Yet another HV power supply project / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

Yet another HV power supply project

Move Thread

LAN_403

klugesmith

Sat Oct 25 2014, 12:38AM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

The other day I unpacked my new plaything, found on ebay. Cathode and anode tanks, and HV cables, from a Bennett 125 kVp x-ray generator.

First time I'd actually seen those big old HV plugs and receptacles.

Started taking one apart, before even trying it to see if it worked.

Here it looks like the primary winding uses thick, flat magnet wire.

Here we see a few of about nine separate wires between secondary windings and the bottom board.

I bet the topology is similar to that in this old thread:

I have learned that these are designed for use with 100 kHz (!) inverters. Is that too fast for IGBTs? The equipment requires 220 volt 100 amp electrical service.
Now looking forward to learning about resonance in HV HF transformer-multiplier systems.
Will try running these dry at a small fraction of their design voltage.
Any ideas where to find a linear power amplifier that can drive a few volts, and a few amps, at 100 kHz?

Sigurthr

Sat Oct 25 2014, 01:47AM

Registered Member #4463 Joined: Wed Apr 18 2012, 08:08AM
Location: MI's Upper Peninsula
Posts: 597

wrote ...
Any ideas where to find a linear power amplifier that can drive a few volts, and a few amps, at 100 kHz?

Look towards CW SSTC drivers. Simple Half-Bridge designs should do fine. No need for complex feedback schemes, just use a 50% duty cycle oscillator fed into the Gate Drivers. These circuits are used for driving nonresonant inductive loads. My own personal designs have topped off at 170V @ 20A @180KHz, using MOSFETs.

Patrick

Sat Oct 25 2014, 02:51AM

Registered Member #2431 Joined: Tue Oct 13 2009, 09:47PM
Location: Chico, CA. USA
Posts: 5639

klugesmith wrote ...

The equipment requires 220 volt 100 amp electrical service.

Wholly crap! can you use this as a ray gun weapon?

Sulaiman

Sat Oct 25 2014, 07:20AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Linear amplifiers;
LM380 can deliver a few watts to well over 100 kHz
TDA2030 simple 5-pin TO220 10W to 100 kHz
LM1876 over 20Wrms to 100 kHz

it's not too difficult to design/build a Class-AB audio-type amplifier for 100 kHz

As above, a hard-switched half- or full-bridge driver

for initial experimants I'd be inclined towards a TO220 TC4421/TC4422 half-bridge
(or the newer TC4451/TC4452)
and experiment with the bridge capacitor(s);
high capacitance for squarewave drive
small capacitance (to be determined, based on frquency and inductance of primary)
to achieve resonance.

klugesmith

Thu Oct 30 2014, 07:50PM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

Time for an update. I have spent my time learning about the transformers, rather than preparing power supplies for them. My posts to this thread are redundant with those at fusor.net.

Starting with the Internet, I found a partial manual for equipment using these tanks.
Bennett HFQ 300, 450, and 600 generators are named for their mAs ratings at 125 kVp.
They are available with Stored Energy option that reduces electrical supply requirement from 220V-100A to 115V-15A.

Theory of Operation says the "Current Amp" units send 100 kHz current (!) to high voltage tanks.

Here is a connection diagram for the smallest configuration.

The -450 and -600 add third and fourth BR-C-Driver modules in parallel, but use the same tanks.
The SE options replace the BR-C's with +/- 211 V battery buses and battery chargers.
The schematic section covers all boards except the tanks and E10136/E10137 drivers.

I would not have thought to use a series capacitor in the primary circuit, as Bennett did. And Thomas in this forum today!

- - -

It was easy to trace the circuit of the anode tank. No voltage multiplication is involved.
The transformer has nine independent secondary windings, each charging a 0.05 uF 7000 V capacitor through a bridge rectifier with 4 discrete diodes.
The capacitors are tied in a series string, with HV output socket on positive end and MA sense connection on negative end.
Secondary winding resistance ranges from 34 ohms on the innermost layer to 41 ohms on the outermost layer.
If the nominal system voltage (125 kVp) is divided equally between cathode and anode, then there's practically no margin in the capacitor voltage rating.

The cathode tank primary winding has DC resistance of 1.3 milliohms and 60 Hz AC impedance of 3.3 milliohms (linear up to 50 A). I think that works out to inductance of 8 microhenries. But then if driven with 400 V, the magnetizing current would change at 50 A/us, which is too high to believe. It was a surprise that 50 A of magnetizing current apparently did not saturate the core. I am about to crank up the test current, after figuring the ampacity of the primary wire.

Ferrite core round sections have diameter of 0.795" (2.02 cm). Core effective length is about 12 inches. If pushed to Bmax of 4000 gauss, I figure it could take about 50 V/turn at 100 kHz or 25 V/turn at 50 kHz. We ought to see signs of saturation at 10 Oe, which would need 250 ampere-turns in the primary.

- - -

I learned plenty after connecting cathode tank primary to a fancy LC meter at work.
Swept the frequency and measured Z and theta. My high school physics teacher, Mr. Koller, taught that when running measurements by hand, chart the data immediately if possible. Then transcription blunders, and interesting regions, can be re-measured or explored with finer resolution.

Below 100 kHz it behaves like an inductor of 8 uH, gratifyingly close to the value figured from DC and 60 Hz milliohms.
A resonant peak at 164 kHz suggests a parallel LC circuit with C = 0.12 uF, perhaps the collective secondary winding capacitance reflected to the primary. I guess the resonant dip at 453 kHz comes from inductance of my interconnection. Surely this ground has been trod many times before -- what do you experts think?

1414698535 2099 FT166826 Primary Z Theta

klugesmith

Sat Nov 01 2014, 04:52AM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

Ta-da! Finally saturated the core of a tanked transformer at 60 Hz, and measured it with the aid of a fluxmeter. In this image, X axis is primary current (100 A/div). Y is flux linkage to a temporary sense winding (0.001 volt seconds per div). Chart can be re-scaled as a BH curve of the core material, after we learn the number of primary turns (7 or 8, I don't yet know for sure).

Before that experiment, I measured the primary wire thickness (rectangular, 0.05 x 0.35 inches) and calculated that 150 A RMS for a full minute would not damage it. That's based on adiabatic heating model. Assume no cooling at all, so the rate of temperature rise is the electric power loss divided by the wire's thermal mass.

Let's illustrate by example. AWG14 building wire can handle 28 A for a minute, 97 A for 5 seconds, 217 A for 1 second, or 2170 A for 10 milliseconds without damage to the insulation.
What those limits have in common is an I^2-t product (current squared x time) of about 47300 ampere-squared-seconds.
One meter of the wire has a resistance of 8.28 milliohms, so total heating, I^2-R-t, is about 392 watt-seconds. The wire weighs 18.6 grams so its temperature will rise about 55 degrees C. That delta-T value makes my I^2-t computations from scratch match tables in the following handy reference. [edit: fixed the link and added another. The same I2T table is replicated in both.]

I^2-t ratings go up as the fourth power of wire diameter, because thicker wire has less resistance and more thermal mass. The flat wire in my transformer primaries is almost equivalent to AWG6. Maybe actually better at high frequency, if derated for skin effect.

Sulaiman

Sat Nov 01 2014, 10:29AM

Registered Member #162 Joined: Mon Feb 13 2006, 10:25AM
Location: United Kingdom
Posts: 3140

Your core looks magic ... no hysteresis visible.
Is there an air-gap ?

klugesmith

Sat Nov 01 2014, 07:53PM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

Thanks for remarking on that, Sulaiman.

The apparent lack of hysteresis was a surprise to me too. Never seen it before, and didn't know if that means the material is really good or magical. Had already planned to repeat and extend the experiment, before holding forth at 4hv about BH curve tracing. Maybe some spurious perturbation is tending to compensate for the effect of hysteresis.

The displayed BH loop was grossly "open" to begin with, when my fluxmeter input was connected directly to primary terminals. That was because of the IR voltage on top of the L*dI/dt voltage. Then I opened the cathode tank for the first time, to wrap a sense winding with known number of turns. Cut my thumb on edge of lid -- first blood for this project.

Back to I2T, just realized why my calc. for 55 degree rise matches the industry table values for 75 degree rise. The tables are based on initial operating temp of 75 deg C and final temp of 150 deg C. So their value for Cu resistivity is about 40% higher than the IACS 20 degree value I was using.

klugesmith

Mon Nov 10 2014, 09:45PM

Registered Member #2099 Joined: Wed Apr 29 2009, 12:22AM
Location: Los Altos, California
Posts: 1716

As a next step, I want to try a monochromatic (sinusoidal) power waveform at high frequency.
Starting at much less than design voltage, to defer the transition from air to oil.

Does anyone here have experience with monolithic amplifiers for car audio systems, such as the TDA7850?
4 channels, 25 W per channel into 4 ohms with low distortion.
In parallel, with some balancing resistors, ought to comfortably put out 100 W into 1 ohm.
It's used in DVD players like one lost in my garage, and in $25 single-board amplifiers (heatsink not included).

Some important details are not found in the datasheet.
Can be found by measurement, but maybe someone here knows the answers and won't mind sharing them.
- matching of the 4 channels
- DC offset voltage at output (not an issue if used with series C)
- Response at frequencies above 20 kHz
- Response at DC (for the lab electromagnet project).

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob