off-line SMPS const current 1-2kW

[Spehro Pefhany]

How would you suggest approaching this? Need a few hundred volts DC at
<5-10A, 5-10% ripple. Need galvanic isolation. Essentially a resistive
load.

Best regards,
Spehro Pefhany

 

[mook johnson]

Do you need a constant current output programmable from 5 - 10 amps with
100+ volts of compliance? Or do you need a current mode DC output supply
that is capable of achieving 5-10 amps?

How much skill do you have? What package does it need to fit in? How much
efficiency is required? Budget?

There are some good tech notes and ICs on the TI website design such a
beast.

Also look at www.Vicr.com for some modules that may do it for you.


P.S.
you should do a better job of hiding you email name. Its listed in your
signature. Prepare for the spammers.

How would you suggest approaching this? Need a few hundred volts DC at
<5-10A, 5-10% ripple. Need galvanic isolation. Essentially a resistive
load.

Best regards,
Spehro Pefhany

http://www.speff.com

 

[CBarn24050]

It depends wether you want to do it the easy but expensive way, or spend money
on a cost efficient design.

 

[Spehro Pefhany]

It depends wether you want to do it the easy but expensive way, or spend money
on a cost efficient design.

More the former than the latter. Small volume. Has to run at a
fraction of full load 24/7.

Best regards,
Spehro Pefhany

 

[Winfield Hill]

Spehro Pefhany wrote...

How would you suggest approaching this? Need a few hundred volts DC at
<5-10A, 5-10% ripple. Need galvanic isolation. Essentially a resistive
load.

If you need only one, an easy solution is to haunt eBay for a few
months, looking for an appropriate Sorensen or Zantrex programmable
switching supply, going for about $300 to $400 per kW.

For example, a Zantrex XHR 300-3.5 does up to 300V and up to 3.5A.
http://www.xantrex.com/products/product.asp?did=223&p=9
http://www.xantrex.com/support/docdisplay.asp?prod=56

Their XFR 300-9 handles up to 9A. Also, made in Canada, I think.
http://www.xantrex.com/support/docdisplay.asp?prod=55

If you want to build your own, the construction techniques in these
supplies bears examination. A PFC input stage to 340V or so, then
a phase-shift-modulated zero-crossing soft-switching PWM H-bridge
driving the dc-dc transformer, ucc3895, etc. Big low-loss diodes on
the output. An elegantly-spare cooling system, with paper shrouds.
Hmm, probably at least a month of engineering time to start. :>)

Thanks,
- Win

whill_at_picovolt-dot-com

 

[CBarn24050]

It's fairly easy to do a push pull supply, if you need pfc then that
complicates things quite a bit. For very low volumes it's cheaper to buy a
ready made unit.

 

[Tony Williams]

More the former than the latter. Small volume. Has to run at a
fraction of full load 24/7.

Small volume? Buy a programmable System Supply
off the shelf Speff. The price might look high
at first glance, but I doubt you could develop
even a few for the same money..... and all the
various regulations (UL, etc) are done for you.

A Lambda Genesys GEN150-10 will do up to 150V/10A,
or a GEN300-5 to get 300V/5A. Programmable via the
F.Panel, or by resistors plugged into the back, or
over an RS232 link.

I played with a GEN80-19 a few months ago (RS232
control in my case) and it worked ok. Current in
CI-mode was quite stable.

Note though that all switchers have a hefty output
capacitor, so the constant current is only valid
for very slow load variations.

 

[John Woodgate]

I read in sci.electronics.design that Spehro Pefhany <speffSNIP@interlog

How would you suggest approaching this? Need a few hundred volts DC at
<5-10A, 5-10% ripple. Need galvanic isolation. Essentially a resistive
load.

I'd buy one, since I don't design SMPS, and this power range is not the
one to learn at. But if I could accept the size and mass, I'd have a
non-switching supply working in a few hours.

 

[Frithiof Andreas Jensen]

How would you suggest approaching this?

Get some better requirements *before* asking:

What is it *for* first of all?
What kind of supply does it run *from* - single phase, three phase, do we
care about PFC?
What sort of control would you like, fixed, manual, programmable?
How do you like weight, volume and audible noise?
Are there any other safety requirements - f.ex. current foldback?
Do you need many?

Right now I would say off the cuff:

Buy one, use a Thyristor bridge, a Variac or a Transformer. The 10% ripple
you can meet with an LC filter and a three phase rectifier.

 

[Spehro Pefhany]

I read in sci.electronics.design that Spehro Pefhany <speffSNIP@interlog


I'd buy one, since I don't design SMPS, and this power range is not the
one to learn at. But if I could accept the size and mass, I'd have a
non-switching supply working in a few hours.

You know, this is probably the most sensible suggestion. I think I can
do it reliably with crude ca. 1970 technology as an alternative to the
current off-the-shelf PSU solution.

Thanks guys, esp. Win for answering the main question and anticipating
the follow-up.

Best regards,
Spehro Pefhany

 

[Nico Coesel]

I read in sci.electronics.design that Spehro Pefhany <speffSNIP@interlog


I'd buy one, since I don't design SMPS, and this power range is not the
one to learn at. But if I could accept the size and mass, I'd have a
non-switching supply working in a few hours.

My idea. With pre-regulation cooling shouldn't be a problem even if
full amperage must be available over the entire output voltage range.

 

[CBarn24050]

Its going to be a BIG unit, you might find a transformer hard to come by as
well.

 

[Genome]

How would you suggest approaching this? Need a few hundred volts DC at
<5-10A, 5-10% ripple. Need galvanic isolation. Essentially a resistive
load.

Best regards,
Spehro Pefhany

http://www.speff.com

I'd be egor and go for a modular approach.

Let's say four modules at 500W per go.

Implement them as current fed push-pull or bridges. Thats a buck converter
on the primary side feeding the push-push/bridge operated at close to 100%
duty. Have some sort of wicked synchronous drive on the buck stage. Your
outputs are just bridge rectifiers with minimum/zero filtering. Phase shift
the clocks on the primary sides and you'll get the ripple down.

Implement current feedback from the secondaries and parallel the beasties
up. More power egor..... add more modules.....

Wicked. Well Wicked. I think I'll go and have sex with myself.

DNA

 

[Spehro Pefhany]

I'd be egor and go for a modular approach.

What's that? Has Igor gotten 'imself more self-esteem?

Let's say four modules at 500W per go.

Implement them as current fed push-pull or bridges. Thats a buck converter
on the primary side feeding the push-push/bridge operated at close to 100%
duty. Have some sort of wicked synchronous drive on the buck stage. Your
outputs are just bridge rectifiers with minimum/zero filtering. Phase shift
the clocks on the primary sides and you'll get the ripple down.

The input filters get really nasty and/or complex at the kW level.
Looking the cap size/voltage/ripple current and a 10 year life, it
looks kinda expensive/troublesome. Unless there's 3-phase power
feeding the thing there's no way around it that I can see.

Implement current feedback from the secondaries and parallel the beasties
up. More power egor..... add more modules.....

That concept would actually fit well in this situation. Maybe I'll
take the hard stuff out of it and keep the modular bit.

Wicked. Well Wicked. I think I'll go and have sex with myself.
DNA

Is it the weekend already? Enjoy. Thanks for the comments.

Best regards,
Spehro Pefhany

 

[Spehro Pefhany]

Its going to be a BIG unit, you might find a transformer hard to come by as
well.

Hmm.. it's well within the size range of SMPS cores that I see. The
larger ones (a few inches on a side) ought to be good for around 10kW.
This is relatively high voltage at reasonable current so it's not like
it has to be wound with bus bars.

Of course 1-2kVA 60Hz transformers are reasonably small, relatively
cheap and common.

Best regards,
Spehro Pefhany

 

[CBarn24050]

No I meant if you use a 60hz transformer it will be big, I have a 3kw telecom
power unit in my garage it weighs 60kg.

 

[Spehro Pefhany]

No I meant if you use a 60hz transformer it will be big, I have a 3kw telecom
power unit in my garage it weighs 60kg.

Sure. Maybe 30lbs (15kg) for 1kVA.

Best regards,
Spehro Pefhany

 

[Genome]

What's that? Has Igor gotten 'imself more self-esteem?

Whoops, did I pissmell myself.

The input filters get really nasty and/or complex at the kW level.
Looking the cap size/voltage/ripple current and a 10 year life, it
looks kinda expensive/troublesome. Unless there's 3-phase power
feeding the thing there's no way around it that I can see.

Comparing like for sort of like I wouldn't have thought there was too much
difference between a lump of iron and the switching approach as far as
current draw is concerned. Like someone else has mentioned power factor
correction would be cute, better form factor.... If you're bothered about HF
ripple then you might consider the same parallel approach on the primary
side.

That concept would actually fit well in this situation. Maybe I'll
take the hard stuff out of it and keep the modular bit.

OK, have fun.

Is it the weekend already? Enjoy. Thanks for the comments.

Best regards,
Spehro Pefhany

http://www.speff.com

 

[Winfield Hill]

Winfield Hill wrote...

Spehro Pefhany wrote...

If you need only one, an easy solution is to haunt eBay for a few
months, looking for an appropriate Sorensen or Zantrex programmable
switching supply, going for about $300 to $400 per kW.

For example, a Zantrex XHR 300-3.5 does up to 300V and up to 3.5A.
http://www.xantrex.com/products/product.asp?did=223&p=9
http://www.xantrex.com/support/docdisplay.asp?prod=56

Their XFR 300-9 handles up to 9A. Also, made in Canada, I think.
http://www.xantrex.com/support/docdisplay.asp?prod=55

If you want to build your own, the construction techniques in these
supplies bears examination. A PFC input stage to 340V or so, then
a phase-shift-modulated zero-crossing soft-switching PWM H-bridge
driving the dc-dc transformer, ucc3895, etc. Big low-loss diodes on
the output. An elegantly-spare cooling system, with paper shrouds.
Hmm, probably at least a month of engineering time to start. :>)

Spehro Pefhany wrote...

Thanks guys, esp. Win for answering the main question and
anticipating the follow-up.

One other suggestion for a quick first-pass on an operating power
supply: purchase a 2 or 3kW switching supply (any voltage) on eBay,
study its design, and modify it for your desired output-voltage.
See http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=3801663856

This means rebuilding or redesgning the dc-dc transformer, a task
you are certainly up to, replacing the rectifier diodes and output
electrolytic with appropriate higher-voltage-rated parts (an easy
task), replacing the relevant feedback resistor, and other stuff.

You can concentrate on Sorensen DCS, DLM and DHP designs that were
the inspiration for the Zantrex supplies, because Sorensen sells
repair manuals with schematics, compared to Zantrex, who refuses.

Once you have a working model, you can boot-start your own design.
One issue, the older designs don't have PFC input stages. The
DCS data sheet says for 3kW: "190-250 VAC, three phase, 14A; or
200-250 VAC, single phase, 20A. Note: Maximum power output of
3kW supplies must be limited to 2.5kW for single phase input."

But if you use 3-phase power, the lack of PFC isn't so important.

Thanks,
- Win

whill_at_picovolt-dot-com

 

[legg]

You can concentrate on Sorensen DCS, DLM and DHP designs that were
the inspiration for the Zantrex supplies, because Sorensen sells
repair manuals with schematics, compared to Zantrex, who refuses.

Sorenson sold the Xantrex product under it's own brand name, even
before it was bought-out by Xantrex. They never designed a HF unit
in-house to meet that form-factor.

RL