AT Power supply circuit breakers

[Robotnik]

I've been handed an interesting project at work (not related to my REAL
job).

Some 12 volt hallogen track lights managed to overheat their factory power
supply. (three 20 watt bulbs, in paralell)

I decided to use an AT PC power supply after having no luck with finding any
12volt power supplys that would output a load of 5 amps... (well, I couldn't
find any inexpensive ones).

The power supply I donated is rated at 9.5 amps on the 12 volt side, I
figure that gave plenty of room (I assume these ratings were maximum, not
rms).

In my own testing at home, they seemed to work fine, running for several
hours.

However, when I put them back in their own environment, the power supply had
shut off sometime during the day.

I'm trying to decide if it was an overheating issue, since it lasted a
while, or if it's an electrical current overload issue.

I'm thinking that heat build- is the problem, the power supply (separated
quite a distance from the lights, they're not heating it any) is in a
somewhat confined space, I think it's heating to a certain point. I ended
up cutting up some cardboard and making a "snorkal" for the air to flow into
the power supply from another location, rather than sucking hot air back
in.

Any thoughts on this?

Thanks,

Nic

 

[Bill Sloman]

I've been handed an interesting project at work (not related to my REAL
job).

Some 12 volt hallogen track lights managed to overheat their factory power
supply. (three 20 watt bulbs, in paralell)

I decided to use an AT PC power supply after having no luck with finding any
12volt power supplys that would output a load of 5 amps... (well, I couldn't
find any inexpensive ones).

The power supply I donated is rated at 9.5 amps on the 12 volt side, I
figure that gave plenty of room (I assume these ratings were maximum, not
rms).

In my own testing at home, they seemed to work fine, running for several
hours.

However, when I put them back in their own environment, the power supply had
shut off sometime during the day.

I'm trying to decide if it was an overheating issue, since it lasted a
while, or if it's an electrical current overload issue.

I'm thinking that heat build- is the problem, the power supply (separated
quite a distance from the lights, they're not heating it any) is in a
somewhat confined space, I think it's heating to a certain point. I ended
up cutting up some cardboard and making a "snorkel" for the air to flow into
the power supply from another location, rather than sucking hot air back
in.

Personal computer power supplies seemd to be designed on the basis
that they are going to be fan-cooled. Did your set-up include a fan?

Adding a vertical chimney to a power supply is better than nothing -
the longer the chimney, the more you will encourage natural
convection, driven by the density difference between the hot air in
the chimney and the cold air outside the chimney - but it is nothing
like as effective as a fan.

Hve you tried measuring the temeprature of the air in the chimney,
just above the power supply?

 

[Ken Taylor]

I've been handed an interesting project at work (not related to my REAL
job).

Some 12 volt hallogen track lights managed to overheat their factory power
supply. (three 20 watt bulbs, in paralell)

I decided to use an AT PC power supply after having no luck with finding any
12volt power supplys that would output a load of 5 amps... (well, I couldn't
find any inexpensive ones).

The power supply I donated is rated at 9.5 amps on the 12 volt side, I
figure that gave plenty of room (I assume these ratings were maximum, not
rms).

In my own testing at home, they seemed to work fine, running for several
hours.

However, when I put them back in their own environment, the power supply had
shut off sometime during the day.

I'm trying to decide if it was an overheating issue, since it lasted a
while, or if it's an electrical current overload issue.

I'm thinking that heat build- is the problem, the power supply (separated
quite a distance from the lights, they're not heating it any) is in a
somewhat confined space, I think it's heating to a certain point. I ended
up cutting up some cardboard and making a "snorkal" for the air to flow into
the power supply from another location, rather than sucking hot air back
in.

Any thoughts on this?

Thanks,

Nic

Does the supply require a nominal load of some value on the 5V line? This is
common for AT supplies - you may need to derate the supply if you don't do
this. Otherwise I'd be going for heat as Bill has already said....

Ken

 

[Robotnik]

Personal computer power supplies seemd to be designed on the basis
that they are going to be fan-cooled. Did your set-up include a fan?

Adding a vertical chimney to a power supply is better than nothing -
the longer the chimney, the more you will encourage natural
convection, driven by the density difference between the hot air in
the chimney and the cold air outside the chimney - but it is nothing
like as effective as a fan.

Hve you tried measuring the temeprature of the air in the chimney,
just above the power supply?

This power supply does have a fan on it, but the power supply is oriented
such that the air flows sideways, left to right. I have I guess a
"backwards chimney" on it, the cool air is being taken in from a different
location, where there shouldn't be as much heat.

Are you suggesting that I put a chimney on the exaust too (or instead)? I
hadn't thought of that. I could even put another fan at the top of the
chimey, or somewhere in it.

And I have not had a chance to check the temperature. However, This is the
Second at power supplly that I have tried, the first one was only rated at 7
amps, and it didn't last a day either. It has a temperature controlled fan
though, and when I figured out it didin't work, it was very veryr warm to
the touch. At first I didn't think twice about the temperature, and that's
why I treid a higher-rated (on current) power supply.

BUt the one I'm trying currently does not have a temp controlled fan, it
stays on constantly (until the breaker trips).


Nic

 

[Robotnik]

Does the supply require a nominal load of some value on the 5V line? This is
common for AT supplies - you may need to derate the supply if you don't do
this. Otherwise I'd be going for heat as Bill has already said....

Well I don't know tons about electronics, I'm not completely sure I
understand your question.... But Both of these ran for a few hours at my
house with plenty of ventaltion, and no problems. I think it operates
without a load on 5v. Do you suggest that I put some sort of load on it?

Nic

 

[Syd Rumpo]

I decided to use an AT PC power supply after having no luck with finding any
12volt power supplys that would output a load of 5 amps... (well, I couldn't
find any inexpensive ones).

The power supply I donated is rated at 9.5 amps on the 12 volt side, I
figure that gave plenty of room (I assume these ratings were maximum, not
rms).

In my own testing at home, they seemed to work fine, running for several
hours.

However, when I put them back in their own environment, the power supply had
shut off sometime during the day.

I'm trying to decide if it was an overheating issue, since it lasted a
while, or if it's an electrical current overload issue.

The inrush current (the initial surge) for an incandescent lamp will
be maybe ten times the normal running current - the filaments have a
positive temperature coefficient and are much lower resistance when
cold. You might get lucky, but I'd find a different solution.

 

[Robotnik]

The inrush current (the initial surge) for an incandescent lamp will
be maybe ten times the normal running current - the filaments have a
positive temperature coefficient and are much lower resistance when
cold. You might get lucky, but I'd find a different solution

So are you suggesting that the resistance will increase as the filaments
heat up?
I see no problem with that, that means they'll draw less current as their
temperature increases.
Turning them on seems to be no problem, the power supply shuts off after a
long period of runing time, even after the lamps are heated to a fairly
constant temperature themselves.

Nic

 

[Ken Taylor]

This
Well I don't know tons about electronics, I'm not completely sure I
understand your question.... But Both of these ran for a few hours at my
house with plenty of ventaltion, and no problems. I think it operates
without a load on 5v. Do you suggest that I put some sort of load on it?

Nic

It wouldn't hurt - if the supply starts wandering from spec. under heat
stress then without a properly regulated 5V supply (no load) it may end up
shutting down. There's usually a minimum load and a hunt on the
manufacturer's website should reveal this. Failing that, find a 6V light
bulb. It's a cheap test.

Cheers.

Ken

 

[Roger Hamlett]

finding supply

So are you suggesting that the resistance will increase as the filaments
heat up?

This is inherent with any resistive load that has a positive temperature
coefficient. Most materials behave this way.

I see no problem with that, that means they'll draw less current as their
temperature increases.

Not really. They draw their _rated_ current when fully hot. The problem is
that when cold, a lamp will typically draw perhaps 10* this current. The
normal supplies avoid this being too much of a problem, by having a fairly
high internal resistance, thereby limiting the maximum current during this
phase.

Turning them on seems to be no problem, the power supply shuts off after a
long period of runing time, even after the lamps are heated to a fairly
constant temperature themselves.

This is because the supply is designed to handle the 'turn on' surge, but is
switching off, perhaps because the thermal cutout is set slightly too low.
Unfortunately a PC supply is not designed to handle this type of surge, and
if it has enough current rating to deal with it, may well significantly
shorten the life of the bulbs (slowing the rate of initial 'switch on' a
little has a large effect on bulb life). One solution is to add a NTC
resistor in the line feeding the lamps.

Best Wishes

 

[Robotnik]

This is inherent with any resistive load that has a positive temperature
coefficient. Most materials behave this way.

Not really. They draw their _rated_ current when fully hot. The problem is
that when cold, a lamp will typically draw perhaps 10* this current. The
normal supplies avoid this being too much of a problem, by having a fairly
high internal resistance, thereby limiting the maximum current during this
phase.

This is because the supply is designed to handle the 'turn on' surge, but is
switching off, perhaps because the thermal cutout is set slightly too low.
Unfortunately a PC supply is not designed to handle this type of surge, and
if it has enough current rating to deal with it, may well significantly
shorten the life of the bulbs (slowing the rate of initial 'switch on' a
little has a large effect on bulb life). One solution is to add a NTC
resistor in the line feeding the lamps.

Okay, I think I see what you mean. Any suggestsions on the value of the
resistor? (supply rated at 9.5 amps 12V, lights in total (eventually) 5 amps
12 volts).

Thanks!

Nic

 

[Robotnik]

Alright, thank you!

Nic

 

[Roger Hamlett]

Turning them on seems to be no problem, the power supply shuts off
after

a but

Okay, I think I see what you mean. Any suggestsions on the value of the
resistor? (supply rated at 9.5 amps 12V, lights in total (eventually) 5 amps
12 volts).

Thanks!

Look in circuits for halogen projectors. You are looking for something like
a 1R to 0.1R unit. This will change resistance from 1ohm when 'cold', to
0.1ohm when 'hot'. The parts are quite common in older slide projectors, and
overhead projectors for exactly this reason. Modern designs tend to use a
switch mode supply, and have 'soft start' built into the supply. With the
cold resistance of the bulbs (which should be about 0.25ohm), you will just
about be 'in spec' for the supply as it starts. The parts are sold as
'inrush current surge limiters', an SL121R008, would probably be close to
what you need. This is a '12mm' diameter type, designed to limit the maximum
current to 8A, with a nominal 1R resistance when cold.

Best Wishes