require advice for hot lm317

I'm looking for some advice if possible

I have been working on the attached circuit for use in decorative lanterns. The 1 watt LED's provide white light while a cluster of 5mm yellow LED's in the middle simulate the flickering of an oil lamp. I thought my circuit was OK and have been trying out various LED configurations using a benchtop PS for a week or so. I now have my LED configuration looking good through the ricepaper and thought before I solder it all up on a PCB I'd try it out using a 12V 1A power plug. The online LED wizards I used to work out my arrays figured I'd be drawing 700mA for the one watt LEDs and 200mA for the 5mm LEDs but actually I measured 350-60 mA and 2-5 mA respectively> I did not expect the 5mm's to draw so little flickering randomly

The short of it is that while I had no heat problems on either LM317 working off the bench top PS at 12.5V I have a serious heat problem on the top LM317 ( 7V ) when hooked up to the 12V, 1A power plug. I did not push it too far after 80C but changed to a larger heatsink for no joy.

When I hooked up my multimeter in line between the power plug and the circuit my overheating problem went away. This leads me to believe my problem is solvable without using an excessivly large heatsink but it's beyond my level of knowledge at present.

I hope my hasty schematic is clear enough, I'm fairly new to all of this.

try putting .1uf, or some small cap values on the inputs and outputs of your lm317 parts.

I have one between the power plug and the 2 lm317s and one each on the output. .1uf or 100?. I'm using the small ceramic marked 101.

use one for each lm317 near its input and output. I mean near, distance does matter.

Hi,

first off why would you want to waste an adjustable voltage regulator supplying the MCU? For the top part use a 100nF cap at the input pin of the LM317. The regulator probably produces excessive heat due to oscillations.

Also you will fry the MCU connecting relatively high load directly to the port pins. The LEDs won't get their nominal current because the chip can't source the required current. Use NPN-driver transistors instead and connect the LEDs to them using the output of the LM317 for LED supply to avoid power fluctuations for the MCU supply.

See attachment

Boncuk

Mikebits, I have one cap infront of the two lm317's, I'll change that tomorrow and see how that goes. and I'll back the ones on the output side closer to the lm317s. Can you have too many capacitors? it is my understanding that one should be just before the PIC.

Boncuk, I didn't realise I was wasting regulators, It was the only way I could figure out getting both 7V and 5V out of a single 12V supply. I am doing something similar with an lm317/6V supply to 4.5V to switch 15 LEDs with a microchip in another lantern we use in or home for a nightlight. This circuit has run every night for 3 months without frying anything (see attached). Thanks for the schematic I'll study it as I hope to learn a bit about NPN transistors in my next project. Note, however that I am switching 4 LEDs per pin for a total of 20 LEDs, I'm not too sure I'd get there with your approach.

With 12v on input and 7v on output of LM317 at current draw of 700 mA you have 3.5 watts of heat dissipation on LM317. That requires a pretty good sized heat sink. When you say it does not get hot when DVM is in series is just means the resistance of the your DVM is dropping the voltage and part of the heat dissipation is in the DVM.

A 0.5 volt drop across ballast resistor is not much margin for variation in LED's as they get warm. There will be variation in the LED current that may destroy them.

justwantin said:

I have one between the power plug and the 2 lm317s and one each on the output. .1uf or 100?. I'm using the small ceramic marked 101.

Click to expand...

101 is only 100pF; the minimum capacitors around a 317 should be 0.1uF, which would be 104

You say you used a 12V 1A power plug. Is this an unregulated wall-type plug? The voltage could be higher than 12V.

I'll have to chase up some capacitors to get that right. However with the given circuit I have two outcomes with two different power supplies. What might be the difference?

What I have on the breadboard was not overheating when powered off my bench top supply at 12 and 12.5 volts but does off a new switch mode power plug fresh out of a blister pack rated at 12V, 1A regulated with short circuit, overload load and over current protection. It also reads 12V on the same multimeter. I just had the circuit on the benchtop PS for 15 min with the same probe that gave me plus 80C yesterday holding at a constant 25C. I didn't measure the temperature, but on the power plug thee heat sink gets uncomfortable in a couple minutes.

The benchtop PS was built from a design on the net and using an lm317 and pot ranges from 1.25_30V supposedly at 1.5A. Voltage is spot on but I have no way of checking the current.

The only differences may be current and that the power plug I'm using is switch mode

I have just done a comparison of voltage readings from three different power supplies on this circuit. Can anyone comment on my reasoning.

The third PS is a power plug for from the base unit for our home phone which is branded Panasonic and I am quessing is unregulated. While rated at 9VDC it measures at 14.3 with not connected. When connected to the 9V power plug, the lm317 regulating the 7V is only marginally warmer compared to when connected to my benchtop PS.

1- Benchtop PS, 12V ?A
2- Unregulated? power plug, 9V 800mA (from Panasonic remote phone base unit)
3- Regulated power plug, 12V 1A
* Measurements are VDC

- - Full Load / 5mm LEDs only / 1W LEDS only / No load
1- 7.6 / 11.9 / 7.7 / 12
2- 8.5 12.9 8.6 13.5
3- 11.7 / 11.9 / 11.7 / 12

Not knowing anything I would take the above to indicate that:
1) The regulated (12V) power plug will supply around 12V no matter what the load so I would have to heatsink for max the capacity of the powerplug,
2) that both my benchtop PS and the 9V unregulated power plug will supply relative to the load so the thermal buildup will not be as great and heatsink requirements less if not marginal, and
3) the regulated 12V power plug is unsuitable for this application with the circuit as is (capacitors being a separate issue).

My aplogies, tabs don't seem to work in adavnced text mode or the table would be easier to read and I may not be quite right in my terminology

12V plug pack

The 12VC plug pack you have been using may be a SMPS and causes oscillations as other members already suggested.
Try putting on the 12 Volt supply a 4700 uF 35 Volts elco to dampen the ripple and see if that works ok.

Justwantin,Have you tried swapping these LM317s to LM317Ts? As far as the data sheet goes LM317 doesn't give more than 100mA,while only LM 317T is rated for 1.5A!

Not knowing anything I would take the above to indicate that:
1) The regulated (12V) power plug will supply around 12V no matter what the load so I would have to heatsink for max the capacity of the powerplug,
2) that both my benchtop PS and the 9V unregulated power plug will supply relative to the load so the thermal buildup will not be as great and heatsink requirements less if not marginal, and
3) the regulated 12V power plug is unsuitable for this application with the circuit as is (capacitors being a separate issue).

Click to expand...

1) True. This is a Good Thing
2) Partially true. Your bench supply and the 9V adapter are inadequate and they don't operate the regulators.
3) The regulated 12V power is the best supply; but the circuit must be designed to handle the current that the LEDs need.

Thank you for the replies.

I have a couple elcos but not a 4700uF I'll see if I can squeeze a trip to Jaycar out of lunch for the caps.

I'm using lm317hvt 1.5A, but talking generically

Your bench supply and the 9V adapter are inadequate and they don't operate the regulators

Click to expand...

Yes, I understand that, they do not operate the regulators, they just supply power. The lm317's regulate the power. Is it necessary to also have a regulated power supply if the 2 linear regulators can do the work within a reasonable range of voltage supplied to the circuit?

It's not necessary to have a regulated supply; it just needs to be powerful enough.

If you use an unregulated supply, then the LM317's heat sinks need to be big enough for its actual output. This output may be a range; maybe 12V fully loaded and 15V with minimum load.

The whole purpose of a regulator is to maintain a constant output voltage over a specified load regardless of a varying input voltage within a specified limit. Due to your mention that the regulator cools down when connected to a DVM indicates an unstable circuit which can be corrected with proper capacitor bypassing.

I might be stuck in a paradigm here having started off with experience using lm317's for quick and dirty charging of batteries from solar cells working on irrigation systems. I've got some caps and will have another play tonight after work.

Acoustic Addict said:

Justwantin,Have you tried swapping these LM317s to LM317Ts? As far as the data sheet goes LM317 doesn't give more than 100mA,while only LM 317T is rated for 1.5A!

Click to expand...

The 100mA version of the LM317 is commonly known as LM317L.

Oh no! ......magic smoke tonight!. Yes the polarity was correct on the 4700u felco. I.couldn't figure out where the smell was coming from but it was the felco with a faint whiff of microchip too. maybe something popped lose and shorted on the board, dunno. Time to back off, take a break and maybe re-breadboard the whole thing this weekend and do a little reading as well.

Thanks all who contributed