Quick question: parallel LM317

[maxlorenz]

Hi,
I run a multi TDA1543DAC. I have some HF noise. One of the possible sources that I want to discard is noise generated by the LM317 regulator running too hot or near/over its current limit.
Typical max I for TO-220 is 1.5A and limit is 1.8A.

I made a test regulator consisting of two units with its pins soldered in parallel and they keep the desired Vout with only the standard pair of R.

By this mean I want to make that each reg passes half the current and *possibly* works more quietly

Does anybody see any inconvenient to this approach?

Thanks in advance...
M

 

[jackinnj]

it's just a little expensive to do it that way, but National Semi shows it in their product folder. (i have done this with their switching regulators --- just stacking one on top of the other.)

the amount of heat which can be dissipated in one of these regulators is a function of current times the differential between in put and output voltages.

 

[pinkmouse]

Would it be less noisy to use a meaty pass transistor?

 

[jackinnj]

it may be less noisy to have the two references operating -- but I haven't demonstrated this (and don't want to spend the time doing so).

 

[maxlorenz]

Thanks guys

I just wanted to be sure...

National Semi shows it in their product folder.

Oh! I missed that, sorry...

Would it be less noisy to use a meaty pass transistor?

A good friend designed a circuit for this application but I haven't got all the parts needed

Anyway, I think I'll learn in the attempt

Best regards,
M

 

[jcx]

Sorry guys but you're all wrong

you may get lucky but you shouldn't literally parallel pins of a pair of feedback amplifiers -which is what the Vregs are

if the internal Vrefs, loop gains don't match very exactly then the low impedance outputs will not share the load equally, with the Vregs only being able to source current at least it may not blow up like opamp or chip amp outputs tied together without sharing/limiting series R, but likely one reg is supplying 90% of the current

pinkmouse' approach is fine if you have a spare ~ 1 Volt of headroom

 

[maxlorenz]

Mmmhh...thanks, I'll study it again.

 

[maxlorenz]

Hi Jcx,
You were right, of course.
I did the test. I can't measure current but the new paralelled reg is hotter than the ancient: 58ºC v/s 38ºC, meaning, I guess that one is doing all the task.

The HF noise is still there, so it is comming from other place.
I would say music is clearer ...but I'm not sure.

As I can't source parts easilly, would it be worthy if I reproduce the regulator circuit, to end with two paralelled paths?
Will they pass a comparable current?

Thanks for your help,
M

 

[BWRX]

You could try this circuit http://www.diyaudio.com/forums/showthread.php?postid=1005645#post1005645 if you want better performance and have a beefy NPN or N channel FET handy.

 

[maxlorenz]

Thanks BWRX,
I'll check my box...

 

[jackinnj]

Sorry guys but you're all wrong

you may get lucky but you shouldn't literally parallel pins of a pair of feedback amplifiers -which is what the Vregs are

if the internal Vrefs, loop gains don't match very exactly then the low impedance outputs will not share the load equally, with the Vregs only being able to source current at least it may not blow up like opamp or chip amp outputs tied together without sharing/limiting series R, but likely one reg is supplying 90% of the current

pinkmouse' approach is fine if you have a spare ~ 1 Volt of headroom

yupp, National shows 0.22R load balancing resistors -- in using the SimpleSwitchers from National, all you have to do is stack the devices. I don't think that these buck/boost regulators get a lot of use in quality audio circuits.