We are currently undertaking EMC pre-compliance tests with a P1 based product here in Melbourne, Australia.
The Device Under Test first failed at an EMC compliance lab. We have come back to our premises and have been able to reproduce the issue using our (newly purchased) Spectrum Analyser, but not in a Faraday cage (as yet).
The emissions tests are failing when there is NO WiFi Access Point to associate with.
The attached spectrum analyser display shows the issues between Marker 1 and above Marker 2.
- Marker 1 = 2484 MHz (top edge of WiFi 2.4 GHz band)
- Marker 2 = 2502.8 MHz
I note that the P1 was in SAFE mode, ie our application firmware was not running. The spectrum analyser was connected directly to the P1’s external antenna connector.
Has anyone got comment on this, especially Particle engineers? This is a real problem for us.
We will repeat this test with a Photon module and report back shortly.
Update: when the P1 powers up, it sends out Probe Requests on different WiFi channels. So far so good.
Problem is that It seems that when there is no Access Point to associate with, it eventually sends a probe on channel 14, which is outside of Australian and American WiFi limits, but within Japanese regulations.
Can anyone comment on this, eg is it a DeviceOS config of some sort or DeviceOS issue?
Both Photon and P1 by default have their country code set to Japan. There is some historical info about this in https://github.com/particle-iot/device-os/pull/942 and https://github.com/particle-iot/device-os/pull/1014.
You may manually set the country code through DCT in order to conform:
$ echo -n "AU0" > cc.bin
$ dfu-util -d 2b04:d008 -a 1 -s 1758:leave -D cc.bin
That is great news and explains it perfectly!
I will perform the change next week and report back via this ticket.
You have really made my day…
Happy to confirm that the Channel 14 Probe Request no longer happens when the country code is set to “AU0”.
(Perhaps more to follow on the noise at 2502 MHz under a separate ticket).