[linux-dvb] DIBcom based dvb-t scanning problems

Patrick Boettcher patrick.boettcher at desy.de
Sat Aug 20 13:22:12 CEST 2005


Hi,
On Sat, 20 Aug 2005, Christophe Thommeret wrote:
> DiBcom based dvb-t owners (DiBcom 3000P/M-C DVB-T and DiBcom 3000M-B DVB-T)
> have reported that Kaffeine scan failed.
>
> Using DVB card "DiBcom 3000P/M-C DVB-T"
> tuning DVB-T to 198500000 Hz
> inv:2 bw:1 fecH:3 fecL:0 mod:1 tm:1 gi:3 hier:0
> polling....
> Getting frontend event
> polling....
> Getting frontend event
> Event:  Frequency: 198500000
> Transponders : 1/6
> Invalid section length or timeout : pid=16
> Invalid section length or timeout : pid=17
> Reading PAT : pid=0
> ( and so on for all mplex )
>
> As you can see, it tunes well.
> But NIT and SDT can't be read (and sometimes even PAT failed).

This is because of the PID filter and the USB buffer when using the USB1.1 
and the USB2.0 boxes on a USB1.1 port. In this case the PID is enabled and 
the buffer (4KB) is filling slowly, when only such small section PIDs are 
requested. It returns when it is full.

> What's really strange is that if they first zap to a channel (created by
> hand), and then scan that mplex (like dvbscan -c), it works !

This is because streaming is in progress (a video PID is fed) and so the 
buffers are filled much fast.

> In addition, dvbscan works fine (with or without -c)! But i can't find 
> what's wrong with Kaffeine.

dvbscan has maybe a longer timeout than kaffeine.

The solution would be to lower the size of the buffers for USB1.1
operation, but for some reason the box and/or usb bus doesn't like smaller 
buffers.

The solution is easy: use an USB2.0 device on an USB2.0 port. ;)

best regards,
Patrick.

--
   Mail: patrick.boettcher at desy.de
   WWW:  http://www.wi-bw.tfh-wildau.de/~pboettch/


More information about the linux-dvb mailing list