Suggestions for 3-rail open-frame power supply

[Lewin A.R.W. Edwards]

Hi all,

I'm trying to decide the most cost-effective way to get the following
rails in a product for low volumes (~1000 per year, but with
components bought typically for 10 piece production blocks only):

* +5V 3000mA (3500mA would be better)
* +12V 250mA or 4250mA depending on how the circuit is built
* +3.3V 1000mA (1250mA would be better)

I can buy an off-the-shelf PSU that will give me all the above and
more, but it's about $120 in small quantities (!)

On the other hand, I could buy a +5V, +3.3V dual-rail supply and get
the 12V cheaply with a simple boost regulator based around the
MC34063.

Or I could buy a a +5V 6A supply and try to derive my other two rails
off that with a pair of switching regulators. The 3.3 would be tricky
though.

Ideally I'd like to get the cost under $40. I know I can get
everything I need by using an ATX power supply, but there doesn't seem
to be a standard form factor for slim ATX-compatible PSUs and our
housing won't fit a big fat brick.

Does someone know of an off-the-shelf, *continuing production*
open-frame or enclosed module that does everything I want, or failing
that are there any comments on which of my above options seems most
feasible?

 

[Walter Harley]

"Lewin A.R.W. Edwards" <larwe@larwe.com> wrote in message
news:608b6569.0402051348.1da9317f@posting.google.com...

Hi all,

I'm trying to decide the most cost-effective way to get the following
rails in a product for low volumes (~1000 per year, but with
components bought typically for 10 piece production blocks only):

You mean they do a production run twice a week and yet they can't order
parts in advance? Sheesh.

* +5V 3000mA (3500mA would be better)
* +12V 250mA or 4250mA depending on how the circuit is built
* +3.3V 1000mA (1250mA would be better)

What do you mean by "or 4250mA depending"? There's a BIG difference between
3W and 50W.

Or I could buy a a +5V 6A supply and try to derive my other two rails
off that with a pair of switching regulators. The 3.3 would be tricky
though.

Why would you use a switching regulator to get 3.3V? Why not just an
ordinary linear? The dissipation will only be about 2W. A small heat sink
would be adequate. And it would be a lot cheaper.

 

[Lewin A.R.W. Edwards]

I'm trying to decide the most cost-effective way to get the following
rails in a product for low volumes (~1000 per year, but with
components bought typically for 10 piece production blocks only):

You mean they do a production run twice a week and yet they can't order
parts in advance? Sheesh.

I did say "typically". The production *averages* to 1000 but there are
quiet periods with just one or two in a week, and during the lulls only
10 pieces max are kept in stock, because the rest of the BOM includes
extremely expensive components. So sometimes we'll order 100 and
sometimes we'll order 10. Anyway, that's a supply chain detail I can't
influence.

* +12V 250mA or 4250mA depending on how the circuit is built

What do you mean by "or 4250mA depending"? There's a BIG difference between

There are alternative ways of building the product. There are two
modules in it, and one of them requires 4A@12V, and comes with a
120V->12V supply of its own. The other module requires 250mA on the 12V
rail. So I can either have two power supplies in there (since we get the
4A PSU free) or build the multirail PSU meatier.

Why would you use a switching regulator to get 3.3V? Why not just an
ordinary linear? The dissipation will only be about 2W. A small heat sink
would be adequate. And it would be a lot cheaper.

Really cheaper? The LDOs seem to cost about $3. The switching controller
IC costs about $0.60 and the passives around it are about another dollar.

 

In sci.electronics.components Lewin A.R.W. Edwards <larwe@larwe.com> wrote:

* +5V 3000mA (3500mA would be better)
* +12V 250mA or 4250mA depending on how the circuit is built

IMHO: You won't meet your cost target with the 12 V/4.25 A option. So
sigh deeply and resign yourself to having two power supplies.

Ideally I'd like to get the cost under $40.

Just looking in the Mouser catalog...

An ST LD1086V33 regulator (3.3 V/1.5 A, 1.3 V) is $1/qty 1. Add a
heatsink and some caps for maybe $1.50 more, making the total $2.50
quantity 1, plus labor to solder it together.

Phihong PSA-45U-201 gives you 5 V, 4000 mA and 12 V, 1500 mA for $37.40
quantity 5 or $34.60 quantity 25. Run some of the 5 V through the
regulator above. Total ~$40 quantity 5, ~$37 quantity 25. A little
tight on your current limits, though.

If you don't mind three power supplies...
A Cincon CFM40S050 gives you 5 V, 6000 mA for $28/5 and $26/25. Run
some of it through the regulator above. Also get a Cincon CFM1002S,
12 V, 830 mA, $16/5, $14/25. Total ~$47 quantity 5, ~$43 quantity
25. This gives you lots of room on all 3 supplies, but goes over
budget. Possibly replace the 5 V supply with CFM2001S, 5 V, 4400 mA,
for $17.50/5 and $16.50/25, for a total of ~$36/5, $33/25.

I don't work for any of the companies mentioned. Standard disclaimers
apply.

Matt Roberds

 

[Lewin A.R.W. Edwards]

Hi,

* +5V 3000mA (3500mA would be better)
* +12V 250mA or 4250mA depending on how the circuit is built

IMHO: You won't meet your cost target with the 12 V/4.25 A option. So
sigh deeply and resign yourself to having two power supplies.

Well, as it turns out, I'm on crack. I didn't think of this, but
someone emailed me to point out that there IS a standard-form-factor PSU
that meets all my needs and more, and is available from multiple sources
to boot - viz, ATX PSUs for 1U rackmount server blades.

The price range on these items is from about $50 to $65, which is out of
budget, though not seriously - the budget was a "wanna" not a "gotta".
However, for $60 I can get the PSU in a 1U casing. If I can sell the
empty casings on eBay for $25, which I think I can (judging from the
fact that they regularly sell for $40!) then I'm under budget.

 

In sci.electronics.components Lewin A.R.W. Edwards <larwe@larwe.com> wrote:

I didn't think of this, but someone emailed me to point out that there
IS a standard-form-factor PSU that meets all my needs and more, and is
available from multiple sources to boot - viz, ATX PSUs for 1U rackmount
server blades.

You know, that's a good idea. Wish I'd thought of it. Good to hear
that you found a solution, though.

Matt Roberds