LM317 as CCS

[deicide67]

This is dead horse and I know it’s not a great CCS due to non-linearity, but I put a 317 set to 45mA on the cathode of my 807 SE amp by passed with a 100uf cap and it starts collapsing as the signal increases. Drops the cathode voltage down to around 2V at full tilt.

I popped in a 250ohm resistor and instantly get way cleaner sine waves at nearly twice the output power before clipping. What causes this? Non-linearity wouldn’t cause what I’m seeing I don’t think.

Thanks!

 

[nigelwright7557]

The 317 is maybe putting the valve on a different part of the load curve to the resistor ?
You need to see if it needs more or less current to behave better.

 

[deicide67]

I think I will put a pot on the adjust pin and try that. Just seems strange. I know they clip funny, but this just seems wrong. Thanks!

 

[abraxalito]

I've had LM317s oscillate when running as CCS's. Particularly when there's a lot of voltage across them - I tend to use series resistors to drop the excess volts, keeping the drop across the LM317 to under 4V.

 

[deicide67]

I’ve done that too. I just assumed 18V is well within the voltage limits of the 317.

 

[trobbins]

There would be a voltage shift across the 317 with change in signal level - that voltage shift has to stay well within the allowable span for the 317 to operate as a CCS - you may need to monitor the device voltage drop via a scope.

 

[sottomano]

Walt Jung about CCS, LM317 incl. cascoded variants, in Part 2

Part 1
Part 2

 

[Refugee1]

A CCS for a SE amplifier needs to have a time constant in the feedback loop that is longer than the lowest frequency required by the bass bin.
If this is not true the bass will be clipped by the CCS starting to regulate before you want it to start working.
Push pull amplifiers are a lot easier.

 

[deicide67]

That’s a thought. I did not pull the feedback to see if it fixed the issue. I’m using Schade (was plate-to-plate) in this amp as well as a touch of global that could be interfering with the CCS.

 

[deicide67]

A CCS for a SE amplifier needs to have a time constant in the feedback loop that is longer than the lowest frequency required by the bass bin.
If this is not true the bass will be clipped by the CCS starting to regulate before you want it to start working.
Push pull amplifiers are a lot easier.

Thanks! I will remove the feedback and see if that helps.

 

[Kay Pirinha]

Are you sure you've wired the LM317 the right way, i.e. a resistor of 1.25 V/0.045 A = 27 ohms between OUT and ADJ, the 807's cathode to IN and ground to ADJ?
Best regards!

 

[deicide67]

Yes. The bypass cap is not pictured here, but the 317 is.

 

[RajkoM]

I use long time CCS with LM317 (72mA and 1K helper resistor between cathode and CCS) in my SET. At first I had problems with modern LM317 (I do not remember of manufacturer). When I took NOS gold pin Motorola LM317, everything was fine.

Just for info.

 

[deicide67]

Thanks! What’s a helper resistor?

 

[RajkoM]

As abraxalito explained in post 4.

It is a powerful resistor (I use 12W Mills) between cathode tube and CCS. The main dissipation is on this resistor and then LM317 worked at about 5V working voltage.

 

[deicide67]

I see. I’ve never heard it called a “helper”. Just a cathode resistor to split the load with the CCS. I’ve done this before as well. Thanks!

 

[RajkoM]

Like this

 

[deicide67]

Yes. I’ve done this. Seems like a good thing to try out. Thanks!

 

[Alllensoncanon]

I popped in a 250ohm resistor and instantly get way cleaner sine waves at nearly twice the output power before clipping. What causes this?

If the LM317 is dead (aka: shorted), 250R then replace the LM317 in the circuit. Of-course there won't be any CCS function coming from the dead LM317.

 

[deicide67]

It’s not dead.