Linear Regulators Sinking Current

[dknguyen]

Linear regulators can't sink current right? If you need it to sink current you have to use an op-amp reference? I remember reading something about this a long time ago but it's so long ago I can't remember if I'm just making it up.

 

[audioguru]

A zener diode is a shunt regulator. It can be boosted by a power transistor.

 

[Roff]

Linear regulators can't sink current right? If you need it to sink current you have to use an op-amp reference? I remember reading something about this a long time ago but it's so long ago I can't remember if I'm just making it up.

You are undoubtedly talking about positive regulators.

 

[on1aag]

Current source and current sink.

Hi dknguyen,

Some expensive exotic regulators can do both, even at the same time.

on1aag.

 

[kchriste]

Linear regulators can't sink current right? If you need it to sink current you have to use an op-amp reference?

If you add a load resistor on the output of a 78xxx regulator you can "sink" some current through the resistor. This is only really practical for small currents as the max "sink" current is equal to Vout/R.

Some expensive exotic regulators can do both, even at the same time.

 

[Hero999]

Shunt regulators can both sink and source current.

Building a linear regulator that can sink current is easy, just use a complimentary push-pull emitter follower on the output rather than the usual single transistor.

 

[Roff]

Shunt regulators can both sink and source current.

Building a linear regulator that can sink current is easy, just use a complimentary push-pull emitter follower on the output rather than the usual single transistor.

That will work if you are building a discrete regulator, but adding it to an integrated regulator is problematic.

 

[KMoffett]

Like polarity, sinking and sourcing are just a matter of reference. With a reference to the supply negative, a positive voltage regulator (78XX) set up as a current regulator can only "source" current to the load. But, switch to a negative regulator (79XX), with the same reference to supply negative, and the regulator is now a current "sink".

Ken

 

[nwpayne]

TPS77133 does

At least up to 5mA, even though the TI people swear that it doesnt.

 

[MrAl]

Linear regulators can't sink current right? If you need it to sink current you have to use an op-amp reference? I remember reading something about this a long time ago but it's so long ago I can't remember if I'm just making it up.

Hello there,


Yes that is correct. Most series linear regulators can not sink current and in fact some of them have a minimum sink current that the user must supply or the regulator wont regulate properly.
The reason of course is because of the internal series pass transistor, who's output is the emitter of an NPN transistor in many cases. This means that if you try to pump current back into the output of the regulator, the NPN transistor simply turns off because the base emitter gets reverse biased. The only current that flows then is whatever resistance happens to be on the output plus the tiny reverse leakage current.
If you suspect that your power supply will have to sink current at some point you have to provide for that yourself in the design, either by connecting a resistor on the output that can take the sink current (and not overload the normal operation of the regulator) or an active current sink.

As other posts have suggested the shunt regulators dont have this problem because they are always sinking current and their problem is that they never source any current. The only current that does get 'sourced' to the output comes not through the regulator but through the external series resistance. This is a problem too because all of the normal current has to go through that resistor so there is always some power dissipation.

This means the choice between what type of regulator to use depends on what it has to do and what power source it has to work with and what tradeoffs can be made. The series regulator is the most typical choice because the load for many applications is almost always present. There are those times though when the shunt regulator works out better even though that means a series resistance of some type has to be present.

Some LDO regulators have a PNP output stage where the collector is the output terminal. These kind might sink a little current because their collector emitter would become forward biased. I suspect the allowed current will be low though because of the internal biasing, and the voltage regulation will be more like a diode in series with a true regulator (so there will be much more variation than usual for a decent regulator).

The question then is does anyone make a series regulator that also can shunt current? There may be one out there somewhere but i havent seen one yet. You can make your own for sure, but it will be much more expensive.

 

[Roff]

Hi MrAl,
You probably didn't notice that this thread is 2 years old.

 

[Sceadwian]

It's nwpayne that originally brought the thread back, this happens a lot with new users that simply randomly browse the forums.

 

[MrAl]

Hello again,


Yeah geeze! Got me again ha ha
Yeah that other user replied and that was to an old thread, so that's why it came up in my browser.
What do we have to do, check the first post in a thread just to make sure it isnt a super old thread? Geeze
It would be nice to see a 'warning' of some type indicating that the first post of the thread was older than say 90 days.

 

[Sceadwian]

MrAl, I've mentioned that exact same solution at least twice in the last few years.

I didn't get anywhere =)

 

[Roff]

Hello again,


Yeah geeze! Got me again ha ha
Yeah that other user replied and that was to an old thread, so that's why it came up in my browser.
What do we have to do, check the first post in a thread just to make sure it isnt a super old thread? Geeze
It would be nice to see a 'warning' of some type indicating that the first post of the thread was older than say 90 days.

The only reason I noticed is that I opened the thread at the beginning. I saw some of my posts, and didn't remember making them, so I checked the dates.
I like the idea of a warning.

 

[BrownOut]

I usually snap to an old thread when I see a post form Hero999.

 

[MrAl]

MrAl, I've mentioned that exact same solution at least twice in the last few years.

I didn't get anywhere =)

Oh i see, well i posted a new thread asking for a warning so we'll have to wait and see if they want to implement this or not. I think it would save some unnecessary posts from occurring.

 

[Sceadwian]

The unnecessary posts occur every few months, it's hardly worth the worry, but I wouldn't be opposed if they implemented it. Again, this was a thread here last year sometime, I was much in the same position as you =)

I personally after reflection don't think anything needs be done, I liked the fact that this thread was touched upon again, as I missed it the first time.