Re: [Discuss-gnuradio] Reconnecting OFDM rx

[Julius Durst]

Hi,

I had a peek inside the S&C block, it's a hier block, so I simply 
rebuilt it with the single blocks it composed of. Aaaaand it works. But 
I did not see anything wrong in the hier block. Maybe it could be 
because of buffers, which get flushed or some do and some do not or 
something like that? Could it be a bug?

So that was the "solution" for the first problem. Now for the second 
one: After reconnecting, every detection lead to the

INFO: Detected an invalid packet at item 192
INFO: Parser returned #f
INFO: Detected an invalid packet at item 240
INFO: Parser returned #f
  ... and so on

error. I figured out that this is a problem of the header/payload demux, 
which is stateful. Because it got no new items/samples when 
disconnected, after reconnection it took the new items as part of the 
old paket, so all in all the wrong header for the wrong payload. I 
solved this by feeding the rx with a null source while it is 
disconnected. I'm still not sure if this could avoided and if it makes 
sense if the HPDemux looked for a new trigger, even if the old paket was 
not complete. But for now I won't go deeper into this block.

So to make it easier than rebuilding the S&C block from its single 
components, is it possible to let gnuradio clear all buffers, even the 
ones inside a hier block?

Or any other ideas what could be the problem with the S&C block?

For anyone interrested or if you want to try the difference between S&C 
and "self made" S&C: http://pastebin.com/r88MKGEv

BR
Julius


On 14.07.2015 19:17, Martin Braun wrote:
> On 14.07.2015 10:03, Julius Durst wrote:
> > Hi Martin,
> >
> > I don't exactly understand what you mean. I connected the Tag Debug to
> > both outputs of Schmidl&Cox, but that does not seem to add any tags even
> > when running in the "normal" mode. I don't think it is supposed to as it
> > doesn't get any keys.
> >
> > When disconnecting the rx, the detect output of S&C stops (no new
> > samples come out of it). When reconnecting the rx, I get a flow of zeros
> > from the detect output (no more detections). This is where I'm really
> > wondering if it is doing what it is supposed to.
>
> Yeah, I meant 1s on the detect output, not tags. But I see you got what
> I meant.
>
> And this is what I meant; the S&C is pretty robust and doesn't care
> about states (it does have some memory, though, for the averaging). It
> shouldn't care less if something was disconnected or not.
> Anyway, if only zeros are coming from the detect output, that's the
> issue right there. Can you have a peek inside that block, see what's
> going on?
>
> Cheers,
> Martin