2.5 kJ Capacitor Bank Complete!
Posted by aonomus on July 12, 2009
After plenty of delays due to more important projects, I have finally completed the capacitor bank control system and construction.
The capacitor bank is a 22,500uF, 450V capacitor bank with a 10kA surge current capacity using solid state switching. A custom built charger was created so that I could use 12VDC (ie: car batteries) to charge the capacitor bank outdoors for more energetic tests. Along with the charger, a control system to monitor bank voltage, and provide safety interlocks to prevent discharge during firing, automatic voltage control, and discharge load control to abort high-current tests were key aspects of the charge controller.
In essence the entire design of the bank is for low inductance, high current capacity for pulse power experiments, to that end, big SCRs, heavy bus bars, and a low inductance layout were key to building the bank.
I decided that the bank should only be 450V maximum to control the maximum current peak for extremely low inductance and resistance loads to prevent damage to either capacitors or semiconductors, and that to parallel 12 large capacitors and split the load across several sets of bus bars, minimal angles were required in the bus bar wiring.
As a discharge load, two 100W incandescent light bulbs in series were used with a small SCR (really any could have sufficed, I had a stud mount SCR handy). A important part of a large capacitor bank is to have a quick way of discharging it in case residual power remains in the bank, or in case you need to abort your test due to safety reasons (eg: bystander in the firing area, etc). The capacitor bank can discharge from full 450V charge to 0V in under 20 seconds using this load.
To accomadate the large puck style SCRs, I had to build custom clamping devices, the goal of which is to ensure sufficient pressure (several kN or hundreds of pounds of force) and good electrical contact.
The internal construction of a SCR is essentially a large wafer of silicon material with the external metal contacts sandwiching the material and a small trough for the gate lead to the center of the wafer. The manufacturers create a metal diaphragm which has to be compressed with the minimum force to push the external contacts together, since I got mine used I could test electrically (ie: low power) the functioning, but I would still require a large surface area contact with good pressure to make sure the entire load was spread across the wafer.
The clamp itself is a pair of thick aluminum bars I bought from metal supermarkets, cut to length, and drilled for 1/2″ thick bolts. Because the entire device is electrically connected, the SCR and connections had to be isolated, and as a quick insulator, some perf-board squares superglued onto the clamp itself with a washer at the end to concentrate the force into the center of the SCR.
The resultant assembly was bolted onto the capacitor bank, two in parallel on the negative side of the capacitors with the positive unswitched.
Additionally you will notice in the above photo a few other minor details:
- Bleeder resistor: 18kohm/50W for extremely slow, failsafe discharge, and to maintain the bank at 0V during storage
- Back-EMF diode: a large R6021225HSYA power diode (1200V, 250A constant, 5-8kA surge) to catch any ringing that occurs, mounted directly across the + and – rails so that voltage reversal is prevented (electrolytics are damaged upon voltage reversal).
- General wiring: two 5ohm resistors are used as gate current limiting to prevent damaging the SCRs, without them the SCRs could be burnt out from the current.
Control and Charging System
The charger and control system was designed specifically with safety interlocks to step up 12V to up to 450V with a moderate charging rate, automatic voltage cutoff, and various other safeties to enhance the professionalism of the project (anyone can trigger a SCR with a 9V battery). The entire device was built into a roomy IP66 case which was laser-cut by some friends from the Hacklab.
Internally, a multitude of relays connect the external power to systems as the key switches change modes, enabling or disabling charging, firing triggers, and so on. The voltage indicator is connected via a resistive voltage divider to the output terminals which remain connected to the capacitor bank during operation. A large relay disconnects the capacitor bank from the charging circuit to prevent any voltage ringing from damaging the charging circuit. Additionally upon close examination, a small loading capacitor (2 capacitors in series) is present on the output terminals just so that the system always has a load to charge, to prevent damage to the circuitry.
The capacitor charger itself is based off of Uzzors ZVS capacitor charger, with several modifications including differing turn ratios, inductance and capacitance values, and a slightly different voltage reference for the comparator. The output of the device is fed into a custom wound flyback transformer with a full bridge rectifier on the output and into a small set of capacitors to present a minimal load to the charger in case no capacitor bank is present.
Some low power tests have shown that the bank has a large amount of peak current capacity, however since I have not gotten my rogowski coil functioning yet, I will need to finish my instrumentation to get actual quantitative data to support the goals of high current, low inductance.
Multiple tests at slowly increasing power levels, the last test at 25% maximum power. I need to create a proper coil to transfer power efficiently into objects to launch.
A slow motion (60fps slowed down) video of the 25% power test showing failure of the connection to the 1 gauge cable.
- Instrumentation: rogowski coil to track current waveform
- Charger: add heatsinks to the MOSFETs just as a added measure of efficiency
- Case: add carrying handles, a safety cover for the positive terminal of the capacitor bank, and a latch for the plexiglass cover