Continuing the discussion from Photon losing Wifi:
I just got some test code running on one of my Photon devices. Previously I had problems with setting up a device. I thought it was something in software. What I learned was disheartening. I am not sure if anybody is studying this problem, but I have an access point across the room. Specifically, it is an Apple Time Machine with WiFi. I loaded the WICED “hello world” program on my Photon. It is called make snip.scan-BCM9WCDUSI09. I am sure Particle developers know it well.
From that program, using Broadcom’s BCM943362WCD4_EVB I see:
WICED Board Snip Scan:
Waiting for scan results…
Type BSSID RSSI Rate Chan Security SSID
0 Infra E8:8D:28:5F:3C:35 -41 216.7 1 WPA2 AES PSK M2MD One
1 Infra 16:6F:3F:0E:71:80 -65 130.0 6 WPA AES PSK M2MD Two
2 Infra 10:6F:3F:0E:71:80 -65 117.0 6 WPA2 Mixed PSK M2MD Two Guest
3 Infra 28:CF:E9:81:94:B8 -69 144.4 6 WPA2 AES PSK M2MD One
4 Infra 4C:E6:76:AE:1B:50 -74 130.0 11 WPA2 Mixed PSK M2MD Three
5 Infra 52:E6:76:AE:1B:50 -70 130.0 11 WPA AES PSK M2MD Three Guest
Scan complete in 1213 milliseconds
From my Photon board in the exact same spot (actually moved and oriented for best signal):
Starting WICED v3.3.1
Platform BCM9WCDUSI09 initialised
Started ThreadX v5.6
Initialising NetX_Duo v5.7_sp2
Creating Packet pools
WWD SDIO interface initialised
WLAN MAC Address : 6C:0B:84:59:35:E0
WLAN Firmware : wl0: Nov 7 2014 16:03:45 version 22.214.171.124 FWID 01-82c3f110
Waiting for scan results…
Type BSSID RSSI Rate Chan Security SSID
0 Infra E8:8D:28:5F:3C:35 -77 216.7 1 WPA2 AES PSK M2MD One
The device Photon device sitting on my desk is intermittent. It will not stay attached to the network. The access point is 15 feet away. One observation is that Broadcom’s PCB antennas are significantly better than the chip antenna. I am wondering if the external antenna restores coverage. Has anybody else done any tests on the RF characteristics and noticed range problems. My original intent was to use this module as the go to market solution, at least until the product proved itself. At this point, I’m not so sure, unless I can find a satisfactory external antenna solution.
One potential issue we saw was that the default antenna behavior (auto) on the photon can sometimes switch rapidly between both antennas. Setting your firmware to use internal only might see better results. Also catching networks during a scan can be hit or miss / timing related (as far as I know), so I would recommend a more comprehensive test.
I am still struggling with RF sensitivity and I am about to pull out another of my Photons to see if it is equally as bad as the first. Before I do that I’d like to see if you have any other comments on my predicament. Certainly someone at Photon has run the snip.scan WICED demo. I have attached an external antenna on the Photon and I have selected either the internal or external or auto without success. I have inserted the antenna selection code:
wwd_wifi_select_antenna( (wiced_antenna_t) value ) into the initialization code for (after wiced_init) to select the external antenna (“WICED_ANTENNA_2”). This appears to be the correct select antenna port because when I remove the external antenna, my Photon does not detect any access points. When I select “WICED_ANTENNA_1”, the typical value for the access point across the room is -71. With the external antenna, I see signals from the other access points in the vicinity, but the best I see is -67. I’m expecting something in the -43 to -44 range of the WICED EVB.
Is the RF sensitivity something that Photon tests on the assembly line? Are you typically seeing values in the -40 range for devices in your offices or is this a problem related to the USI module? My external antennas are similar to those that you sell (but not the exact one) and the cable length is about 6 inches.
This topic has been one that has plagued this Photon since the beginning. I first noted the difficulty attaching to my network immediately after receiving the Photon. (previous post) I noted that when I moved the Photon next to my access point it would attach just fine. I really want to use the Photon in my production solution because the value of FCC type acceptance and an easy path to market make sense for my product introduction, but I’m worried that either there is a problem specific to the USI module.
Have you got a router which will tell you at what speed each node is connected to the wifi point ?.
Because my photon
s which are both outside and upstairs both show about 68 - 75 rssi mark ( inbuilt ax ) But are connected to the router at different link speeds , is that your problem ? you dont see a different rssi value , just a different link speed ?.
I see values on the Scan program that show 216.7 for one router, 117 for the other on the Photon. When I use the WICED EVB, I see 216.7, 144.4, 117 and 130 for each of my different access points. I would expect to see something very similar given the the program is exactly the same on both hardware implementations.
@Chuck The specific issue we had in a previous FW version was that when the antenna is in ANT_AUTO mode, this tells the broadcom radio to toggle the RF Switch controls back and fourth rapidly between internal and external antennas looking for the best signal. It is not supposed to stop on either connection, but just keep switching. What was observed is that it would sometimes actually stop on the EXTERNAL antenna (u.FL) while trying to connect to the WAP which for most people do not have an antenna connected to it, so after a timeout in code it would reset things and try again… sometimes repeatedly.
We now default to the ANT_INTERNAL antenna if no WiFi.selectAntenna() calls have been made that would persist a user choice (even if WiFi.selectAntenna() is not called in future applications).
By setting the antenna to your preferred choice, you can ensure the best signal.
Please give that a try and see if you see improvement.
Resolved my problem. I pulled out another Photon from my stash and it appears to work correctly. I guess that the first one I received from Spark/Particle.io was one of the early ones before testing was well understood. I have three from an early batch and two from a recent batch. I don’t know which group the second Photon came from, but hopefully Particle is actually testing the RF sensitivity before they ship them.
We do! We test to make sure the wifi RSSI is at or better than a prescribed value, otherwise we fail the board.
I know this is an old topic but after spending all day yesterday I think this is relevant.
I’m installing Photons throughout my house (see here: https://medium.com/home-wireless) and upstairs I figured this was too far away from my router so I hooked up a router I have around (I’ve got a few). I was getting pretty bad RSSI figures - in the middle -70s, (my software sends it to the server) but they were all working so whatever.
Then two nights ago one of them randomly crashed, I thought, and rather than take it out of the wall and reset it I decided to test a full power-off power-on cycle for all of them - they’re on a common DC line. Well that didn’t go well at all. Apparently my (cheap) current-limited supply dips for a second when the load is applied and half of them didn’t boot. I’ll leave out the various issues that implied (including increasing reset times with a cap) but one photon in particular just never came up.
I thought it had a public key problem so removed it, reset the public key, reinstalled it… nothing. Finally I stuck a router in the same room so it could connect (which it did) but it was showing -71dB on a router 4 feet away! Wow. I brought a different Photon upstairs and right next to that one it was showing -59dB.
a) replaced the photon upstairs with the second one and it’s connecting to the farther router (which I couldn’t even talk to) at -65dB.
b) carried the ‘bad’ photon down to my office and it’s connecting to my network at -83dB whereas commonly they connect at -53dB.
I guess this is just a ping to ask you to please more carefully test wireless performance. I looked at the schematic and layout for the Chip antenna and personally it’s a little strange (I used to be manager/lead architect for an RF EDA software product at Agilent). This was just a quick peek but I wonder if you’d like me to pass it along to a good RF consultant I know who might be willing to take a longer peek at it, or did you do a thorough RF analysis of the front-end?
Thanks for the feedback! It sounds like you have a photon that’s experiencing slightly worse RF performance than your other photons. Is it possible that you have different versions / models of photon? We’ve made RF improvements over different versions of the Photon ( you can see some notes here - https://github.com/spark/photon ).
Although we do lots of testing (both for certifications and for our designs in general), we always welcome helpful / constructive feedback, and I’d be happy to share any RF notes you might have with our hardware team.