Re: [Discuss-gnuradio] GMSK2 with BasicRX and BasicTX

[Eric Blossom]

On Tue, Feb 28, 2006 at 04:41:26PM -0700, Brian Johnson wrote:
> Hi,
> 
> What sort of problems can I expect if I use the GMSK2 examples with the
> Basic RX/TX boards?

You mean problems like "it doesn't work" ???

The basic problem is that the transmitter will be blasting all the
time, making reception *really* difficult.

> The information I have found in this discussion group suggests that the
> GMSK2 examples were developed only for the Flex 400 daughter boards.  Since
> we currently only have the Basic RX and Basic TX boards I tried it with them
> anyway.  

We've got an idea on how to fix this by ramping the transmitted DAC
value down to 0 when there's no data in the FPGA Tx fifo.  If you'd like to
solve this problem, we'd appreciate a patch.

> I modified the example to use the new carrier sense MAC code, but still had
> a lot of packets lost on the ping.  After digging into it a little I found
> that the packets were being lost on the receive side, and with a little
> experimentation found it could be corrected by putting a "time_sleep(0.1)"
> before "self.fg.send_pkt" in the MAC code  (need some time to complete the
> os.read before sending the packet?)

When I tried the carrier sense MAC, if found that my test for "ok to
send" was not sufficient.  I believe that problem has to do with my
(insufficient) kludge to check for carrier.  I suspect that it would
work better if we really checked the level, instead of checking the
correlator state.  The other issue could be excess latency in the pipeline.

> With this I am now getting < 5% packet loss on the ping between 2 USRPs and
> can successfully transfer files between the USRPs using ftp.

That's great!  Glad to hear it!

> I realize that these examples were written to utilize the auto
> transmit/receive feature that is only in the Flex 400, but other than that
> are there any problems with using these boards for GMSK that I just haven't
> run into yet?  I do get significantly more packet loss if I ping each USRP
> from the other one at the same time - would that be corrected with the auto
> transmit/receive feature?

I think that cleaning up the actual carrier sensing would help, as
would turning off the transmitter (or at least generating zero's out
the DAC which would be close enough to "off" to fix the problem).

> Brian

Again, glad to hear of your success!

Eric