LDO Regulators as Current Sources

[Speakerguy]

I could hook this up on my breadboard, but I only have one type of LDO regulator at home and it wouldn't give me a true answer with a sample of n=1.

Do adjustable LDO and quasi-LDO regulators, which normally need an output capacitor sized appropriately for stability, operate OK as current sources? Do I need to throw a compensation cap in there somewhere or will this setup be inherently stable like it is with non-LDO types? I've only ever used LM317s and LM338Ks as quickie current sources, and I haven't seen the issue addressed in any LDO data sheet that I've read.

 

[dknguyen]

Good question. I don't know since I don't know exactly why linear regulators have stability problems in the first place. That said, check out the REG101, 102, and 104 regulators from TI. THey are LDOs and have adjustable versions, and from the internal block diagram they work the same way as the LM317, except that they require a ground pin. So they probably work as current sources too.

DId you notice that the LM317 DOES NOT need capacitors for stability? Just letting you know...that might be why they caps aren't needed for the current source. I don't go around building current sources so I do not know if anyone has ever used a linear regulator that needs caps as a current source. THe one time I needed a current source, I found the LM317 and looked no further.

EDIT: It also seems that the LM338K also needs no capacitor. It might be that the design of linear regulators which allows them to be wired as current sources inherently do not need capacitors. Which LDO did you have in mind?

 

[Speakerguy]

I've read that the standard linear regulator topology (such as used in LM317, 7805, etc - pretty much any regulator chip w/ the usual 2.2V-2.5V dropout) is inherently stable and doesn't require capacitors; however, I had a 7815 a while back that had a >1MHz oscillation without an output capacitor. I put a capacitor on the output and sure enough that oscillation went away (it was a few volts AC at >1MHz on top of the correct DC output voltage). I think I even posted about it on this forum, as the data sheet for the exact part even said "no output capacitor needed". IIRC these type of regulators use dual NPN pass transistors fed by a PNP. There was some app note I saw this in that talked about the typical topology of regular dropout, quasi-LDO, and true LDO regulators. I should dig that up.

The LDO's I believe use the output cap as a compensation capacitor as part of a feedback loop. I'd guess it gives the internal error amplifier more phase margin and thus stability. But I'm not sure if that's needed if using it as a simple current source

 

[dknguyen]

WHere would the cap even go in a current source? In parallel with the resistor?

 

[Speakerguy]

Yea, or maybe to ground. I really have no idea.

This wasn't the app note I was looking for, but it does talk about the types of topologies there are for LDO and quasi-LDO. It looks like the output cap on LDO regs does affect the phase margin of the error amp (P1 on the graph).

https://www.electro-tech-online.com/custompdfs/2008/04/ldo.pdf

 

[eng1]

The LDO's I believe use the output cap as a compensation capacitor as part of a feedback loop. I'd guess it gives the internal error amplifier more phase margin and thus stability.

You are correct. The output capacitance introduces a zero in the feed-back loop and the phase shift of +90° compensates for a pole by assuming suitable values of C and the ESR of the capacitor.
Non-LDO regulators don't required additional capacitors to achieve stability (because they have a single low-frequency 'dominant' pole); LDO regulators have an additional pole at low frequency.


Back to your original question, couldn't you just build a discrete current regulator with a couple of transistors? Which current (and performance) do you require?