LM393: device behavior question

That statement is correct. But please note that outside the valid input range, the electrical characteristics (especially propagation delay) are no longer guaranteed.

For low-voltage applications, better use a low-voltage comparator with rail-to-rail inputs like the LM393LV.

Hi Jim,

As Clemens mentions, it is correct and I would also refer you to LM393LV which is well suited for low voltage applications and pin compatible with LM393

Hi Jim,

When both inputs are outside the common mode, the "expected" behavior is that the output will go low. But because the common mode range is being violated, the performance is not guaranteed. Please see section 2.2- 2.5 of this app note to understand about the input structure of this family:Application Design Guidelines for LM339, LM393, TL331 Family Comparators (Rev. C) TI can only recommend operating within the input voltage range. 

Please note that LM393 is undergoing PCN and that for the new die, the output will actually go high. This is also mentioned in the app note. I would highly recommend using LM393B so that you are aware of any issues that may occur. 

Hi Jim,

From the resistor values you have selected, your channel 1 would have hysteresis values of VL = 2.97V and VH = 3.29V. For channel 2, you have hysteresis values of VL = 10.78V VH = 11.76V. VL and VH are the switching thresholds for the output of the comparator from high to low and low to high respectively.

C1189 and C1191 are fine to keep since they are not on nets that affect hysteresis.

Attached are images of the transfer characteristics of your circuit.

Hi Jim,

Decreasing the R3 and R8 resistances is one way of reducing VH, but keep in mind that the hysteresis of your circuit is a function of more than just R3 and R8. Since you are using an inverting comparator with hysteresis, you may refer to Analog Engineer's Circuit: Inverting Comparator With Hysteresis Circuit to help with your design.

To find the variations in VL and VH from ideal, you will need to account the tolerances of your resistors. For the worst-case values, you will need to refer to the electrical characteristics of the LM393 to find the maximum input offset voltage and apply it to the positive input of your comparator. In addition, you will need to take the maximum input bias current over temperature and multiply it with the equivalent resistances at your IN+ and IN- nodes, then apply those voltages to the respective nodes.

Hi Jim,

Attached are two TINA files for you to use. Could you simulate them to confirm that this is the behavior that you want? Keep in mind that this is an idealized model of the behavior, and that it doesn’t account for nonidealities like input offset voltage, tolerance, and input bias current. In addition, you have a few degrees in freedom in altering the resistances per your tolerance and power requirements.

From your specifications, it seems like you want a 7mV threshold window for your 3V3 monitor, and a 24mV threshold window for your 12V monitor. Keep in mind that these windows are small, and the fact that you are using a voltage divider to maintain common mode requirements scales these windows down.
JimTsai3V3.TSCJimTsai12V.TSC

Hi Jim,

The transfer characteristics and transient simulation look good to me. I'm going to note that the simulated values might not be the exact values that you've stated here, but they are close enough such that it should be fine. For example, the simulated value for 3V3 is VH = 3.07V and VL = 2.92V. I’ve edited the timestep of your transient simulation so that you can see the simulated values better.

I also noticed that the capacitors on the feedback path (C1188 and C1190) on your updated schematic have not been removed yet. As suggested before, it’d be best to remove them as to not add delay to your feedback path.

PQ4000_P3V3_final_solution_updated.TSC PQ4000_P12V_final_solution_updated.TSC