Tuesday, February 28, 2006

Changed Relay K4 and SCR on M03

I received the SCR and relay this morning, I installed them and test the system again. However, the high voltage still could not be controlled. Mike helped to check the circuit board again.

Craig loaned us the used computer (400 MHz, Pentium II, better than our old one), which can be used to grab the wave froms from the oscilloscopes. I installed Tektronix WaveStar and successflully obtained the wave forms, much faster than the old machine.

1 comment:

Randy said...

I have been looking at the "blog" and noticing the many weeks of work that have been done by Song and Mike troubleshooting Prometheus and, in particular, resolving the issues associated with the main DC charging supply system. I have some suggestions and thoughts on (1) improving the present system, and (2) a better long range fix.

(1) Improve the Present Charging System

The system utilizes an SCR based command system wherein the North and South Bank Capacitor stacks (1.5 uF total each) are charged through a series array of ten 10K resistors. These resistors have been failing (burned due to excessive power). This can be due to the main SCR not turning off after charging the capacitors and/or the Thyratrons of either bank not recovering immediately after switching and therefore drawing excessive current at the onset of the next charging cycle. I would propose that the North and South Banks each be charged through their own respective resistor strings helping to isolate the failures. Furthermore, these resistor strings could be sized to better absorb whatever SCR charging failure is occurring. Some details for this scheme are listed below:

Presuming that the maximum repetition rate is kept to 0.4 Hz or slower, then the maximum value of the charging resistor string for simple RC type charging can be determined as:

5RC < 2.5 gives R< 333K. The closest resistor value for ten 225 Watt wirewound resistors would be 30K for each power resistor. A total fault (i.e. either or both thyratron bank(s) stays shorted) would result in a power of 30kV x 30kV/300k = 3000 watts which is still in excess of the string rating of 2250 watts. This is, however, much better than the present situation of a dissipation of 30kV x 30kV/100K = 9000 watts. Clearly, when the SCR charging, timing circuitry, and Thyratron Banks are all working correctly, everything is fine. The actual situation of how much excess power must be dissipated by the resistor string during a fault condition is determined by the duty cycle of the fault but designing for maximum fault conditions is best practice.

Another way that the present SCR system can charge the capacitors is via Constant Current Charging by varying the "phase angle" of the SCR versus time during the charging sequence. In order to charge the 1.5uF capacitor to 30kV in 2.5 seconds requires a constant charging current of I > CV/t = 1.5uF x 30kV/2.5 = 18 ma. From limited waveforms that I have seen, the system is actually being charged at a much higher constant current since the power transformer has a rating of 200 ma on the secondary which is historically necessary for repetition rates of several hertz. However, the present charging time can be up to 2.5 seconds (0.4 Hz) and it may be possible to alter the SCR feedback scheme to operate at lower constant current charging. This would be better to insure any possible problem with Thyratron bank recovery after switching. The actual charging waveforms probably lie between the extremes of RC and Constant Current charging but I do not at present have access to the exact details of how the system operates when correct.

In summary, if one is to keep the present SCR charging scheme that has worked well but with failures over the last 18 years, there are two "simple" changes that can make the system more reliable and robust. These are (1) Provide a separate string of charging resistors for each Thyratron Bank and (2) alter the maximum charging current that the SCR can deliver.

(2) Replace the DC Charging System:

My long term (maybe sooner rather than later) recommendation is to replace the SCR, its control boards, and the Transformer (plus diodes and resistors) such that each Thyratron Bank is charged and controlled by an independent Gamma High Voltage, commercial charging supply. At Los Alamos and Sandia I have used supplies from Gamma over the last 20 years on many systems with great success. These supplies are ultra reliable, fault tolerant, easy to use, and quite affordable (a few K per supply). A suggested minimum Gamma supply would be model the RR40-25 which is a 40kV, 25 ma, 1000 watt, rack mounted unit; the cost of this supply is $3900. Some savings could be had by buying a 30kV supply but this is right at the maximum voltage that is now utilized; one should have some safety and operating margin. This supply would charge the 1.5uF Thyratron capacitor bank to 30kV in t = CV/I = 1.8 seconds. The supplies have excellent constant current control, current and voltage meters with remote readouts, remote current and voltage control, and interlocks. I think that the interfacing would be quite straight forward and eliminate a lot of "home made" Beta boards and parts and all the associated problems that have occurred in the DC charging portion of Prometheus.

OK -- so there's a very brief summary of some recommendations -- all of this can be modeled in detail and compared to measured waveforms before any commitment is made.