Hi,
I am trying to create small current detect circuit using LM2903 (comparator).
Objective is to detect if current is lower than 4 uA. (Range of I1 is 0uA to 1800 uA)
R4 is there to build up the Voltage which is input(+) to comparator. R1 and R5 are voltage divider to give Vref (equivalent to 4 to 5 uA). The circuit is working fine during simulation. When I test the physical circuit at 25C ~ 85C it is working too.
But when I test it at -40C, hysteresis become zero. When input(-) approaches to threshold (input+) output started to oscillate, switching 1 and 0.
why does hysteresis width change across temperature? Initially i created the circuit with two diodes where R4 and R5 are right now. When hysteresis became zero at -40C, i blamed two diodes and replaced them with resistors. Now same thing happen with new design too.
Datasheet said Input Voltage range is -0.3V to +36V. So it should work, shouldn't it?
Please help, I have no idea where went wrong.
Thanking in advance.

 

hi,
The lower temperature specification of the LM2903 is -40C.

Is it necessary to have the LM2903 at the same temperature as the current sensing element.?
If you want high accuracy its usual to maintain the comparator at a known, fixed temperature and expose the temperature measuring element to the changing temperature.
E

 

Thanks for the reply.
Yes, it is necessary LM2903 to expose -40C.
The point is the whole circuitry including LM2903 should work at -40C environmemt temperature.
So do you suggest I should look for another comparator that can work at lower than -40C?

 

hi,
Look at the top of page #3 of the attached d/s. LM193. -55C thru +125C
Also consider military spec versions.

Contact National Semiconductor, they may send you free samples.

E

 

Contact National Semiconductor, they may send you free samples

No big deal but National Semiconductor is not around any more. They were acquired/engulfed by Texas Instruments some years ago. But TI may help with free samples

 

I glanced at the datasheet on Digikey, and it only has specs for 5V, and doesn't show much detail across temperature, so there could be some variation in the 2903. I think there could also be some variation in the rest of the circuit across temperature, such as your voltage and current sources. If you don't need a very fast response, I'd recommend adding some capacitance, perhaps between the 2903 plus and minus inputs. You might also make sure the 2903 is decoupled appropriately so its power is stable.