How I became Accidental Experimentalist !

This long blog describes my successful foray into experiments to prove my model and my insistence on using only components available in the local market. It led to a unique development, improvement of machine to international standards and publication in top rated journal in spite of efforts made by colleagues to put barriers!

While working at Saha Institute of Nuclear Physics(SINP) in Kolkata(1986-95), I was engaged in nuclear fusion research. The device that we worked with was called Tokamak. It is an experimental nuclear fusion reactor whose concept was first given in Russia and is still the main contender for a commercial fusion reactor. The device in SINP was built by Toshiba(Japan), according to the given specification. I came back from USA and joined SINP around end of September 1986.

After few months, the Japanese supplied the machine and they came to install it. My specialization was in modeling and simulation. However due to my knowledge of computer I started developing data acquisition system for the machine as it was not available with the machine. The discharge durations were expected to be very small, only a few milliseconds to may be maximum of 20 milliseconds or so. While I was working on this, the machine was getting installed by Japanese engineers. After sometime, when the machine was operational, my data acquisition system was also integrated with it. Due to this, I used to be around whenever the experiment was going on and observe the experimental results.

In parallel, I was also working to get computer systems installed in SINP on which I could develop or adapt existing codes to model the discharge behaviour. During my PhD days, I had used Supercomputers for my research and there was no supercomputer available in SINP(In fact there was no computer available at all) to run my model. I took some effort in trying to procure a basic computer system for SINP. I managed to procure a couple of computer systems based on Motorola 68020 processor running Unix OS. After effort of several months, I could run one of the sophisticated models that had been developed for Tokamak called “BALDUR”. This code had nearly sixty thousand lines of program. It was very difficult to understand all aspects of it and there were thousands of parameters that one could adjust. It was not possible to try changing all the parameters and I had to leave many of the standard parameters to their default values. With all this I found it very difficult to reproduce the behavior of the Saha Institute Tokamak(or as we called it “SINP Tokamak”) discharge.

One of the parameters that we usually put in the model is “impurity level”. Default value for this was around 5 to 6 per cent. That was the standard level of impurity that most of the machines used to see around the globe. As I tried to reproduce the discharge behavior, I found it almost impossible to get anywhere close to the experimentally observed behavior by trying to vary some of the usual parameters.

Then I decided to write a much simplified code by removing all the spatial dependency and only keeping time dependence. Such codes were referred to as “zero dimensional” code. I could get this code of about few hundred lines working after few months. As I had written this code completely by myself, I had full understanding of the code. Thus it was easy to start playing with various parameters. As mentioned above, one of the parameters was impurity level in the discharge and after lot of effort, I found that if I put very large amount of impurities like 50-100 %, I would be able to reproduce the discharge behavior. Based on this, I understood that it is likely that lot of impurities were coming in during the discharge by interaction of the plasma with the vacuum vessel walls. So, although the vacuum was extremely high of the order of 10^-8 Torr, the vacuum vessel wall could have lot of adsorbed impurity. When the plasma was created, it would go and hit the walls and bring out adsorbed impurities on the wall. Among these oxygen was one of the most dangerous impurities.

Based on this knowledge, I requested some of my experimental colleagues to try to see if they could develop a system to clean up the inner wall of adsorbed impurities. This technique is typically referred to as “Discharge Cleaning”. But, one of the scientists, who was working on something like this, told me that it will take him nearly 2-3 years as he needed to import some components such as high power vacuum tubes. I was in a hurry to try to clean the machine wall to test the validity of my model and could not afford to wait so long.

It was at this stage that I decided that I should personally get involved in developing a discharge cleaning system, in spite of the fact that I had no experience with experimental developments.  When I talked to other scientists in my department about this, they warned me to not get into experiments as I would not be given any funds for such developments. However I had made up my mind to go ahead. So I started to look at various discharge cleaning methods in literature to find out which would be the most suitable one for me to develop.

One of them, which was found to be very efficient in mid 70s, was called “Taylor discharge cleaning”. It was also referred to as audio frequency discharge cleaning system. In this, an oscillating plasma current was created with electron temperatures less than 5 to 6 electron volts. In the original work, Taylor had created about 20 milliseconds discharge and repeated it at the interval of 500 milliseconds. This system was found to be very effective in cleaning adsorbed oxygen, which is the most dangerous impurity in the early stages of discharge. I thought – let me try to make something similar since this seemed to be the most efficient system developed so far.

However, I wanted to make sure that it can be done in short time. Taylor discharge cleaning system used high voltage vacuum tubes to create the necessary oscillating voltage on the Tokamak Ohmic coil(primary). This would create electric field to breakdown the hydrogen gas and create plasma discharge. Vacuum tubes with that kind of power and voltage had to be imported since it was not available in India. That would have taken more than one year to import. So, I decided to design a system where I would use only those components which were available in the local market of Kolkata. This meant that I had to completely avoid using vacuum tubes and instead create a discharge using transistor – based power supply.

The transistors can not handle very large voltages. So, this was going to limit the way we could produce the required loop voltage of about 25-30 Volts by transformer action. SINP Tokamak Ohmic Coil has 52 turns so this would have required a power source of 1500-2000 Volts. Power transistors available at that time were restricted to 50-70 Volts range, which was way too less. Thus only way I could create possibility of gas breakdown, if I did not use the Ohmic Coil of 52 turns of the SINP Tokamak. Instead I decided to wind a single turn primary coil in parallel to Ohmic Coil and use that to create the discharge.

I had absolutely no experience with developing any such system and no one in the world had done anything like this so I was not sure if this would work or not. To get the necessary power to create the discharge I needed to have large numbers(150-200) of transistors in parallel. I had seen transistors being used in power supply which were controlling magnets in Variables Energy Cyclotron Centre(VECC). VECC was in the same campus as SINP and I had good relationship with Scientists and Engineers there. So I went to the engineers in VECC to discuss if I could use something similar to create an audio-frequency discharge cleaning system in SINP Tokamak. They were not very familiar with the pulsed systems but suggested that there are some old transistor mounting copper plates lying as scrap and may be I could try to make a small system to carry out trial experiments to get some feel about possibility of my concept working. So, I borrowed a used copper plate and mounted 7 power transistors(most probably 2N3773) on it and connected them in parallel, I had to put a little balancing resistance in each of them in series so that in case a transistor switches on early it would not take the load full current and burn.

As mentioned above nobody uses a single turn primary because the coupling is going to be very poor. But, this machine was having iron core. So, I expected that while coupling may be poor but it will not be too bad. So, my first task was to see whether I could create sufficient loop voltage on the plasma side by giving 50 volts or so on the primary side of a single-turn coil. So, with those 7 transistors in parallel on an old copper plate I created an oscillating voltage connected to a makeshift single turn primary coil and then measured to see if was creating enough electric field to breakdown the gas. Luckily for me, I saw that I was getting enough voltage and the loss was not too bad. However, after trying it few times, I found that the transistors were destroyed. That was very surprising to me. But, then after discussing with VECC engineers I realized that I needed to put something to take care of the fact that the load was inductive. So, when you switch the transistor on or off, there will be a high voltage spike that could destroy the transistor. So, I had to put a protection to make sure that this would not happen.

With the first small scale trial I was encouraged to go ahead and create a sufficiently large power supply which could give enough power to create a proper plasma discharge and could be used to clean the inside wall of the machine. So I did some calculations and came up with the requirement that I would need about 150 transistors working in parallel along with a sufficient large DC power supply of 20-30 KW to make it possible. So, from market, I purchased large number of 2N3773 transistors. I borrowed more transistor copper plates from VECC which also had a water-cooling channel built-in to keep the transistors cool. This would have been very handy in case the transistor started to heat up. So, I used 5 different copper plates each having 30 transistors in parallel and then all of them in parallel together. Each of them had the balancing resistance and protection from back EMF due to inductive load. I also borrowed an old three-phase transformer along with a 3-phase rectifier set from VECC that was kept as a spare one. As I brought the transformer, I used a 3-phase variac to control the AC voltage. This variac was lying around for long time in the lab and no one was using it. After putting together full system, I got ready to try out the system. Just when I was about to put on the system, one of the Scientists came and said that the particular 3-phase variac belongs to him and he wants to use it. I asked him to wait for couple of hours so I can finish my trial.

In fact that 3-phase variac was lying around in very bad condition with dust all over and so on. So, I had got it cleaned and made proper contacts to make it workable.

In spite of several requests he insisted that he has to use it right away and I must give it to him. So, finally I took it out and gave it to him. Of course it was simply to stop me from trying out this discharge cleaning system. I had seen a similar 3 – phase variac lying unused in the workshop, although of slightly lower capacity. So, I went to the workshop and requested to borrow that 3-phase variac. They said it is not in working condition but I could take it at my own risk. I brought it and over next 2-3 days got it repaired. It was of lower capacity but was worth trying since procuring a new one would have taken lot of time. And since it was going to be a pulsed discharge, with lower duty cycle, it may survive.

Finally with repaired variac and rest of the system, I was ready to try out creating a plasma discharge for cleaning purpose. As I said this system was unique and no one had tried anything like this. Due to this the chance of its success was very uncertain. However I found that I could get the desired plasma discharge but it was not lasting for sufficiently long time. However I was happy that at least I could break down the Hydrogen gas and create a plasma discharge. Now, my next task was to prolong the discharge to last for about 60 milliseconds. So, I made some modifications and after few trials, I could sustain the discharge for as long as I wished. Then I made the provision that I could create 60 millisecond discharge to be repeated every 1.5 second. After getting all the systems ready, now I was in the mood to test it and see whether the discharge cleaning system that I had developed could clean the impurities that were adsorbed in the wall of the vacuum vessel.

This was going to be a long trial because I did not expect that cleaning for 1 hour or 2 hours is going to have an impact and I may have to run it for several days. I finally planned in one of the evenings that I would carry out the experiment whole night. One of the PhD scholars offered to stay with me in the night. And sometime in the evening, I started the discharges. Initially, I was conservative, so instead of trying every 1.5 second, I set it to repeat at 6 seconds interval. This went on for quite some time. System was working ok. Discharges were being created regularly without fail. Then I reduced the interval to about 4 second to create more number of discharges per minute and increase the cleaning efficiency. My target was of course to slowly reduce the interval to 1.5 seconds for which it was designed. After couple of hours I reduced it to 2.5 second. By this time, it was nearly 4.30 a.m. In the enthusiasm, I did not realize that I had not ensured that all the power supplies that were being used, were designed for the discharge to take the load of creating discharges at 1.5 sec interval. But, in the hurry to speed up the cleaning process, I set the system to create discharge every 1.5 second. By this time it was close to morning 5 o’clock. After few minutes, may be 5 to 10 minutes, the discharge system stopped working. I went inside the power supply room and I saw that few electrolytic capacitors had burst due to overheating. And then I realized the foolishness I had done. Probably if I had continued at slower rate of 2.5 second, I could have continued working with the system for long- long time. Nevertheless, the damage had been done!

I shut down the system, went home and slept for about 2-3 hours and came back to office again. I was feeling little frustrated that I did something stupid due to which some of the capacitors got heated up and burst. The whole system became unusable. I would have to try to procure capacitors again and may take several days or to several weeks before I could proceed further. But out of curiosity, I just went and put on the machine to create the main plasma discharge hoping that with several hours of cleaning there might be some effect that I could notice.

When I put on the machine, I was surprised to see that the discharge had improved completely and now started to behave more like a typical Tokamak plasma discharge! That proved that what my model was predicting about too much impurities inside the machine was true. It also gave me confidence that my unique discharge cleaning system was working well. After this, I went back and developed the system properly so that it could be used on regular basis and it continued to be used for more than 10 years. That work was also published in “Review of Scientific Instruments(Rev. Sci. Instrum., Vol. 65 (1) Jan . 1994, p 135-139)”, which was one of the most respected journals for this kind of work globally. This led to the possibility of other developments and I could get the machine to work to international standards.

Experience gained on this machine proved very useful for me to improve the bigger Tokamak “ADITYA” in Institute for Plasma Research(IPR), Gandhinagar. I would share those in my other blogs.

I must also add here that just like some people tried to put barriers, there were many who went out of their way to help make things work too!