This module is designed to measure currents with a noisefloor down to 5fA/sqrtHz(@1G V/A). It is also optimised to send very little energy to the sample (M1b does even better but has lower bandwidth). For that reason it's input resistance is not "as low as possible" but fixed at a function of the [V/A] setting (see calculation below).
There is a postgain switch [x100ac x1 x100dc] giving a max total output of 100G V/A.
When doing lock-in or noise measurements as a function of dc-current the [x100ac] makes it possible to amplify the ac-signal or noise while maintaining a large dc-sweep range.
The postgain can also be used to obtain a lower Rin/ higher speed at high V/A settings. Example: 1G V/A could be exchanged for 10M V/A with postgain [x100dc] (you pay with higher noise).
This unit is implemented with ground-lift, this solves problems with thermovoltages that are usually present in cryogenic setups. The zero reference pin(2) should be connected to a cold-ground-wire or otherwise to a room temperature ground.
Other use is hooking up a small dc-voltage (S3m module) to pin 2 and doing a 1-wire resistance measurement via pin4 on a sample that is grounded with it's other contact.
2=ref-in means that you can (and should) connect pin 2 to a reference ground that can have a difference dc-level (up to some 100mV) e.g. a wire from cold-ground inside the fridge. This helps to remove excess offset by thermovoltages.
Usual procedure to determine offset voltage: If you put your voltagebias on 0V there should be no current, otherwise tune it to a value where you find I=0, the voltage that you have now is the offsetvoltage in the system. You can compensate by taking this into account.
Rin calculation
Rinput is: 100ohm + 1E-4 * [V/A] for [low-Noise]
Rinput is: 100ohm + 1E-5 * [V/A] for [low-Rin].
Example: at 10M V/A the input resistance is 1.1kohm @[low-Noise] or 200ohm @[low-Rin]. These values are accurate within 1% so a possible voltage drop can be subtracted in the data.