Weird LM317 behaviour

[Murdock]

I have a weird acting LM317. Or at least I think it is weird as I'm quite a novice when it comes to regulated power supplies.
I wanted to build a little PSU for some measurement mics I got. They need 150V polarization voltage and 30V for the impedance converter circuit.
I still had a unused G7 psu board and power transformer here which I missused for that purpose. The power transformer had a 135V/50mA and a 9V/500mA secondary.
As the impedance converter circuit only draws 5mA max and the capsule almost nothing I thought I could just use the 135V for both using the LM317 to regulate the 30V.
Modified the circuit a bit and came up with a crude supply.
After some troubleshooting everything seems to work well. I know, it probably is not a good circuit but I hacked it together to find out if the mics work.
But what puzzles me is that the regulated output is only quiet when the Input-Output Differential Voltage is almost nothing...
You can see in the circuit I opted for an output voltage of 28.75V (100 Ohm and 2k2 Ohm). R6 is a poti.
When the input voltage is over 31V I get some kind of white noise on the mic output. I go below and everything is quiet.
I thought and read that the minimum recommended input-output differential voltage should be 3V-5V.
So can someone explain what's going on?

 

[JohnRoberts]

what is the max input/output voltage across the 317

JR

 

[Murdock]

I can vary it with the Poti R6 up to 40V. So max Input-Output differential voltage can be around 11V.

And I forgot that there is a 10uf electrolytic across the reg output.
Edit: when I simulate that circuit in spice it seems the LM317 can't handle 100 Ohm for the "upper" resistor. No matter the input the output is always around 1.25V lower. When using the more common 240 Ohm it seems to work "normal".
Is 100 Ohm too low for the LM317?

 

[TwentyTrees]

Check the datasheet, but from memory 200-240 ohms is recommended there.

 

[Newmarket]

Additionally it's advantageous to have a noise reduction capacitor from the ADJ terminal to 0V ?

 

[Murdock]

Check the datasheet, but from memory 200-240 ohms is recommended there.

I know that 240 Ohm is the recommended value here. But I read that the minimum load current for the LM317 is around 10mA. As the mic only draws around 1mA (5mA max) I thought I use 100 Ohm to get 12.5mA load for the regulator to be happy... And I read several post from other people using 100 ohms without a problem.

Additionally it's advantageous to have a noise reduction capacitor from the ADJ terminal to 0V ?

That could be, but that doesn't explain the weird behaviour or does it? Then it would always be noisy or?

 

[ccaudle]

My first suspicion with noise like that is oscillation on the regulator output. When the in-to-out difference is low the device is out of regulation, once the voltage differential is high enough the device starts running and has enough gain to oscillate. Three terminal regulators are pretty robust, but enough inductance in the layout might get them oscillating.

 

[Matt Syson]

The 100 ohms should be fine and dive a reasonable 'standing current' to make the 317 regulate. Adding a capacitor across the 2k2 (as shown in the schematic improves the ripple performance but you should add a 'reverse connected' dipde across the 100Ohms so taht if the output were ever shorted the capacitor current won't go back into the adjust terminal. I expect it would prefer to see more capacitance on the output and possibly even a small 'stopper' resistance of a few ohms or more depending on the HF performance of the output capacitor. You should also prevent the input to output voltage from ever being greater than 39 Volts so either a 39 or 36 Volt zener on input to ground or other technique.

 

[jrg.roberts]

Could you supply DC/AC and ripple voltage at each major point?