mistakes on MOD-ZIGBEE and MOD-ZIGBEE-UEXT boards.

[raetos]

Hi to all,

looking the MOD-ZIGBEE and MOD-ZIGBEE-UEXT schematics we discovered that there are at least two mistakes refering to the Microchip reference design http://ww1.microchip.com/downloads/en/DeviceDoc/70329b.pdf on page 8.
First of all,  if you look at the balun formed by components L1,L3,C2,C14, it is clear that L1 is inverted with L3 and vice versa.
Another minor issue is on the capacitors c19 and c18 that are 27pF instead on reference design are 18pF.

As result, when we try to connect the two modules, we can reach no more than 2-3 meters of range.

We would like to ask to the Olimex staff how can we overcome these issues?

Thanks in advance.

[LubOlimex]

Hello raetos,

Thanks for the feedback! Indeed, I noticed your report in another thread and investigated. Clearly L1 and L3 are inverted which is a mistake in our design. The Microchip documentation on the subject hasn't changed since 2008, so it must be our error. Whether this unfortunate swap causes problems to the communication is another issue which we have to investigate further. We are continuing to test.

The capacitors on the quartz might cause disturbance but quartz crystals have different additional parameters, each manufacturer of quartz crystals recommends the capacitor values. Still if reverting the two inductors doesn't bring much change we will try with different capacitors.

Note that it is very likely that we won't be able to reach the transmition distance of Microchip's RF modules since they usually use tighter PCB traces in their manufacturing.

I will keep you informed about the outcome. You can try swapping L1 and L3 if you have a person talented enough with the soldering iron.

Best regards,
Lub/OLIMEX

[rjp]

If it turns out that 2-3m is the limit of communication, can we get a refund or swap on the mismanufactured boards?

[olimex]

all boards we test make at least 40-60 meters even with the swapped inductors, with inductors on correct places this rise to 60-80 meters
to have just 2-3 meters I guess there is lot of noise in the 2.4Ghz band

[raetos]

Hi to all,

I have to say, for sake of clarity, the following things:

1) Our first test was conducted in a room, so in a space closed and quite small. In our test there were 1 coordinator and 1 end device. I don't think that the short range was due to the noise because, the coordinator chosen the channel after an energy scan. Infact the less noisy channel was chosen (energy level was near to 0).  Instead, I think it was related to the environment, antennas orientation, and for sure, to the mentioned component issue (L1, L3 are swapped).

2) I made others tests in an open space, I reached  a range of about 30-40 meters.


In addition, I think the issue with the swapped inductors could be,  dangerous also for the mrf24j40, because there is impedance mismatching between its output stage and the antenna (matching circuitry doesn't work properly).

[olimex]

the reason we didn't catch this earlier is because the range is too big to catch, in our test setup the devices under test are at 15 meters from repeater and the tests we did with all devices we had in stock show min 40-60 meters range even, so all devices passed the test even with these swapped inductors
all devices in our stock now are fixed and the schematics and users manual also

[raetos]

This isn't just a problem related to the distance covered, but also of reliability. I repeat, again, the output stage of MRF24J40 is not matched with the antenna.That could bring to the break down the transceiver.

You said that for the new module the issue is fixed, but for the modules that are already sold?

[olimex]

I do not want to argue, but how big the mismatch is and how did you measured it so you assume this will cause reliability issues?
MRF IC do not get even hot with 1 degree above the ambient temperature i.e. not working in overloaded mode, how it will get broken then?
did you ever saw such module to fail?
the mismatch is minimal otherwise the module will not tansmit/receive with only 20 meters of difference toward microchip design