TLV1822-Q1: Confusing behavior with hysteresis

Hi John,

I'll be here to help you with your issue. First of all, there's nothing to forgive, and it's definitely not obvious to me.

Your calculations for the hysteresis and simulations seem correct, so I have a couple of questions about your physical circuit.

1. You said that the signal looks correct at the filter. Could you please show a picture of what you observed at the filter (voltage across C?)?

2. How is the 2.5V for Pin 6 (IN-) generated? Could you please probe Pin 6 (IN-) and make sure that it is a clean 2.5V?

3. What is the port "TO_MICRO"? Is it an MCU GPIO? If so, could you make sure that the GPIO is configured as an input HIGH-Z instead of pullup/pulldown mode?

1:  Here a capture showing the previous signals along with the filtered value IN_filter

2. Here is the circuit for the 2.5V reference:

it looks pretty clean:

3.  The pin does connect to a micro, it is set for high-Z, no pullup/pulldown

Thanks!

Hi John,

Thanks for the follow up. These pictures and the reference generator check out to me. Could you please do these additional tests to help isolate the problem:

1. Could you tell me how the 100Hz square wave is generated? Our calculation for hysteresis assumes low output impedance from the IN_raw source, so any high impedance from the IN_raw source would cause the voltage in the non-inverting terminal of the comparator to be higher than expected.

2. Could you please verify that all the resistors on the board are actually the right values as shown in the schematic? Changes in any of the resistor values would cause our calculated voltages to be different.

3. Could you please try removing the feedback resistor (1MEG) and see if applying the same square wave would cause the comparator to switch properly? This is to check if the device operates properly. It would be helpful to make sure that the voltage at Pin 5 under this configuration swings from 0 to 10V as a sanity check.

I will do those tests as soon as I can.

1.  I am using a Siglent SDG2122X signal generator with the load set to HiZ.

Hi John,

I see. Setting the load to Hi-Z wouldn't change the 50Ω output impedance of the signal generator, so the signal source of IN_raw should still be low impedance. Please do the other tests and let me know the results.

Thanks,

Ho

2. All of the resistors are within 1%

3. I get the same result, very strange, I am going to try another 1822 part

With the part removed the signal goes to ground on the pin 5 pad.

Hi John,

Thanks for the follow-up. I really appreciate all of your actions in debugging this issue.

To clarify:

John Lagerquist said:

With the part removed the signal goes to ground on the pin 5 pad.

Do you mean that with the signal generator turned on (0-10V 100Hz) and the 5V DC source turned on, the voltage at the Pin 5 pad is ground (unexpected)? Or is it ground with all the sources turned off (expected)? 

What is that screenshot you've provided showing? Is that with the device removed and the sources (Siglent and 5V DC) turned on?

Just curious, what are you using the other channel of the TLV1822 for? How is it configured?

John Lagerquist said:

3. I get the same result, very strange, I am going to try another 1822 part

This is strange. With the feedback resistor removed, the device should switch due to the 0-10V crossing the 2.5V reference. Could you please check the board to ensure that the device is soldered on correctly and that the pinout corresponds to what you expect?

With the part removed and board and sig gen powered it's the waveform i last posted.  It does get to ground.

I have tested it on multiple different boards.  I will check the pinout\layout.  The other channel is configured identically.

I get the result on the other channel.

Hi John,

What's generating the 5V rail? Is it connected to a DC source? I don't expect Pin 5 pad to go all the way to ground as there is a DC path to 5V through the pull-up resistor. That source should pull the Pin 5 pad to some voltage above ground. The simulation lines up with my calculations. The voltage when IN_raw goes to ground should be ~1.39V.

Hi John,

John Lagerquist said:

I get the result on the other channel

Could you clarify what you mean? Do you observe the same result with the other channel where the comparator does not switch and stays high, or does it switch properly which is different from what we've seen?

John Lagerquist said:

2. All of the resistors are within 1%

I'm still getting a feeling that the resistor values aren't exactly what we see on the schematic. Did you measure them to confirm that they are all within 1% of the values, or is this just the tolerance of the resistors from the delivery packaging? A good sanity check would be to physically disconnect the 5V rail and remove the TLV1822, short IN (IN_raw) to GND, and measure the resistance from the Pin 5 pad to GND using a multimeter. This should give a result of 402KΩ + 1.62KΩ. You can subsequently remove the 402KΩ, and confirm that it is indeed 1MΩ across the Pin 5 and Pin 7.

It would also be helpful to do a "voltage map" of the circuit. With the TLV1822 and all resistors populated. Apply DC 10V at IN and DC 5V at the pullup and measure the voltage at Pin 5 to GND using a Digital Multimeter. Then apply a DC 0V at IN DC 5V at the pullup and measure the voltage at Pin 5 to GND. Write down those voltages to see if they correspond to what we would expect. With these DC values, we should expect Pin 5 to not go all the way to ground @ IN = 0V.

Once again, I really appreciate your patience and the work you've done to help debug this issue.

With the TLV1822 removed and unpowered we measure 404K at pad 5.

With everything populated measuring at pin 5 with a DC source

0.0V: 2.59V

5.0V: 4.55V

10.0V: 8.03V