An 868MHz project I am involved with uses a Xemics 1205 Tranceiver
taken from a reference design. The output of which is fed to a PCB
mounted SMA connector at the end of about 1.5" of PCB trace. The
trace was calculated to be 50Ohms but not measured although a
transmission test from the point of the reference design and the SMA
connector shows a 2db drop.

Unfortunately the whole thing is placed in a metal box with the
Antenna passing through the box onto the board. The aerial
manufacturer has told us that most energy is emmitted in the box and I
guess we also loose receiver sensitivity. Performance wise we set
the transmitter to transmit at 15db and get a an ERP of .2mw remote
devices with a PCB aerial in plastic cases have an ERP of

We decided to change the SMA connector to a bulk head mounted sma
connector, However the range now goes down from a reasonably
respectable indoor range of 25m+ (can't get further without passing
through a metal wall) to a less inspiring 4 feet. It appear that
adding 2-3" of coax with the SMA on the end is enough to do this. We
have managed to get about 20Feet if we use a Yagi aerial.

The Xemics part is a direct conversion device and produces IQ signals
to monitor. With the extra cable the signals look good with full
amplitude at 2 feet but at 4 feet either the I or Q signal vanish but
strangely not both as I believe the same input signal is mixed with
the local oscillator and a 90 degree phase shifted on to produce the
IQ signals.

Could the affects I'm seeing be a reflection of the local oscillator
off a poorly matched PCB trace or is there some other strange things I
should be looking for, apart from a good RF engineer to work it out
for me.

Many thanks for any help or pointers

Paul

Hi Paul,

About your 868 MHz project (probably european [ETSI] SRD application).

The datasheet gives impedance requirement that should function when
connected directly to 50 Ohms. I saw (little bit surprised) the
datasheet doesn't give a reference design.

If the design is to be terminated with 50 Ohms, and you run a coaxial
cable from bulkhead to PCB (both sides bread of cable connected), it
should be OK (irrespective of cable length). I assume that the antenna
that you are using is a 50 Ohms antenna (a quarter wave?). The antenna
will us the metal casing as ground plane avoiding RF currents inside
the housing.

Sometimes, circuits may become instable (out of band or just out of
band) when changing the cable length. You may check this by adding a 3
or 6 dB attenuation. Normally spoken, adding 6 dB will reduce the
range with about a factor 2 (assuming free space, not close to
ground).

Other things can be interference on the IF (directly onto the chip)
from a switching regulator or digital HW. Is it possible that you are
close to other applications (for example 2W UHF RFID, or UHF broadcast
stations). You should see that on the analog I&Q outputs, If you move
with the antenna, with no transmitter present, the output noise should
not change (so not become more or becoming less).

When this a commercial application, and have some budget, you may hire
a specialist. Many of the single chip TRX are difficult to get in
line with EN300220, though the datasheet does mention compliance.

Best Regards,

Wim
PA3DJS.

On Mar 14, 8:17 pm, "Wimpie" wrote:
Hi Paul,

About your 868 MHz project (probably european [ETSI] SRD application).

The datasheet gives impedance requirement that should function when
connected directly to 50 Ohms. I saw (little bit surprised) the
datasheet doesn't give a reference design.

If the design is to be terminated with 50 Ohms, and you run a coaxial
cable from bulkhead to PCB (both sides bread of cable connected), it
should be OK (irrespective of cable length). I assume that the antenna
that you are using is a 50 Ohms antenna (a quarter wave?). The antenna
will us the metal casing as ground plane avoiding RF currents inside
the housing.

Sometimes, circuits may become instable (out of band or just out of
band) when changing the cable length. You may check this by adding a 3
or 6 dB attenuation. Normally spoken, adding 6 dB will reduce the
range with about a factor 2 (assuming free space, not close to
ground).

Other things can be interference on the IF (directly onto the chip)
from a switching regulator or digital HW. Is it possible that you are
close to other applications (for example 2W UHF RFID, or UHF broadcast
stations). You should see that on the analog I&Q outputs, If you move
with the antenna, with no transmitter present, the output noise should
not change (so not become more or becoming less).

When this a commercial application, and have some budget, you may hire
a specialist. Many of the single chip TRX are difficult to get in
line with EN300220, though the datasheet does mention compliance.

Best Regards,

Wim
PA3DJS.

Thanks for the Info. We do use UHF and VHF systems as well however I
don't believe this is the problem here as turning of the transmitter
causes the IQ signals to go to close to DC. There is some ripple due
to ripplet from a switcher though we have reduced this to 5mA and a
large cap by the transmitter reduces this further. In addition the
switched is i a separate compartment to the transceiver.

Cutting the tracks at the tx module to the SMA connector and wiring
coax directly seems to do the job, so I guess that the PCB package did
not do a good job of calculating 50 Ohm. Mind you I think it was
probably my fault I should have copied the module and tracks were on
the inside surrounded by planes where as from the module to the SMA
connector the tracks were on the outer layers of PCB. I assume the
50Ohms would also be affected once the board was in the metal box.

Paul.