Support for Ecowitt / Fineoffset WS90 (Wittboy GW2001 station)

[koalabi]

I'm a new (and enthusiastic) user of rtl_433 that I would be happy to use with my recent Ecowitt Wittboy (aka GW2001). So far, with the latest version of rtl_433, it is not detected (using a afael Micro R820T tuner) but I discovered there seemed to be a WH24 not far away.

I have seen a mention of a forthcoming ticket to support the Wittboy but apparently it has not been created yet; hence, this one.

Is there anything I could do or provide to help supporting the GW2001? Currently it doesn't seem to be even detected by rtl_433 but I wonder whether it is possible to detect the weather station and capture raw byte streams using another tool ...

Thanks for any input.

Best regards

[zuckschwerdt]

Ah ok, 433.92 M vs 868 M. Do you plan to use two receivers or are you planning to hop between both bands? Since 868 is twice 434 the usual antennas should be good for both bands.

That it works without but not with the antenna on 868 M could be overloading (but not at 10m I'd say), or that the antenna on 868 M is jammed by something else, or that without antenna the frequencies don't match well (but without you get a wider band by some aliasing).

Try to grab a sample or at least observe the band at 868 M and inspect with https://triq.org/pdv/ or upload here as zip. Use -S all to get a sample or e.g. -w myfile_868M_1024k.cu8 -T 30 (the _868M_1024k.cu8 part should match what you get as g00… sample filenames. Or just brute-force try offset frequencies, e.g. -f 868.3M

[Grisen80]

In order to grab a sample I guess I need to migrate everything to a separate linux box (running as HA addon at the moment.) Before I go down that route I will try some brute-force. But anyhow I am leaning towards returning the 868 unit and order the 433 version instead. I read above that godric got that working. Sticking to 433 would also have the benefit that I could avoid hopping.

[Grisen80]

Haha. Magic. Set the frequency to 868.1 and it worked (also worked fine w 868.2 but not 868.3). With hop_interval set to 90s and the external antenna attached, I now get sensor readings from both bands reported in the logs. Great thanks @zuckschwerdt

[zuckschwerdt]

Try also with added -M level -M noise to see where (roughly) the WS90 is sending and what the noise levels are.

[Grisen80]

Not sure how to interpret this but it hoovers around these values

[rtl_433] Humidity : 74 % [rtl_433] Wind direction: 104 [rtl_433] Wind speed: 0.0 m/s [rtl_433] Gust speed: 0.0 m/s [rtl_433] UVI : 0.0 [rtl_433] Light : 330.0 lux [rtl_433] Flags : 82 [rtl_433] Total Rain: 0.0 mm [rtl_433] Supercap Voltage: 0.4 V [rtl_433] Firmware Version: 132 [rtl_433] Extra Data: 3fff000000------0000ff8ff90000 [rtl_433] Integrity : CRC [rtl_433] Modulation: FSK [rtl_433] Freq1 : 868.3 MHz [rtl_433] Freq2 : 868.4 MHz [rtl_433] RSSI : -0.1 dB [rtl_433] SNR : 15.0 dB [rtl_433] Noise : -15.1 dB [rtl_433] [Auto Level] Current noise level -14.7 dB, estimated noise -13.3 dB [rtl_433] [Auto Level] Current noise level -14.6 dB, estimated noise -14.6 dB

and a few secs later [rtl_433] [Auto Level] Current noise level -14.6 dB, estimated noise -14.4 dB [rtl_433] [Auto Level] Current noise level -14.6 dB, estimated noise -14.6 dB

[zuckschwerdt]

The sender is around 868.3 – 868.4 M (FSK has two frequencies). If you hit one dead on it won't work (the receiver is conceptually deaf at the center), that's why -f 868.3M didn't work.

The noise level is intense, a quiet band is easily below -20 dB. But only an SNR of 9 dB is needed, you still have a slight margin left. Investigate the band with some SDR GUI if possible to look for the source of the noise. Or try to maybe edge out the noise by using e.g. 868.5 or higher.

[Grisen80]

Good input. Currently, I have the WS90 set up just for testing. Its' final location will hopefully be much more quiet. Will keep an eye on the noise level.

Again, many thanks for all your help.

[gdt]

What's the status of this issue? Is the WS90 working with today's git master? Is someone working on a PR?

[sdalu]

Looking at the code is there a reason to use mV for battery and V for supercap? why not use mV for both?

[gdt]

@sdalu I don't know but your comment is almost certainly going to be lost over time. If you would like to look at the codebase and file a bug that some particular thing should be different, or better yet a PR to regularize that would be more likely to lead to progress. The norm in rtl_433 is about battery_ok 0-1, and supercap vs battery is interesting.

But more importantly, does this work or not with today's git master? Why should the issue remain open?

[buelowp]

This works, and it works well. I expect we could close this. I may agree with the V v mV comment, but that should be a new bug/enhancement.

It's logging data for me today to Adafruit IO and the rain is very accurate. After I moved it to it's final location, it's been counting very well.

[jpochmara]

The data from the WS90 is in different units. The battery voltage is reported in 20mV units, whereas the supercap voltage is reported in 0.1V units. Converting the supercap voltage to mV would be misleading about the resolution of the data.

[sdalu]

@jpochmara that the same unit V in both case, resolution is 20mV for battery, 100mV for supercap. I fail to see why this would be misleading. Or at least if we consider there is some misleading because to use mV you need to have at least a 1mV resolution, both supercap and battery should use V in these case.

Is there a kind of general rule for consistency in rtl_433 for unit and unit prefix ?

[gdt]

The battery voltage stuff is now in #2653; it's off topic here.