[linux-dvb] Some thoughts and questions

Manu Abraham abraham.manu at gmail.com
Sat Sep 29 16:39:06 CEST 2007


Hi,

Wolfgang Wegner wrote:
> Hi,
> 
> On Sat, Sep 29, 2007 at 04:18:31PM +0400, Manu Abraham wrote:
>> Johannes Stezenbach wrote:
>>>> 1) LNB drift
>>> - LNBs have a constant error plus a temperature drift
>>>   (e.g. +/-1MHz error, +/-3Mhz drift for a temperature range
>>>   of -40 ... +60 °C -- cheap no name equipment usually worse)
>> This is the old LNB, the one's we use are generally based on PLL's have some 15 - 20k drift
> 
> hmm.... I am not sure how old our LNBs (at work) are, but there and with
> my LNBs at home (around 2 years old) I have measured offsets of around
> 1-2 MHz.
> 
>>> - sw zig-zag is by no means stv0299 specific and is used by
>>>   (almost?) all DVB-S demod drivers
>>
>> hmm, i didn't mean swzigzag, but as you see from that discussion, 
>> it was the drift that i am looking at
> 
> Is it really a difference? I am not sure if I understand it correctly,
> but IMHO this is or could be handled as the same issue.
> 
> For low symbol rates, the problem is that you can not get the frontend
> locked to the transponder if the offset (probably being caused by drift
> or simply constant frequency offset because of the cheap crystal in
> frontend PLL and/or LNB) is too high. You can only overcome this by
> zig-zag-scan.
> 
> 
>> My point being to have zigzag specific to the demod, since each device of the 
>> devices which implements zigzag does it in a different way.
>> The computation being different
> 
> Why is the computation different? I did not look into it yet, but at
> the moment I can only see stepsize and transponder bandwidth (symbol rate
> in case of DVB-S) as the parameters the algorithm depends on.
> 

A quick generalized comparison. (looked at 2 devices from 2 vendors)

devices from STM does Fs / 2 zigzag, where Fs is variable.
Intel uses Foff = (err1 * mclk / 65536 + err2 * srate / 262144)

Although point to be noted is that dvb_frontend is not doing Fs/2 zigzag either. It does
"some" zigzag.

> [...]
>>> IIRC Andrew de Quincey spent significant time optimizing the
>>> zig-zag code and the parameters for various frontends.
>>>
>> I do remember the time where he spent so much time optimizing the
>>  swzigzag for the STV0299.
> 
> It seems my thinking is too easy above, I will have to look into the
> code to get enlightenment, I hope. :-)
> 
>> You have any cases of broadcasters doing it in a non-standard way ?
>> ie inverted transmission ?
> 
> I do not remember the transponders, but there are some - not on
> Astra/Hotbird, though. Maybe I can give an example on monday.
> 

Thanks,
Manu




More information about the linux-dvb mailing list