Mac Michaels wrote: > On Wednesday 12 July 2006 23:50, Trent Piepho wrote: >> On Wed, 12 Jul 2006, Mac Michaels wrote: >>> The original code written for analog mode was designed >>> to tune quickly by increasing the pump current then >>> switch back to the more stable current value after the >>> signal was acquired. This makes switching channels that >>> are far apart >> I think there is some confusion here. It sounds like you >> are talking about this code in tuner-simple.c, >> default_set_tv_freq(): case TUNER_MICROTUNE_4042FI5: >> /* Set the charge pump for fast tuning */ >> config |= TUNER_CHARGE_PUMP; >> >> And then there is more code a bit later in the same >> function to poll the PLL and turn the charge pump to low >> current after it locks. > > You are correct. I was thinking about the code above when I > replied to the list. My comments do not apply to the > original discussion. > > --Mac > >> ** This is totally different code that was being >> discussed. ** >> >> What was being discussed was code in tuner-core.c, >> set_type(). This code changes the "Aux Byte" that >> effects the AGC take-over-point. Totally different than >> the charge pump current. >> >> The set_type() function has never had an existing PLL >> message to modify. It's called on module load to set the >> tuner type, not to actually tune to something, so having >> a message wouldn't even make sense. What would the >> message be? >> >> There is very very similar code in the dvb-pll version >> that does the same thing, except it is called via the >> function that does tuning. It works correctly because it >> takes the tuning command and modifies it to set the aux >> byte. It does this on every tune, not just on module >> load. >> >> It's clear what happened. The working dvb code was >> cut&pasted into the v4l tuner driver, but put in the >> wrong function, and not adjusted to work properly in this >> new spot (thus the use of uninitialized data). >> >> So, there is some code in the v4l tuner driver that >> writes a few random bytes to the PLL when the module >> loads, which doesn't actually do much of anything to the >> PLL. Just remove it. Put something to set the aux byte >> on every tune call, like the dvb code does, if you want >> to be totally correct. But it's not necessary, as the >> tuner has been working fine all this time without it. I don't have a lot of time to type out a long email explanation right now, but I think the patch speaks for itself. This is what I did to rectify the situation, in the case of the LG-H06xF: http://linuxtv.org/hg/~mkrufky/v4l-dvb?cmd=changeset;node=b67541b1a123 Trent and I were talking about this yesterday, and he had some interesting ideas about applying this logic to the Thomson DTT761X tuner, and possibly some others. I would like to have this patch merged first, and if we move forward with Trent's idea, we would probably end up changing this again. Any comments? -- Michael Krufky _______________________________________________ linux-dvb@xxxxxxxxxxx http://www.linuxtv.org/cgi-bin/mailman/listinfo/linux-dvb
