bmp/bme280 first read

[johndoe71rus]

First read data bmp/bme280 get error

dofile("bmp280.lua")
bmp280.lua:7: bad argument #2 to 'format' (number expected, got nil)
stack traceback:
[C]: in function 'format'
bmp280.lua:7: in main chunk
[C]: in function 'dofile'
stdin:1: in main chunk

all next readings correct.
I have bme280 and bmp280, check booth.

Code http://paste.ubuntu.com/24710875/
logs http://paste.ubuntu.com/24711075/

i can wait long time to start script. Problem only with the first reading. After power up

[marcelstoer]

I can confirm that. Just ran into this yesterday.

[devsaurus]

What is returned by init()? Should be nil in the error case.

[marcelstoer]

My test bed is really simple (snippet from our examples). Note that I'm on the 1.5.4 branch while the OP tests against https://github.com/nodemcu/nodemcu-firmware/tree/22e1adc4b06c931797539b986c85e229e5942a5f i.e. doesn't have your #1887 changes yet.

print(bme280.init(6, 7))

H, T = bme280.humi()
print(string.format("T=%d.%02d", T / 100, T % 100))
print(string.format("humidity=%d.%03d%%", H / 1000, H % 1000))

Run for the 1st time right after power up this yields

2 -- i.e. it's a BME280
stdin:1: attempt to perform arithmetic on global 'T' (a nil value)
...

NOT running bme280.init() again but only querying the sensor for humi after that works fine.

[devsaurus]

Maybe you hit the initial measurement phase with the read commands.
Does it help anything to insert a ~100-200ms delay between init() and humi()?

[marcelstoer]

A delay of even 10ms fixes this for me:

print(bme280.init(6, 7))

tmr.create():alarm(10, tmr.ALARM_SINGLE, function()
    local H, T = bme280.humi()
    print(string.format("T=%d.%02d", T / 100, T % 100))
    print(string.format("humidity=%d.%03d%%", H / 1000, H % 1000))
end)

Is there a sensible way for the module to correct this internally or do we just need to document this behavior?

[devsaurus]

The delay depends on the selected oversampling rates (ref. chapter 9.1 in the BME280 datasheet) and varies between some tens to hundreds of milliseconds.
I think the user script should at least get a chance to spend the time with something more meaningful than waiting. So I'd vote for the doc solution (and have consistent examples which poll after init until data is delivered).

[eyaleb]

I expect that the correct way to handle this is to check the status register (0xF3) for zero after a forced read or after a cold start.

[johndoe71rus]

So, "universal" code looks like this?

alt=217 -- altitude of the measurement place
mode = nil
mode = bme280.init(5, 4)
tmr.create():alarm(200, tmr.ALARM_SINGLE, function()
if mode == 1 then
    print("bmp280")
    T, P, H, QNH = bme280.read(alt)
    local Tsgn = (T < 0 and -1 or 1); T = Tsgn*T
    print(string.format("T=%s%d.%02d", Tsgn<0 and "-" or "", T/100, T%100))
    print(string.format("QFE=%d.%03d", P/1000, P%1000))
    print(string.format("QNH=%d.%03d", QNH/1000, QNH%1000))
    MHG = P / 133,3
    print(string.format("pressure = %d.%02d", MHG/10 , MHG%10))
    -- altimeter function - calculate altitude based on current sea level pressure (QNH) and measure pressure
    P = bme280.baro()
    curAlt = bme280.altitude(P, QNH)
    local curAltsgn = (curAlt < 0 and -1 or 1); curAlt = curAltsgn*curAlt
    print(string.format("altitude=%s%d.%02d", curAltsgn<0 and "-" or "", curAlt/100, curAlt%100))
elseif mode == 2 then
    print("bme280")
    T, P, H, QNH = bme280.read(alt)
    local Tsgn = (T < 0 and -1 or 1); T = Tsgn*T
    print(string.format("T=%s%d.%02d", Tsgn<0 and "-" or "", T/100, T%100))
    print(string.format("QFE=%d.%03d", P/1000, P%1000))
    print(string.format("QNH=%d.%03d", QNH/1000, QNH%1000))
    MHG = P / 133,3
    print(string.format("pressure = %d.%02d", MHG/10 , MHG%10))
    print(string.format("humidity=%d.%03d%%", H/1000, H%1000))
    D = bme280.dewpoint(H, T)
    local Dsgn = (D < 0 and -1 or 1); D = Dsgn*D
    print(string.format("dew_point=%s%d.%02d", Dsgn<0 and "-" or "", D/100, D%100))
    -- altimeter function - calculate altitude based on current sea level pressure (QNH) and measure pressure
    P = bme280.baro()
    curAlt = bme280.altitude(P, QNH)
    local curAltsgn = (curAlt < 0 and -1 or 1); curAlt = curAltsgn*curAlt
    print(string.format("altitude=%s%d.%02d", curAltsgn<0 and "-" or "", curAlt/100, curAlt%100))
else
    print("no device")  
end
end)

[johndoe71rus]

i don't know why panic happend
paste.ubuntu.com/24891194

paste.ubuntu.com/24891211 sensors.lua
first time it go to print("ok") but then panic because P is nil

[johndoe71rus]

I have to replace

bme280.init(bmeSdaPin, bmeSclPin) 

to

i2c.setup(0, sda, scl, i2c.SLOW) -- call i2c.setup() only once
bme280.setup()

?

this work in master or i need dev firmware?