Hi,

(1) The stock RFX-type board does have inadequate uplink/downlink
isolation for full duplex cellular, but this can be improved greatly
by removing the C202 capacitor.

I can also confirm that even without this mod, it works fine with a single RFX for desktop testing provided some code modification.
The performance will not be optimal but you can cover a big room without issues.
 
(2) With the C202 capacitor removed, it should be possible to run
OpenBTS on a single RFX-type board without crosstalk creating a
performance limitation.  The current OpenBTS transceiver does not
support this, but that is probably due to suboptimal gain settings in
the transceiver code.  It has not (yet) been a priority for Kestrel
to correct this problem, but the OpenBTS transceiver really is open
source and anyone who thinks it is really important to fix this has
been welcome to do so all along.

I posted a patch to do that on the mailing list if anyone wants to try it.
You can find it in the list archives :

http://sourceforge.net/mailarchive/attachment.php?list_name=openbts-discuss&message_id=866320f70910151107w44c1395ew2213b6e73579b0e3%40mail.gmail.com&counter=1

I'll try and make the tranceiver configurable to support both config and/or support dbsrx as a RX board as well.
It should even be possible to 'remerge' back 52M and 64M Tranceivers into a single code base with just an init test or #if #endif

    Sylvain