Could use help with an overheating power supply…

[Revolvr]

Hi all,

I am building a simple bench power supply with a 12VDC and 5VDC output.

A 24VDC wall wart is fed into a 7812 regulator for 12VDC output. The output of the 7812 is also fed into the input of a 7805 regulator to provide a fixed 5VDC output. The 5VDC output also supplies power to 2 common ground LED panel meters; each draws 60mA.

Problem I am having is this. When nothing is hooked up to the outputs, I see 12V and 5V as expected. When I provide 5V to one panel meter, both regulators heat up quite a bit. But I have confirmed the actual current drawn is 60mA. When I connect both panel meters I have verified a current draw of about 120mA. However both regulators quickly become too hot to touch and will shut down after a few minutes. The inputs to the panel meters have not been connected.

Any ideas? Have I committed some faux pas by chaining two regulators? Any debugging ideas? I have broken the circuit in a few places to verify current draw and checked the voltages. Everything seems normal except the heat generated.

Thanks,

-- Dan

 

[dknguyen]

I don't think anything is wrong with chaining regulators (non-switching ones at least). The only thing is 60mA seems kind of high for LEDs, but maybe they are different kinds of LEDs than what I'm used to.

BUt to the things that you are actually asking about- the limit on the regulator is 1A right? But is this at 6V or 12V? There is 6x more energy being dissipated when clamping down 12V->5V than when clamping 6V->5V. Maybe you need a heatsink.

 

[Revolvr]

Yes the regulators should be able to provide 1A no problem. And they are both heatsinked, and they both heat up equally - as far as my fingers can tell. The wall wart used can supply up to 1.6A.

-- Dan

 

[dknguyen]

Try connecting a moderate ohm resistor rather than your panel meter and see if they overheat. Maybe you missed something in your panel meter specs.

It's not so much the heating that's the problem (your fingers probably can't handle the temperature the regulators can), but the fact that it's getting hot enough to shutdown the regulators is weird.

 

[mvs sarma]

Hi Revolvr,
The input is too high for the load of .12A (120mA)at 5V.
7805 expects 5+say 3to 4 extra=8 to 9 V
7812 expects 16V.
power loss across 7812 appears roughly .12*(24-12)=1.44 watts.
and across 7805 is 7*.12=.84 watts.
thus if you have another adopter with 9V Dc output you may perhaps use 7805 directly and with a small aluminum sheet as heatsink.

Alsao

Do you use 12V output for any purpose other than as input to 7805?

this load is not foreseen in above calculations!! whether you current measurements double checked and are correct?

 

[Revolvr]

Hi Sarma,

I agree with your power calculations, but this is well within the 78xx series envelope. The max input voltage for both is 35V. They should be able to dissipate about 8W with heatsink at 25C ambient. I can't imagine 1.4W would get too hot to touch.

Also I eliminated the panel meters as a factor by driving a 33 Ohm load. My Fluke showed 123mA current. Both heatsinks quickly became too hot to touch, but didn't shut down after about 10 min.

-- Dan

 

[JTECH]

Just a simple question: Did you include the input and output decoupling caps on the regulators? (Mounted as close as possible to the devices, 0,1mfd will do). Sometimes (without these caps), simple circuits tend to oscillate at high frequencies (unwanted!!) and this can heat up things very fast!

 

[eng1]

They should be able to dissipate about 8W with heatsink at 25C ambient.

This depends on the heat-sink you chose. If you use a small TO-220 type, I doubt that it's enough. And you're considering just the current for the meters... what will happen when you'll connect a load to the PSU? Decide the max output current, first.

 

[Revolvr]

Good thought. I do have 0.1mfd ceramic caps on input and output of each regulator.

I had spare regulators so I breadboarded a copy of the circuit. 24VDC in, 7812 out to 7805, out to 33 Ohm load, giving about 120mA current. This should eliminate any soldering or circuit errors. Circuit behaves the same. The 7812 heats up quickly to be too hot to touch. The 7805 heats up a little slower. (Both without heatsinks in this case).

I'm building this not just because I need a power supply, but to learn too. So perhaps the problem is my expectations on the amount of heat generated with a small current due to the 12V drop on the 7812. Perhaps I just need a bigger heatsink and this is to be expected?

-- Dan

 

[Nigel Goodwin]

Perhaps I just need a bigger heatsink and this is to be expected?

Yes, you need bigger heatsinks - with none at all you can only drop fairly small voltages across them and not pull much current.

 

[Revolvr]

These are TO-220 packages and I am using typical small black aluminum finned heatsinks.

-- Dan

 

[Nigel Goodwin]

These are TO-220 packages and I am using typical small black aluminum finned heatsinks.

Only suitable for low power use - they really aren't very good heatsinks!, and you're dropping a LOT (too much!) voltage across the 7812.

 

[Revolvr]

So what would you recommend to get a 12V and 5V output from 24V in? Other parts, bigger regulator, voltage divider?

-- Dan

 

[dknguyen]

Not a voltage divider- it will waste a disproportionate amount of current due to a resistor being connected to ground. You could use a switching regulator. There are some nice cheap ones I just found out about at

www.dimensionengineering.com

They have the same profile as TO-220s, but of course they cost a little bit of money. Maybe just get two wall-warts?

 

[Nigel Goodwin]

So what would you recommend to get a 12V and 5V output from 24V in? Other parts, bigger regulator, voltage divider?

You need to reduce the heat dissipated by the regulator - there are various ways of doing this:

Use a more suitable input voltage! - this will reduce the heat.

Use a bigger regulator, TO3 instead of TO220, but you still need a large heatsink, and the heat dissipation is still the same - it's just better able to transfer it to a heatsink.

Lower the voltage entering the regulator - a pre-regulator could be used, or a simple LARGE WATTAGE resistor. This keeps the heat dissipated the same (as it's got to), but it transfers some of the heat to the pre-regulator/resistor.

Use a switch-mode regulator, these are FAR more efficient, but much more complicated as well.

 

[Revolvr]

OK, well I'm learning and that was one objective. Looks like I might need to redesign the hardware.

I wanted to use the 24V supply because I wanted to also add an LM317T variable regulator to provide 1.5-20V or so for various projects - in the same case. That would have the same heating problem especially at low voltage so I'll need to solve that also.

Thanks everyone, as usual this is a very helpful site.

-- Dan