DC-DC conversion from 1 Kilovolt DC down to 12VDC (isolated)

Don't know this company but you might want to contact them about it. If the don't have it maybe they can tell you who does:

Also, keep in mind that AC input switchers can generally also be used for a DC input.

If you don't find anything and you must roll your own then the best bet might be a half-bridge forward converter. Unless the input voltage level fluctuates a lot. If you don't want to use HV-FETs you can cascade them but that can become iffy during fault conditions. Check out the ST PowerMesh series:

Have fun, and wear protective gear in case something goes *KAPOW* :-)

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

google find something like this:

et_Railway.php#10

if you put the output of multiple converters in series wouldn't they automatically share the load?

-Lasse

--
I'd connect the 1KV DC to the center tap of the primary of a
transformer and switch the ends push-pull to get the AC needed, then
I'd rectify, filter, and regulate the output from the secondary to get
my 12V.

You're in around the right voltage range for a 600VAC-input supply with a bit of margin.

I've never used them, but these guys claim they can do up to 1200VDC input:-

On a sunny day (Wed, 22 Jun 2011 12:10:46 -0500) it happened John Fields wrote in :

You are aware that if one tap goes to zero the other goes to *two* kV?

On a sunny day (Wed, 22 Jun 2011 08:39:34 -0700 (PDT)) it happened Terry wrote in :

Even in the 198ties this was no problem. I did something like that, but not all the way up to 1000 V, with 2 TV output transistors (good for 10A / 1500 V each).

+HV |______________________ | | T1 === | | ------- transformer --- | ======== | T2 secundary === | | | | | /// diode /// diode /// | | ---------- | L | === --- | /// I wound my own transformer on a big EI core. Same for the inductor L, needs airgap.

To drive those transistors I wound some transformers on potcores, added some RC in the base to speed things up, and a current sense in the emitter of T2. Used some chip for the driver, cannot remember which one, but had to add flip flops to get the correct drive waveforms. These days a PIC can do all that :-) (Joerg now panics).

Why choose that topology? It subjects the transistors to double the input voltage.

A half-bridge inverter would be soooo much easier.

--
Yup, my bad. Thanks! :-)

Just curious, but where do you get the 1KV dc supply from?

John

Well, there's DC transmission-line switchgear to use as a model. Basically, the HV is switched by pulse-triggering (transformer input or optocoupled) SCRs that are stacked in series to achieve the desired voltage standoff. The rest is straightforward AC handling (step-down transformers and such). A few HV capacitors are required (they block the current when they charge up in the (+) half-cycle, which is why the SCRs turn off).

No, they do the opposite, i.e., try to compete as to who's "in charge," and end up blowing each other up.

Sorry. Rich

Rich is right- the input of a SMPS with a constant load looks like a negative resistance- so, say A & B are in series with equal voltage input and with output a constant load, and the voltage goes up a tad on A, then the current on A drops, so the voltage goes up a bit more... eventually (milliseconds) A fails (by short most likely) and the full voltage appears across B..

Go to the Miller welding site and download a manual for a DC inverter welder that runs on 480 volts. They take the 600+ volts and switch it down with half bridge pair to about 30 volts at many amps.

Here is the manual for one that runs on 480 three phase:

That may give you some ideas.

tm

2kV is not that bad. A couple of layers of 500v/mil insulation will do. And SCR drive. Too bad there are no small GTO's availble.

Cheers

Pico electronics Their website is down, give them a call.

Cheers

Kind of like what gang members do :-)

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

Not really a good idea.

Just split the primary, connect extra switch transistors between the sections (so everything is in series). Drive with multi winding gate/ base transformer, keep drive waveform 50/50 ish. Isolation is down to transformer design, there are bobins with seperate concentric sections available.

I beg to differ :-)

... and that's where the problem is with a split primary. "ish" isn't enough. Veer off that 50/50 only ever so slightly and *PHUT* ... *BOOM*

On a 1kV DC the result won't be pretty and fusing that much DC isn't trivial with catalog parts.

Either that or get certified wire from Rubadue or other places like that. There's also pretty good insulating material these days that can be wrapped around the primary. In the old days we used oil paper and stuff like that. Just like we used hemp and pipe dope when mounting faucets and nothing ever like (unlike today).

Winding prim/sec on top of each other has the benefit that you can have a shield foil (or shield winding if you must) between the two.

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

But Jan Panteltje's geometry works OK, with a flying capacitor +HV |______________________ | | T1 === | | ------- transformer --- | ======== | T2 secundary === | | | | | /// diode /// diode /// | | ---------- | L | === --- | ///

and if you replace T1 and T2 with series pairs of triac-output optocouplers (600V each) the whole thing kinda works. Bypass the triacs with RC so the voltages divide even. Of course, you need to excite the LEDs with some suitable pulses, but maybe a rechargeable battery (the output was 12VDC, right?) would suit the application.