Re: [PATCH v1 2/3] iio: ltr501: Add interrupt rate control support

[Lars-Peter Clausen <lars@metafoo.de>]

On 04/01/2015 07:45 PM, sathyanarayanan kuppuswamy wrote:
>
>
> On 04/01/2015 08:15 AM, Lars-Peter Clausen wrote:
>> On 04/01/2015 05:02 PM, Daniel Baluta wrote:
>>> On Wed, Apr 1, 2015 at 5:39 PM, Lars-Peter Clausen <lars@metafoo.de> wrote:
>>>> On 04/01/2015 04:04 PM, Daniel Baluta wrote:
>>>> [...]
>>>>>
>>>>>
>>>>>> +static const struct iio_chan_spec_ext_info ltr501_ext_info[] = {
>>>>>> +       {
>>>>>> +               .name = "intr_persist",
>>>>>> +               .read = ltr501_read_intr_prst,
>>>>>> +               .write = ltr501_write_intr_prst,
>>>>>> +               .shared = IIO_SHARED_BY_TYPE,
>>>>>> +       },
>>>>>> +       {},
>>>>>> +};
>>>>>> +
>>>>>
>>>>> Would be nice to standardize persistence attribute
>>>>> (IIO_CHAN_INFO_PERSISTENCE).
>>>>
>>>>
>>>> If I understand the behavior correctly it causes that the event needs to be
>>>> triggered at least n times before the event is reported by the chip. In my
>>>> opinion 'persistence' is not a good term for that. I'm not sure what a
>>>> better term is but I think it should go more in the direction of ratelimit
>>>> or something.
>>>
>>> I've seen this term used for many devices:
>>>
>>> *  TSL25911 ambient light sensor [1]
>>>
>>> [ One set of thresholds can be configured to trigger an interrupt only when
>>>   the ambient light exceeds them for a configurable amount of time
>>> (persistence)
>>> ]
>>>
>>> * TAOS TCS34725 ambient light sensor [2]
>>> [
>>> The interrupt persistence filter allows the user to define the number
>>> of consecutive
>>>   out-of-threshold events necessary before generating an interrupt.
>>> ]
>>>
>>> * Avago SAPDS-9950, Sensortek STK3310
>>>
>>> I think the TSL25911 datasheet best describes this parameter, as the
>>> amount of time
>>> that ambient light should exceed a threshold until an interrupt is
>>> generated.
>>
>> Ok, that makes more sense. I misunderstood the initial description as that
>> the signal would have to go first above the threshold then below the
>> threshold, and this for a number of times. Whereas it needs to exceed the
>> threshold for a certain amount of time before the event is triggered. If
>> it goes below the threshold before the persistence interval no event is
>> triggered and the counter is reset.
>
> Yes, it needs to cross the threshold n number of times before a event is
> generated.

Wait. It needs to cross the threshold or it needs to stay above the threshold?