[SOLVED] [moved] Linear regulator error amplifier input & output voltages & compensation type?

[moved] Linear regulator error amplifier input & output voltages & compensation type?

Hi,

The questions are for learning purposes, mainly. I have been reading about both points, and read some related threads here, but haven't found answers yet. I'm having difficulty understanding something and it's beginning to annoy me.

Regarding linear (LDO or Standard) regulators and the error amplifier:

As the error amplifier - which seems to be a difference amplifier - always appears to be balancing the reference voltage to a scaled down output voltage: 1.2 or 2.5, etc. volts...

a) Does the input difference tend to be in the range of a few millivolts?

b) If so, how is the error amplifier output voltage scaled to the driver transistor or directly to the pass device? How much gain is appropriate for a design made with individual ICs? Is there some "normal" limit to line regulation that can be predicted? Is the output off set to match the transistor base emitter/MOSFET turn on to saturation voltage?

Trying to find the above info., I found info. about compensation types 1, 2 and 3, so I'm only just beginning to understand what poles and zeroes are in this respect...

2) Do any and all linear (not switching) regulator error amplifiers need some type of compensation to avoid oscillation? If so, for a non-engineer, how could I assess a circuits needs?

If questions like this, without needing to refer to a specific design, are able to be answered in general, overview way, great.

Thanks

Re: Liner regulator error amplifier input and output voltages and compensation type?

a. the conduction will be adjusted so the 'scaled down' output voltage, whether by fixed internal divider or from an adjustment pin, exactly matches the internal reference voltage. The natural variation is output voltage that gives their 'tolerance' (+/-) is due to minor differences in the divider resistors and the reference voltage.

b. not necessarily, it depends on the way the feedback is applied and the internal circuit of the regulator. Bear in mind the difference is 'calculated' by a high gain amplifier and it is essential it provides negative feedback or it would make an oscillator with the power rail as the tuning components!

Brian.

Re: Liner regulator error amplifier input and output voltages and compensation type?

Hi,

So does the amplifier not output millivolts but anything within its supply rail and specs range (e.g. ~0 - ~4.7V for rail-to-rail) to balance the inputs, if you say it's a high gain amplifier?

This is what I'm annoyed (at self) about: from the simplified schematics I've seen it looks like a difference amplifier that is a voltage follower, so that could output the same millivolts it sees at the inputs, or output a voltage proportional with gain, but then what point in the datasheet of an op amp do I need to refer to to have some idea of the output voltage - is it open loop gain characteristic? Or do I set the gain as I see fit or am (in)capable of calculating? I saw that some error amplifiers had 80dB gain, a point I'd need to look into further to get the grip on dB as voltage.

Because the general images are simplified, it leaves a lot of guesswork as to what's missing or not from the error amplifier stage (excepte pdfs about compensation). From the little I understand, open loop is not the same as the schematics for compensation types, and not the same as a non-inverting (or inverting) amplifier. Back to reading.

Negative feedback means type 1, type 2 or the RC Christmas tree that is type 3, please?

Re: [moved] Linear regulator error amplifier input & output voltages & compensation t

here is sim in free ltspice of linear regulator...play around with compensation r and c and see its stability/response change.

Do load stepping to see if you can get it to go unstable aswell. You kind of get the kind of value ranges of the compensation r and c which bring about a nice slow stable feedback loop....then you can check it with a gain phase analyser.

- - - Updated - - -

of course, the jist of it is that there is an error amplifier with negative feedback in there......and it is connected up so that its inputs go to equal values....and that is the regulation state.

- - - Updated - - -

so the long and tall of it is that.....if you see an opamp with negative feedback then there is a good chance that you can expect its input voltages to be equal (inverting and non inverting inputs)

Re: [moved] Linear regulator error amplifier input & output voltages & compensation t

b) If so, how is the error amplifier output voltage scaled to the driver transistor or directly to the pass device? How much gain is appropriate for a design made with individual ICs? Is there some "normal" limit to line regulation that can be predicted? Is the output off set to match the transistor base emitter/MOSFET turn on to saturation voltage?

Click to expand...

There are two problems, the first is static voltage regulation the other is transient response and stability.
Infinite loop gain and infinite bandwidth around the feedback loop would be ideal, but the problem becomes one of cumulative delays (or phase shift) around any practical feedback loop.
The aim is to have as much dc gain as possible at zero frequency to provide excellent static voltage regulation. There simply cannot be too much !

As soon as you introduce some kind of step perturbation to this, there will always be some delay around the feedback loop, and final correction takes a finite time.
If the delay around the loop is sufficient, it is possible for the negative feedback to become positive feedback at some higher frequency due to total combined phase shift.

It becomes necessary to roll of the gain as frequency rises, so at the point where it would become unstable, the total loop gain has been reduced to less than one.
That is the basis of compensating any feedback amplifier, so that there is maximal low frequency closed loop gain, and also freedom from oscillation and instability at higher frequencies by deliberately rolling off the loop gain as frequency rises.

2) Do any and all linear (not switching) regulator error amplifiers need some type of compensation to avoid oscillation? If so, for a non-engineer, how could I assess a circuits needs?

Click to expand...

Yes, compensation is always a requirement for any closed loop feedback system.

Circuit needs are assessed by measuring the total delay (phase shift) around the whole loop and ensuring the high frequency gain is reduced sufficiently before the negative feedback can become positive feedback.

This becomes complicated because the phase delay is never all in one place, it is a cumulative effect of multiple causes, so the phase frequency response can sometimes become quite lumpy and irregular.

Re: [moved] Linear regulator error amplifier input & output voltages & compensation t

An opamp is frequently used as an error amplifier in a linear power supply. An opamp can have a DC and very low frequencies voltage gain from 90dB (thirty thousand times) to 130dB (millions).

Re: [moved] Linear regulator error amplifier input & output voltages & compen

Hi,

Thanks for Spice model. How do I import this to LTSpice IV, please? Do I need to chane the notepad name to .asc or similar? I'll try a few things like that, anyway.

- - - Updated - - -

...Finally found the way of turning it from .txt into a Spice model..., thanks all the same, now I'll try to learn what I can glean from it, great.

Re: [moved] Linear regulator error amplifier input & output voltages & compensation t

do you want me to add a load stepping circuit to it?, so you can see the transient response, then you can see if a big transient knocks it into instability...here it is below, in ltspice again