Issue with temperature units

hek

@TroyF
Strange. Try clearing eeprom an re-include sensor.

RJ_Make

I have the same problem (but reverse), and the only fix is to hard code the conversion..

Replace this (for the Temperature sketch) from @hek

 float temperature = static_cast<float(static_cast<int((gw.getConfig().isMetric?sensors.getTempCByIndex(i):sensors.getTempFByIndex(i)) * 10.)) / 10.;

With this

 float temperature = static_cast<float>(static_cast<int>((sensors.getTempFByIndex(i)) * 10.)) / 10.;

and this is what I used for the Humidity & Temp. Sketch

	//float temperature = dht.getTemperature();
	float temperature = dht.getTemperature()*9/5 + 32;

	if (isnan(temperature)) {
		Serial.println("Failed reading temperature from DHT");
	} else if (temperature != lastTemp) {
	lastTemp = temperature;
	if (!metric) {
		temperature = dht.getTemperature()*9/5 + 32;   //dht.toFahrenheit(temperature);
	}
	gw.send(msgTemp.set(temperature, 1));
	Serial.print("T: ");
	Serial.println(temperature);
	}

EDIT: Keep in mind you will need to make the changes necessary to convert to Celsius

TroyF

Strange. I actually just left it alone and a few hours later it was reading in Celsius like it should have been. Not sure why it took a while to start reading that way, but it's fine now. Thanks to everyone for their responses.

legeantvert

I have seen the same issue with 1.4 when using DHT or Dallas ...
I have try to clean EEPROM, change on Vera from metric to Imperial and back to metric without any succes...
So hardcode temperature in C was the solution for me...

jocke4u

I have the same problem with DallasSensor

Startup says:

sensor started, id 2
send: 2-2-0-0 s=255,c=0,t=17,pt=0,l=3,st=ok:1.4
send: 2-2-0-0 s=255,c=3,t=6,pt=1,l=1,st=ok:0
read: 0-0-2 s=255,c=3,t=6,pt=0,l=2:M                          <<<==== Metric <<<
send: 2-2-0-0 s=255,c=3,t=11,pt=0,l=17,st=ok:Light+Temp Sensor
send: 2-2-0-0 s=255,c=3,t=12,pt=0,l=3,st=ok:1.0
send: 2-2-0-0 s=0,c=0,t=16,pt=0,l=3,st=ok:1.4

And I trace:

Serial.println(gw.getConfig().isMetric);

which output 0, not 1

Switching to Imperial I get:

sensor started, id 2
send: 2-2-0-0 s=255,c=0,t=17,pt=0,l=3,st=ok:1.4
send: 2-2-0-0 s=255,c=3,t=6,pt=1,l=1,st=ok:0
read: 0-0-2 s=255,c=3,t=6,pt=0,l=2:I                          <<<==== Imperial <<<
send: 2-2-0-0 s=255,c=3,t=11,pt=0,l=17,st=ok:Light+Temp Sensor

and the trace says 0 as expected.

Then I tried to trace in MySensors.cpp and added (not workin though ;) ):

} else if (type == I_CONFIG) {
    // Pick up configuration from controller (currently only metric/imperial)
    // and store it in eeprom if changed
    isMetric = msg.getByte() == 'M' ;
    debug(PSTR("isMetric=%s\n"), isMetric);
    debug(PSTR("cc.isMetric=%s\n"), cc.isMetric);
    if (cc.isMetric != isMetric) {

But then it started to work :D

Removed the lines above and the problem is back....so something there...
Don't have time right now to test....

hek

@jocke4u

Are you saying adding debug prints makes it work?

jocke4u

@hek

Yes for some strange reason :)
However I used %s and not %d in the debug but have now used %d as below

} else if (type == I_CONFIG) {
    // Pick up configuration from controller (currently only metric/imperial)
    // and store it in eeprom if changed
    isMetric = msg.getByte() == 'M' ;
    debug(PSTR("isMetric=%d\n"), isMetric);
    debug(PSTR("cc.isMetric=%d\n"), cc.isMetric);
    if (cc.isMetric != isMetric) {
        cc.isMetric = isMetric;
        debug(PSTR("Updating EEPROM\n"));	
        debug(PSTR("isMetric=%d\n"), isMetric);	
        debug(PSTR("cc.isMetric=%d\n"), cc.isMetric);						
        eeprom_write_byte((uint8_t*)EEPROM_CONTROLLER_CONFIG_ADDRESS, isMetric);
        //eeprom_write_block((const void*)&cc, (uint8_t*)EEPROM_CONTROLLER_CONFIG_ADDRESS, sizeof(ControllerConfig));
    }

Gives the output

sensor started, id 2
send: 2-2-0-0 s=255,c=0,t=17,pt=0,l=3,st=ok:1.4
send: 2-2-0-0 s=255,c=3,t=6,pt=1,l=1,st=ok:0
read: 0-0-2 s=255,c=3,t=6,pt=0,l=2:M
isMetric=0
cc.isMetric=1
Updating EEPROM
isMetric=0
cc.isMetric=0

So it seems that "metric" flag is not unpacked correctly, or?

Yveaux

@jocke4u said:

l=2:M

I find it strange that length is reported as 2 while only a single character seems to be sent. The protocol version 1.4 is correctly reported as length 3.

hek

@Yveaux

uint8_t MyMessage::getByte() const {
	if (miGetPayloadType() == P_BYTE) {
		return data[0];
	} else if (miGetPayloadType() == P_STRING) {
		return atoi(data);
	} else {
		return 0;
	}
}

Payload type will be P_STRING when controller sends the data. Yes, why does it get size 2?!

hek

Hmm.. we should probably not use atoi in case of String. Payload type should not matter in the getByte.

hek

Yep, this is a bug.

I've also found that the newline character is included and transmitted on all outgoing messages from gateway! So payload length 2 was correct, and wrong ;)

How is it possible no one has seen this before (me included)?

jocke4u

Sounds good anyway....one bug less :)
I just setup a sensor with DHT22/AM2302 and get the same results here.

hek

Fix checked in. I'm creating a more comprehensive release note in the 1.4 thread.

Yveaux

@hek said:

How is it possible no one has seen this before (me included)?

At least I saw it ;-)
I personally tend to read over all the details in the serial api... This time I noticed.

Be glad you nailed it!

jocke4u

Thanks @hek and @Yveaux I can confirm it's working on both Dallas sensor and DHT22.
Great support!!

Nuubi

Hmmm... how can one override this statement to be METRIC?
float temperature = static_cast<float>(static_cast<int>((gw.getConfig().isMetric?TemperatureSensors.getTempCByIndex(CHILD_ID_TEMPERATURE_FIRST+i):TemperatureSensors.getTempFByIndex(CHILD_ID_TEMPERATURE_FIRST+i)) * 10.)) / 10.;

hexenmeister

hello,

I must confess I have not read this thread completely. Too much for my bad english)

We try MySensors support for FHEM (www.fhem.de) to impementieren. I noticed that when the I_CONFIG message is answered correctly with 'M' for metric, then the sensor sends in Imperial. Is not surprising, everything is interpreted as a string from Gateway, but also byte checked. My solution proposal: In MySensors.cpp:
(isMetric = msg.getString()[0] == 'M' ; instead of isMetric = msg.getByte() == 'M' ;)

			} else if (type == I_CONFIG) {
				// Pick up configuration from controller (currently only metric/imperial)
				// and store it in eeprom if changed
				//isMetric = msg.getByte() == 'M' ;
				isMetric = msg.getString()[0] == 'M' ;
				if (cc.isMetric != isMetric) {
					cc.isMetric = isMetric;
					eeprom_write_byte((uint8_t*)EEPROM_CONTROLLER_CONFIG_ADDRESS, isMetric);
					//eeprom_write_block((const void*)&cc, (uint8_t*)EEPROM_CONTROLLER_CONFIG_ADDRESS, sizeof(ControllerConfig));
				}
			} else if (type == I_CHILDREN) {

Best regards,

Alexander

hek

Welcome @hexenmeister!

What version of MySensors library did you download? The current version in gihub/master already has getString()[0].

hexenmeister

Thanks for the quick reply.
I've downloaded the source here: https://github.com/mysensors/CodeBender/blob/master/MySensor.cpp
Since it is still old.

Now I see that https://github.com/mysensors/Arduino/blob/master/libraries/MySensors/MySensor.cpp would be better.

hek

Oh.. you downloaded from codebender. That explains it.

Unfortunately it's still a bit too messy to update the public libraries over there (have to contact the codebender-people every time). Nowadays we keep a copy among in the MySensors-account which gets copied if you clone one of the examples.

Github always contains the latest.

Issue with temperature units

13

11.7k

11.2k

113.1k