USRP RIO Channel Isolation?

[rocketengineer]

I have 2 USRP RIO and 2 USRP 2922's and if I run the interactive spectrum analysis on the same channel as the channel that I'm generating the QAM signal then the spectrum will show that signal even with a 50 ohm load on each output port.    I can't find any port Isolation details in the datasheet for any of the USRP's but I can see from the USRP block diagram, that the TX and RX ports on each channel, do have an internal wire connected with an RF switch.

I also notice that if I generate a QAM signal on Channel 0 using the RIO, and terminate all the SMA ports with 50 ohm, that the RX2 port on Channel 0 and Channel 1 can see that signal.

I'm baffeled by the purpose of having all the channels and ports internally connected to each other.  

Is there a method to getting the ports to be isolated from each other?  I would think that the channels would be isolated but I don't think they are also.  

How would you generate a QAM Transmitting signal on Channel 0 and get the real external signal on the other channel, when running an RF cable from one channel SMA port to the next channel RF port and then still see the signal be shown with the cable connecting those two port and then when disconnecting the cable, the signal disapear?

[EccentRick]

Hi rocketengineer,

Which USRP-RIO model are you currently using, and which block diagram are you referring to? On the USRP-2940R, you can see in the block diagram a switch on the "TX1/RX1" port to allow it access to the transmission section of the diagram or the receiving section, depending on the mode. There shouldn't be any circuitry connecting the two channels of the USRP RIO (RF 0 and RF1), except for the FPGA that they both interface with (see http://www.ni.com/white-paper/52119/en/). 

Are you seeing the same power levels when connecting RF0 and RF1 with a cable as you do when not using a cable?

[rocketengineer]

I used the two example basic VI's that you include with the USRP.  One is the "niUSRP EX TX Continuous Async on the Fly.vi" file that is generating the signal and the other is the "niUSRP EX Spectral Monitoring (interactive).vi" file that is detecting that signal and showing the result on the magnitude spectrum.

If I go from the same device Channel 0, TX1 to Channel 0, RX2, I get about -57 (no cable) to -29 (loopback cable with 20 dB pad) but I have a 20 dB pad protecting the RX2 port and so that is showing about 28 dB of total isolation.

(28 dB)

Now if I go and do the same test from Channel 0 to Channel 1 then my signal, with no cable, starts out at -70 (no cable) to -29 (with cable), giving me about 60 dB of isolation because I have the 20 dB pad on that port also for protection.

(60 dB)

If I go to a totally different USRP RIO then my signal starts out reading a -104 (no cable) and then when a external cable is connected, I read a -29, which is showing about 95 dB of isolation.

(95 dB)

Is that the same type of isolation results that you get normally with the USRP RIO's? 

[DollaBrillz]

These readings are largely dependent on your particular application environment so it is not something that is explicitly specified. This being said, are you concerned about leakage or crosstalk? 

[rocketengineer]

Yes the cross talk makes it harder for me to determine the amount of gain a PA has when placed in series between the port generating the signal and the port receiving the signal.  I use a cable to get the readings without the PA and then connect the PA and measure the levels to get the approximate gain.

[kevins33]

Hi rocketengineer,

We do not spec the isolation for the USRP, because it is dependent on the particular environment (as mentioned above). What particular USRP RIO are you using? It seems like your method for determining the gain of the PA is a good one.