Keeping Shunt current output consistent

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
AX tech editor
Joined 2002
Paid Member
That looks OK.
You don't need R8.
there's another improvement you can do: feed the FET CCS not from the raw rectified DC, but from Vout!
This should work if Vout is like 5V or more above the reference voltage.
I am not 100% sure it will start up that way (I have done some things like that and it *may* not start up) but when it works it gives you a very good reference and very good Vout.
Of course the right side of R2 goes to the top of the shunt voltage reg, the wire hanging in the air can be deleted.

jan
 
That looks OK.
You don't need R8.
there's another improvement you can do: feed the FET CCS not from the raw rectified DC, but from Vout!
This should work if Vout is like 5V or more above the reference voltage.
I am not 100% sure it will start up that way (I have done some things like that and it *may* not start up) but when it works it gives you a very good reference and very good Vout.
Of course the right side of R2 goes to the top of the shunt voltage reg, the wire hanging in the air can be deleted.

jan

Great idea! Thanks for that, worked perfectly!

I found Walter Jungs article on current sources:

http://www.audioamateur.com/media/jung2779.pdf

Cascoding a DN2540N5 depletion mode mosfet with the LM317 looks like a good thing to do also. I will play around with that a bit.

I have another project and I was wondering if I could get the same circuit to provide about 1.2 amps. Looks like the LM317 is OK for that (with a heatsink). I could not figure out anyway to use Walters Jungs circuits to increase performance of the LM317 at such high amperage. Any suggestions?

Thanks Jan!
 
AX tech editor
Joined 2002
Paid Member
There are other regs like the 317 that can provide 3 or even 5A; don't remember the type # off the top of my head.
They all can be used similarly. Like the 317, when you know the internal ref voltage between adj and out is 1.25V, and you know the current you want it to source, simple V = I/R will give you the resistor value (your R2).
So you can use any integrated reg for this, but check that the internal ref is also 1.25, may be different but I think 1.25 is pretty universal.

Edit: The LM1085 is a low dropout positive voltage regulator with a maximum dropout of 1.5V at 3A of load current. It has the same pin-out as the LM317. Or the LP38513 for better hf preformance. Or the LMS1085 and 1087 (5A). The sky is the limit! ;-)

jan
 
Last edited:
Thanks!

I am looking at the circuit now and I have the Jfet current source hooked up to the Vo of the LM317 as you suggested. It is working perfectly. Voltage output looks cleaner on the scope. : )

Do I really need a Jfet curret source there?

Will a resistor work just as well? Or a current limiting diode?

I do not understand the need for a Jfet if I am hooked up to the rock solid clean voltage output of the LM317?
 
R8 reduces the dissipation of Q5.
It can be linked across for low voltage operation or made very big for high voltage operation.
Keep the space on your PCB to allow for flexibility in your applications.


The raw rectified DC that this was connected to is about 12 volts max.

But I am currently connecting the jfet directly to the Vo of the LM317 which is about 3.4 volts per Jans advice and it is looking good.

I do not underestand the need for a Jfet at this point. A simple current limiting resistor should be fine instead?
 
AX tech editor
Joined 2002
Paid Member
The jFET looks like a very high impedance so the current it sources into the ref diode is very nearly independent of any ripple and such on the voltage that feeds it. That will help making the reference cleaner.

You have understood the issue perfectly!

Now that the jFET is supplied from a rock stable supply in itself, the difference between a resistor and a jFET may be very small if any. My bet would be that there would be no audible difference.
So yes, a resistor will work well here, but since you have the jFET already in, and it is very cheap, why not use a belt AND suspenders ;)
And it's a bit more flexible. If you change voltage levels you don't need to adapt the resistor, since the jFET will automagically keep the current it sources the same.

jan
 
Hi Jan,

Yep, me again! ;)

That confirms what I am hearing. The LM317 CCS is not sounding as good as what I had before.

I tried using a cascode DN2540 as Walt Jung advises in his CCS article, but I just do not have enough DC voltage to keep the LM317 (Vo-Vin) > ~3 volts.

I was at about (Vo-Vin) equal to 1.5 volts, but even with that it sounded better than just the LM317 on its own.

I was going to try a current limiting diode feeding then LM317. These diodes are suppose to have quite a bit of high frequency rejection from what the data sheet says.

Any other ideas are of course most welcome! :)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.