I've decided to take a different approach for the current source. Based on a document from Analog, I am going to try the following:
I have to see if I use multiple set resistors, copy the entire system for each range, or use a DAC/set of resistor deviders to switch the reference voltage (instead of the current shunt resistor).
I don't think the opamp buffering the voltage reference is necessary, but I've added it in case I want to put multiple (3 or 4) of these sources in parallel driven by the same reference.
Stuff I'm going to now look at: Either figuring out how I can turn the source off if the current can't flow (to prevent a inrush when it is connected again) or put a diode, or set of diodes in parallel with the DUT as suggested by schmitt trigger. In order to do so I will have to do some maths, so I know what realistic ranges are in terms of voltage drops and currents. I would like to keep the voltage over the DUT as low as possible, but of course the limit here will be the offset of the measuring differential amplifier. Once I know this, I can look at the leakage current at that voltage, and see how much it will throw off the final measurement. It can be that it could be calibrated out, but I would rather not do to much calibration.
Hi,
Have you already checked the project of this guy?
https://youtu.be/anE0jDeBuxo?list=PLUMG8JNssPPxGPivQllTex7AEVIIeHFd0
I've looked at it before. The thing I don't like is that he has no ranges and that he seems to be too confident that getting the accuracy he claims is easy. Just because I'm using 0.1% parts does not mean I get 0.1% accuracy. (All parts I'm using are at least close to an order of magnitude more accurate than my target, and I am starting to think even then I wont pull it off.