[linux-dvb] Lost lock problems over extended periods of time

Andrew de Quincey adq_dvb at lidskialf.net
Thu Jun 9 01:08:25 CEST 2005

On Wednesday 08 June 2005 23:04, Andrew de Quincey wrote:
> > Incidentally, my latest test was to set the tone every 60 seconds in the
> > frontend loop. This didn't work though - it still lost the lock, and went
> > into the patch I posted earlier and regained it simply by setting the
> > tone.
> My mistake - there was a bug in this code which meant it wasn't actually
> doing the above. I've corrected it and am waiting to see if it still loses
> the lock.

Nope - its just lost it and ran the auto-regain-by-setting-tone code.

So to summarise my current findings:
*) Setting the tone every 60 seconds does not help.
*) Setting the tone when the lock is lost causes it to be regained in under 
*) If you don't set the tone/reg0x08 on lock loss, you will never regain the 
*) The contents of register 0x08 of the stv0299 stay constant.
*) It cannot be noise/traffic on the i2c bus changing other registers since 
the only thing needed to fix the issue is to set register 0x08 again.
*) No i2c traffic to the stv0299 besides reading the lock status occurs after 
the initial tuning attempt.
*) Only a few cards are affected. Cards tuned to the same channels on other 
sites are fine. Cards using the same multiswitch at the same site are fine.
*) At a site with the issue, channels can be manually reassigned between 
cards/machines until a stable situation is found. This is very time consuming 
*) Temperature is unlikely to be a cause.
*) This problem occurs across a whole range of stv0299 based cards (we have 
several vintages of them from the last few years).
*) There is no pattern temporally to the lock loss.
*) The problem occurs across different multiswitches by different 
*) The signal is perfect (BER==0) up until the time the lock is lost.
*) The problem _may_ affect stv0299 based Twinhan cards as well, but this 
needs verified.

This is sounding awfully like the tone being generated is drifting or stopping 
completely - due to an unknown cause. Unfortunately I do not currently have 
the means to measure the tone output during one of these incidents.

More information about the linux-dvb mailing list