Need "constant voltage-drop device"

Try "voltage regulator".

Or maybe even "Zener diode".

Good Luck! Rich

--
-----BEGIN GEEK CODE BLOCK-----
Version: 3.1
GAT(E P) dpu s: a++ C++@ P+ L++>+ !E W+ N++ o? K? w-- !O !M !V PS+++ 
PE Y+ PGP- t 5+++)-; X- R- tv+ b+ DI++++>+ D-? G e+$ h+ r-- z+ 
------END GEEK CODE BLOCK------

power transistor with a reversed zener in the emitter lead, resistor from collector to base, resistor from base to anode of diode. Collector and emitter are your 2 terminals.

make an "active zener" - I would use a 100A~300A, 1200V IGBT with a suitable zener from C to G, and some gate impedance too (say

10k//100n//12V zener). Voila, a giant zener that can dissipate (with a large heatsink) you 2kW. BTDT for a 300A 750Vdc dynamic brake (in which case the DB resistor was in series with the IGBT collector).

you might want to use a resistor too, and greatly reduce the heatsink size (by letting the resistor get effing hot):

K-----+----------------+ | | __/ [R] R = 90V/20A = 4R7 or so, rated 2kW 90V /\\ | | | | / c +----+-----+---|| | | | \\ e __/ [10k] [100n] | 12V /\\ | | | | | | | A-----+----+-----+-----+

Cheers Terry

This sounds like another good ASCII-schematic opportunity. But before we get started, perhaps John W could regale us with details about his application. It's not everyday that you find someone willing to burn up 2000 watts of power, even if 20kW is available. John, we hope you're ready to take part in the engineering conversation. TVS "super zeners" are easy to get and use (individual standard 5kW parts will drop 100V and 20A for 2ms without trouble, series stacks will do much better). Active zener diodes are easy to make with modern power MOSFETs.

But we have questions, like, what's the expected duty cycle for your super-zener, and how long? What's your accuracy spec, is 5 to 10% OK? How about 1%? Are there transient-response issues we should consider?

--
 Thanks,
    - Win

OK, sure. I will try to explain the background of the problem:

The application is that we own a very good quality (and VERY expensive)

30 KW regulated power supply that was designed to put out anywhere from ~150 VDC to 1500 VDC at up to 20 amps. And it works OK within its design range, but occasionally we need outputs as low as 50 VDC, and the supply becomes unstable at those very low settings. (And we can't afford to just go out and buy another supply.)

So, I thought that I might be able to "trick" the supply into thinking that it is working into a 150 volt load while actually applying only 50 to the load. (And, by the way: The load is an ionized gas plasma, and its impedance can vary rapidly over a wide range.)

Duty cycle would be "continuous" i.e. several hours at a time. Accuracy is not super-important; 5% would be fine, and even 10% might work.

'Transient-response' is an interesting question! The existing power supply has a lot of very complicated (and proprietary) circuitry built into it to maintain constant output voltage even while the load impedance is changing abruptly. I'm not sure, but I think that it also has circuitry to momentarily turn the output voltage OFF for a few microseconds if the load impedance goes below some critical threshhold (i.e. if a localized, intense arc forms in the plasma.) I suspect that an 'active' voltage dropping device with a finite response-time might confuse the control circuitry quite a bit, but I'm not sure. At the present time, we are using resistor banks for the voltage-dropping, but this is not really satisfactory because the current can vary so much, depending on the gas pressure & composition.

Hope this helps to clarify the problem.

You are describing a basic switching regulation circuit. You just need a similiar one at the second stage. However, 20A would be tricky to build with solid state. Google for mechanical switching regulators, for example.

A long time ago I saw a circuit for a wide range, high power DC supply. It consisted of a preregulator, which was a phase-fired SCR rectifier, to provide a raw voltage to the input of a second linear regulator, which would control the output essentially to zero. The duty cycle of the pre-regulator was set by the difference between the input and output of the second regulator. Your application would require a rather hefty linear section, with 20 amps at 100 V drop, but 2000 watts is not unworkable. This should work OK as long as the main supply will not create voltage surges high enough to damage the components of the second regulator. Large capacitors should be able to sustain rapid load impedance changes, as long as there is no need to change the supply voltage rapidly.

A more efficient approach might be to use a switching power supply with a

150 VDC output (maybe several in series), and wire it in reverse, to subtract its voltage from the output. You may need to use large output capacitors on this supply to handle surges, and probably a big series diode on the output of your main supply to block reverse current.

A resistive load across the output may also help to stabilize the supply at lower voltage levels. It is probably a PWM system that gets down to very low duty cycle when running at its lower limit, and an extra load will increase the duty cycle to a more stable condition.

Paul E. Schoen

Thanks very much to everybody who offered constructive suggestions.

I have ordered some parts to try out the IGBT circuit suggested by Terry Given. If that doesn't work, I will look into Paul Schoen's suggestions. I suppose I could just as well use a stack of car batteries rather than a stack of switching power supplies, Right? (We happen to already have about 10 car batteries sitting around here, left over from another project.)

john w.

Hi John,

email me, and I will send you a schematic of an actual dynamic brake.

Cheers Terry

Yeah. But don't forget the reverse-current protection on your supply.

I don't know what the inductance of a car battery is. I guess it's probably pretty low.