Poor Performance

[cody]

thanks, i'm looking into it. maybe, it has something to do with frequency re-centering, which is off in current 8921 drivers. i don't know what demodulator is inside FireDTV-S2, but i guess it does frequency re-centering by default. in any case if frequency re-centering is on then the measured frequency offset is not the real frequency offset measured as offset from the frequency sent as part of the tune request, which then unfortunately doesn't rule out frequency offset in your environment. anyway, having the ability to test on the same frequencies as you did should help me to understand what's going on in reality.

[Derrick]

in any case if frequency re-centering is on then the measured frequency offset is not the real frequency offset measured as offset from the frequency sent as part of the tune request, which then unfortunately doesn't rule out frequency offset in your environment.

I don't understand what you're saying, but you mentioned again my environment. I can tell you that 4 out of 5 of my receivers get it right. You can guess which one doesn't Suppose you buy a new car radio. On the crowded FM-band you want to listen to your favorite station, but every time you tune to the channel a station next to it either below or above comes up.

Dunno what you mean by re-centering. This topic is about narrow band carrier acquisition. A tuner tries to find a carrier by sweeping a frequency search range (depending on signal bandwidth = f(SR)) around the entered frequency. If energy is detected in the middle of the range the tuner should examine this signal and eventually stop there but not on another signal at the edge. I don't know the algorithm used by TBS for carrier acquisition, but it doesn't work correctly!

[cody]

ok, let me, try to clarify what i mean by 're-centering': i mean that your hardware not just scan the auto-frequency acquisition range and estimate offset from there, but changes the point ('center') from which it estimates the offset and i will even try to give a rough example to make it more clear: tune to X with +/- 2 MHz auto-frequency acquisition range then without re-centering X-2, X-1, X, X+1, X+2 are tried, with re-centering any of those 5 possibilities will be tried with another additional +/- 2 MHz auto-frequency acquisition range, i.e. changing the frequency center, which can effectively mean that no matter you acquired lock it's on frequency far outside the +/- 2 MHz search range. so, that's why i think your other hardware could use 're-centering' and test with such hardware without some precise debug information from it similar to the one i added to 8921 driver doesn't give any clear information and rule out that there is no frequency offset added by your environment, i.e. 're-centering' could masks/hides the real problem and the source of the frequency offset. another way to rule out that is measure your environment with spectral analyzer for estimating if there is or there isn't and what exactly is the figure of frequency offset in your environment.

@ all

is there someone else having something similar with 8921, i.e. need to change the frequency to acquire lock?

thanks,
cody

[Derrick]

@cody, why don't you simply admit that your algorithm is badly designed?

There are 3 carriers with identical modulation parameters within the auto-frequency acquisition range. Your algo thinks that the wanted carrier is found on X-2 or X+2 and stops there without evaluating the correct one in the center. The so called re-centering method as described by you will only lengthen the already very slow search but won't fix the problem.

so, that's why i think your other hardware could use 're-centering' and test with such hardware without some precise debug information from it similar to the one i added to 8921 driver doesn't give any clear information and rule out that there is no frequency offset added by your environment, i.e. 're-centering' could masks/hides the real problem and the source of the frequency offset. another way to rule out that is measure your environment with spectral analyzer for estimating if there is or there isn't and what exactly is the figure of frequency offset in your environment.

Again you're trying to blame my environment. I've already showed that there is nothing wrong. A single carrier locks on the supposed frequency with only very small offset. You were talking about generators and analyzers that you can use. You've also said that you can receive 10E and will check these frequencies. Please do and stop talking nonsense

[Derrick]

Hello, is anybody there?

[steven]

Hello, is anybody there?

Hi Derrick
Cody because some of personal reason will be duty off some day,when he come back ,i will mention him about this
thanks

[enb141]

I'm having the same issue with low FECs on Hispasat 30.0 W, I hope this can be fixed.

[Derrick]

..any new development concerning the issues with the 8921? If not I'll dump it

[Derrick]

When can we expect a driver/firmware that works according to the specs published by TBS?