Wait for conversions in continuous example

[WizardTim]

Previously polling rate (apparent sample rate) was limited purely by I2C speed, script will now wait a full conversion cycle until it next reads the conversion result.

Requires #61 to be merged first.

With RATE=250 and i2c_baudrate=1000000 (1 MHz)

Old

Took 0.139 s to acquire 1000 samples.

Configured:
    Requested       =   250    sps
    Reported        =   250    sps

Actual:
    Polling Rate    =  7171.07 sps
                       2868.43%
    Repeats         =   966
    Conversion Rate =   243.82 sps   (estimated)

New

Took 4.009 s to acquire 1000 samples.

Configured:
    Requested       =   250    sps
    Reported        =   250    sps

Actual:
    Polling Rate    =   249.41 sps
                         99.76%
    Repeats         =    25
    Conversion Rate =   243.18 sps   (estimated)

---

Also constrains polling rate at any other valid sample rate, eg. RATE=3300 and i2c_baudrate=1000000 (1 MHz) New

Took 0.309 s to acquire 1000 samples.

Configured:
    Requested       =  3300    sps
    Reported        =  3300    sps

Actual:
    Polling Rate    =  3234.39 sps
                         98.01%
    Repeats         =    25
    Conversion Rate =  3153.53 sps   (estimated)

[WizardTim]

This pull request resolves #27, #35, #43.

New

Took 10.003 s to acquire 33000 samples.

Configured:
    Requested       =  3300    sps
    Reported        =  3300    sps

Actual:
    Polling Rate    =  3298.86 sps
                         99.97%
    Skipped         =     2
    Repeats         =   444
    Conversion Rate =  3254.48 sps   (estimated)

Plot of time.monotonic() error in comparison to time_next_sample at ADC read (3,300 sps jitter target is 303 μs)

Bugs: Will often skip first 1 or 2 samples, I suspect this is due to the following line however I don't think it's worth the trouble to change it. https://github.com/adafruit/Adafruit_CircuitPython_ADS1x15/blob/c513e9ec6681e7f458a08d439f268b7c27585b6a/adafruit_ads1x15/ads1x15.py#L180

[WizardTim]

New Fix it skipping 1-2 samples on the first acquisition due to setup time, still skips 1 sample every so often with >1000 samples

Took 0.303 s to acquire 1000 samples.

Configured:
    Requested       =  3300    sps
    Reported        =  3300    sps

Actual:
    Polling Rate    =  3298.52 sps
                         99.96%
    Skipped         =     0
    Repeats         =    27
    Conversion Rate =  3209.46 sps   (estimated)

[caternuson]

Sorry. Just now seeing this.

So this essentially looks like you've implemented something like suggested here? https://github.com/adafruit/Adafruit_CircuitPython_ADS1x15/issues/35#issuecomment-501747400

[WizardTim]

Sorry. Just now seeing this.

So this essentially looks like you've implemented something like suggested here? #35 (comment)

Yes pretty much although this implementation is a bit more advanced to make sure the sample rate doesn't drift or over-sample to catch up. It also uses a blocking while loop instead of time.sleep(SOME_DELAY) in order to achieve better timing at the cost of power consumption.

[caternuson]

Yep. Thanks. Looks like you also tested it, so I'm fine with merging. If this was useful for you, hopefully will be for others trying to use fast read.