I need an opamp which is able to drive unlimited load capacitance with high output current

Hi GerdF,

I think that the ability of an op-amp to control capacitive load depends only on correction method, but not on additional power amplifiers. You can use any additional output stage which can provide required output current. If circuit will be unstable you can add external correction components.

I did a little experiment, look at the result

Currents through emitter resistors

There is a peak current at the beginning of the transition process. I think that this is due to the fact that the capacitor is ideal. I added 10 milliohm resistor in series with capacitor to model the impedance of a real capacitor.

Currents through emitter resistors:

You can try it in nature and report about results.

I hope that this help you.

Regards,

Kirill

Hi Kirill,

thanks for your answer.

I am currently not on my computer with LTspice, but I suspect this circuit would oscillate with a very low frequency if allowed to run for a longer time.

I think making the sine source a pulse one with a pulse duration of a few milliseconds and a period of say 10 seconds and running the transient simulation for a few seconds would show instabilities.

Regards,

Gerd

Hi Gerd,

Transient response, 400 uF and 50 ms duration without output stage:

With output stage:

Some materials about capacitance load

https://www.analog.com/en/analog-dialogue/articles/ask-the-applications-engineer-25.html

https://www.analog.com/en/analog-dialogue/articles/techniques-to-avoid-instability-capacitive-loading.html

But with small capacity frequency response is bad

1 uF

100uF

I have to admit that I have never dealt with such applications. But it seems to me a very interesting and entertaining problem. Could you tell me more about the required output waveform and the characteristics of the load?

You said that load have capacitance in thousands uF range and I will simulate circuit at similar conditions. As far as I understand you need an op-amp which not required external compensation for capacitance driving and I would like to design simple circuit which not required compensation and which has simple output stage which can provide high current. If you clarify the requirements, maybe I can help you with something.

Regards,

Kirill

Hi Kirill,

First, thank you very much for your efforts.

Indeed it is a very interesting behavior. I think at very small load capacitances it would be very stable, because no relevant phase shift would be added.

At very high capacitance loads it seems the output resistance and the load capacitance build the dominant pole and thus provide the loop compensation.

Only when the load capacitance is in medium ranges, the internal compensation pole and the external rout-cload pole are too close to each other and introduce excessive phase shift while the loop gain is still above 1. Due to the isolation by the output buffer the automatic increase of the internal copensation effect with increased capacitive load capacitance does not work then.

I think the circuit's behavior gets even more interresting if a certain DC output current flows, putting one of the transistors on by default.

In the application, there capactitve load can be variable, because the user can switch in either small, medium or very large capacitors. The DC current, however, should be negligible.

The amplified or buffered signal is mostly DC voltage but the DC value changes from time to time and then the transition should be quite fast, hence the demand for higher output current than the AD817 can provide by itself.

I wonder if an output current boost like the one in figure 85 of AN87 would work with the AD817.

If the resistors in the supply line are kept low enough to only activate the transistors when the required current exceeds like half of the AD817's capability, there would be no extra quiescent current and maybe no stability issues once the signal has settled again after a jump of the DC value.

Hi Gerd,

In other words, you need fast recharging of capacitors. I think this configuration is able to provide the required parameters

The phase margin is no less than 5 degrees at all conditions. Sounds like you meant it. An op-amp takes over part of the load and provides the necessary correction. If you want I can provide the Bode plots.

Regards,

Kirill

Hi Kirill,

Sorry for the late reply. I have quite a bit of headache for the recent days and that makes my thinking capabilities vanish to a great extent.

Yes, that is like the circuit in the AN87. But there they have capacitors across BE of the output transistors, stealing the AC current from the output transistor's bases above about 2.5 MHz. Don't know if that would be important, though.

As there is virtually no DC current load, the booster transistors should always be off unless there is an actual voltage step. During such a step the stability should not really matter as long as any possible peaking settles as soon as the booster transistors go into off state after the capacitance has been charged to the desired voltage.

I wonder if the 68 ohms resistors should be of a lower value in order to somehow transfer the AD817's output current limit into a booster transistor base voltage/current limit.

Bode diagrams would be nice, especially as I don't feel like being able to do them myself (looking at a screen for too long does not really help with headaches).

Regards,

Gerd

Hi Gerd,

Now 7 ma current flow through resistors and transistors are turned off. In fact, we see the frequency responses of op-amp itself. 

It's our local time of 5am and it's not conducive to my thinking. We are in almost the same conditions.

I wish you health!

Kirill

Hi Gerd,

I have a few thoughts about capacitors. I reason as follows:

1) Current consumption of an op-amp increases and at a certain point in time quickly turns on the external transistor.

2) Now an op-amp does not required a large amount of current and current consumption decreases.

3) External transistor starts to turn-off

4) The process can be repeated.

In my opinion, these capacitors should bring the nature of switching processes to exponential. I decided to check it out.

Look at the current through R4

Rise

Fail

There is some optimal capacity value at which the processes are weakened as much as possible but I do not understand how to find it analytically. I found a simpler solution. I understood that now output stage of an op-amp and external stage interact with each other directly. I decoupled theirs by using resistor

Regards,

Kirill

Dear ADI specialists,

I am interest what term "stable with unlimited capacitive load" means from the datasheets? Based on simulation results, the phase margin no less than 5 degrees at very large capacity. Formally, this means that oscillations does not occur in circuit but this value is very close to 180 degrees, therefore we can expect to see the ringing at transient response. In other words, stable means that oscillations cannot occur at any cases but the appearance of ringing is possible?

But ringing absent at large capacitive load. Why?

Thanks,

Kirill