• State Resolved
  • Locked Locked
  • Replies 5 replies
  • Subscribers 64 subscribers
  • Views 2955 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

LM4040: Precision issue

Olivier Frenette5
Prodigy 90 points
Part Number: LM4040

Hi,

we recently received a 200 board batch from our external assembler and are having an issue with the output voltage of the LM4040. We are using the LM4040AIM3-3.0 which has a precision of +- 0.1% at room temperature. Out of 10 random boards I measured, 8 were out of spec range. A precision of +-0.1% would give me a +-3mV tolerance. This is the circuit I am using. The output goes to a high input impedance pin of a microcontroller. Although there is a note with a larger tolerance, this is only to remind me of the tolerance over the whole temperature range.

These are the voltages I measured : 

- 3.003

- 3.006

- 2.993

- 3.010

- 2.985

- 2.995

- 2.988

- 2.986

- 3.003

- I have a supply of 3.316 (measured).

- I have removed some LM4040 to test them isolated from my circuit and showed the same voltage.

- I have verified the marking of all LM4040 and showed the correct one (RKA).

- I have tried removing or adding different capacitor values.

- I tested at room temperature.

- I used 2 diffrent calibrated multimeters.

Any comments or suggestions on why the voltages are out of spec are well appreciated :). So far I am thinking either a bad batch of LM4040 from TI, or bad assembly from the assembler.

Thank you!

Olivier

  • 0
    TI__Expert 3835 points

    Hi Oliver,

    0.1% is the initial accuracy which we test to industry standards in perfect conditions. There are a lot of errors to take into consideration in applications. One is the difference in actual cathode current to nominal cathode current. This multiplied by the dynamic impedance (spec in EC table) can cause a slight voltage shift. This error is more dramatic in extreme ambient temperatures.

    Since that is not the case, I believe the largest contribution to your error would be the temperature coefficient. Although your device is not going through the full range of thermal hysteresis (and you are testing in room temperature) , the device will go through some thermal-mechanical stress due the nature of soldering the package to the circuit board. The maximum temperature coefficient at grade A is ±100ppm/C .

    At industrial range ΔT = 65°C :

    A-grade: ±0.1% ±100 ppm/°C × 65°C = ±0.75% (±22.5mV)

    Since there is no temperature coefficient or temperature range specification from the nature of assembling and soldering, taking worst case conditions is necessary.


    Hope this helps,
    Ethan

  • 0 in reply to Ethan T
    Prodigy 90 points
    Hey Ethan,

    thank you for your reply. I am not sure I understand fully the change of precision with respect to temperature.

    I understand there will be some drift on the voltage if the temperature of the LM4040 is changed. However, when the LM4040 is brought back to room temperature, I would expect the voltage output to be the same as before changing the temperature. Is that correct?

    Thank you,

    Olivier
  • 0 in reply to Olivier Frenette5
    TI__Expert 3835 points
    No, not necessarily. This is something called thermal hysteresis. The Vref will shift based on thermal cycles it goes through. On the datasheet a thermal cycle is defined as as a change from room temperature to the minimum temperature to the maximum temperature and finally back to room temperature (25C to -40C to 125C to 25C). The data sheet spec this shift to be 0.08% (800ppm) typical.

    Since there is no spec of thermal hysteresis when soldering on to a PCB, this shift may be larger. The heating of the part when soldering onto a PCB and any subsequent solder reflows will cause shifts in the reference voltage. Thermal hysteresis occurs in die stress and therefore a function of the package, die-attach material and molding compound, as well as the layout of the IC itself.

    My apologies, this is probably the answer you are looking for. The temperature coefficient is regarding to the temperature change in operation, but since you are operating at room temperature, the voltage shifts is most likely due to thermal hysteresis.

    Ethan
  • 0 in reply to Ethan T
    Prodigy 90 points
    So even if buying a 0.1%, in the end, the minute it gets soldered to a board, I have to assume it has a +- 22mV tolerance in this case?
  • 0 in reply to Olivier Frenette5
    TI__Expert 3835 points
    Yes, 0.1% is an initial accuracy specification, but it truly depends on soldering techniques and the assembly process. ±22mV was taken from an equation of temperature coefficient, and I don't expect that much of a shift from thermal hysteresis. This is an expected error and the nature of voltage references as described in my answer above.

    Ethan