LM393B: Strange waveform when LM393B powers up at 0- -30C

Frank(EP) Liu

Part Number: LM393B
Other Parts Discussed in Thread: LM339, TLV4021, TLV7042, TLV7041, TLV7021, TLV7022, TLV7081, LM393LV, LM139

Hi,

My customer uses LM393B as voltage detection.

when there is input power and ambient temperature is at >0C, it works normally as below.

CH1: Vo

CH2: IN+

CH3: IN-

ch4: Vcc(VBUS)

when there is input power and ambient temperature is at 0 - -30C, LM393B works strangely as below.

Vo has strange pulse but IN+ is lower than IN-.

the schematic is as below.

from Datasheet, LM393B should work at >2V and -25C.

Can you help check what happen?

BR,

frank

over 2 years ago

0 Clemens Ladisch over 2 years ago

Guru 243090 points

This circuit uses invalid input voltages; see the V_IVR specification in the recommended operating conditions.

This is explained in more detail in the Application Design Guidelines for LM339, LM393, TL331 Family Comparators:

The specified LM339 Family input voltage range (V_ICR) is 0 V (relative to the negative supply pin) to V_CC – 1.5 V at room temperature. However, the actual upper input voltage range reduces by -4 mV/°C at cold temperatures. Therefore the specified LM339 Family full temperature range common mode range is 0 V to V_CC – 2 V to account for this reduction. The V_CC – 2 V range is strongly recommended for use in all designs.

Also, the second comparator's inputs must not be left floating.

To avoid that pulse, you must use a comparator with rail-to-rail inputs. If you do not want to change the board, then the TLV7022/TLV7042 in the DSG package would be possible replacements (but do something about the floating inputs). If you can change the board, consider a smaller, single-channel device like the TLV7021/TLV7041/TLV7081, or even a comparator with integrated reference voltage like the TLV4021.

0 kai klaas69 over 2 years ago

Guru 140200 points

Hi Frank,

remove C133 and especially C127. Low pass filtering the +input of comparator when introducing hysteresis is counterproductive. The positive feedback shall help to move through the no man's land at the trip point with increased speed. But low pass filtering ruins this effort.

And I don't like C148. The TS2431 becomes instable when the filter capacitance doesn't fit with the cathode current. See figure 15 of datasheet. Why not increasing C148 to >1µF? This would also help for the power-up issue.

Kai