LM317 based pwrsupply oscillates?

[Gripp]

Hi all!

First I must say that my former post on this subject. http://diystompboxes.com/sboxforum/viewtopic.php?t=25066 was rather confused due to sleep deprivation and frustration. Sounds familiar :wink:

Of course the pwrsupply, built like this http://www.geofex.com/FX_images/pedschm.gif minus Rbatt plus an extra 0.15 uF cap at the end, has some peculiarities to it.

When I scoped it loaded by the orange squeezer (draw ca 4 mA) I saw a train of spikes with fundamental freq of 50kHz ca 5mV ppk plus additional higher harmonics being visible after each spike (lower in amplitude). All this not present when powered by a battery. The pwrsupply seems to regulate as it should, output 9.6 V loaded (have only loaded it with 20mA max so far) and unloaded.

So the orange squeezer must be extra sensitive to pwrsupply and the noise observed in my pwrsupply must be the cause of the higher opamp and/or Jfet noise or there is some other fundamental difference between a good pwrsupply and a battery that I don't understand.

Now to the questions. Is this 50k signal due to oscillation in the pwrsupply or? Is it common? How do I get rid of it in the best possible way?

Thoughts to those living in Ivan's path. Hope all is well.
Best!
/Pelle Garpebring

[spongebob]

When I scoped it loaded by the orange squeezer (draw ca 4 mA) I saw a train of spikes with fundamental freq of 50kHz ca 5mV ppk plus additional higher harmonics being visible after each spike (lower in amplitude).

Did you see this on the output of the orange squeezer or only on the power supply?

How do I get rid of it in the best possible way?

You could try adding a 10uF bypass cap from the LM317 adjust pin to ground, this is said to improve ripple rejection, according to the datasheet. Not sure if this helps your problem, but it can't hurt!

What kind of transformer did you use for the power supply?

[Gripp]

Thanks Spongebob!

My mind wasn't exactly present when checking my problem out yesterday either....  tried to drill and solder an EA trem at the same time. I didn't scope the output of the squeezer, only the supply line in  :oops: . None of my other pedals sounds affected in a bad way by this powersupply though. Just the squeezer having much more noise (hiss) when driven by it compared to a battery of the same voltage.

Will try the extra cap on the adjust pin and will definitively scope the output of the squeezer also. I'm just having trouble understanding what such a high frequency signal is doing there and what it's doing to the squeezer. I thought that ripple would cause hum, ie 100Hz here in sweden.

I'm using a 20VA 12V toroid transformer originally intended for illumination duties.

Best!/
Pelle Garpebring

[yano]

Look at it this way, you say the signal is 50kHz? Humans only hear up to around 20kHz, or maybe 22kHz. Since that is out of human hearing range, its not a problem. That is probably what the lm317 operates at.

For instance MAX1044 has a pin called boost, when you connect Vcc to it, the operating frequency of the chip its increased by a factor of 6, which moves it out of auditory range. With the pin disconnected, there is an audible whine.

Perhaps you could reduce the noise if you filtered the higher frequencies out, but I'm not really sure how much of an effect that would have.

[mikeb]

The LM317s aren't switching cap devices like the MAX1044 etc, so don't have a frequency of operation. I'd try simply subbing in a different LM317 - sometimes these regulators just don't work as they should 'out of the box'. Also make sure that the  bypass cap is as close to the leads of the unit as possible.

Mike

[niftydog]

you really need to scope the circuit more thoroughly. Input the the regulator should be pretty smooth thanks to the 1000uF cap. It should also be a decent amount above the required DC output voltage.

Then, the output should be smooth as a babys bum.

50kHz is too high to be ripple and by my guess, too "precise" to be oscillations. Normally you wouldn't get pulses and hamonics, you'd just get a whole crap load of noise and hash... basically just a mess.

Check out the data sheet for the LM317 and implement any extra caps it might suggest.

I'm wondering if it's coming in on your supply line. Sometimes there's odd little "communication" type signals imposed on top of the mains at much higher frequencies.

[R.G.]

The LM317, in common with other three terminal regulators, is a feedback system that can become unstable under certain conditions.  This may or may not be what is happening to you.

The National Semiconductor app notes for the 317 states that you need a decoupling cap of 0.1uF if the 317 is more than 6 inches of conductor away from the power supply filter cap. It's good practice to just put one there anyway. The output really should have a 10uF or more cap at the output pin.

If neither of those fix your problem, then it's probably not the LM317, and it's likely power line junk, or oscillations from something else showing up there.

And yes, over-audio oscillation sometimes sounds like aggressive hiss back in the audio channel.

[Gripp]

Thanks for all the helpful tips guys!!

I'll add the caps as per the LM317 datasheet although no component is more than 6 inches away from another. I built this on a really small piece of perfboard.

Mikeb:
I will order a new LM317 if everything else fails. I knew about the LM317 being an ordinary regulator and not a switching one. I did not however, know about the failrate of those and I havn't got to understanding their operation...yet :wink:


Niftydog:
Will definitively scope some more. I did scope the output of the supply filter caps when I had just built it and at the time (unloaded) it looked pretty smooth.

Input the the regulator should be pretty smooth thanks to the 1000uF cap.

Thats what I thought too. But I'm just beginning learning and relearning electronics here but I'm trying pretty hard. It's been 12 years since I had some at high school :wink: And yes, the output from the supply filter caps was ca 15V (directly out of my head). And that should really be enough.


RG:

And yes, over-audio oscillation sometimes sounds like aggressive hiss back in the audio channel.

That's exactly what I was suspecting. Thanks for confirming it. Not saying that it's the case here though. I have the 10uF cap on the output of the regulator as per the schematic and I also added an 0.15uF film cap. I looked at a number of power supply schematics out there and saw that it was more or less practice to add a 0.10uF (only had a 0.15uF) at the output to "help" the electrolytic cap at higher frequencies.

Well, back to the workbench tonight. Hope I have the parts at home...haven't been stocking up much.

Have a nice weekend all!
Best!/
Pelle Garpebring

[Paul Perry (Frostwave)]

Perhaps try a different manufacturer for the LM317, if you can. In my experience, power regulators are the one thing that can vary in behavior between manufacturers, though you would expect them to be simple..