My -- limited -- experience shows that the telos/micaz radio chips are quite a bit more reliable vis the mica2 900Mhz version. I have not done a 433Mhz comparison though. You might be having multi-path or antenna placement issues, as these would be frequency dependent. As Angelo suggests, comparing different device pairs in slightly different positions might elucidate.
The tweaks you have done so far are exactly what? Tuning each freq? MS Paul Johnson wrote: > Hi, > > My apologies in advance if this question has already been answered in > the archives. I've searched for anything similar but never came up with > any relevant hits. > > I am experiencing unusual packetloss with the Mica2 (CC1000) platform. > Initially, I saw approximately a 20% loss. After doing some digging, I > realized that simply setting the operation frequency to 900 MHz does not > modify the other hardware configuration presets. After tuning the radio > using one of the 900MHz presets, i was able to improve that figure to > somewhere around 3% packet loss, but any other tweaks seem ineffective > at removing these last few percentage points. > > The environment in which I am performing the tests is very controlled, > and I have verified with a cheap commercial spectrum analyzer > (Metageek's WiSpy for 900MHz) that there is no external interference > that could be causing the packet loss. To the best of my ability I have > determined that for some reason or another, the CC1000 Radio is not > always picking up the preamble of the transmitting node. (I have > checked to see if the packets just fail CRC by modifying the SNOOP > implementation, but the CRC check is only rarely the cause of packet > loss. It seems as if the receiving node never locks on and decodes the > packet.) > > I have tested my application on MICA2's 433 MHz nodes, 900MHz nodes, and > the TelosB 2.4 GHz nodes and I have only experienced this problem with > the MICA2 900 MHz nodes(The 433 MHz, and 2.4 GHz nodes experienced 0 > packetloss during the test operation). > > By looking at the CC1000 datasheet, i've been unable to figure out what > setting would need to be tweaked to correct this issue. So far, the > best candidates i've found are the PEAK_LEVEL_OFFSET in the MODEM2 > register(Derivation of this value is dependent on frequency, however i > wasn't able to determine if the preset value matches the derived value), > and the MODEM1 register which deals with the average filter and > locking/settling time(In the CC1000 Documentation it says "If the data > is Manchester coded, there is no need to lock the averaging filter..." > however it appears that it is being locked in the radio presets). I > have tried tweaking these parameters to no great effect. > > Has anyone experienced similar problems before, or does anyone have any > ideas as to what is causing the packetloss issue in only the 900 MHz? > > Thanks, > -Paul Johnson > _______________________________________________ > Tinyos-help mailing list > [email protected] > https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help _______________________________________________ Tinyos-help mailing list [email protected] https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
