Win's next 10kV project, a 1us ramp

[Joerg]

Hello John,

Only if you honor abs max specs. I'd try it at 10KV. There's a lot of
vacuum inside one of those jugs.

Be careful. I once built a tube RF amp and being a student it had to be
ultra low budget. So I let the finals have about 50% higher plate
voltage than their data sheet said. One of them arced over. Big blue
flash. That took the plate transformer, diodes and other stuff with it.

Regards, Joerg

 

[Winfield Hill]

Joerg wrote...

Hello Win,


You can get very large cores up to almost three pounds of weight,
just have to make sure it's available in the right material:
http://www.fair-rite.com/newfair/pdf/LT.pdf#page=136

Wow, up to 6-inch diameter by 3/4-inch thick. Nice.

 

[Winfield Hill]

Robert Baer wrote...

For very low leakage inductance and have good HV insulation,
use coax cable for the toroid winding, splitting the shield
primary for parallel input and 1:N ratio where N is the splits.

I can't quite picture what you're saying, can you expand? What
do you mean, splitting the shield? Are you saying the coax will
have 10kV across its dielectric for the last primary section?

 

[Winfield Hill]

John Larkin wrote...

Only if you honor abs max specs. I'd try it at 10KV. There's
a lot of vacuum inside one of those jugs.

OK, where do I get an 833A datasheet? Where does one get 833As?

 

[Winfield Hill]

Winfield Hill wrote...

John Larkin wrote...

OK, where do I get an 833A datasheet? Where does one get 833As?

OK, I see, Svetlana. Hmm, I wonder what the plate characteristics
look like extended to 10kV. And I note the positive grid voltage
(+100V, to 200mA grid current) to get 1A plate currents below 1kV.
The performance curves aren't nearly as elegant as the glass tube!

 

[Fred Bloggs]

Robert Baer wrote...



I like the constant-current into a capacitive load that a
flyback coil / transformer can deliver. If the transformer
has a 10:1 step-up ratio I could use 1kV MOSFETs to switch
5 to 10A on the primary... But I keep worrying about the
transformer's leakage inductance. If the secondary + load
was say 60pF total, then that'd have to be less than 17uH
for a say 5MHz bandwidth. Hmm, 17uH doesn't look so hard.

Once you get the transformer close, you can remove the shortfalls with a
little feedback.
View in a fixed-width font such as Courier.

..
.. 1KV
.. |
.. .----+---. ^
.. | | / |
.. [R4] [R5] / |
.. | | / |
.. | + | / |
.. +--||----+ .-----+------------. -- |
.. | C3 | | | | |/
.. | | (o | |
.. ||- )||( | CL ===
.. .----||< )||( [R3] |
.. | ||- o)||( | |
.. | | | ( | |
.. | | | | | |
.. | --+--------+--+------------------'
.. | 1:10 |
.. | |
.. | .--[R1]--. |
.. | | | |
.. | +---||---+ | comparator
.. | | C2 | |
.. | | /| | C1 | |\
.. | /| | /-|-+--||----+------|+\
.. | /+|--+---< | | | >---.
.. '-< | \+|-. | .-|-/ |
.. \-|--. \| | [R2] | |/ |
.. \| | | | | |
.. | --- --- THRESH |
.. 0+ - ^ - |
.. | | |
.. | | |
.. | | terminate ramp |
.. -Vprog ---- <----------------------'
..
..

 

[Winfield Hill]

Fred Bloggs wrote...

Robert Baer wrote...


I like the constant-current into a capacitive load that a
flyback coil / transformer can deliver. If the transformer
has a 10:1 step-up ratio I could use 1kV MOSFETs to switch
5 to 10A on the primary... But I keep worrying about the
transformer's leakage inductance. If the secondary + load
was say 60pF total, then that'd have to be less than 17uH
for a say 5MHz bandwidth. Hmm, 17uH doesn't look so hard.

Once you get the transformer close, you can remove the shortfalls
with a little feedback.
.
. 1KV
. |
. .----+---. ^
. | | / |
. [R4] [R5] / |
. | | / |
. | + | / |
. +--||----+ .-----+------------. -- |
. | C3 | | | | |/
. | | (o | |
. ||- )||( | CL ===
. .----||< )||( [R3] |
. | ||- o)||( | |
. | | | ( | |
. | | | | | |
. | --+--------+--+------------------'
. | 1:10 |
. | |
. | .--[R1]--. |
. | | | |
. | +---||---+ | comparator
. | | C2 | |
. | | /| | C1 | |\
. | /| | /-|-+--||----+------|+\
. | /+|--+---< | | | >---.
. '-< | \+|-. | .-|-/ |
. \-|--. \| | [R2] | |/ |
. \| | | | | |
. | --- --- THRESH |
. 0+ - ^ - |
. | | |
. | | |
. | | terminate ramp |
. -Vprog ---- <----------------------'

I don't understand how you get a ramp. It's not a flyback,
how do you get a constant current into the load capacitance?
Is that supposed to be a 10MHz bandwidth servo loop for the
high-current MOSFETs gate? BTW, 10A 1kV MOSFETs, such as a
mtw10n100E, have Ciss = 3500pF. Ouch! I wouldn't go there!

 

[Fred Bloggs]

Fred Bloggs wrote...

Robert Baer wrote...


I think i suggested a flyback transformer.
Adjusting the supply voltage on the inductor (primary) can
adjust the slope / slew rate; maybe to 1% without too much
trouble and the amplitude can be controlled by the gated
time - again (perhaps) within 1%. One can easily have
multiple kV insulation between primary and secondary (ies).
And multiple drivers are not that big of a deal either.

I like the constant-current into a capacitive load that a
flyback coil / transformer can deliver. If the transformer
has a 10:1 step-up ratio I could use 1kV MOSFETs to switch
5 to 10A on the primary... But I keep worrying about the
transformer's leakage inductance. If the secondary + load
was say 60pF total, then that'd have to be less than 17uH
for a say 5MHz bandwidth. Hmm, 17uH doesn't look so hard.

Once you get the transformer close, you can remove the shortfalls
with a little feedback.
.
. 1KV
. |
. .----+---. ^
. | | / |
. [R4] [R5] / |
. | | / |
. | + | / |
. +--||----+ .-----+------------. -- |
. | C3 | | | | |/
. | | (o | |
. ||- )||( | CL ===
. .----||< )||( [R3] |
. | ||- o)||( | |
. | | | ( | |
. | | | | | |
. | --+--------+--+------------------'
. | 1:10 |
. | |
. | .--[R1]--. |
. | | | |
. | +---||---+ | comparator
. | | C2 | |
. | | /| | C1 | |\
. | /| | /-|-+--||----+------|+\
. | /+|--+---< | | | >---.
. '-< | \+|-. | .-|-/ |
. \-|--. \| | [R2] | |/ |
. \| | | | | |
. | --- --- THRESH |
. 0+ - ^ - |
. | | |
. | | |
. | | terminate ramp |
. -Vprog ---- <----------------------'

I don't understand how you get a ramp. It's not a flyback,
how do you get a constant current into the load capacitance?

The servo regulates output dV/dt to Vprog. The FET discharges C3 through
the primary at relatively constant current.

Is that supposed to be a 10MHz bandwidth servo loop for the
high-current MOSFETs gate? BTW, 10A 1kV MOSFETs, such as a
mtw10n100E, have Ciss = 3500pF. Ouch! I wouldn't go there!

That's doable- the only high frequency response required is at turn on,
and from there on out it is relatively low band tracking compensation
for the build up of magnetizing current and IR drop in the primary.

I like it because it measures the most important parameter directly, and
it does leave open the possibility of equalization of the fine details
via the Vprog input and an AWG, which even at 500MHz are relatively low
tech these days.

 

[John Larkin]

Winfield Hill wrote...

OK, I see, Svetlana. Hmm, I wonder what the plate characteristics
look like extended to 10kV. And I note the positive grid voltage
(+100V, to 200mA grid current) to get 1A plate currents below 1kV.
The performance curves aren't nearly as elegant as the glass tube!

I'm sure there are other transmitting tubes with higher rated plate
voltages, and maybe lower max current. Probably ceramic, more rugged
but not as pretty.

John

 

[Fred Bloggs]

I'm sure there are other transmitting tubes with higher rated plate
voltages, and maybe lower max current. Probably ceramic, more rugged
but not as pretty.

John

Oh yeah- for sure- check Eimac. Some decades ago I used one to generate
a 1A current pulse into a 20KV load- the tube plate dropped to 1.5KV at
this current, gird control voltage in the 350V range. It was a $600 tube.

 

[Ancient_Hacker]

Look up any of the radar modulator tubes. The old WW2 715-ABC can
handle like 3 AMPS at 20KV. At a short duty cycle of course. They're
relatively cheap. Still being made under some other number IIRC.

 

[Michael A. Terrell]

OK, where do I get an 833A datasheet? Where does one get 833As?

http://broadcast.rell.com/tubes.asp is a good source for high power RF
components. I'll bet that you can find something there.

--
Service to my country? Been there, Done that, and I've got my DD214 to
prove it.
Member of DAV #85.

Michael A. Terrell
Central Florida

 

[Fred Bloggs]

Fred Bloggs wrote...

Robert Baer wrote...


I think i suggested a flyback transformer.
Adjusting the supply voltage on the inductor (primary) can
adjust the slope / slew rate; maybe to 1% without too much
trouble and the amplitude can be controlled by the gated
time - again (perhaps) within 1%. One can easily have
multiple kV insulation between primary and secondary (ies).
And multiple drivers are not that big of a deal either.

I like the constant-current into a capacitive load that a
flyback coil / transformer can deliver. If the transformer
has a 10:1 step-up ratio I could use 1kV MOSFETs to switch
5 to 10A on the primary... But I keep worrying about the
transformer's leakage inductance. If the secondary + load
was say 60pF total, then that'd have to be less than 17uH
for a say 5MHz bandwidth. Hmm, 17uH doesn't look so hard.

Once you get the transformer close, you can remove the shortfalls
with a little feedback.
.
. 1KV
. |
. .----+---. ^
. | | / |
. [R4] [R5] / |
. | | / |
. | + | / |
. +--||----+ .-----+------------. -- |
. | C3 | | | | |/
. | | (o | |
. ||- )||( | CL ===
. .----||< )||( [R3] |
. | ||- o)||( | |
. | | | ( | |
. | | | | | |
. | --+--------+--+------------------'
. | 1:10 |
. | |
. | .--[R1]--. |
. | | | |
. | +---||---+ | comparator
. | | C2 | |
. | | /| | C1 | |\
. | /| | /-|-+--||----+------|+\
. | /+|--+---< | | | >---.
. '-< | \+|-. | .-|-/ |
. \-|--. \| | [R2] | |/ |
. \| | | | | |
. | --- --- THRESH |
. 0+ - ^ - |
. | | |
. | | |
. | | terminate ramp |
. -Vprog ---- <----------------------'

I don't understand how you get a ramp. It's not a flyback,
how do you get a constant current into the load capacitance?
Is that supposed to be a 10MHz bandwidth servo loop for the
high-current MOSFETs gate? BTW, 10A 1kV MOSFETs, such as a
mtw10n100E, have Ciss = 3500pF. Ouch! I wouldn't go there!

Maybe it's not so much what you turn on but what you turn off. Here's a
way to get around an unattainable turn-on characteristic, bit it throws
more farads into C3 requirement.
View in a fixed-width font such as Courier.

..
.. 1KV
.. |
.. .----+---. ^
.. | | / |
.. [R4] [R5] / |
.. | | / |
.. | + | / |
.. +--||----+ .----+-------+------------. -- |
.. | C3 | | | | | |/
.. | | (o --- | |
.. ||- )||( | | | CL ===
.. .----||< )||( | | [R3] |
.. | ||- o)||( --- | |
.. | | | ( | TOE | |
.. | | | | | | |
.. | --+--------+--+----+--------------------'
.. | 1:10 |
.. | .------'
.. | .--[R1]--. |
.. | | | |
.. | +---||---+ | comparator
.. | | C2 | |
.. | | /| | C1 | |\
.. | /| | /-|-+--||----+------|+\
.. | /+|--+---< | | | >---.
.. '-< | \+|-. | .-|-/ |
.. \-|--. \| | [R2] | |/ |
.. \| | | | | |
.. | --- --- THRESH |
.. 0+ - ^ - |
.. | | |
.. | | |
.. | | terminate ramp |
.. -Vprog ---- <----------------------'
..
..
..
.. TOE
..
.. latch switch off after 1A + Delay
.. ---+-------------------------
.. |
.. o /|
.. |- - < |--------+
.. o \| |
.. | |
.. +--------|\ |
.. | | >-[R]-+
.. [1] 1V-|/ |
.. | ===
.. | C |
.. ---+----------------+---------
..
..
..

 

[Ken Smith]

I can't quite picture what you're saying, can you expand? What
do you mean, splitting the shield? Are you saying the coax will
have 10kV across its dielectric for the last primary section?

I've never done this at 10KV but this is what I think he means to do:


============= =============== ===============
Center =============================================================
============= =============== ===============
A B A B A B

You get a 3:1 turns ratio if you put this on a core and hook the "A"s
together and the "B"s together.

 

[Tim Williams]

OK, where do I get an 833A datasheet? Where does one get 833As?

Tube G33ks to the rescue
http://www.mif.pg.gda.pl/homepages/frank/
Frank has a big database, and Duncan has several connected, including Frank.
http://tdsl.duncanamps.com/tubesearch.php

Tim

 

[Robert Baer]

Robert Baer wrote...



I can't quite picture what you're saying, can you expand? What
do you mean, splitting the shield? Are you saying the coax will
have 10kV across its dielectric for the last primary section?

Just for visualization, assume one had ten inches of coax in the
winding and running left to right.
Cut or part the shield every inch, giving 11 lengths of shield; the
left ends connected together and the right ends connected together -
making the primary.
The inner conductor is the secondary and one then could have 1:11
turn ratio.
Max voltage difference would be from low end of primary to "far" end
of secondary, so the coax should be rated for that value.

 

[Robert Baer]

I've never done this at 10KV but this is what I think he means to do:


============= =============== ===============
Center =============================================================
============= =============== ===============
A B A B A B

You get a 3:1 turns ratio if you put this on a core and hook the "A"s
together and the "B"s together.

Yup!

 

[Robert Latest]

On 7 Jan 2006 05:50:21 -0800,

However, we need a repeatable ramp,
hopefully without significant time jitter, since we'll be signal-
averaging the data from thousands of scans.

What is it you're measuring?

robert

 

[YD]

John Larkin wrote...

OK, where do I get an 833A datasheet? Where does one get 833As?

http://www.mif.pg.gda.pl/homepages/frank/sheets/049/8/833A.pdf

- YD.

 

[Winfield Hill]

Ancient_Hacker wrote...

Look up any of the radar modulator tubes. The old WW2 715-ABC can
handle like 3 AMPS at 20KV. At a short duty cycle of course. They're
relatively cheap. Still being made under some other number IIRC.

I found 715B and 715C datasheets. That's an impressive tube!