Linear voltage regulator: how to make good use of "sense" and "ground sense"?

[NeoY2k]

Linear voltage regulator: how to make good use of "sense" and "ground sense"?

Hi!

There are some linear voltage regulators (integrated or not) that have "sense" pins and even "ground sense" pins for some.

These should be routed the nearest to the load.

Do you have ressources on how to make good use of these options as I guess they could make things much worse...

Example: My topology is:

regulator=>ferrite bead=>bypasse cap(to ground)=>several loads (each one with their own small bypass cap).

I though I'd put the "sense" after the ferrite bead and the "ground sense" where the ground of the cap is.

Then, the regulator could "try" to compensate for things the cap would have left in. BUT it could also pick up more noise and then disturb regulation.

I could then use a filter on the sense line, eventually being careful not to shift the phase too much if I expect high speed regulation, ie like Walt Jung's 2 is supposed to act (but how fast is a regulator...?).

Is it worth it in the end, even a very small bit?

Thanks,
Nicolas

 

[bear]

Probably a good idea to move this question over to the Solid State section??

_-_-bear

 

[unclejed613]

or better still, to the "POWER SUPPLY DESIGN" forum.....
but to answer the question. the sense lines go to the load. before or after a ferrite bead doesn't really matter because the bandwidth of the regulator is only a few hundred hertz, usualle enough to remove ripple harmonics, but not much beyond that. using sense lines is a technique used mostly in industrial applications (which covers an extremely broad spectrum). long runs of wire are a fact of life in industrial equipment (anywhere between 3 and 20 feet) so some means of compensating for voltage drop in the wiring makes sense (no pun intended). usually i use open frame regulated supplies with short wires of adequate guage wire, so i just tie the sense lines to the output terminals. some "tricks" can be done with the sense lines (within the limits of the power supply of course). one of them is putting a couple of diodes in series between the + output and the + sense line to bump up the voltage a little. this is useful with a 12V supply if you want to bump it to 14.1V to use as a gel-cell charger without going in and changing the Vout adjustment. 3 1N4004 diodes in series between the output and sense lines will "program" the output to 14.1V, and still maintain regulation.

 

[jan.didden]

The sense lines should indeed go to the load, as close as possible. There is quite a lot to say about where to ground the rectifier capacitors, the opamp negative supply, the reference return, etc.

You might need some filtering on the sense lines as a good regulator is a quite wide bandwidth device with very high freq response. (The few 100 Hz mentioned above isn't correct).

You may look up 'super regulators' on Walt Jung's website www.waltjung.org . I did an article with Walt some years ago that had a separate part (part 2) about exactly these issues.

Edit: It's actually Part 3: http://www.waltjung.org/Regs.html

Jan Didden

 

[benchtester]

Linear Technologies has a regulator with sense pins LT1087. However, they don't "recommend it for new designs." So it may be hard to find.

Their parts LT1083 through LT1086 are good three pin regulators with lower noise than plain three terminal regulators.

I don't use the ferrite beads on the linear regulators.

Instead of one regulator with several loads, I would suggest giving each load its own regulator. So you could use simple three pin regulators located where the sense lines would ordinarily go. I think this would show the greatest improvement.

Best wishes for you project!

Edit: Be sure to use a good grounding practices, for example star grounding.

 

[NeoY2k]

Thank you very much.

Janneman, I didn't read all the Jung's papers, only the last rev of his regulator. I'll read all the others.

I was wondering: what is the voltage drop across this regulator? I'm designing my power supply, if I could avoid making another one to test these regulators...

In fact, I'm designing a DAC and I provided lines for remote sensing to play with it. I think I put the sense where they had to be, now things will be up to the regulator board... I see a lot of prototyping xD

I'll read further, and keep you informed when I'll be baking and testing some boards.

Thanks again

 

[martin clark]

To answer the original question - yes, it is worth trying even with run-of-the-mill adjustable 3-pin voltage regulators. Subtleties in your implementation can make for huge difference in the results! Here's Werner Ogier's excellent overview:

http://www.tnt-audio.com/clinica/regulators2_impedance1_e.html

The diagram labelled 'Bad (left) and correct (right) wiring of the LM317' should be of interest.

 

[unclejed613]

you can see what i meant about the frequency response of regulators in the graphs. the impedance turnover in most of the graphs were at about 400-500hz. a "semi-discrete" regulator using an op amp with good bandwidth as the error amp might be a good solution. even with a good regulator using sense lines, you still want to use local bypassing of any noise sensitive or noise producing circuitry. i see a lot of equipment taking a dive because of dried out bypass caps at the regulator or at the "local" level, sometimes both. error correcting regulators are amplifiers. instead of driving a speaker load with audio, they are driving a complex load (semiconductors, relays, sometimes motors, etc...) with DC. as with any amplifier. you want to get the benefit of feedback, across as high a bandwidth as possible.