Interesting Power Supply Company

Commenter windmill at Talk Polywell has brought to my attention an interesting power supply company Diversified Technologies Inc. Here are a couple of short (under 10 pages) papers that explain the technology.

Solid State High Voltage DC Power Distribution & Control [pdf]

Here is the key point from the above [pdf].

The largest cost components in this design are the semiconductors (IGBTs). Because of their widespread use in locomotive engines, subway cars, elevators, and a wide range of electrical motor drive and power supply systems, these devices are evolving at a rapid pace, especially in comparison with vacuum switch tubes. In the last decade, we have seen the switching speed and power handling capability of IGBTs increase by an order of magnitude (200 kVA to 4 MVA), at essentially constant prices. This puts high power electronics, for the first time, on a favorable, long term cost reduction path. This is the equivalent of the computer industry’s Moore’s Law of continually higher performance per unit cost, but applied to power systems.

Today, a 100 kV, 2MW buck regulator, with a series switch, can be built for approximately $500k USD. This cost will decline due to increased semiconductor performance and decreased manufacturing costs. In contrast, estimates for the equivalent conventional approach are $2- 3M USD, and show no trend towards cost reduction.

Quite so. IGBTs with a voltage rating of 6,500 Volts and a 600 Amp current rating are now off the shelf.

A Solid-State Switch for 13.8kV Power Distribution [pdf]

The company claims to be able to make power conversion equipment that costs in the range of 10¢ a watt in production quantities. That is a very good number. Diversified claims specifications for their supply technology that are very not too bad. An adjustable 100 KV DC supply can deliver 1% regulation and .1% ripple. That is just the ticket for Polywell Fusion experiments using D-D. For pB11 at the resonance peak I'd like to see tighter regulation. Say .1% regulation and .01% ripple. I have some ideas.

Fusor Power Supplies

I think a supply in the 15KV to 30KV range with an available current of 30 mA would be good for general purpose experimentation and neutron generation. It should be adjustable, regulated, and current limited. It should be immune to the usual lab accidents such as shorts and current bursts. It should have an emergency fast trip.

I have some candidate mfgrs:

Glassman High Voltage Inc.
Spellman High Voltage
Universal Voltronics

Power Supplies - Update #1

My thinking on the power supplies has evolved.

What I'm considering now is a 400 or 500 VDC power bus (maybe two of them) SCR regulated from a 3 phase power supply. 12 pulse for reduced filtering requirements.

The bus will supply the power for 100 V @ 25 A converters (parallelable) for the magnet supply, and 1,000 V @ 25A converters (series amenable) for the high voltage.

If each converter has its own phase trigger, the phases can be staggered so that the effective ripple frequency will be the drive freq. (around 30 KHz) times the number of converters, times 2 (for full wave rectification). For a 1,000 A @ 100 V magnet supply that puts the ripple up at about 2.4 MHz. Easy to filter. A 20 KV @ 25 A supply would have 20 converters. That would give a 1.2 MHz ripple. A full up 80 KV @ 25 A supply would have 4.8 MHz effective ripple.

Update: 08 July 007 1856z

On the HV side we will also need 24VDC (nominal) @ 25 A to power any controls on the HV side including UC2901 feedback modulators. Plus 120VAC @ 60 Hz 10 KW for misc power. The 24VDC will be battery backed to supply surge currents. There will also be quench SCRs (into resistive loads) for shorting the magnet supply and the HV supply. The resistors to be sized so that the supply output is discharged in .1 second or less to 1/10,000th of the maximum voltage or current.

Update: 09 July 007 0757z

I have been thinking about safety issues. Most notably what happens when utility power is lost? First off if the LN2 pump is pumping it should keep pumping for a few minutes after power loss. Which means battery backup. Second any water cooling must continue for a few minutes after power off. The water chiller need not be backed up. The control computer needs to be battery backed up. In fact it should operate from the battery bus at all times so power glitches do not affect its operation. The same will be true of the data collection computer. In addition a backup generator with a 48 hour gasoline supply should be provided. All important loads should have automatic transfer switches to the backup generator. Power transfer should be sequenced to minimize the surge load on the generator.

HV Power Conversion

Light Triggered SCRs [pdf] an overview.

Power Conversion a book (the above link is an excerpt.

Eupec Information - semiconductors.

GvA-leistungselektronik power semiconductors and modules.


Monograph - Light triggered SCRs 8kV [pdf] used in Celilo Converter Station of the Pacific Northwest-Southwest HVDC Intertie by Bonneville Power Administration in Portland/Oregon/USA.

Magnet Power Supplies

My plan for the magnet power supplies (100V, 1,000A max) is to use an induction heating power supply to supply the power and use a ferrite transformer for voltage isolation. Followed by rectification and filtering.

Some candidates for the induction heating supply can be found at: Induction heating power supplies.

Since supplies of up to 1.5 MW are standard it may be a good way to develop the high voltage as well.