I'm trying to make a relay with a very long time division (rx for X seconds, tx for X seconds, repeat).
Attached is my block diagram implementation. It's a bit cheeky, but it works (for just raw signal, no demodulation). Anyway, the flow graph's behavior works on my laptop, but my goal is to have it on a smaller embedded system (an Odroid XU4 running lubuntu 14.04, which actually has a more badass processor than my laptop).
(Additional info: Ettus B200, running over usb 3 port, XU4 uP: 8 core[4 @ 4GHz, 4 ])
Moving the same flow graph to my odroid stops working after ~10 rx/tx switches. the tx/rx LEDs on the board stop lighting and there's no console error thrown, so it's a bit hard to track down.
Isolating just receive (replaced usrp sink with null source) works with expected behavior, but Isolating the transmit (replaced usrp source with noise source) throws the error:
Got a ctrl packet with unknown SID....
which leads me to believe that muting and un-muting the transmit stream has some un-intended software implications.
Currently my only thought is to re-write my own valve block to replace the deprecated valve, but I was wondering if any of the experts here on the mailing list could enlighten me to what under-the-hood stuff I'm breaking.
Thanks and, as always, any help would be much appreciated!
University of Michigan '17 | B.S.E. Computer Engineering
Re: Implications of continuously un/muting tx stream
confirming this. NEVER use Throttle with a hardware block –
that's why GRC will actually warn you not to do that if you add a
Throttle block to a flow graph that contains a block that's marked
as hardware block.
On 13.05.2017 13:19, Murray Thomson
Have you tried without the throttle block? With
that source I don't think you need one.