LM317-based regulated PSU: how does this thing work?

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Jump to Latest
F

Fred Dieckmann

diyAudio Retiree
Joined 2002
Confessions of an audio drama queen?

"Fred, we had been on the productive posts, until someone told us that this is not a soft start circuit.

Hm, do you remember who that person was?"

I believe most everyone has moved on to talking about circuits to provide improve the regulator performance. I once again advise people
to read the data sheet for the regulator and reread the thread after getting an understanding of how the regulator works. I don't know what page you on and what your agenda is. It sure isn't helping anyone and most of all is making you look more silly with every post. Sorry you are too busy being a drama queen the join the others who want to get on on with improving the regulator circuit. :bawling:
 
jan.didden

jan.didden

AX tech editor
Joined 2002
linearaudio.net
Paid Member
millwood said:



Jan, I am pretty sure that I don't understand regulation. more than that, I am also sure that the guys at National don't have a clue about regulation either. This is what they said about that capacitor. Pay attention to Note C.
This is straight from National's datasheet, Page 4.

since it can be hard to read, please allow me to type it up here: "Cadj is used to improve ripple rejection; it prevents amplification of the ripple as the output voltage is adjusted higher." Isn't datasheet wonderful?

Hope it helps.
Click to expand...

This will probably get me the first price for arrogance (oh well, being always second isn't that great either), but I think they got it wrong, or at least are not clear enough. Those caps are wonderfull to suppress line ripple and input noise, whatever, as long as the load is constant. And that is what they say.

BUT, when the load starts to vary, and introduces ripple by the varying load current through the non-zero output impedance, these caps are preventing the rgeulator to do what it is paid for - regulate. That is not addressed in the data sheet at all. Sorry.

Yes, data sheets can be wonderfull. For those who understand them.

Jan Didden
 
M

millwood

Disabled Account
Joined 2003
janneman said:
The ripple on the output needs to be fed back (remember feedback?) to the adjust pin to make sure that the ripple is counteracted by the regulator giving out more or less current to keep the output stable. With a cap on the ref, that feedback doesn't work, the reg works essentially as a follower, load regulation is all but absent. Basics, really.

Jan Didden
Click to expand...


maybe the basics is a little bit different than perceived here?

Is it possible that this feedback isn't that feedback? R1 is there to generate a constant current flow through R2 and establish a voltage on the Adj pin. so to the extent that there are ripples on the output pin, they will be reflected, dollar for dollar and euro for euro, on the Adj pin. With a cap from the Adj pin to the ground, hopefully we will get rid of such ripples and hold the voltage on the Adj pin constant, because of Vref, achiving constant output.

Is that plausible?

here is what spice said about using the cap. I injected a 1vp signal to Figure 3 of the national datasheet:

with a 10uf cap: output ripple is 200uvp.
with a 1pf cap (essentially no cap): output ripple is 4mvp.

Which of the two has better regulation?
 
Eva

Eva

Member
Joined 2003
janneman :

Floating regulators are very tricky since the voltage reference is relative to the output, not to ground, so it's better to think twice before writing...

The capacitor from 'Adj' to ground causes the regulator to work as a 1.2V reference follower at high frequencies, so all the AC error voltage present on the output is feed back to the Adj input without being divided by the voltage divider that sets the DC operating point

In theory, the capacitor could enhance both line and load regulations by a factor of 'R-gnd-to-adj' / 'R-adj-to-out'

You can imagine the floating regulator as an opamp whose non inverting input is the 'Adj' terminal and whose inverting input is connected to the output through a 1.2V reference

Just a stupid drawing...
 

Attachments

  • floatreg.gif
    floatreg.gif
    1.7 KB · Views: 685
M

millwood

Disabled Account
Joined 2003
Re: Confessions of an audio drama queen?

Fred Dieckmann said:
I don't know what page you on and what your agenda is.
Click to expand...

I am still on Natonal's datasheet, and my agenda is to rid the forum of mis-information.

Yes, Fred, that IS a soft start circuit. Yes, Fred, it was implemented correctly. Yes, Fred, those National guys really know what they are talking about.

janneman said:
This will probably get me the first price for arrogance (oh well, being always second isn't that great either), but I think they got it wrong, or at least are not clear enough.

Jan Didden
Click to expand...


Jan, you are starting to sound like Fred, :)
 
M

millwood

Disabled Account
Joined 2003
Eva said:
The capacitor from 'Adj' to ground causes the regulator to work as a 1.2V reference follower at high frequencies, so all the AC error voltage present on the output is feed back to the Adj input without being divided by the voltage divider that sets the DC operating point
Click to expand...

Bingo!

Finally, someone giving the National guys their due respect, :).

What is "confusing" is that the "divider network" really isn't a divider network. R1 is there to generate a constant current flow into the lower resistor so that a voltage can be established on the Adj pin, in order to adjust output voltage.
 
jan.didden

jan.didden

AX tech editor
Joined 2002
linearaudio.net
Paid Member
millwood said:



maybe the basics is a little bit different than perceived here?

Is it possible that this feedback isn't that feedback? R1 is there to generate a constant current flow through R2 and establish a voltage on the Adj pin. so to the extent that there are ripples on the output pin, they will be reflected, dollar for dollar and euro for euro, on the Adj pin. With a cap from the Adj pin to the ground, hopefully we will get rid of such ripples and hold the voltage on the Adj pin constant, because of Vref, achiving constant output.

Is that plausible?

here is what spice said about using the cap. I injected a 1vp signal to Figure 3 of the national datasheet:

with a 10uf cap: output ripple is 200uvp.
with a 1pf cap (essentially no cap): output ripple is 4mvp.

Which of the two has better regulation?
Click to expand...

You don't give up, do you?

Everything above: "is that plausible" is nonsense. You don't get rid of the ripple with the cap. How can that be? Does the extra current in the cap somehow magically induce the circuit to counteract the ripple?

But this is a REGULATOR for Pete's sake. I really get very tired now. It can only regulate if the output deviations, caused by the varying output current through the non-zero output impedance, is somehow fed back as information to the input of the error circuitry which is at the adj pin. With a cap, that info doesn't get there. It does a great job keeping out ripple, but it doesn't regulate. So, to answer your question: the one with the 4mV is the better regulator.
With the cap, it basically is a voltage follower, which has been confused with a regulator more often on this forum than I care to remember.

Jan Didden
 
M

mwh-eng

Member
Joined 2004
I trust the Spice analysis! I agree that reading the application note is important. Careful reading of application notes usually saves me from some trial and error. Sometimes I don't breadboard and start with PCBs. Using Pspice, Spice, or equivalent with good models and reading application notes, usually saves me time and money.

Someone has probably already mentioned that Linear Technology offers a useful free Spice called SwitcherCAD. I use Pspice, but my friend prefers SwitcherCAD.

http://www.linear-tech.com/new/design_tools.html
 
Eva

Eva

Member
Joined 2003
millwood said:


two problems I have with it is that a) it has a positive feedback loop; and b) R1 should somehow be linked in the 1.2v loop.
Click to expand...

Yes, floating regulators have positive and negative feedback at the same time, but negative feedback has higher magnitude

The purpose of the capacitor is to reduce or eliminate positive feedback at AC in order to enhance AC regulation
 
F

Fred Dieckmann

diyAudio Retiree
Joined 2002
he is just messing with millwood

"With the cap, it basically is a voltage follower, which has been confused with a regulator more often on this forum than I care to remember."

The ADJ is the feedback terminal for the regulator. Bypassing the resistor to ground decreases the AC gain while leaving the same DC gain, chosen to get the desired Vout from the 1.2 volt reference voltage. Bypassing the feedback resistor with the cap increases the amount of feedback and therefore lowers the output impedance, lowers the noise, and increases the PSRR like would be expected since is essential an op amp with a pass transistor and internal voltage reference. The curves in the data sheet bear these trends out. Am I the only one that sees this? Surly not ........... John, Nelson, Walt?

"a) it has a positive feedback loop":whazzat: :smash:

Am I the only one that is sober? Or maybe you have conspired behind the scenes to drive me mad. This has gone from silly to surreal. A Salvador Dali painting is a better example of reality than the stuff I am reading here. Am I on Candid Camera. :confused:

YOU PEOPLE ARE FREAKING ME OUT :bigeyes:
 
M

millwood

Disabled Account
Joined 2003
Richard C said:
Negative feedback dominates via the 1.2V ref.
Click to expand...


Eva said:


Yes, floating regulators have positive and negative feedback at the same time, but negative feedback has higher magnitude
Click to expand...


I think that's right. My 2nd concern wouldn't be valid as the non-inverting end will be essentially 1.2v lower than the output so the R1 is essentially in a loop with the 1.2 Vref.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Top Bottom