justinbroweK.C. Lee referred me to the MAX44284, which was helpful. Its specs were close to those of the MAX9923F that I was looking at, but it smaller and almost half the price at scale!
| v offset | i bias | gain error | CMRR | Vo min | $pk | |
| MAX44284 | 20.5uV | 80nA | 0.15% | 145db | 1mV-20mV | 1.19 |
| MAX9923F | 10uV | 1pA | .5% | 140db | 10mV | 2.1 |
Maxim also offers a spice model of the MAX44284f on their website, so I set up a simulation just to confirm the problem I noticed the other day.
The amplifier doesn't start to behave until the supply current reaches ~40uA. which is pretty far outside my spec.Instead of using a current sense amplifier, I am going to use a instrumentation amplifier - probably the max4209.
Instrumentation amplifiers are not cool with common mode voltages outside their supply voltage range, so this design will prevent people from power-debugging devices with supply voltages > 5v, but I think that is acceptable. This is a tool for optimizing the battery life of low-power devices and most power-sensitive battery operated devices are going to have supply voltages <5v.
I had to use the reference pin to add in a 0.2v offset voltage. This is going to take a small bite out of my ADC range, but the ADC I am using has some programmable amplification options that will allow me to counteract that loss a bit. I think I am going to move forward with this design.
Discussions
Become a Hackaday.io Member
Create an account to leave a comment. Already have an account? Log In.
You could have a 40uA+ constant current sink to put the current sense in the linear region. All you have to do is to subtract that amount in the firmware as part of the calibration process.
The other thing is that if your instrument have its independent battery), then you could just float the whole circuit for positive rail sensing. Use ground sensing (relative to the instrument) directly with a high resolution ADC so you don't even deal with opamp.
Are you sure? yes | no