2002-10-16
I patched the driver and dvbtune program to scan through QAM values. I tuned my Hauppage to the HTV YLE 283.000MHz stream. In ves1820.c I used 36.000MHz as the IF frequency instead of 36.125MHz, so there is a -0.125MHz offset implicitely used there. With the CVS IF freq I would have had to tune to 282.875. Are you confused enough yet? :-)
My test patches are at http://www.iki.fi/hyvatti/dvb/ .
Modify dvbtune.c accordingly when testing different registers. The command line I used was:
./dvbtune -f 283000000 -s 5900 -qam 128 -I 0 -v 512 -a 650 ./dvbtune -m
I started going through the
Reg1: I found the useful values to be 63..118. I did not do ves1820 reset when changing this value, so the results may be unreliable. The average is 90, so I used that one in the following tests.
Reg9: Now I used FE_RESET after changing QAM values. I found that values between 206 and 131 lock reliably. Centre value=168, so I used that one in the following tests.
Reg5: Values between 61..36 are reliable. About 48 is the centre value, so I used that one.
Reg8: Values between 57..34 are reliable. 46 is the centre value, so I used that one.
Reg1 again, now with FE reset: reliable between 93..74, centre value: 84
Reg9 again: 110..199, centre value = about 154.
So without further iterations I found the values:
{ QAM_128, 128, 84, 48, 46, 154 },
Now changing to QAM_128 channels seems more reliable than ever. These values seem to depend on tuning, changing tuning will probably result in different values and different values result in different optimum tuning.
Maybe we could make a scanner program that goes through these registers in iterative manner? After we distribute this program to developers around we could get some statistics..
I will not be able to further test this for about a week, but good luck to others..
Regards,
Jaakko