[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [OT] ATX-PSU and amperage on connectors...

On Mon, 03 Mar 2008, Michelle Konzack wrote:
> I have tested last weekend some of the Step-Down-Converter chips (I
> have ordered several Evaluation Boards) and was realy surprised about
> the quality of the Maxim and National Regulators...
> They are all without any exception better as all of my 78 AC-PSUs (200W
> to 900W) I have here and tested under a REAL environement...

Well, it looks like what I learned on graduation near ten years ago still
holds.  If you are going to look for the best chips to do something, you
will find stuff that is MUCH better than what you get in consumer goods.

And it is not even that more expensive, it's just that every cent counts
when you are doing runs of hundreds of thousands parts at a cut-throath
margin :p

> So, if I have found a Reseller for InfraRed soldering Equipment, I will
> order free samples  :-)  of the chips and build my first 24V-DC-PSU.

That should be trivial to do in Europe.  I don't know about cheap, but you
should be able to just directly ask the manufacturers of professional
soldering equipment for a number of resellers, and start asking around for
prices.  Now, if you want it used for much better prices, then yes, it will
take some doing to find one :-)

> The4 waving is much more less then at the AC-PSUs (the ATX12V spec is
> talking about 60-120mV) I have always (all Voltages) less then 30mV.

Make sure they are very stable under spiked loads, that's when the RAM chips
start flipping bits :p  Some don't work well in either extremes of the load
curve either, and that can be a factor (you might end up needing two 5V
regulators, one that powers up the boards, and another that powers up the
disks, for example, to have the more critical load (boards) that has also
less variance, on a more "tunned" regulator).

> > BTW: if this is the kind of stuff you are going to leave in the field inside
> > an airtight box, did you remember to add some EMP shielding and surge
> > protection?  Otherwise, the first time you get an electrical storm dropping
> > a lightning bolt near your box, the entire thing will fry.
> I will use the normal ATX-PSU case and a special Mini-ATX-PSU for 1U
> racks, but for this I have not found a specification about the size.

That won't do any good if you don't have good surge supressors and the
entire system (PSU+everything else) is not inside a grounded good-enough
faraday cage for high-frequency equipment.

> Currently I have my Opteron 140 running on a test installation...

Just remember that if you use mobile CPUs or chipsets, than can do even more
drastic power load changes than an Opteron (which doesn't even try to
conserve power very much, to begin with).

BTW, depending on what you need to do with the embedded computer, have you
looked at the stuff from routerboard.com and similar vendors?  They are much
friendlier to embedded and custom designs than a standard PC motherboard.

> I have had the same problem with ma IBM TP570...
>                              ....and solved it with a glue pistole too!

Do you have any photos of the glue on the planar card that we could add to
thinkwiki?  The harware hacks section of thinkwiki looks too feeble right
now :-)

> Most peoples have never seen such professional equipment and of course
> you should never tell them how much it cost...  they will never believe
> it (my experience; Most I have bought on real or internet auctions, so
> the Insurance price would be arround 200.000 Euro where I have payed
> only 30.000 Euro)

Yeah, this kind of stuff has only two type of buyers: those who *have* to
get it new at could care less for the list price, because it will pay itself
off in less than six months, and those who will buy it only at less than 20%
of the price for a new equipment.  And that means you always have to sell
them dirty-cheap when you don't need them anymore :-)

> > I'd kinda expect to find ready-made DC-DC PSUs that do it for the telecomm
> > market, but they won't be cheap at all.
> Not more neccesary, since currently I am using a "National" Step-Down
> Regulator which make 25A on 12V and it works heavyly stable...

By ready-made, I mean in a dust-tight, air-tight, water-tight extruded metal
container with a few wires sticking out of it, and a 5-year warranty with a
better than 20-year MTBF :-)

> > A proper UPS needs a bit more.  If it is a DC UPS, it is easier (no need for
> Right, but since I have 20 Sonnenschein G120 (12V/120Ah) Batteries,
> which mean 24V with 1200Ah and they are charged from a Solar-Charger
> and a "Windmotor 1803 Furlmatic" they must not be whatched.

They won't last nearly as long as they could if the charger supervisor
doesn't do thermal compensation (if you can't somehow make sure the
batteries are always at 25°C), proper burst and float charging control, etc.

> I am measuring only the INPUT voltage and the consumation and then from
> all other voltages the (V and A), and then I transfer the data over I²C
> to my currently used DS80C411 (also Evaluation Board)...

There are good lead-acid gauge chips out there I think (it is at least three
years since I looked at battery gauge chips), that can do it all for you.

If you can monitor each battery, instead of the bank, that's much better
too.  You get to notice one is going bad before it damages the entire
battery bank.

> Oh, I have already tried out my Multi-Cemistry Charger (also Evaluation
> Board) to charger NiCd, NiMH and Li+ accus with it...  Then I have put
> an Alkaline on it and the charger has shutdown immediatly...

Nice!  Although I have a big thing against "multi-chemistry" stuff: I'd
rather have an specific algorithm that does the absolutely best for that
chemistry, when I can get away with this.

Sometimes your need for versatility overrides the improved performance of
the chemistry-specific stuff, of course.

  "One disk to rule them all, One disk to find them. One disk to bring
  them all and in the darkness grind them. In the Land of Redmond
  where the shadows lie." -- The Silicon Valley Tarot
  Henrique Holschuh

Reply to: