LM317K regulator current problem

[smokyLED]

I'm making a bench supply out of the 5A Constant voltage/current circuit on page 16 of the LM317 datasheet http://cache.national.com/ds/LM/LM117.pdf.

I've replaced the MJ4502 transistor with a MJ2955 TO-3 transistor.

Thing thing works fine for current up to 1A, adjusted via the pot. But once u put a load on there with current adjustment more than 1A, the 2955 transistor just shorts out and dumps the whole 5A across it. If u leave this going long enough it smokes the 3.3ohm resistor as well.

I find this strange since the transitor is rated to 15A, and the circuit is meant to go up to 5A (though i've only supplied it with 2A max while testing).

I've got the transitor on a small diamond shaped finned heatsink, though this seems insufficient. Is this my problem?

Also for short crcuit protection, diodes can be added to the circuit. Namely diode acoss the out and input of the 317 regulator with the negative end connected to the input, and another across the adjust and output - negative end connected to the output.

This circuit lacks these diodes, so I take it that is is not short circuit proof? Whould I be able to add them into the circuit without affcting the operation of the opamp for current limiting?

Should the 3.3ohm resistor be more then ur standard 1/2W variant? i.e. 5W

 

[audioguru2]

I'm making a bench supply out of the 5A Constant voltage/current circuit on page 16 of the LM317 datasheet http://cache.national.com/ds/LM/LM117.pdf.

I've replaced the MJ4502 transistor with a MJ2955 TO-3 transistor.

This is your 1st problem. The wimpy MJ2955 can dissipate only about half the amount of heat as a powerful MJ4502. You should have used two MJ2955 transistors in parallel but with 0.47 ohm equalizing emitter resistors.

Thing thing works fine for current up to 1A, adjusted via the pot. But once u put a load on there with current adjustment more than 1A, the 2955 transistor just shorts out and dumps the whole 5A across it. If u leave this going long enough it smokes the 3.3ohm resistor as well.

The MJ2955 gets way too hot, has thermal runaway and nearly melts.
The schematic shows R1 as 33 ohms which is wrong. I think it should be 1 ohm so the boost transistor carries 4A.

I find this strange since the transitor is rated to 15A, and the circuit is meant to go up to 5A (though i've only supplied it with 2A max while testing).
I've got the transitor on a small diamond shaped finned heatsink, though this seems insufficient. Is this my problem?

This is your 2nd problem. You don't say what is your output voltage but if it is 5V, then the MJ2955 has 30V across it and if your R1 is 3.3 ohms then the transistor has 1.7A through it if the load is 2A. 30V times 1.7A is 51W! Your little heatsink is good up to about only 10W.

If the output voltage is very low or shorted then a simple calculation of 35V times 4.4A results in 154W of heat to get rid of. A single MJ2955 can dissipate only about 60W on a huge heatsink without an insulator, and about 90W adding a high-velocity fan.

The LM317 won't melt, when it gets too hot (35V times 1A makes 35W which is like a soldering iron) then it simply shuts-down. With your 3.3 ohm resistor then its max current is only about 0.4A and its heating is 14W. With my recommended 1 ohm resistor its max current is 1A.View attachment 37781

 

[audioguru2]

OOps!
If you add 0.47 ohm equalizing emitter resistors to two paralleled MJ2955 transistors, then R1 should be about 2 ohms/4W.

 

[smokyLED]

thanks audioguru! 

so i'd have to run 2x 2955 in parallel with 0.47 ohm resistor. 5W variant?

and change the base resistor to 2 ohm 5W

i was testing with 12V before.

 

[smokyLED]

oh, the whole circuit was powered by a bench supply @ 35V 2A (current limited)

the -6v is fed from another 10v transformer feeding a 7805 regulator. the ground pin of the reg is connected to the -6v, and the output is connect to the circuit's ground. is there a better way to do this, like using the same supply as the circuit?

 

[audioguru2]

Hi Smokey,
Each 0.47 ohm resistor will have a max of 2A and will dissipate 1.88W. A 2W resistor will get very hot so use 3W or 5W resistors.

Use the simple capacitor-coupled charge pump circuit in our 0V to 30V project for your -6V supply. It uses a zener diode for regulation.

View attachment 37783

 

[smokyLED]

wierd thing about that circuit is that i'm not getting any voltage across the zener diode

 

[audioguru2]

C3 sould have more than 6V.

 

[smokyLED]

i got 34V across C2, no voltage anywhere else

 

[audioguru2]

Then D6 is open or backwards and doesn't conduct.

 

[smokyLED]

ok, it seems to work a little now.

the input voltage slowly drops and the voltage across the zener slowly rises. didn't measure the voltage for long enough before to see this happen.

but with only a voltmeter connected across the zener and ground, the voltage seems to drop by itself. once it hits zero, the voltage doesn't rise back up unless u connect the voltmeter between the input side of the 82 ohm resistor and ground.

how does a charge pump actually work?

 

[audioguru2]

'ow do dey do dat?
A charge pump works by charging a capacitor when the transformer terminal swings positive, then discharging the capacitor into the load when the transformer terminal swings negative. I didn't calculate the actual voltages of the circuit which depend on how much the capacitor can charge during its short time when its charging current is limited by R2. Also C2 is in a voltage divider with R2 and R3 during its discharge which reduces the voltages in the circuit. ;D

View attachment 37784

 

[ante1]

Hi Audioguru,

If you don’t have the LM301A op-amp, which op-amp would you recommend?

 

[audioguru2]

Hi Ante,
I just noticed that National's power supply circuit uses their old LM301 opamp with a 41V total supply, but their max for it is only 36V. Shame on them!

The LM301 works with its inputs at its positive supply, but I don't think that feature is required in this circuit. The TL081, TL071 and LM411 opamps also work with their inputs at their positive supply but their max supply is again only 36V.

I would use the MC34071 with its 44V supply rating. It is fast so the feed-forward 75pF caps and the diode at pin 8 wouldn't be needed.
I would probably change the circuit a bit so it has a regulated output voltage down to zero volts which might be easy since the MC34071 has inputs that work at its negative supply. ;D

 

[smokyLED]

got it fixed. stablized the zener voltage by sticking a 47uF cap across it

 

[audioguru2]

Hi Smokey,
Don't you have 26VAC as an input to the full wave bridge rectifier that I showed?
Then C2 will be pumping up and down, charging C3 through D6 at 50 or 60 times per second.
Without a load, the zener diode should get warm with the negative current through it and have a voltage across it of a steady
-6V. With the tiny load of the negative supply for an opamp then it will hardly change.

Maybe your bridge rectifier is blown or connected wrong. Replace it with two diodes for D3 and D4 on my sketch.
Make certain that D5 and D6 are good by measuring them on a diode test function on a digital meter and make certain that they are connected the right way around.
Try replacing the zener diode since maybe it shorts when it warms up.

 

[smokyLED]

yea it's all good now. just needed a cap across the zener. i'm using a bridge rectifier instead of the 4diodes

 

[smokyLED]

thanks for all the help u've give audioguru

 

[ante1]

Hi Audioguru,

So you noticed the flaw, I thought I found you sleeping on your watch (fat chance)! ;D Would an LF411, LM318 or the OP07C work here, I don’t have any MC34071?

I would probably change the circuit a bit so it has a regulated output voltage down to zero volts which might be easy since the MC34071 has inputs that work at its negative supply. ;D

Will I see a schematic of your mod? 8)
Please?