How to set the current limit of an LT3081 using a DAC?

Regarding a digital potentiometer, the AD5272-20 is a 20kohm digipot (I2C) with 10-bit resolution (1024 steps). You could put it in parallel with a fixed resistor to adjust the range and resolution of the net resistance on the I_LIM pin.

Regarding using a resistor, I wonder if you can make an op-amp based negative resistance circuit to skew the effective resistance on the I_LIM pin : https://electronics.stackexchange.com/questions/52860/negative-resistance-impact-on-voltage-source

I'm working on a somewhat-similar personal project and I'm using the LT3086, which has a more explicit current-limit circuit. I'm tentatively using the AD5272 and an analog switch (ADG5401) as a hi/lo range toggle.

Thank you for your suggestions. I'm afraid AD5272 will not be suitable in my application because it's only rated for 5V whereas I need something that can withstand at least 30V.

But I'm curious to learn more about your idea with negative resistance. Would you please draw up a quick schematic to show what you mean by skewing the effective resistance?

The link you provided shows an example of this concept being the Howland current pump which is something I was also simulating and considering to prototype as a current source. Something like the following:

Is this what you mean? Or were you thinking of a different configuration?

I haven't prototyped this circuit yet, and the reason is that it is very susceptible to mismatches in resistors. Perhaps it requires a precision difference amplifier, something I don't have around atm. Also I'm not sure how safe this is since the datasheet for LT3081 suggests an absolute maximum difference between ILIM and OUT pins to be 200mV.

Here you can download the above LTSpice schematic: howland-cs-ilim.asc

That circuit show was what I had in mind, but after more thought I don't think it will work. The circuit would have to be referenced to ILIM, not GND, because the "negative resistance" is applied between the ILIM node and the circuit negative reference (ground, in this case). 

You're right about being concerned for safe operation. I had forgotten about the 200mV differential limit between ILIM and OUT.

I'll have to give this some more thought.

I came back to this project recently, and after some experiments, it turns out indeed the 6uA coming out of the ILIM pin of the regulator is not quite constant, but there is a temperature dependent tiny current added on top of the 6uA that boosts up the current limit voltage as the IC temperature rises. I verified this by feeding the 1uA/C out of the TEMP pin directly to the ILIM resistor (or to the middle of it using a pot) and observed that the output current now remained steady in current limiting mode as the IC temperature changed over time.

So this complicates the matter, and it mostly voids the attempt of feeding extra current to a small resistor at the ILIM pin. This method would be fairly unreliable for current limiting because the output current in current limiting mode would be all over the place, and it becomes difficult to calibrate. 

I came up with a different solution that I want to share here, hoping to get feedback from the analog experts on this site, as well as to potentially help future readers who may be interested in controlling the current limit of this regulator digitally. I essentially added another opamp to monitor the IMON signal that simply brings down the set voltage as the current monitor value increases above a certain threshold. Here's what I came up with:

The compensation capacitor values can of course be tweaked as it depends on the opamp used, or a snubber network can be used at the opamp output instead. I built the above circuit with the OPA2189 precision opamp, and used a switch mode pre-regulator with a differential feedback as shown in CN0508 to keep the VIN 2 volts above VOUT. In reality, it seems to work very well. The current limiting is reasonably fast at a hundred microseconds or so, and it recovers without any overshoot. Best of all, finally I get a current limiting facility that doesn't drift around.

It's worth noting that with this circuit, one can easily detect when the supply is in current limiting by sensing the output of either opamp with a comparator. In contrast, the LT3081's own current limiting has no good way of having an indicator signal like this, to light up an LED for example when the device is in current limit.

Some readers may look at this and think, well at this point, why not just use a series pass transistor and just build the supply from scratch?  The LT3081's here are used purely as an emitter follower any way. That would be an alternative, but then the frequency compensation of the voltage control loop would be annoying, and it would then be sensitive to pcb layout and may oscillate with different output capacitors and so on. The LT3081's regulate the voltage very nicely and they are quite stable. And they have all the requisite protections built in.

I'm open to comment and suggestions about how to improve the circuit above, and I hope that it can potentially help someone who may be interested doing a similar thing. 

lt3081-ilim-digital.asc

did you build it?

Yes, I did in fact build it, it works quite reliably. It's one of my better supplies. 

I should add here that the value of R22 and C8 are very important in determining the dynamic behavior of the current limit loop. For example, with LT6015, 10k and 500p seem to work great in simulation. You have to look at the phase margin of the amplifier and how fast you want the current limit loop to react.