search

letscontrolit / ESPEasy

Easy MultiSensor device based on ESP8266/ESP32
http://www.espeasy.com
Other
3.3k stars 2.22k forks source link

LoopTimerSet_ms won´t work as expected, if PN532 enabled #4170

Closed kuppe234 closed 2 years ago

kuppe234 commented 2 years ago

Hello, I have the "RFID - PN532" connected on the ESP32. When I use "LoopTimerSet_ms" the loops are very different in its lengh. I use it to blink with the onboard LED (GPIO-2) in a rule. The same applies to all GPIOs to which I have connected an LED.

Example:

On System#Boot do
  looptimerSet_ms,1,500,60
endon

On Rules#Timer=1 do
  GPIOtoggle,2
endon

When I switch off the plugin, "looptimerset" works as expected, the LED flashes steadily. In a firmware I checked out and built on April 01 this year, this problem does not occur yet. What could be the cause?

Build: normal_ESP32_4M316k (ESP_Easy_mega_20220802_normal_ESP32_4M316k.bin) I2C: SDA: GPIO-5, SCL: GPIO-4 Plugins: only PN532 Wifi Status LED: none GPIO boot states: all on default

TD-er commented 2 years ago

This is the cause: https://github.com/letscontrolit/ESPEasy/issues/4103

In short, some code does actually "wait" for a sensor to complete. We're in the process to make this more asynchronous exactly for this reason you're describing (and others like WiFi connections getting dropped etc.)

kuppe234 commented 2 years ago

Thank you for the explanation. But I think it can also be due to "IDF 4.4". On 27.04.2022 there was a "Merge pull request #3845 from TD-er/build/ESP32_IDF4.4". Since then, the timers no longer work properly (tested with various git checkouts).

TD-er commented 2 years ago

Can you give some screenshots of the timingstats page running on both builds? And maybe even more important, can you test if it still persists in one of the last GitHub Actions test builds? (using a newer SDK version of ESP32)

kuppe234 commented 2 years ago

Here are 2 screenshots, one as of 20/04/2022 and one as of today. These are the identical configuration with the same hardware.

The RFID reading process is slow (1..5 seconds) in the current firmware, but there is no delay in the one from 20.04.

tonhuisman commented 2 years ago

Here are 2 screenshots,

You can drop the screenshots in the Github web editor, so we don't have to access an unknown external host to view that...

TD-er commented 2 years ago

I just received the same sensor today, so I can see what's the difference among builds.

TD-er commented 2 years ago

I have connected the PN532 and indeed it takes forever to get a reading. Apparently the I2C performance has taken quite a hit in the latest ESP32 SDK version. See also: https://github.com/espressif/arduino-esp32/issues/6553

Jason2866 commented 2 years ago

The i2c driver is refactored in Arduino => 2.0.3. It is now not amymore fault tolerant to bad (wrong) written i2c implementations. Maaany i2c libs are just garbage. Theo had to refactor some we use in Tasmota. After this was done speed was the same as before.

TD-er commented 2 years ago

Thanks for the heads up. I will have another look at the timings but it seems this PN532 is a bit of a lost cause regarding I2C compliance.

N.B. another thing I noticed is that it seems switching I2C frequencies is also not working anymore.??? (on ESP32)

Jason2866 commented 2 years ago

We gave up (i think it was 2 years ago) to get PN532 reliable running via i2c. Iirc clock stretching is a mess how it is done with the PN532. Andre Thomas burned a lot of time with... The i2c driver code was big and ugly and still no joy. So he decided to remove it completely. Luckily the device can be connected via serial too. So he wrote a driver for serial connect. Via serial it runs stable. Had a uptime for 6 month. Just ended because i did a update of Tasmota.

TD-er commented 2 years ago

I now seem to have it running stable, but I will for sure keep it in mind when the next reported issue shows up for this plugin.

As a matter of fact, I will link this conversation in de code to remind me :)

TD-er commented 2 years ago

Just merged the code to fix this. @kuppe234 You can try it with the latest code in the mega branch. Latest GH Actions build including this code

kuppe234 commented 2 years ago

I have test the new mega branch. It worked as expected. Here you have the actual Timings: normal_ESP32_20220812

The repetition rate for reading is also the same as for the April-Branch:

703864 : Info   : PN532: New Tag: 1458541861 39
703889 : Info   : EVENT: RFID#Tag=1458541861
703907 : Info   : ACT  : publish,ESP_Easy/RFID/Uhrzeit,"2022-08-12 11:03:53"
703921 : Info   : ACT  : publish,ESP_Easy/RFID/Seriennummer,1458541861
703959 : Info   : ACT  : let,1,1
703965 : Info   : ACT  : publish,ESP_Easy/RFID/Zugang_gestattet: ,nein
703977 : Info   : ACT  : tone,3,300,600
703986 : Info   : ACT  : publish,ESP_Easy/RFID/Status,1
703998 : Info   : ACT  : TimerSet,5,10
704176 : Info   : PN532: Old Tag: 1458541861 40
704476 : Info   : PN532: Old Tag: 1458541861 41
704779 : Info   : PN532: Old Tag: 1458541861 42
705085 : Info   : PN532: Old Tag: 1458541861 43
705387 : Info   : PN532: Old Tag: 1458541861 44
705694 : Info   : PN532: Old Tag: 1458541861 45
705995 : Info   : PN532: Old Tag: 1458541861 46
706299 : Info   : PN532: Old Tag: 1458541861 47
706609 : Info   : PN532: Old Tag: 1458541861 48
706912 : Info   : PN532: Old Tag: 1458541861 49
707215 : Info   : PN532: Old Tag: 1458541861 50
707519 : Info   : PN532: Old Tag: 1458541861 51
707823 : Info   : PN532: Old Tag: 1458541861 52
708128 : Info   : PN532: Old Tag: 1458541861 53
708630 : Info   : RFID : Removed Tag