LM317 power supply - anomalous behavior

[rodolfok]

Hi, I've made a variable voltage power supply with LM317 and I'm testing it.

The scheme includes two ranges, first connecting one and then two power supplies in series (two identical switching PSU for Laptop PC), to switch the input voltage to the LM317 between 20V and 40V and obtain outputs in the ranges 1.3-17V and 16.5-35V. To change the ranges, I use a switch that commutes the series connection of the power supplies on/off and sets/removes the short-circuit of two resistors (R2, R8) which are used to change the LM317 divider's ratio.

The anomalous behavior consists in the fact that if you set the minimum voltage of the second range (at 16.5V, i.e. with the input of LM317 at 40V) and then switch on the first range (at 1.25V), where only one power supply (20V input) is feeding the LM317, the output voltage instead of reaching 1.25/1.3V settles at 19V and remains there until you rotate the potentiometer a bit, increasing the value of R5. This only happens when the potentiometer is at the zero resistance and only by switching from high to low range. During this phenomenon I have a current of approx. 100mA on the branch of the divider (R7-R2-R8-R5), where in this position R8 is short-circuited and R5 is rotated to ground.
I have done other tests, putting a capacitor on the potentiometer and adding another one closer to the input of the LM317, there seems to be no oscillations but the situation does not change.

I have verified, however, that if I delete completely the "PCB 2" block (composed of TR1, TR2, R3, R9, see attached complete schematics), the above mentioned phenomenon does not occur, i.e. when switching from pos.2 to pos.1 at the minimum voltage the value settles correctly at 1.25V.

Someone would know where the problem could be?

[Benta]

The series resistance of R2 + R7 can not be higher than 120 ohms. I can't read the schematic, so I'm not sure of the values you're using.

[Damianos]

Welcome to the forum!

I guess that the lower end of R5 is somewhat burned!
Think that, when it is in the high range, at what voltage the capacitors are charged. Then, by switching to the low range, C2 is discharged on R5 (plus some current through R7-R2).

[rodolfok]

Hi,
I also tried to short circuit R5 and the behavior is the same. While the output stays at 19V, the voltage on the Adj pin is about 20mV, so I have about 19V across the Out and Adj pins of the LM317. This problem does not occur if I disconnect the power transistor; I tried to change them for a single PNP, but it's the same. The only way is to turn off the power supply before changing the range. Maybe I could use a switch with an "off" position between the ranges, or permanently connect the two 20V units in series. The original idea was to have two ranges to limit the voltage dropout and lower the power dissipation thrgough the power BJT.
The two values of R2 and R7 are the actual measured values, and chosen to have about 1V overlap in the ranges.

[Damianos]

Is there a load at the output? Try initially 100mA. Also try without the C3.
Is the R1 in good condition or all the current is passed through the transistor(s)? --> Looking again at the schematic diagram: If the value of R1 is 68 Ohms, it is too high. Decrease it at least one order of magnitude , so that the regulator has the control...

It seems that the excess voltage between the Output and Adjust pin(*) results in a locked condition in which the regulator draws enough current to keep active the transistor(s).
(*) there is a 6V zenner between these pins...

[rodolfok]

Hi Damianos,
thanks for the hint! Lowering the R1 value (39Ohm now) did it, the supply is working fine now, with or without load.
Cheers
rodolfo

[mikerj]

Hi Damianos,
thanks for the hint! Lowering the R1 value (39Ohm now) did it, the supply is working fine now, with or without load.
Cheers
rodolfo

Why have you made R1 so large?  This resistor should typicaly be a few ohms or lower.

[iMo]

I would recommend to use something like this with your power part - see below:

[Zero999]

The circuit has some flaws you should beware of.

The short circuit protection circuit won't work, when the output voltage falls below 1.25V. In order to make the LM317 go to 0V, the adj pin needs to be taken 1.25V below the 0V rail. If the output of this circuit is short circuited, a large current will flow, resulting in destruction of Tr1.

Even if the current limiting circuit is fixed. If the high voltage range is selected and the current limiting kicks in, the power rating and safe operating area of Tr2 will be exceeded. The output voltage should be sensed and a relay used to switch to the appropriate input voltage.

C1 is a bad idea. If power is suddenly applied to the circuit, the current spike will cause R1 to drop >0.7V, which will turn on Tr1 and Tr2, even if there's a very light load on the output of the circuit, resulting in nearly the full input voltage being applied to the output, for a short time.

[rodolfok]

Hi,
could you suggest and possibly sketch the modifications to apply (only if it doesn't eat-up too much of your time)?
Thanks for the comments.
rodolfo

[iMo]

The OP does not plan to go below 1.3V output. The over-current protection should be placed somewhere around TR1 and TR2 instead, afaik.

Imagine a short. The over-current protection kicks, and the output voltage drops to aprox. 1.3V.
The large current will continue to flow, however.

With 20V input, 1.3V output and say 3.25A current (the notebook power supply limit) the TR2 will dissipate (not counting the voltage drop at R4)

P = (20 - 1.3) * 3.25 =  60.8W (with 40V input P=126W) of heat.

[Zero999]

The over-current protection, provided by the mains power supply can't be relied upon to protect the regulator circuit. The PSU might be able to provide double the rated short circuit current, possibly even more for short periods of time. It's seldom specified.

Yes, the over-current protection needs to be before the driver transistors. A relay or power transistor, can also be used to switch the input from one voltage to another, depending on the output voltage.

The attached circuit shows both current limiting and input voltage tap switching.