This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM2903 - Output behavior at VCC < VICR + 2.0V

Guru 19775 points
Other Parts Discussed in Thread: LM2903, LM393-N, LM393, TLV3402, TLV3401, TLV1701

Hello Team,

We understand that LM2903 requires VCC > VICR + 2.0V ( VCC > VICR + 1.5V @ room temp) to guarantee the device operation in full temperature range. I believe the device internal current source needs this voltage to work correctly. However, could you please tell us the output behavior when this condition is not met ?

Our customer had tested the operation at below schematic.

When customer decreased the VCC voltage down to VCC < VICR +1.5V, VOUT had switched from HIGH to LOW level even IN+ > IN-.
VCC = 2.5V were the threshold when VICR was 1.75V. VCC - VICR = 0.75V

However, instead of decreasing VCC, they increased VICR (by increasing IN+ voltage) to 3.0V at VCC=3.3V. As a result, VOUT kept HIGH level even VCC - VICR = 0.30V 

How come the threshold is different even using the same sample ?

If VCC > VICR + 1.5V, can you commit LM2903 would always output the correct output level ?

Best Regards,


  • Hello Takushi,

    *IF* one of the inputs is still within the valid common mode range (Vcc - 1.5V), the output will be correct. This is mentioned in note 5, on page 5, of the LM393-N datasheet.

    " (5) Positive excursions of input voltage may exceed the power supply level. As long as the other voltage remains within the common-mode range, the comparator will provide a proper output state. The low input voltage state must not be less than −0.3V (or 0.3V below the magnitude of the negative power supply, if used)."

    But if BOTH inputs are outside the valid common mode range, then the output is indeterminate and there is no guarantee what the output will be.

    You may find that the output goes low in most of the cases, but there will always be the possibility it will go high. There is no guarantee - and we do not guarantee operation outside specifications.

    Keep in mind that the VCM "threshold" changes over temperature and devices - so it will vary. You do not want to design too close to the margin. Keep one of the pins well within the VCM limit.

    The LM393/LM290x input stage also has the unique ability (because it does not have ESD protection diodes) that the inputs can go up to 32V without damage, regardless of supply voltage. So it is possible to go above the supply - even though it still violates the VCM spec.

    "(4) The input common-mode voltage or either input signal voltage should not be allowed to go negative by more than 0.3V. The upper end of the common-mode voltage range is V+−1.5 V at 25°C, but either or both inputs can go to 36 V without damage, independent of the magnitude of V+."

  • Hello Paul-san,

    Thank you for your prompt response.

    Firstly, please allow me to confirm the common-mode range.

    > "As long as the other voltage remains within the common-mode range, the comparator will provide a proper output state. "

    What do you mean by the common-mode range, here ?
    I was thinking that common-mode range (VICR) were VICR = {(IN+)+(IN-)}/2. Is my recognition wrong ?

    Secondly, our customer tested the operation in the following condition.
    - TA = Room temperature
    - VCC=3.3V, IN- = 1.5V, IN+ = varied from 2.0V to 3.0V.

    In this case, would you say BOTH inputs are outside of the valid common mode range ?

    Best Regards,
  • Hello Paul-san,

    Please allow me to ask you another question.

    Is there a better comparator which does NOT have VDD and common mode voltage range limitation like LM2903 ?

    It seems TLV3402 has open collector output and better characteristics. I don't see any limitation to the common mode voltage range, but, how about this device ?

    Best Regards,
  • Hello Takushi,

    Q1: No. This is a common mistake and is perpetuated by some texts related to op-amps. This "average" assumes the inputs are always very close together, as with an op-amp under feedback. Comparators are designed to have wide differential input voltages.

    The common mode range limit applies to EACH pin INDIVIDUALLY. So VICR = VS - 1.5V for EACH input.

    So for a 3.3V supply, the maximum input for EITHER pin is 3.3-1.5=1.8V. So the legal input range is 0 to 1.8V on a single 3.3V supply.

    As long as one input is below 1.8V, the output should be correct.

    Q2: IN- is within the common mode range (1.5V < 1.8V). IN+ is outside the common mode range. BUT, because the IN- input is still within the common mode range, the output should be valid.

    Q3: The TLV3402 is unique in that it is a "Rail to Rail" input, but also can extend the input voltage up to 5V above the positive supply voltage (16V max).

    There are many devices that would fit your <2V requirement. Any particular reason you chose the LM2903? Other than cost?? :^)


  • Hello Paul-san,

    Thank you for the detail explanation. I am clear now. It seems LM2903 family is a little bit difficult device to use. One may not be able to understand clearly by just reading the datasheet.

    Our customer had chosen this device because LM2903 were registered product in their company, also it was low cost.

    Their requirement was
    - Absolute maximum ratings for power pin(VCC) is > 12V.
    - Output circuit is open collector/drain
    - Input pin could accept voltage higher than VCC

    It seems TLV3401 would be good for their application. I also found TLV1701, however, this device could not accept higher voltage than VCC.

    Could you advise us if there is much better comparator device than TLV3401 ?

    We are grateful for your help.

    Best Regards,
  • Hello Kawai,

    The TLV340x series is the only one where the inputs can (legally) go above the rails.

    Just beware the TLV340x is micro-power (470nA supply), so it is slow (55us), and can only sink 500uA.

    You will have more choices if you drop the above-rail input requirement. Can they divide-down the input signals so that they do not need an over-the-rails input?


  • Hello Paul-san,

    I really appreciate for your great support on this.

    Your information helped me very much. I will ask customer for their detail requirement and see if they could consider TLV340x or divide down the input signals.

    Best Regards,