LM2903: LM2903D parts not operating at -40°C

HI Jon,

Without a schematic or problem description, it's hard to say.

But, generally when things go wrong at cold, it is because the input voltage range is being violated.

The input range shrinks at cold. Room temp input range is VCC-1.5V, but VCC-2V over temperature - particularly at cold.

So make sure the voltages at BOTH the inputs are at least 2V below V+ maximum.

Please see section 6.4 to 6.7, and section 6.7 and 6.8 of the LM339 family application note. That may explain everything.

 Application Design Guidelines for LM339, LM393, TL331 Family Comparators

Usually this is because the customers have violated the input range, and the output just happened to "flop" in the correct direction (output low) and they never noticed. The circuit never really worked...

The new die has different behavior in "outside datasheet behavior" and that behavior finally get's noticed.

See section 3 of the appnote to see which die you have. you CANNOT reliably tell by datecode - you have to see the label or run the lot code.

"TI" is the TI identifier (replaces TI logo on newer devices).

The "4B" is the date code, or 2024 November.

"M" is the manufacturing site code.

The "A2JE and "A6RM" are the partial lot code.

The "A2JE" is the Ji3 die (PCN #1) - that goes high when both inputs are violated.

I cannot get a hit on the "A6RM". Please confirm the full top marking of that device. 

The digits on the back identify the package mold used during assembly and has nothing to do with the die.

Please get a schematic, a problem description, and possibly scope photos.