, followed later by Duhamel Automatisme.
Paul Dagleish remembers:
The first automatic temperature controller was designed and programmed by me. I baptised this controller the PTC (Precision Temperature Controller). It was the first temperature controller totally piloted by microprocessor - the Motorola MC6809 which had an 8-bit bus and 16-bit internal registers. The control programme was a simple loop continually monitoring the four temperature measurement channels [2], displaying their values, calculating the appropriate power using a PID algorithm in order to reach and maintain the heater value. The loop communicated with the main instrument via a standard RS232 serial port. [Translator's note: a distant ancestor of the Ethernet Lightning ports on modern computers.] A huge step forward in removing experimental errors was the introduction of thermometer calibrations for each individual cryostat. This was achieved using simple plug-in modules containing the calibrations on EPROM. Each module, one for the sample holder and one for the cryostat regulation block, contained two sets of calibration, for the platinum and carbon thermometers. The module for the regulation block also contained optimum PID values for the given cryostat. This meant that, for the first time at the ILL, sample temperatures could be automatically set without human supervision. I spent some time visiting possible manufacturers around Europe and finally chose ASL (Automatic Systems Laboratories Ltd) in England [...]. ASL was at that time at the forefront of temperature resistance measurement with their automatic AC Bridge invented by their PDG, Peter C. F. Wolfendale.
ILLSEC - the PTC Mark 2
The 1988 annual report mentions the design and development of a new "ILLSEC" temperature controller (ILL Sample Environment Controller) [3]. This was an improved PTC, as Paul Dagleish explains:
The updated version, ILLSEC (ILL Sample Environment Controller), was also designed by me with help from “le gone lyonnais”, Guy Drevetton. It had much better electronics and a proper real-time operating system, the Microware Systems Corporation OS9 running on the MC68032 32-bit microprocessor. Each measurement channel had its own 16-bit microprocessor so that temperature measurement could be much more rapid and accurate. Each ILLSEC contained the complete calibration and control parameters for all the ILL apparatus. Each sample environment apparatus (cryostat, cryofurnace etc) was fitted with a system that transmitted its unique identity to the ILLSEC thus reducing the chance of calibration errors sometimes introduced by not using the correct calibration modules. The ILLSEC was built by Duhamel Automatisme in Domène.
Eddy Lelièvre-Berna adds that the new controller was a real success and that some of its characteristics cannot be rivalled by the modern industrial controllers:
The ILLSEC could store over 100 calibration curves, whilst the famous LakeShore 340 controllers are limited to 40. They delivered over 110 W at 50 Ohms whilst the LakeShores block at 50 W. The ILLSEC could also control the temperature from 2 different thermometers with different temperature ranges, something which can still not be done with commercial devices in 2015.
There are still two ILLSEC being used in the ILL's cryogenics labs, and nobody dares turn them off, in case they never revive!
Computing for success
An essential contribution of the ILLPTC controller was that it could be controlled by the instrument. In other words, the scientist could define in advance a series of actions, including changes in temperature, which the instrument then performed in his absence. At last quiet nights and weekends!
It still was necessary to adapt the spectrometer control program (MAD) and it was John Allibon who did it by creating DTI (Digital Temperature Interface). The computer of the instrument could then ask for a temperature, wait for it to stabilize, perform controlled up and down ramps in temperature ("Cramps" in ILL jargon), during or between two measurements, etc. And there was an innovation: a PID algorithm (proportional corrector, integrator, differentiator) external to that of the ILLPTC allowing the setpoint of the controller to be adjusted, to bring the sample to temperature as quickly as possible. Finally, the user had computerised monitoring of the temperature and cryogenic fluid levels, marking the end of those pen recorders too often out of ink or paper.
The DTI was first developed for PDP11, before being migrated to VAX-VMS, Unix and ultimately Linux, as ILL's instrument control systems evolved. The ILLPTC - MAD - DTI was such a success that it was adopted by various other laboratories: University of Århus (DK), University of Helsinki (Fi), Risø (DK), HMI (Berlin).
Nowadays
Fifteen years later, with technology changing, a complete revamp of ILL's controller technology would be necessary. The ILL therefore decided to switch to Lake Shore's industrial controllers. There were plans to enhance these with the automatic recognition of connected devices, as already available on the ILLSEC. Although SANE developed a prototype, staff shortages have meant that the project was never completed, to the great regret of ILL's scientists.
References
- Société Meric, 22 bd Jean-Jaurès, 91290 Arpajon. Cited in the ANVAR 4532 dossier.
- ILLSEC Handbook, Paul Dagleish, 18 August 1997.
- A single measurement point is not enough to ensure good measurements and good control of the sample temperature. The controller has to manage multiple sensors distributed within the cryostat.