Do you have a question? Post it now! No Registration Necessary.  Now with pictures!

Threaded View

I'm trying to find a PSU that has a very low minimum load requirement for
the +12V rail. This isn't for a PC, but the equipment does have an ATX power
connector and uses 5V (10A), 3.3V (15A), but normally nothing on the +12V
(although equipment options can put this up to 4A).

In my experimenting, a couple of hundred mA on the +12V is enough to get the
PSU running ok, but all of the datasheets that I've managed to find suggest
1A min load.

Any suggestions?


Grumps wrote:
Quoted text here. Click to load it

    The few hundred mA shows what is needed to keep the 12v value under
control (it will go higher of the drain is zero). The true minimum
current value will rise, usually due to power supply heating under heavy
other loads, or a hot ambient environment. The 1A value in the specs is
put there just as CYA by the vendor so that the 12v regulation value
will never change for any load including zero external (1A + whatever
you load it).

    If the 12v you require need not be teribly well regulated, then you
fractional amp load is good enough. If you demand excellent regulation,
then add a few 100 ohm 2 watt resistors across the 12v terminals.  Do
supply a minimum load.

    The only way to determine whether you are heading for a problem with a
tiny current load is to hook up a DC voltmeter on the 12v terminals with
no load at all. Raed the value exactly. Then let the power supply run
for a long period of time, preferably with a  normal or a high load by
other components and in a warm environment if possible. Let all get
totally warm, then read the voltmeter exactly. If the voltage is
acceptable, then all is OK. Likewise for your few hundred mil load. If
the 12v drifts too high, ot possibly develops noise, then the external
load is reaquired.

    Angelo Campanella


Angelo Campanella wrote:
Quoted text here. Click to load it

Thanks, but that's not the whole story.
I have a few PSUs here (good makes, and no-names). If the +12V has no load,
then the +3.3V can drop to 2.5V. Increasing the 12V load gradually brings
the 3.3V rail up to spec.

I have managed to find a PSU with no minimum load requirement. BluTek Power,
are they any good?


Grumps wrote:
Quoted text here. Click to load it

I have no idea. I am proceeding just on old design experience. The only
way to find out is to run some load tests on them.

    Ang. C.


On Tue, 1 Apr 2008 15:39:46 +0100, "Grumps"

Quoted text here. Click to load it

It seems your PSU has a shared feedback arrangement, where
there are two voltage divider resistor feedbacks to a
comparitor circuit.  IF you want/need to keep using this
PSU, what you would do is trace the 5V feedback, find it's
series resistor, add a potentiometer in parallel with that
resistor, and completely remove the 12V rail's resistor from
the circuit (or instead of the parallel pot on the 5V, put a
series pot in the circuit before or after the 12V resistor.

Essentially you are changing the bias of how much current
flows through each feedback.  If 12V feedback is removed
entirely, which makes sense to do since you aren't using it
at all (except to power a fan??, which I would think you'll
want when loading to 10 & 15A, then the parallel POT gives
you the adjustment to increase the current to the point that
it regains regulation at the desired level for 3V & 5V.  It
could be mathematically resolved what POT value to use, but
I  can't resolve all the variables not knowing the exact
topology nor the values of the voltage divider circuit in
this feedback loop.  Maybe I shouldn't have written anything
at all about it, but if you have some spare PSU and are
safety minded you might be able to hack out the solution
from what I've mentioned... and it might help to get the
datasheet for the PSU controller chips as they may have a
reference circuit not so unlike what is in your PSU.  A
TL494 datasheet might give a classic example and was a very
popular controller for quite a while - might be what's
actually in your PSU if it's an older or cheap design.

Quoted text here. Click to load it

It would help if you linked to the model you're talking
about.  Generally, unless the unit has independent
regulation then their omission of a 12V rail load is either
an oversight, or a bad sign that the PSU has no 12V
overvoltage or overcurrent shutdown protection.  Some psu
have internal loading from a small power resistor but this
is typically far less than a few hundred mA load in a PC


kony wrote:
Quoted text here. Click to load it

Thanks. I may crack open the unit(s) we have here and investigate, but I'd
rather have an off-the-shelf part.
The first on this page from BluTek is what I found.


On Wed, 2 Apr 2008 10:25:13 +0100, "Grumps"

Quoted text here. Click to load it

Their 500W unit seems impossibly rated for current
capability, has poor airflow for that current rating.  It
puts all their spec's into question, it isn't a large enough
unit to produce power claimed and do so with independant
regulation so there would have to be sizable interal loads
or else it was only a representative example.

By that I mean it represents what they can custom design per
a target system which is a known load above the actual
minimum, and yet that still doesn't validate their rating.

Since I have no hands-on experience with these and I don't
recall ever seeing an online review, they are a wildcard.  I
would not assume it can actually keep any rails within a
given spec without any load on them, and yet they could
still be primary or wholey regulated based on 5V feedback
which could be of benefit - and yet they also list 0A
requirement for that rail.  Something has to give, it is not
at all likely one could randomly load any single rail to the
max they list without having to put a load on any other rail
unless that rail were the 5V and their overvoltage shutdown
had a relatively high(er) threshold than a decent ATX PSU
normally would.

You dont' mention the specifics of your project nor
cleanliness and voltage tolerance required, but offhand I
would look for some old OEM ATX PSU design that has at least
240W continuous rating and no more than 14A 12V current
rating.  The higher the 12V to 5V current ratio the more
likely you'd need a load on the 12V rail, unless you did a
modification to the feedback circuit as mentioned

The easier answer is what you already know, you can simply
add the load to the 12V rail and use just about any ATX PSU,
or of course you could seek a PSU that only had the 5V and
3.3V output instead, some industrial PSU which might be
lower cost at an online surplus electronics 'site.  3

1)  Good current:rail rating, ample reserve current per
rating, downside is it's open-frame.  If your project had a
suitable, isolating (from AC high voltage) enclosure and you
didn't need ATX PSU connectors, it's the one I'd pick though
it may not have a remote turn-on logic level signal, rather
would turn on/off by simple switch to AC or your added relay
control, etc.

2)  Long and skinny, too big for most uses except a 1U rack
case.  Otherwise well suited though it's two small fans
probably run at fairly high RPM producing a bit of noise.

3)  Less optimal due to higher 12V current ratio, but
probably still managable, enclosed and not as long as #2,
has ATX connectors and a larger fan which might be more
reasonably throttled back to low RPM if noise is excessive
(but should start out being quieter than #2).

I've no idea how old any of these 3 examples are, nor the
storage conditions of their surplus.  Hopefully the contacts
aren't tarnished or corroded but if they're pushing 10 years
old it might be time to overhaul with new capacitors before
deploying for a lengthly period of time... but we dont' know
your project requirements.

Certainly other PSU would do the job, a typical 200W from a
roughly 1989 era Compaq, Dell, Gateway or HP PC PSU would be
a reasonable match as well, somewhat similar to #3 above
would be typical of those.  Essentially the ideal PSU from a
PC would be an earlier ATX generation when systems weren't
expected to have more than a half dozen amp load on 12V rail
from 2 or 3 hard and optical drives plus a modest fan... but
again, given some hacking you could ultimately make just
about any ATX PSU work if you reverse engineer and modify
the feedback circuit or put a load on 12V rail.


Quoted text here. Click to load it

Thanks again for your comprehensive replies.
The BluTek unit is pretty much out of the running as the company are
unresponsive to email and phone messsage.
I'll check your other links later today.


On Tue, 1 Apr 2008 13:47:10 +0100, "Grumps"

Quoted text here. Click to load it

You're wanting a PSU that regulates based on mostly, if not
solely, 5V voltage level.  That was more common a few years
ago when PSU were designed to accomodate systems using more
5V current than 12V current.  There are ways to hack a PSU's
feedback to make it do what you want but they are a bit
complex to describe in a usenet post, particularly not
having the unit here to reverse engineer.

Mainly, if your PSU mostly regulates from 5V rail feedback,
the 12V voltage level can float high and it wouldn't be a
problem, until that voltage gets to a level excessive enough
that it trips the overvoltage protection circuit.  You'd
need to add a load to the PSU you want to use if it can't
run with that much of a 5V load before raising the 12V
current too much.  Similarly, a PSU with a very low 12V
current rating (relatively, compared to some of today's PSU
that have 16 or 18A, even multiple times this much) compared
to it's 5V current rating would require less if any load on
12V to keep it under the protection shutoff threshold.

1A minimum load is for ATX regulation spec, you would be
able to make due with a lesser load, I suggest trying
something closer to a 47 Ohm, 10W power resistor you bolt to
the inside of the PSU case, preferribly near an airflow
path.  That'll give you a quarter amp load, which as you
mentioned is close to that couple hundred mA you found
results in acceptible operation.

Site Timeline