• +44 (0)1793 784389

Case Study: PPM Custom Solutions

Oil tank HV assembly

Repetitive Pulser – 55 kV, 3 kA at 250 Hz

The Customer’s Requirement

A requirement came to PPM for an equipment to be conceived, designed, commissioned and serviced, that could stress large passive components with fast voltage and current pulses. The requirement was for unsupervised 24/7 operation, variable repetition rate up to 250 Hz, and up to 55 kV amplitude pulses, with a short circuit current of over 3 kA.

There were lots of ways this could be done which might work and be realisable. A technology review and specification development was started, where various switch technology, power supplies and configurations were considered. After four iterations of refining requirements with the customer, developing risk-assessments and analysing specifications of the sub-systems, a system level specification was delivered. The document totaled 100 pages and became the basis for the living build document. Major components like the switch were selected and details such as HV cable heating and the overall layout of the equipment to help with room sizing, were formalised. The risks associated with use and servicing of the equipment were evaluated and mitigated.

Detailed Design of the System and Components

The detailed design work was started. This included design of the oil tank; the mechanical support structures; insulation; capacitor assembly; thyratron cage; thyratron driver; snubber resistors; current limiting resistors; power supply control specification; sensing and instrumentation. By the end of the build process, a complete set of documentation totalling 300 pages had been produced. This would enable future servicing and maintenance of the machine.

To make the system realisable and easier to service, much of the HV and high power components were created by combining an array of lower voltage parts. For example sourcing a single 20 nF, 40 nF and 100 nF 60 kV capacitor wasn’t practical (in this case a requirement of the test was for variable energy storage options) as no manufacturer would want to start a winding machine for one or two parts. To solve this problem a different approach was required. There were many 2 kVdc standard parts available from catalogue distributors (figure 1) which could be utilised with appropriate techniques in parallel and series arrays. This created problems such as dynamic voltage sharing between the low voltage parts which were in series; current sharing for those in parallel (lesser concern in this case); and also the number of solder joints required. Each problem was an opportunity for a mistake or weakness, but with care these issues were overcome.

2kV 1uF film capacitorStandard 2kV 1uF film capacitor.

The Switch

The thyratron selected for the main switch was an E2V CX1525 (now obsoleted). This device is rated for 70 kV and 3.5 kA max anode current and looks like something from a sci-fi movie. The construction is a ceramic tube with a number of electrodes along its length. This particular thyratron is a tetrode configuration (a device with 4 electrodes). For a thyratron to work properly it has to be configured with an ideal set of conditions, a bit like a BJT but add temperature and pressure! The device has a hydrogen reservoir, which requires a low voltage heater supply to release the hydrogen gas into the tube. It is this hydrogen gas which is ionised and conducts when the device is triggered – the more current in the heater the higher the gas pressure. This pressure affects the point at which the gas breaks down and conducts (Paschen’s law).

In this particular device there are two “grids” and each has a specific signal applied, with specific timings to ensure reliable operation. This triggering system is called “double pulsing” and should result in a lifetime improvement of up to 5 times over simpler triggering techniques. A major design consideration is the grid spike that the trigger circuit will see when the device ionises. The driver has to be designed to tolerate this ~2-3 kV spike, which typically increases with the age of the thyratron. The E2V “Hydrogen Thyratron Preamble” is a useful document, and a highly recommended read! Even with prior knowledge of this spike, it took two iterations of the protection circuit to deliver 24/7 operation over weeks, months and years. At the time of writing the thyratron had completed over 4 billion pulses at over 25 kV and 2 kA peak current, with no signs of degradation.

Thyratrons in the non-conducting state can withstand reverse voltages. However transient reversals after conduction can cause arcing and lead to early failure of the device. This was overcome by providing a simple freewheeling diode which could block the required voltage in the forward direction and support the freewheeling current in the reverse direction. This was realised using an array of readily available parts (the trusted UX-FOB diode). A dynamic voltage sharing network was designed to make sure the array performed robustly under the fast pulses.

Capacitor assembly HV dividerClose up of capacitor assembly, HV divider. The tyratron protection diode assembly can be seen in the background.

The Thyratron Control Box

The thyratron required some auxiliary power and trigger pulses to operate. The thyratron control box housed the adjustable LV transformers used to provide current for the heater and reservoir (hydrogen source). These are simple devices that just need to be kept cool for them to work indefinitely. The box also housed custom PCBs to monitor the currents heater currents; perform the “fast” over current shut off and house the thyratron trigger generation system supplied by North Star High Voltage. The fast overcurrent detection was designed to inhibit the trigger signal in the event that the previous current pulse exceeded a certain value. This would prevent repetitive over current events. This was tested by applying a dead short to the output of the system and gradually increasing the comparator reference voltage until the required short circuit current was reached. If this condition was met, a fault signal was sent to the HMI to inform the user that pulsing was inhibited due to an overcurrent condition.

The Power Supply

Given the capacitive energy store in this system and the repetition rate required, a standard HV PSU designed for biasing was not really appropriate. The voltage requirement of 55 kV was outside the range of TDK Lambda capacitor charging units. After some searching, the power supply selected was an NWL PowerPlus. This is normally deployed as an electrostatic precipitator power supply in industrial processes. Its topology makes it suitable for capacitor charging, and with control changes implemented by NWL to provide a constant current and constant power output, it provided all the functions required. The controller also provided the “end of charge” signal used to trigger the thyratron through the driver board. The power supply was capable of sustaining the required 250 Hz continuous repetition rates and showed no signs of instability.
A diode was also used to protect the PSU from voltage reversals. This was possibly a belt and braces move, but for the sake of a small number of additional components, the safety this offered was hard to resist.

Oil tank HV assemblyOil tank and HV assembly. The PSU can be seen on the far right hand side.

Safety for 24/7 Operation

The make this test set up appropriate for 24/7 operation, with minimal user input, lots of sensors were required. The following events and parameters were monitored by the supervisory PLC:

  1. Arc detection (dual optical sensors)
  2. Smoke detection
  3. 8 fibre optic temperature sensors
  4. Cooling water flow rates
  5. Cooling water temperature
  6. Voltage waveform
  7. Current waveforms
  8. Fast over current detection

Some of these were obvious go-no-go choices – if you detect an arc event, turn it off! The same goes for smoke! The flow rates are useful in a number of ways (you require a minimum rate to achieve the cooling required), but if there is a difference in the inlet and outlet flow it means you probably have a leak. The HMI was able to capture continuous, 250 Hz samples of the voltage and current waveforms and store them to hard drives for analysis off line. Alternatively you could select a “normal” envelope and only capture waveforms which fell outside that normal shape.

The Human Machine Interface

To enable the system to be used safely and repeatedly, a HMI was designed that combined the monitoring of the oil tank (temperature, cooling flow rates, oil level), test management (each shot was captured in a results database) and test set up (thyratron warming, PSU settings, capacitor selection and overcurrent detection reset). The HMI ran on a remote PC, allowing galvanic separation, and a big grounded enclosure to mitigate the radiated emissions from the unit under test. The PC running the HMI was accessible from a remote location, so the system could be monitored at any time.

Repetitive pulser systemThe Human Machine Interface.

Conclusion

After nearly 4 billion pulses and counting, the system is still performing as intended. There were a couple of learning points on the way.

  1. The system is only a good as the weakest link. In this case that weak link was the components used to manage the grid spike from the thyratron. Until the system was used this was of unknown amplitude and the components needed to be reinforced for continuous operation.
  2. The impedance mismatch between the load and the source meant the output voltage divider was lost during early testing, due to reflections. With some more careful measurement this could have been avoided. The issue was resolved by the addition of a snubber assembly to reduce this effect.
  3. Screw terminals are not perfect and they can get worse over time. The low voltage connections into the thyratron control box used 50A terminal blocks. After many hours of operation and numerous bumps and scrapes, the resistance of the connection increased and overheated This result in the housing of the block melting and conductors coming into contact with the grounded case, halting operation.

CT snubber assemblyClose up of CT, snubber assembly and bypass resistor.

PPM Power is here to work with customers to solve high voltage and power electronics problems. We are technology agnostic and in this case proved that thyratrons are sometimes the best way to meet a given application’s needs. For all of your power requirements, please contact our technical support staff by emailing sales@ppm.co.uk, calling +44 (0)1793 784389 or using our live chat window.

Capacitors

custom DC link capacitor and busbarCustom Capacitor Solutions

power ring homepage buttonPower Ring

APCS standard test kitStandard Test Kits

Ceramic Capacitors

More Information?

Telephone +44 (0)1793 784389 or email: sales@ppm.co.uk