Reading DS18B20s is a bit erratic

[pscs]

I'm trying to use the BAS to read two DS18B20s, but it seems erratic from Python on a rPi 4B. I've just updated to firmware 2.01 which has made it work at all, but it's still not great.

I do the 'megabas.owbScan(0)' at the start of my program, which seems to start a background task on the I/O processor to scan the devices (the documentation isn't clear what's actually going on). (It seems quite slow at scanning (much slower than when I scanned from my own software on a Pico))

Then, every second or so, I do:

x = megabas.owbGetSensorNo(0)
y = megabas.owbGetRomCode(0, 1)
z = megabas.owbGetTemp(0, 1)
print(f"Temp Sensors - {x} {y} {z}")

x = megabas.owbGetSensorNo(0)
y = megabas.owbGetRomCode(0, 2)
z = megabas.owbGetTemp(0, 2)
print(f"Temp Sensors - {x} {y} {z}")

Usually, it works fine, and prints
Temp Sensors - 2 [40, 2, 239, 149, 240, 1, 60, 241] 23.06
Temp Sensors - 2 [40, 105, 252, 149, 240, 1, 60, 248] 23.93

which looks OK with the correct IDs and reasonable temperatures.

But, occasionally, it'll get the IDs messed up, and print:

Temp Sensors - 2 [40, 105, 252, 149, 240, 1, 60, 241] 23.06
Temp Sensors - 2 [40, 105, 252, 149, 240, 1, 60, 248] 23.87

or

Temp Sensors - 2 [40, 2, 239, 149, 240, 1, 60, 241] 23.75
Temp Sensors - 2 [40, 2, 252, 149, 240, 1, 60, 248] 24.75

or similar

So, the IDs seem to have random bytes copied from the other sensor. The temperatures look OK and are different for the different probes, as expected.

I don't know if this is a race condition on the I/O controller or something.

Any ideas?

[alexburcea2877]

Hi,

The problem might be the I2C communication between Raspberry and card. Can you add few millisecond delay between reading functions?
The ID's are changed only during scan process. I will investigate the scanning time.

[pscs]

I've added a short delay (10ms) and it seems better, but not perfect.

As a possibly related point, I've just had a case where it didn't read the ID correctly at all

paul@raspberrypi:~ $ megabas 0 owbscan
OK
paul@raspberrypi:~ $ megabas 0 owbidrd 1
0x263c530000000000
paul@raspberrypi:~ $ megabas 0 owbidrd 2
0xfd3cc8e3817bec28
paul@raspberrypi:~ $ megabas 0 owbidrd 3
0xd3c74e381a71a28
paul@raspberrypi:~ $ megabas 0 owbidrd 4
0x8c2ce08181692928

The 'owbidrd 1' is wrong. I tried doing another owbscan, and even rebooting the Pi and it didn't fix it. The ID of probe '1' always came back with lots of 0000s

A full power-cycle did fix it:

paul@raspberrypi:~ $ megabas 0 owbcntrd
5
paul@raspberrypi:~ $ megabas 0 owbidrd 1
0x263c53e3818c7028
paul@raspberrypi:~ $ megabas 0 owbidrd 2
0xfd3cc8e3817bec28
paul@raspberrypi:~ $ megabas 0 owbidrd 3
0xd3c74e381a71a28

I couldn't see any way to reset the I/O controller on the Megabas board, without power cycling it, which is a bit of an issue as it meant opening up the enclosure to pull the plug, which isn't 'user-friendly' and is hard to automate.

And sometimes it reads one probe as 0x8c3cece381692928 and other times the same probe as 0x8c2ce08181692928 (so far, always the same two IDs). The other probes seem to be consistently correct (I don't know what the correct ID of this probe is, as I've been relying on the Megabas to tell me the ID)

I'm not sure if this is the ID scanning code in the I/O controller, or something else.

[alexburcea2877]

Hi,
There is a way to reset the Megabas card without cycling the power. I could do this available on command line but let me try to find the problem first .