TLV1822-Q1: one channel of comparator stops switching above approx. 18 kHz

Hi Christian,

Is the output on a different source? The output is swinging 5V coming from UB151, but your input is swinging 15V.

UB151 is 15V?

 that last pic it is clearly crossing the reference, was there a time when it was functioning above 18kHz?

Best,

Michael

Hi Michael,

UB151 is in the range of 14 ... 16.5 V, typical 15.5 V. GND1 and UB151 are also the supply rails for the source of the frequency signal. And the output of the comparator is switching the LED current of an optocoupler. Everything of this described circuit is related as one potential to GND1 and powered by UB151. The UB151 voltage is bypassed by 11 µF at the supply and by 100 nF close to the comparator.

And yes, the input signal is crossing the reference, but the comparator stops switching the output if the frequency rises. The TLV1822 is newly designed in our circuit and I realized this behaviour when testing it. Formerly we used LM2903D instead and there was no problem.

The open-drain output has to sink a current of about 3 mA by 5K1 at 15.5 V. In the textual description of the datasheet in chapter "8.4.2.2 TLV182x-Q1 Open-Drain Output" the recommended current range is defined to 0.1 to 1 mA. But in the "6.7 Electrical Characteristics" the output voltage is defined at a condition of 4 mA. Might this be a problem? But then the lower threshold channels should behave similar and the frequency should have no influence.

I'm wondering, if the behaviour might be related to the power-on reset, because during this time the output is also high impedant. But again, I would expect both channels with the same behaviour.

Best regards,

Christian

Hi Christian,

The 4mA of sinking current wouldn't be an issue, these could sink 20mA if you really wanted too but with other tradeoffs.

I did a quick test in our labs and saw an output with a 9V ref swinging from 0-15V ranging from 10kHz - 30kHz and saw an output, also with a 6V ref.

You mentioned another device in a similar circuit, are you able to troubleshoot this circuit? If you are able get scope shots that probe the input and output pins?

The comparator on the bottom with a ~6V ref is working? Because the only difference between these 2 would be the 6V and 9V ref, would you be able to swap these inputs and test?

Best,

Michael

Hi Michael,

here are some more detailed scope screen shots:

1. channel (yellow) V+ at pin 8 (to V- at pin 4)
2. channel (blue) IN+ at pin 3 or 5 (to V- at pin 4)
3. channel (red) IN- at pin 2 or 6 (to V- at pin 4)
4. channel (green) OUT at pin 1 or 7 (to V- at pin 4)

Channel 2 with pins 5, 6 and 7 for the low threshold works at 21 kHz:

with a rise propagation delay of 2.5 µs:

and a fall propagation delay of 560 ns:

Channel 1 at pins 1, 2 and 3 shows a rise propagation delay of 560 ns:

and a fall propagation delay of 3.6 µs:

but when reaching 18 KHz the channel starts the strange behaviour:

And when reducing the frequency it still skips edges:

And a picture of the part with test connections:

best reagrds,

Christian

Hi Christian, 

Thank you for the pictures. 

I used a 9V ref with both a push-pull comparator and an open-drain comparator w/5k pull-up resistor to drive the input and got different results, some that reflect your issue. My setup started failing at 30kHz using the open-drain comparator output to drive IN+, which makes me believe this could be input related. Are you able to drive the input with better rise times temporarily?

CH1 (yellow): IN+

CH2 (green): OUT

Vcc-Vee = 15V

Input driven with PP comparator (1st pic at 40kHz), 2nd pic 30kHz input

Input driven with open-drain comparator (1st pic at 20kHz), 2nd pic 30kHz input with skipping 

Best,

Michael

Hi Michael,

maybe I could temporarily decrease the slew rate, but for the final design we chose this filtering with this slew rate to be able to receive 25 kHz with duty cycle of 25 ... 75 % but to have suppression ability against high frequent noise.

Your tests show, that the type with push-pull output does not show this behaviour, but the type with open-drain has similar behaviour, right? Do you have an explanation for it?

Is it related to the long propagation delay of above 2 µs? The datasheet does not define a guaranteed max. delay, only typical values.

I think we will change back to the old LM2903QDRQ1 for the frequency detection function.

We also use the TLV1822QDRQ1 comparator for an overcurrent detection functionality. Now I have doubt, if it will always work properly. Here we cannot change back to the old one, because we need the low bias input current and the rail-to-rail ability.

Best regards,

Christian

Hi Christian,

Michael is OOO today. He will get back to you tomorrow.

Thanks,

Ho

Hi Christian,

Yes, it looks like it's related to the propagation delay. 

Looking into this further, TLV1822 suffers with both input types. It suffers from conditions with very large voltage underdrive (<2V) that increases the delay dramatically and effecting the toggle frequency.

I tested the setup with our faster TLV1832/TLV1842, and it did not have issues with the same input voltages and frequency range of your setup. 

There's also TLV1872 which has a level shift on the output, and you could remove the pull-up resistors.

Please let me know if you have further questions.

Best,

Michael

Hi Michael,

I need an automotive type with AEC qualification. I will have a look at the list of automotive comparators on the TI website. Thank you for your support!

Best regards,

Christian