Pulsed power supplies

[Francisl]

This might be a little premature, but I saw that Scandia National Laboratories has a lot of information on pulsed power supplies that may be useful for DPF units. This one looked interesting. Does anyone have an idea how well the different units would work in a continuous mode?
I know that capacitors have a limited lifetime and I was trying to find a way to extend it.

[Charles Griffin]

Benjamin Franklin had pretty good reliability with glass “Leden-Jar” capacitors that he connected in parallel that looked to him like a battery of cannons … hence the name “battery”… Glass is a pretty durable and reliable insulator … A lightning arc is usually made up of at least three cycles or pulses of declining amplitude starting at about 200 amperes each with the duration of each pulse-wave cycle of about 200 milli-seconds … The negtively-charged electrons in the lightning arc surge from the ground, atracted up through the clowds of water vapors that have positive charged water molocules that have been stripped of negtive-charged electrons by high velocity winds-aloft … The electron pulses of the lightning strike over-shoot through, past the positive-charged clowd and are observed as “sprites” extending toward the ionosphere … the positive-charged, hydrogen-ion-protons (of ionized water-vapor) accelerate toward the concentrated negative-charged electrons (that compose the lightning-arc) and … these accelerated, positive-charged, hydrogen-ion-protons have the kenetic-energy to impact each-other and fuse together to form helium ions as evidenced by the resulting pulses of X-rays and gamma-rays recorded by the gamma-ray detectors of the Fermi-Telescope satilite… all very similar to the LPP fusion reactor …

[delt0r]

The trick with cap lifetime is to downrate. By getting caps that are rated at 2x the used voltage you get lifetime multipliers on the order of 10000. This is too expensive for a experiment that may only do 5000 shots its whole life, but well within the capitol cost/running cost trade off of a commercial unit. You pay 5-10x more for the caps and get 10000x more lifetime (note that ESL is not a huge problem for DPF). The switching is perhaps the hardest bit, but by no means unprecedented. Thyratron fire many times per sec for radar applications and although there initial life times was measured in 1000 of pules modern units get a million or more (in some cases).

Sometime i think people forget what it costs to buy energy. Replacing 10 thousand dollar switches once a week will still be economic under many circumstances (yes bigger than 1MW–there is a reason power stations are as big as they are). My power and gas last year for just me was 1000EU.

We can even work it out. A 100MW plant produces 1.6 million dollars of energy per week at typical domestic rates (10c kWh). Remember even for traditional nuclear, fuel costs are negligible. Replacing things regularly that can be somewhat recycled can still work fine. Note that at 10Hz there are 315 million cycles in a year.

ps I forgot, you should look up self healing capacitors. Pretty cool. It is what the NIF is using.

[asymmetric_implosion]

Capacitors at the level required for a large plasma focus typically have short lifetimes. Over specification of caps is the only solution. Repetition rate operation further stresses the caps due to effective series resistance in the capacitor which leads to heating. Typically, you need caps that are specifically designed to operate a repetition rate. Caps like the ones used on FF-1, Z and other large pulse power systems are not agreeable to high repetition rate operation. The pulse power solution being worked on at Sandia and other labs is called the linear transformer driver (LTD). The LTD is a transformer based high current driver. Currently, 1 MA modules are the norm but 2 MA modules have been designed. There isn’t a theoretical limit on the current output of an LTD. It is practically limited by the size of the transformer core.

The LTD is different because it is a step down transformer. It uses a large number of small capacitors (small caps are easier to build at high voltage levels to allow derating. Check out the General Atomics website as an example) coupled together with a transformer core. The 1 MA modules are already demonstrated at 1 Hz for hours on end in a test load. Sandia has designs for a 100 MA z-pinch driver that operates at 0.1 Hz. The downside to the LTD is the number of capacitors and switches. In a typical module, you have 40-80 caps with 20-40 switches. In the 100 MA design you have 500,000 switches. One of the biggest weaknesses of a pulse power system is the switches. You can’t use switches like Thyratrons or solid state switching for current reasons (Thyratron) and cost (solid state). This is a huge engineering challenge for any pulse power fusion system.

[Author]

Posts