search

RobTillaart / DHTNew

Arduino library for DHT11 and DHT22 with automatic sensor recognition
MIT License
99 stars 15 forks source link

error reading am2302 over longer wire #29

Closed Jurand2020 closed 4 years ago

Jurand2020 commented 4 years ago

When using longer wire for am2302 (like 1-5m) there is always error on read. Checked with oscilloscope - the signal looks fine. Grove_Temperature_And_Humidity_Sensor library had no problem with the same hardware setup. I've used ESP32 DevKit V1.

RobTillaart commented 4 years ago

line 213 even change it to 50

RobTillaart commented 4 years ago

if i reduce timeout value in line 213 I get the DHTLIB_ERROR_SENSOR_NOT_READY too

Q: what length is your wire currently?

Jurand2020 commented 4 years ago

still no-go: Sensor not ready, -999.0, -999.0, 21081, 0

Jurand2020 commented 4 years ago

line about 1m

RobTillaart commented 4 years ago

line 266 count = 3; change to 1 ==> so after one detection of state it returns...

Jurand2020 commented 4 years ago

You get problems if you would do while (micros() < (start + timeout))

Yes, you are right.

Jurand2020 commented 4 years ago

line 266 count = 3; change to 1

no change

RobTillaart commented 4 years ago

checked my breadboard I have my sensor powered with the 5V pin. changed it to 3V3 and going to do some test runs.

RobTillaart commented 4 years ago

3V3 still works with short lines I need to find a 1m wire to recreate your config

RobTillaart commented 4 years ago

2 mtr cable does work here ... (edit: flat wires no utp/stp)

DHT_endless.ino
LIBRARY VERSION: 0.3.3

35  1   OK, 55.5,   26.6,   4926,   22
1040    2   OK, 55.5,   26.6,   2,  22
2040    3   OK, 55.4,   26.6,   4954,   22
3045    4   OK, 55.4,   26.6,   2,  22
4045    5   OK, 55.8,   26.7,   5000,   22
5050    6   OK, 55.8,   26.7,   2,  22
6050    7   OK, 55.5,   26.7,   4999,   22
7055    8   OK, 55.5,   26.7,   2,  22
8055    9   OK, 55.4,   26.7,   4902,   22

TIM CNT STAT    HUMI    TEMP    TIME    TYPE
9060    10  OK, 55.4,   26.7,   2,  22
10060   11  OK, 55.3,   26.7,   5003,   22
11065   12  OK, 55.3,   26.7,   2,  22
12065   13  OK, 55.1,   26.8,   4952,   22
13070   14  OK, 55.1,   26.8,   2,  22
14070   15  OK, 54.8,   26.8,   4852,   22
15075   16  OK, 54.8,   26.8,   2,  22
16075   17  OK, 55.1,   26.8,   4954,   22
17080   18  OK, 55.1,   26.8,   2,  22
18080   19  OK, 55.2,   26.8,   4897,   22

TIM CNT STAT    HUMI    TEMP    TIME    TYPE
19085   20  OK, 55.2,   26.8,   2,  22
20085   21  OK, 55.2,   26.8,   4903,   22
21090   22  OK, 55.2,   26.8,   2,  22
22090   23  OK, 55.2,   26.8,   4899,   22
23095   24  OK, 55.2,   26.8,   2,  22
24095   25  OK, 55.0,   26.8,   4900,   22
25100   26  OK, 55.0,   26.8,   2,  22
26100   27  OK, 54.8,   26.8,   4846,   22
27105   28  OK, 54.8,   26.8,   2,  22
28105   29  OK, 54.6,   26.8,   4804,   22

TIM CNT STAT    HUMI    TEMP    TIME    TYPE
29110   30  OK, 54.6,   26.8,   2,  22
Jurand2020 commented 4 years ago

are you using ESP?

RobTillaart commented 4 years ago

Yes ESP32 (tried to upload photo but that crashed ...)

Jurand2020 commented 4 years ago

there is sth about my sensor. I've got it all soldered and it is hard to replace the esp.

Jurand2020 commented 4 years ago

fact is I'm using thin wires inside (0,24mm kynar), but it has no more than 3cm each

RobTillaart commented 4 years ago

there is sth about my sensor. I've got it all soldered and it is hard to replace the esp.

mmmm, can you check the soldering?

Jurand2020 commented 4 years ago

funny is that it works with the previous method. can it be that communication ovelaps somehow?

Jurand2020 commented 4 years ago

connections checked with multimeter. It is still running with grove libs

RobTillaart commented 4 years ago

I tried 3 mtr cable - on bread board and it still works

note the failures in the log below was when reconnecting cables (runtime)

848036  847 SNR,    -999.0, -999.0, 1175,   22
849037  848 OK, -999.0, -999.0, 2,  22
850037  849 SNR,    -999.0, -999.0, 1175,   22

OK  CRC     TOA     TOAB    TOC     TOD     SNR     BS  UNK
815 0   0   0   0   0   34  0   0   

TIM CNT STAT    HUMI    TEMP    TIME    TYPE
851039  850 OK, -999.0, -999.0, 2,  22
852039  851 SNR,    -999.0, -999.0, 1174,   22
853040  852 OK, -999.0, -999.0, 2,  22
854040  853 SNR,    -999.0, -999.0, 1175,   22
855041  854 OK, -999.0, -999.0, 2,  22
856041  855 SNR,    -999.0, -999.0, 1174,   22
857042  856 OK, -999.0, -999.0, 2,  22
858042  857 SNR,    -999.0, -999.0, 1174,   22
859043  858 OK, -999.0, -999.0, 2,  22
860043  859 OK, 66.6,   26.0,   4850,   22

TIM CNT STAT    HUMI    TEMP    TIME    TYPE
861049  860 OK, 66.6,   26.0,   2,  22
862048  861 OK, 67.4,   26.0,   4850,   22
863053  862 OK, 67.4,   26.0,   2,  22
864053  863 OK, 69.0,   26.0,   4956,   22
865058  864 OK, 69.0,   26.0,   2,  22
866058  865 OK, 67.3,   26.0,   4800,   22
867063  866 OK, 67.3,   26.0,   2,  22
868063  867 OK, 65.0,   26.1,   4850,   22
869068  868 OK, 65.0,   26.1,   2,  22
870068  869 OK, 63.0,   26.1,   5105,   22
RobTillaart commented 4 years ago

connections checked with multimeter. It is still running with grove libs

OK, that means the hardware is OK

Propose to go back to what worked for you:

You stated that version 0.3.2 worked if you added just one line (digitalWrite(datapin HIGH) around line 204.

So I am going to try that here with my 3 mtr cable.

RobTillaart commented 4 years ago

Version 0.3.2 with 3 mtr cable fails here with BITSHIFT error. so problem situation recreated.

Jurand2020 commented 4 years ago

wow. I'm not the only one ;) btw, If I set the type before read(), then the results are:

dhtnew_test.ino LIBRARY VERSION: 0.3.3

  1. Type detection test, first run might take longer to determine type STAT HUMI TEMP TIME TYPE OK, 0.0, 0.0, 20, 22 OK, 0.0, 0.0, 2, 22 OK, 0.0, 0.0, 2, 22 OK, 0.0, 0.0, 2, 22

  2. Humidity offset test STAT HUMI TEMP TIME TYPE Sensor not ready, -999.0, -999.0, 94, 22 OK, -999.0, -999.0, 3, 22 OK, -999.0, -999.0, 2, 22 OK, -999.0, -999.0, 2, 22

Jurand2020 commented 4 years ago

no error nor correct values

RobTillaart commented 4 years ago

Adding the one liner digitalWrite(datapin HIGH) makes it work with 3 mtr cable.

This is the simplest fix for the problem. As we now understand how/where the problem is started I propose to use this minimal fix to update the library. As we understand it is a timing problem I propose to add an explicit delayMicroseconds() too

  // REQUEST SAMPLE - SEND WAKEUP TO SENSOR
  pinMode(_dataPin, OUTPUT);
  digitalWrite(_dataPin, LOW);
  // add 10% extra for timing inaccuracies in sensor.
  delayMicroseconds(_wakeupDelay * 1100UL);

  // HOST GIVES CONTROL TO SENSOR
  digitalWrite(_dataPin, HIGH);    // <<<<<<<<<<<<<<
  delayMicroseconds(2);   // <<<<<<<<<<<
  pinMode(_dataPin, INPUT_PULLUP);

  // DISABLE INTERRUPTS when clock in the bits
  noInterrupts();

If that works for you that will be the fix.

The earlier rewrite with waitFor() will get additional testing from my side as this would replace the "hard to tune" timeout loops

RobTillaart commented 4 years ago

wow. I'm not the only one ;) btw, If I set the type before read(), then the results are:

setting the type manually will not force the sensor to cooperate, (nevertheless a good experiment!) it only forces the wake-up time of the sensor to be set

Jurand2020 commented 4 years ago

If that works for you that will be the fix.

Yes, this works here.

RobTillaart commented 4 years ago

OK then I will make a new version of the lib containing those two line. Thank you very much for helping with this investigation, it has learned me some new things about fast reading digital signals and slow rising wires :)

expect the new version in< 5 minutes.

RobTillaart commented 4 years ago

Master upgraded to 0.3.3 with the fix. I will remove the test_29 branches as these did not solve the problem (have a local copy here to investigate further)

RobTillaart commented 4 years ago

testrun with 3 mtr cable starts OK - will get a night run

TIM CNT STAT    HUMI    TEMP    TIME    TYPE
2545332 2540    OK, 59.2,   26.1,   4846,   22
2546337 2541    OK, 59.2,   26.1,   2,  22
2547337 2542    OK, 59.0,   26.2,   5051,   22
2548342 2543    OK, 59.0,   26.2,   2,  22
2549342 2544    OK, 59.0,   26.2,   2,  22
2550342 2545    OK, 58.9,   26.2,   4995,   22
2551347 2546    OK, 58.9,   26.2,   2,  22
2552347 2547    OK, 58.7,   26.1,   4996,   22
2553352 2548    OK, 58.7,   26.1,   2,  22
2554352 2549    OK, 58.5,   26.2,   4894,   22

OK  CRC     TOA     TOAB    TOC     TOD     SNR     BS  UNK
2549    0   0   0   0   0   0   0   0      <<<<  not bad stats to get started  :)

TIM CNT STAT    HUMI    TEMP    TIME    TYPE
2555358 2550    OK, 58.5,   26.2,   2,  22
2556358 2551    OK, 58.5,   26.1,   4895,   22
2557363 2552    OK, 58.5,   26.1,   2,  22
2558363 2553    OK, 58.4,   26.1,   4801,   22
2559368 2554    OK, 58.4,   26.1,   2,  22
RobTillaart commented 4 years ago

Nightly run was OK sofar, > 40.000 calls with 3 meter (10 feet) cable, zero fails no guarantee but it gives some confidence

40997916    40897   OK, 62.6,   22.7,   5146,   22
40998921    40898   OK, 62.6,   22.7,   2,  22
40999921    40899   OK, 62.6,   22.7,   5147,   22

OK  CRC     TOA     TOAB    TOC     TOD     SNR     BS  UNK
40899   0   0   0   0   0   0   0   0   

TIM CNT STAT    HUMI    TEMP    TIME    TYPE
41000926    40900   OK, 62.6,   22.7,   3,  22
41001927    40901   OK, 62.6,   22.7,   5143,   22
Jurand2020 commented 4 years ago

Looks good on my setup too.