transformer secondary coupling

[tm]

How about taking the easy route and get some *old style neon sign
transformers. I think a standard rating was 15kv at 30ma, so it may take
series/parallel combo to get to 50kv at 50ma. I'll let you pick the
rectifiers.
I ran a couple in series one time and had sparks coming out of the
ground, yikes!
Mikek

The high voltage secondary on a NST is center tapped and grounded to the
case.

You could use a full wave CW multiplier.

 

[Jamie M]

1B3& 3A3 were the most common. 3AW3& 3B2 are a couple of the
others. I have 23 new 3A3 on hand if anyone wants some. 5 are GE
branded, 18 are Westinghouse branded.

Hi,

Is it practical to make a HV fullwave rectifier with 4 thermionic
diodes? I'm not sure how the filament heating supplies would work
due to the floating voltages.

cheers,
Jamie

 

[Jasen Betts]

Hi,

Is it practical to make a HV fullwave rectifier with 4 thermionic
diodes? I'm not sure how the filament heating supplies would work
due to the floating voltages.

They'd need isolation, one could be on each end of the
main HV winding, but for the positive rectifiers you'd
need a fully isolated supply

____ . . . . . .
:__ `--------------------------+-----. |_____
:_ `----------------------------|-----' | . |
:_ | . . . . . . |
:_ . . . . . . | |
:_ .-------------. |______| |
__ :_ | .----------' | . |
3:_ | | . . . . . . |
3:_ | |(+) (-)|
3:_ | | . . . . . . |
__3:_ +--|----------. |_____ |
:_ | +----------' | . | |
:____|__|____ . . . . . . | |
:____|__|__ | | . . . . . . |
:: ____| | | `-----------------|-----. |______|
:_______| `-------------------+-----' | .
. . . . . .

 

[Tim Williams]

Is it practical to make a HV fullwave rectifier with 4 thermionic
diodes? I'm not sure how the filament heating supplies would work
due to the floating voltages.

No -- you really want a doubler. The reason most TVs go half-wave is
because the positive and negative pulses are asymmetrical, so where the
positive peak is 20kV, you only gain maybe 1kV grabbing the negative peak
(which is when the sweep tube / HOT is conducting), not worth installing
another diode to make a doubler. (Triplers and quadruplers have been
variously popular, however.)

Your PSU will work best with a 50% duty cycle into a doubler, since that
maximizes magnetic capacity, conduction angle and minimizes components and
voltage drops.

Doublers are also quite tolerant of leakage inductance (which you will
inevitably have in ample supply) and work well on resonant circuits (where
the peaks are symmetrical, even if the conduction angle isn't as high as a
square wave's).

FYI, you can purchase rectifier stacks from Surplus Sales. Voltage drop is
much less than vacuum style, though switching losses compensate a bit. They
are often specified for oil submersion service.

Tim

 

[Jamie M]

No -- you really want a doubler. The reason most TVs go half-wave is
because the positive and negative pulses are asymmetrical, so where the
positive peak is 20kV, you only gain maybe 1kV grabbing the negative
peak (which is when the sweep tube / HOT is conducting), not worth
installing another diode to make a doubler. (Triplers and quadruplers
have been variously popular, however.)

Your PSU will work best with a 50% duty cycle into a doubler, since that
maximizes magnetic capacity, conduction angle and minimizes components
and voltage drops.

Doublers are also quite tolerant of leakage inductance (which you will
inevitably have in ample supply) and work well on resonant circuits
(where the peaks are symmetrical, even if the conduction angle isn't as
high as a square wave's).

FYI, you can purchase rectifier stacks from Surplus Sales. Voltage drop
is much less than vacuum style, though switching losses compensate a
bit. They are often specified for oil submersion service.

I guess a stack of 600V silicon carbide diodes would work well as they
can have less reverse recovery losses? I guess its pretty easy to make
a diode stack, just series diodes on a PCB (optionally in oil).

cheers,
Jamie

 

[Jamie]

I guess a stack of 600V silicon carbide diodes would work well as they
can have less reverse recovery losses? I guess its pretty easy to make
a diode stack, just series diodes on a PCB (optionally in oil).

cheers,
Jamie

when do that, you should be bridging the string of each diode with a
high value R and small cap, otherwise, the recovery time error could
force you to pay the price.

Jamie

 

[Tim Williams]

when do that, you should be bridging the string of each diode with a
high value R and small cap, otherwise, the recovery time error could force
you to pay the price.

Schottky diodes don't have recovery, but they DO have highly nonlinear
capacitance, which behaves much the same way. Don't forget the caps!

It's too bad they don't make any 0.5 or 1A SiC schottky diodes in DO- cases.
I'd love them for this sort of use. Cree and Infineon only make TO-220 and
related packages, go figure.

Tim

 

[Neon John]

Is it practical to make a HV fullwave rectifier with 4 thermionic
diodes? I'm not sure how the filament heating supplies would work
due to the floating voltages.

Sure. Older X-ray machine use vacuum tube diodes. But that's not
necessary. Newer but still obsolete X-ray machines use solid state
diode sticks. The ones I have are rated at 150,000 PIV at 1 amp. Each
one is a module about 8" long and contains about 15 diodes in series.
I got 'em for a dollar a pop from a local X-ray service company.

John

John DeArmond
http://www.neon-john.com
Tellico Plains, Occupied TN
See website for email address
Feed by Giganews

 

[Neon John]

No -- you really want a doubler.

That seems to be quite a lot of work and parts when it is much easier
to use a modern epoxy-impregnated ignition coil (I use GM HEI coils)
and make the high voltage directly.

For my home-made X-ray machine, my power supply consists of a simple
555 oscillator running at about 1kHz. This drives the gate of a 1200
volt, 40 amp IGBT (what I had laying around). The emitter is grounded
and the collector goes to the ignition coil primary. The other side
of the primary goes to 3 45 volt MeanWell power supplies connected in
parallel with an external filter cap.

The coil is immersed in oil because it will flash over the outside of
the coil otherwise. Also in the oil is one X-ray machine diode stack
and a 0.1uF, 150,000 volt X-ray machine filter capacitor.

I haven't gotten around to measuring the current output but it is
enough to drive a rotating anode X-ray tube and produce exposures
comparable in speed to commercially made machines.

For more current output, coils can be connected in parallel.

John

John DeArmond
http://www.neon-john.com
Tellico Plains, Occupied TN
See website for email address
Feed by Giganews

 

[josephkk]

How are you thinking to avoid being grilled by X-rays? Even the tube
rectifiers generate an appreciable amount of X-rays at those voltage
and power levels.

Rats. You beat me to it. X-ray tube power supplies was worrying my mind
as an alternate way to find a supply. Almost onto the X-ray issues of
E-beam welding.

?-]

 

[Jamie]

Schottky diodes don't have recovery, but they DO have highly nonlinear
capacitance, which behaves much the same way. Don't forget the caps!

It's too bad they don't make any 0.5 or 1A SiC schottky diodes in DO-
cases. I'd love them for this sort of use. Cree and Infineon only make
TO-220 and related packages, go figure.

Tim

Schottky diodes are nice due to their low Vf, they seem to be kind of
hard to get in large Vrb these days. I did recently see some
article on schottky up in the 1200..1500 volt range but I do not know
who is making them atm..

Which brings up another subject. We have at our facility the first
x-linking Irradiation machine, still in operation, ever made! Yes, it's
old and still has most of its original style electronics. We are in the
process of making it into a capital project and updating it and one of
the ideas is to maybe try out a schottky electron emitter instead of the
tungsten.

Should be interesting to see what chances will be needed to the beam
section.

Jamie

 

[Jamie M]

How are you thinking to avoid being grilled by X-rays? Even the tube
rectifiers generate an appreciable amount of X-rays at those voltage
and power levels.

Rats. You beat me to it. X-ray tube power supplies was worrying my mind
as an alternate way to find a supply. Almost onto the X-ray issues of
E-beam welding.

?-]

Hi,

thin lead sheet can block 99.9%+ of 50kV and lower x-rays afaik!

cheers,
Jamie

 

[Tim Williams]

Schottky diodes are nice due to their low Vf, they seem to be kind of hard
to get in large Vrb these days. I did recently see some
article on schottky up in the 1200..1500 volt range but I do not know
who is making them atm..

Pretty much Cree and Infineon, and I read something about Dow and a couple
others growing the SiC wafers (raw material, not the finished chips).

Note that schottky diodes suffer increased losses when constructed for
higher breakdown. A 40V diode might drop 0.7V at rated current, while an
80V diode might drop 0.9V. A comparable (100V) junction diode might drop
1.0V, so the conduction savings is modest. "Super barrier" schottkies go up
to 300V, using some interesting MOS structures to improve breakdown; these
structures increase the voltage drop further, so that a 300V schottky has
essentially identical conduction losses to a 400V junction diode.

Anything beyond there (600V, 1200V and more) is SiC at this time. As far as
I know, these are fabricated pretty much like a 40V Si schottky, but with
accordingly higher bandgap and breakdown, the voltage drop is much higher.
Resistivity is particularly high, so much so that a SiC schottky looks worse
than a high speed junction diode in most uses (Vf ~ 2V at rated current,
IIRC).

The main advantages to schottky are for rectification under 20V, and high
frequency (over 500kHz or so), where recovery losses dominate.

When evaluating a design, also keep in mind that schottkies have
tremendously higher capacitance. A low voltage (50-200V), high speed
junction diode can recover faster (~50ns) than a schottky takes to recharge
(could be >100ns at < half rated current); this, as well as the strong
nonlinearity, can actually worsen performance.

Tim

 

[josephkk]

Schottky diodes are nice due to their low Vf, they seem to be kind of
hard to get in large Vrb these days. I did recently see some
article on schottky up in the 1200..1500 volt range but I do not know
who is making them atm..

Not normal schottky diodes then, maybe SiC versions or not schottky at
all.

Which brings up another subject. We have at our facility the first
x-linking Irradiation machine, still in operation, ever made! Yes, it's
old and still has most of its original style electronics. We are in the
process of making it into a capital project and updating it and one of
the ideas is to maybe try out a schottky electron emitter instead of the
tungsten.

Wow, that should be a museum piece instead.