LM358: The issue we meet when TI LM358P replace On's LM358N

Welcome to the e2e forums,

What are the units for the waveforms and what node was measured?

The 3k and 20k resistor for the TL431 equate to a T129 output of 19.1V, but supply voltage is only 15V. That seems odd.

Regardless changing the op amp should have been fine unless pin 7 voltage is under 0.8V and pin 6 is above 1V

Hello Ron,

         Sorry for my mistake, the power supply for humidity sensor is 27V, see drawing below

pin6&pin7 are all around 2.2V

The waveform are SW generated from algorithm. same humidity change, only change IC from on LM358N to TI LM358P, the amplitude is lower.

see waveform below_humidity sensor balanced, yellow is pwm out from mcu, blue is LM358N pin1 input

So Ronald please help analysis what cause the difference, the circuit inside 2 ICs?

Hi,

what circuit are R1, R2 and IC1 referring to?

Kai

Hello the block diagram is like below, R1, R2 are bridge resistors, IC1 is amplifier.

Thanks for the feedback :-)

Kai

user6272222,

IC1a (LM358 CH2) requires alternating output current flow direction on the edges. All LM358 could have an time delay if the current level exceeds the NPN constant current sink in the LM358 classic design. For a test, try adding a resistor from pin 7 to ground, 10k or less. This keeps LM358 CH2 in a class A mode. Let me know if this makes a difference. 

For more information about how the LM358 internally works see Design Guidelines for Devices with LM324/LM358 Cores

Hello Ronald,

         I will check your suggestion_ try adding a resistor from pin 7 to ground, 10k or less.

But it is an old board with PTH part, we will not modify it anymore from HW side.

Someone recommend LM358N nopb below, what's your idea? Can we get 5~10 pcs samples to evaluate first? 

lm358-n.pdf

At this point in time, the reason for the difference is still unknown, therefore trying LM358-N is just guessing. 

Try here and click 'request sample' 

https://www.ti.com/product/LM358-N#order-quality

Or here is a direct link.

https://samples.ti.com/store/samples/en/p/product/?p=LM358N/NOPB

Note: sample availability might or might not vary per region

Hello Ronald,

         adding a resistor from pin 7 to ground, 10k or less. What I shall check? IC waveform or using hotpot test delta caused by humidity change?

Tomorrow we will start, please guide, thank you.

In the hot pot comparison test, test to see if added resistor fixes the TI's delta being lower than on's with same humidity change.

Hello Ronald,

         We did the test, the noise on stable status is high, is not acceptable. please see attached file.

JT469ti add10k from pin7 to GND.7z

So far, ON had about 65 unit delta result, TI delta was about 35 units. The test experiment had a delta of about 40 units and more noise. The noise is probably the result of the parasitics of the method use to add the resistor. The normal noise is significant which means the measured signal is small. This circuit could be sensitive to minor changes.  The circuit it also part of a larger regulation loop so artifacts of the op amp section with the loop dynamics could play a role.  My 2nd theory is that the propagation delay of the comparator may be a reason for the delta difference. It would be somewhat difficult to debug this further. A lot more internal measurements will be needed. Is this something that can be done?

Hello Ronald,

         We suggested out cooking team re-do algorithm using TI LM358P， since the hot pot humidity delta can meet >=30, will let you know the process.

Thank you. Let me know if you need help with any new data gathered.