Current sense to digital output

Q 

I have a high side current sense resistor of 2.5 milliohm, with 0..15A current.
I want to set up a system that will light an LED when the current is above
approx 200mA. I was thinking of using the AD8237 and the ADCMP341. My thinking
is the voltage at 200mA and above across the sense resistor will be 500uV and
greater. So the output from the AD8237 ( at G=1000) can be 0.5V which will
output a high on the ADCMP341 (as it is above the 400mV internal reference).
Below 200mA will result in a low condition. The LED current need only be a few
mA so the 5mA output current of the ADCMP341 should be enough. I have attached
a block schematic. The entire system needs only to work at approximately 10Hz.
The LED is a charging status indicator. The design is cost sensitive and there
is ample PCB area available. It is an all analogue system. There is no
microprocessor in the design. I would like an accuracy of 10% (+/-20mA). The
system power available is +5V and ground only. I will not be able to perform
any calibration of the system at the production stage to compensate for an
offset. If I am not on the correct path, can you suggest an alternative design
for
inside the green box as in the attached pdf?

 

A 

AD8237 is a good choice. You can take G=1000, set it to high bandwidth mode and
still get 1kHz bandwidth, which will help to settle quickly. Or I guess G=800
just to match the 400mV threshold as close as possible. 1% resistors would be
fine, the error budget will be dominated by the offset anyway. The ±75µV offset
voltage of the AD8237 translates to a ±15% error at 200mA. Increasing the sense
resistor would improve this, so there is an inherent accuracy vs power
tradeoff. I would also look at AD8293G160. It is used in many systems like this
and it has lower offset (50uV), but it has a fixed gain of 160, which is only
80mV at 200mA in. But if the shunt resistance is increased, maybe it is worth a
look.

Either way, I would come up with a circuit to clamp the output below 5V. For
AD8237, this could be as simple as 1 or 2 low-leakage diodes between OUT and FB
to bypass the gain resistor at high Vout. With the AD8237 in a gain of 800 and
15A in, the output will want to go to 30V. At 500% overdrive with a micropower
CMOS amplifier, it might take seconds to recover unless you prevent the
saturation in the first place. One more thing, you may want some input
protection resistors for the AD8237 because load or even trace inductance can
cause voltage spikes for any rapid change in current. The AD8237 has a very
high ESD rating (8kV HBM with no external components), but no long term
integrated overvoltage protection.

Also, maybe it’s just not shown in the diagram, but I would consider setting
some slight hysteresis for the comparator.