ALERT/RDY default mode is active low, but example sketch for continuous mode thinks it's active high (due to a misleading diagram in the datasheet)

[bjfischb]

The continuous.ino example has the following code and comments in setup():

  pinMode(READY_PIN, INPUT);
  // We get a falling edge every time a new sample is ready.
  attachInterrupt(digitalPinToInterrupt(READY_PIN), NewDataReadyISR, FALLING);

I think the comment saying that new data is ready on the falling edge of ALERT/RDY is based on diagram 7-8 in the datasheet, which shows the ALERT/RDY as being active high, with the sample being ready at the end of the 8usec positive pulse on the ALERT/RDY pin.

I was looking at the waveforms on an oscilloscope, though, and I saw that the ALERT/RDY behavior is active low, with a low-to-high transition at the end of the 8usec pulse.

I took a closer look at the datasheet and it says that the polarity of ALERT/RDY can be adjusted with the COMP_POL bit. The datasheet says that the default value of COMP_POL is 0b'0, which corresponds to active low.

It's confusing because the name COMP_POL suggests it's controlling only the comparator. However if you read the datasheet carefully, it does say that COMP_POL also affects the polarity of ALERT/RDY when it's being used to indicate the completion of a continuous sample.

As the continuous.ino example is written, the interrupt flag gets set on the falling edge of COMP_POL, which is the start of the 8usec pulse. The data isn't actually ready until the end of the 8usec pulse. It doesn't make a practical difference because it takes at least 8uSec to issue the I2C command to the ADS1115 that requests the data.

But if anyone else besides me hooks up an oscilloscope to their system, they will see the opposite polarity on ALERT/RDY vs. what's described in the example.

[caternuson]

I was looking at the waveforms on an oscilloscope, though, and I saw that the ALERT/RDY behavior is active low, with a low-to-high transition at the end of the 8usec pulse.

So your scope trace looked like the datasheet figure 7-8, but with opposite polarity?

[bjfischb]

Yes, exactly.

If you read the text in the datasheet, it says COMP_POL can be used to get the waveform shown above. But, the default is the opposite, and that's what I saw on my scope (since COMP_POL wasn't changed from its default in the library)

[caternuson]

Thanks. Confirmed.

While running the continuous example on a QT PY M0 and connecting logic scope to ALERT pin - it idles high with the assert being low:

The pulses are the expected ~8us width, but with opposite polarity from datasheet figure 7-8:

[caternuson]

Yah, that is a confusing figure. It does say "active high" in the parans, but that is easily overlooked. And that is the non-default setting for COMP_POL.

[caternuson]

I think this is a simple fix of just changing FALLING to RISING in the example. But have you seen any indications that this issue reaches into other things? Like the library code itself?

[bjfischb]

I don't think there's any problem with the library code itself. Like you say, changing the interrupt from falling to rising in the example will synchronize the interrupt with the completion of a conversion.

Thanks for the quick action on this. It's not urgent, so can wait until you feel the need to do a code release for other reasons.

[caternuson]

OK, updated that example with #99.

Thanks for catching this.