Interference issues in new office - HELP!

Go To Last Post
11 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi,
we use ZigBit modules and dev boards (2.4Ghz) since 3 years and they worked great, no issues. 4 weeks ago we moved and since then we "sometimes" have HUGE issues to transmit even a single packet. It does not always happen, only sometimes, but when it happens, it pretty much prevents us from sending/receiving anything!

Our setup:
1) 1 Coordinator, 3-4 end nodes .
2) Distance is small, about 1-5 meters, direct sight.
3) We transmit data all the time, about 100 bytes per second, battery life is not an issue for us.
4) We use stack version 1.05

The problem:
1) It happens at random times during the day, MAYBE more frequently at around 5pm, but not sure.
2) When it happens, for half an hour, hour or even longer, messages get lost or come in with 5 seconds delay or so.
3) Suddenly, the problem can go away and everything works fine again, and then come back.
4) I do find 90 (!) wireless networks nearby, which is much more then in the old place, but all of them seem to use channel 11, with a few exceptions.

What we tried to fix it:
1) Allow the modules to choose the channel. But for whatever reason the coordinator always uses channel 15, or sometimes 16, but never anything else.
2) We tried to increase the coordinators TX power (it was set to 0 and we set it to 3), but I can't say yet whether that made any difference or not.

I'm I doing something stupid or forgot something?
If anyone has a good idea what I could try,please let me know.

Many thanks for any help,
Tappo

Last Edited: Fri. Oct 16, 2015 - 02:28 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

tappo wrote:
1) Allow the modules to choose the channel. But for whatever reason the coordinator always uses channel 15, or sometimes 16, but never anything else.
15 is one of the ZigBee channels that is placed between WiFi channels, so it is likely to be quiet even if there are a lot of WiFi networks around.

tappo wrote:
I'm I doing something stupid or forgot something?
I don't think so. Recently we also see this behavior - at night some channels are just not usable anymore. I think there is WiFi access point somewhere nearby and at night it does some heavy data transfer (like backup or something). Not much we can do about it.

I suggest you to monitor LQI and RSSI of received frames and see if you can spot any patterns.

NOTE: I no longer actively read this forum. Please ask your question on www.eevblog.com/forum if you want my answer.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

It's not hard to scan channels for an rssi display, then you could comma separate to a daily log file. The 15dot4 sniffer on raven usb stick or Sam7s makes a nice graph that you can periodically dump to a movie https://www.avrfreaks.net/index.php?name=PNphpBB2&file=download&id=19917. Can't remember the name of the free screen dump software I once used to record cruise ship webcams (you draw a box and specify the frame interval, worked very well)

case 'e':
  PRINTF_P(PSTR("Energy Scan:\n"));
  {
    uint8_t i;
    uint16_t j;
    uint8_t previous_channel = rf230_get_channel();
    int8_t RSSI, maxRSSI[17];
    uint16_t accRSSI[17];
					
    bzero((void*)accRSSI,sizeof(accRSSI));
    bzero((void*)maxRSSI,sizeof(maxRSSI));
					
    for(j=0;j<(1<<12);j++) {
      for(i=11;i<=26;i++) {
        rf230_listen_channel(i);
        _delay_us(3*10);
		RSSI = rf230_rssi();
        maxRSSI[i-11]=Max(maxRSSI[i-11],RSSI);
        accRSSI[i-11]+=RSSI;
      }
      if(j&(1<<7)) {
        Led3_on();
        if(!(j&((1<<7)-1))) {
          PRINTF_P(PSTR("."));
        }
      } else
        Led3_off();
      }
      watchdog_periodic();
    }
	
    rf230_set_channel(previous_channel);
    PRINTF_P(PSTR("\n"));
    for(i=11;i<=26;i++) {
      uint8_t activity=Min(maxRSSI[i-11],accRSSI[i-11]/(1<<7));
      PRINTF_P(PSTR(" %02d: %02ddB "),i, -91+(maxRSSI[i-11]-1));
      for(;activity--;maxRSSI[i-11]--) {
        PRINTF_P(PSTR("#"));
      }
      for(;maxRSSI[i-11]>0;maxRSSI[i-11]--) {
        PRINTF_P(PSTR(":"));
      }
      PRINTF_P(PSTR("\n"));
    }
  }
  PRINTF_P(PSTR("Done.\n"));
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi guys,
many thanks for your answers.

We still can't identify the RF noise, so I have 2 questions:

1) Could it help if we use a ZigBit AMP module as the coordinator? Are AMP modules compatible with normal modules in the same network?

2) Could/should we use a directional antenna instead of an omnidirectional one to focus directly on the area we are interested in?

Many thanks for your help so far, this is a hard one to crack for us,
Tappo

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

ZigBit isn't my thing, but a couple of questions:

Does the Coordinator directly communicate with each remote unit, or do they relay info (packets) through each other to get back to the coordinator.

The point being to try to determine if the problem is originating at the Coordinator, or if the final relay node may also be the culprit, or if their is an RF noise source that trashes the entire band.

Does the system reallign itself, or are you manually rebooting the network?

Can you remove one remote node from the system for a day or two, (or three...)? Again the goal being to see if the network is healthy if a particular unit is not participating within the network?

(Not a very efficient debugging method...)

JC

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes you can use an Amp as a co-ordinator, or you can simply add an amp between the co-ordinator and the end devices as a repeater, this may help by adding more paths for the mesh to use.

I purchased a RF Spectrum anyliser off a company in TW called ELECTROSMOG model ED15SA, it gives a simple readout of each chanel on the WiFi band, and shows what is active. then you can look for the source by simply walking around and watching the meter rise and fall, I think it cost around 50 USD,
google it :)

Also, Tappo, where are you in the world? can you fill in your profile?

Thanks

_________________________________

www.proficnc.com
_________________________________
Go Aussie Go!!!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi everyone,
thanks for the replies. We have spectrum analysers we used to find the source, and it looks like it IS coming from WiFi networks, even though we use Channel 15 which is supposed to be good. 26 seems to wkr a little better.

IMPORTANT QUESTION:
We are trying to buy a ZigBit AMP development module for testing, but can't find a supplier who has them in stock here in the US. Any suggestions?

LESS IMPORTANT QUESTION:
Any suggestions on a small directional antenna we could buy to test this?

Again, our application works on short distances, between 1-6 meters, all line of sight. 1 Coordinator, and between 3-6 nodes, star network.

Many thanks,
Tappo
PS: I filled out my profile. We are in SF, California.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

tappo wrote:
Any suggestions on a small directional antenna we could buy to test this?
Texas Instruments has an antenna kit described in Antenna Selection Guide (AN058); the YAGI may work but it's not small though that kit has one that's smaller and somewhat directional. Another possibility could be this antenna for spectrum analyzers: http://www.spectrumspot.com/edu/lpa.html

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I posted a simple PCB design for using the Zigbit amp a while ago, I will try and find a link.
It works well, and just has the LDO and some caps, then 3 headers for JTAG, Analogue, and digital.

[url]
https://www.avrfreaks.net/index.p...
[/url]

try looking here

_________________________________

www.proficnc.com
_________________________________
Go Aussie Go!!!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hi,
thanks a lot for the answers and the board design. In order to blast at the full 20 dB, do I need to set something special in the bitcloud stack?

Thanks,
Tappo

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

No, if you set 0db in the power, it will be 17 db, and if you set 3db it will be 20db, the PA value is fixed in Bitcloud also remember that the Amp can recieve better as well which may help.

_________________________________

www.proficnc.com
_________________________________
Go Aussie Go!!!