HV isolation DC/DC conversion - please help

[JT]

I have here the highest isolation DC/DC converter IC I could find: the
6kVDC isolation DC/DC converter NMS0515 (5Vin, 15Vout, 67mA output),
and it's giving me a headache.

I need 15Vdc and about 100mA with 30kV isolation from the input. The
thing is, I will only need that type of isolation for 100us at a 1kHz
rep. rate. Analogous to Peak power vs. RMS, is there not some finite
electrical break-down time required? How does it scale?

Thanks,

JT

 

[Dave]

JT wrote:

I have here the highest isolation DC/DC converter IC I could find: the
6kVDC isolation DC/DC converter NMS0515 (5Vin, 15Vout, 67mA output),
and it's giving me a headache.

I need 15Vdc and about 100mA with 30kV isolation from the input. The
thing is, I will only need that type of isolation for 100us at a 1kHz
rep. rate. Analogous to Peak power vs. RMS, is there not some finite
electrical break-down time required? How does it scale?

Thanks,

JT

If its rated at 6kV, I woudl not attempt to run it beyond that even for

short periods. Things like resistors and transformers have ratings based
on thermal limits (although resistors sometimes voltage too). But here,
the situation is very different.

Can't use just get a transformer wound and use a conventioal
bridge/capacitor/regulator? I don't think getting a transformer woulnd
to do that would be hard. Transformers are regularly made to produce
much hgiher than 30kV, so one could be wound. But it would not be cheap.

Is it a one-off experiement, or something expected to last? If the
former, and the consequences of failure are not large, then perhaps just
taking a chance might be sensible. But 30 kV is a lot more than 6kV, so
I would not have too much hopes.

 

[JT]

Yeah, this is for a more permanent installation. The conventional
transformer/bridge/cap was the next option. But I'm going to need 30.
That's the problem. I was thinking small ferrite toroidal cores, each
wound with a bunch of turns for their secondaries. Then I could loop a
30kV insulated wire through them all and charge pump it. But who
supplies cores and winding services?

Unless I can find off-the-shelf small form factor isolation
transformers. Any ideas?

 

[Mochuelo]

On 8 Aug 2005 16:02:57 -0700, "JT" <jt.steenkamp@usask.ca> wrote:

I have here the highest isolation DC/DC converter IC I could find: the
6kVDC isolation DC/DC converter NMS0515 (5Vin, 15Vout, 67mA output),
and it's giving me a headache.

I need 15Vdc and about 100mA with 30kV isolation from the input. The
thing is, I will only need that type of isolation for 100us at a 1kHz
rep. rate. Analogous to Peak power vs. RMS, is there not some finite
electrical break-down time required? How does it scale?

Thanks,

JT

Are you sure you need to transfer energy across the isolation
interface, or only information? If the latter one, you could feed the
receiver with a battery, and use fiber optic at the information
channel, or just a homemade optical link.

I find it curious that you need isolation only during 10 % of the
time. Maybe with more information we could find a solution not by
answering your question, but by approaching the problem in some other
way.

 

[JT]

The project is an IGBT switched Marx-generator for plasma ion
implantation. It charges caps in parallel and switches them in
series. Hence, you get a HV step-up only during the pulse.

I do infact use fiber optics to control the switching. It's driving
the IGBT's that need energy and the drivers need to float in between
the caps during the pulse.

 

JT <jt.steenkamp@usask.ca> wrote:

I have here the highest isolation DC/DC converter IC I could find: the
6kVDC isolation DC/DC converter NMS0515 (5Vin, 15Vout, 67mA output),
and it's giving me a headache. I need 15Vdc and about 100mA with 30kV
isolation from the input.

How about a solar cell and a headlight bulb? Crap efficiency, but
isolation off the charts - you've got a few inches of vacuum (and/or gas
fill) and at least an eighth of an inch of glass working for you, and
then you can put the solar cell further away for more isolation (at a
reduced power output).

Matt Roberds

 

[Robert Baer]

JT wrote:

Yeah, this is for a more permanent installation. The conventional
transformer/bridge/cap was the next option. But I'm going to need 30.
That's the problem. I was thinking small ferrite toroidal cores, each
wound with a bunch of turns for their secondaries. Then I could loop a
30kV insulated wire through them all and charge pump it. But who
supplies cores and winding services?

Unless I can find off-the-shelf small form factor isolation
transformers. Any ideas?

For a measly 30, DIY winding is far cheaper than paying a service.

 

[Robert Baer]

JT wrote:

The project is an IGBT switched Marx-generator for plasma ion
implantation. It charges caps in parallel and switches them in
series. Hence, you get a HV step-up only during the pulse.

I do infact use fiber optics to control the switching. It's driving
the IGBT's that need energy and the drivers need to float in between
the caps during the pulse.

Nominally, a DC supply is used to do the charging; high voltage

resistors (or resistor chains) are used to connect the plates in parallel.
Then at the ground electrode, a small arc is produced to break down
the series capacitive chain.
This small arc is usually configured as a coaxial insulated wore in
the ground electrode; the tip of the wire at the "tip" of this main
electrode.
Light pipes may be used to transport the UV from the initiating spark
to the other gaps to speed up and help initiate ionization.
No isolation is needed.
Check out how a standard Marks generator is configured.

 

[Joerg]

Hello JT,

If this needs to look professional and you really have to make 30 looks
at series resonant conversion. The "air gap" could be 1/4", 1/2", or
whatever you need. But there doesn't have to be air in the gap, you
could use a nice insulating material.

Probably pot core halves would be easiest here. For ease of winding you
can often buy split bobbins for these. Another hint: De-burr everything.

Regards, Joerg

http://www.analogconsultants.com

 

[JT]

Replace the sparkgaps with IGBT's then look at IGBT switching
requirements and you'll see what I mean.

 

[JT]

Good idea, I considered this too. Unfortunately, when considering
spacial constraints, number of solar cells needed, power requirement
per unit, I think this one's out the window.

 

[JT]

Thanks Joerg,

I'll look into it.

JT

 

JT <jt.steenkamp@usask.ca> wrote:

Unfortunately, when considering spacial constraints, number of solar
cells needed, power requirement per unit, I think this one's out the
window.

How about these:

- Use a battery in series with a phototransistor as the "output" side,
and an LED as the "input" side. Perhaps use rechargeable batteries
and recharge them when the equipment is otherwise idle, then open
the recharge circuit with a high-isolation switch when the equipment
is in use.

- Use a "flying capacitor" - charge up a cap from the "input" side
and then use a DPDT relay or equivalent transistors to switch it
over to the "output" side.

Matt Roberds

 

[JT]

Thanks for the suggestions Matt. In the short term, I think
batteries are the way to go. As for the flying cap - I like it, but
I ran some numbers, and for a 1h experiment, I'm going to need a
1.68F, 18V cap for every switch. I found some at Vishay, physically
not too big. I'll have to think a little bit about the most
convenient charging method...But a 1.68F cap gives me the creeps.
Isn't this all a little unorthodox?

 

[Frithiof Andreas Jensen]

"JT" <jt.steenkamp@usask.ca> wrote in message
news:1123554166.902574.317160@g49g2000cwa.googlegroups.com...

The project is an IGBT switched Marx-generator for plasma ion
implantation. It charges caps in parallel and switches them in
series. Hence, you get a HV step-up only during the pulse.

I do infact use fiber optics to control the switching. It's driving
the IGBT's that need energy and the drivers need to float in between
the caps during the pulse.

Stpid Question:

Why is it not possible to run the IGBT drivers off the not-so-HV voltage
present across each IGBT switch? Presumably, you need bleeders.

If not, you can stack transformers with lower isolation ratings and feed the
one on ground potential. The advantage is that the ground capacitance
decreases up the stack, which is probably what you want, and
losses/regulation becomes progressively worse.

This is easiest done by toroids and HV-Wire - the supply is taken off the
toroid, one loop of HV wire goes throught the toroid for the next stage and
on loop comes from the lower stage. You end up with two loops of wire
through each core except the top one (and you just might want to connect the
conductor of each HV cable loop to a potential betweent the two stages, to
deal with voltage distribution and providing a convenient place to split the
primary when something blows up).

Making the secondaries resonant can help to improve regulation.

You Understand??


Alternatively, use a flyback topology with a primary winding made out of HV
wire and the many secondaries being ring cores. Good regulation, but
capacitance will hurt more and more as you go up the stack.

In any case, I would buy some ready-made toroid inductors and design around
those for a first shot.

 

[Joerg]

Hello JT,

Thanks Joerg,

I'll look into it.

Unitrode App Notes are IMHO the best in that respect. Now via Texas
Instruments.

Regards, Joerg

http://www.analogconsultants.com

 

JT <jt.steenkamp@usask.ca> wrote:

In the short term, I think batteries are the way to go.

If you're going to run it for a long time, rechargeables are probably
best, but an alkaline D-cell is something like 15 Ah @ 1.5 V at low load
currents. Since you need 0.1 Ah, you could run your experiment for
over six days straight on a set of batteries. A set would be 300
batteries at around $1 each. Or, you could use carbon-zinc cells for
about half the capacity and half the cost.

As for the flying cap - I like it, but I ran some numbers, and for
a 1h experiment, I'm going to need a 1.68F, 18V cap for every switch.

See your local car stereo shop. Theirs may only be rated to 15 V, though.
If you spend a few more bucks you can get one with a l33+ bar-graph
voltmeter on top.

I found some at Vishay, physically not too big.

Cooper Powerstor capacitors might also work, except that you'll have to
stack them up as they are only 2.5 V each.

I'll have to think a little bit about the most convenient charging
method...But a 1.68F cap gives me the creeps.

If you don't like the cap, I just had another goofy idea. How about a
small permanent magnet motor being run as a generator. Couple the motors
to a common shaft and use a small AC induction motor to drive the lot.

Isn't this all a little unorthodox?

Probably. I think the only other ways to go are to wind your own
transformers, as has been discussed, or change the design.

Matt Roberds

 

[Robert Baer]

JT wrote:

Thanks for the suggestions Matt. In the short term, I think
batteries are the way to go. As for the flying cap - I like it, but
I ran some numbers, and for a 1h experiment, I'm going to need a
1.68F, 18V cap for every switch. I found some at Vishay, physically
not too big. I'll have to think a little bit about the most
convenient charging method...But a 1.68F cap gives me the creeps.
Isn't this all a little unorthodox?

As the doctor might say, take two Farads a day...

 

You can drive IGBTs directly from a tramsformer, some experimenting is
of course needed. Have a look at how a conventional current transformer
works, with luck a dazzling flash of inspiration will hit you.

 

[Ken Smith]

In article <3p_Je.3017$Wi6.2188@newsread2.news.pas.earthlink.net>,
Robert Baer <robertbaer@earthlink.net> wrote:
[...]

Unless I can find off-the-shelf small form factor isolation
transformers. Any ideas?

For a measly 30, DIY winding is far cheaper than paying a service.

You use the *wrong* coil winder or you aren't nearly well enough paid.



--
--
kensmith@rahul.net forging knowledge