This is the mail archive of the cygwin-talk mailing list for the cygwin project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

RE: SNR calculations (was Re: setup.exe missing from FTP site)


On 26 May 2007 22:47, Brian Dessent wrote:

> Linda Walsh wrote:
> 
>>>   I prefer to express that as an SNR of -18dB :-)
>> ---
>>      Could you please show your work!
>> 
>>      I arrived at -9dB. Work:
>>          For each doubling/halving of power one adjusts by ~(+/-)3dB.
>>              (note: "3" is approximation of 10*log(2) or,
>>                     to 7 decimal places, 3.010900)
>>                 so 1 unit / 2 units of signal would be -3dB,
>>                    1 unit / 4 units: -6dB,
>>                    1 unit / 8 units: -9dB
>> 
>>      From my limited understanding, -18dB would be a reduction of 6
>> factors of 2, or a 1:64 signal/noise ratio.
> 
> Typically you use dB to compare the amount of power in a signal relative
> to a fixed reference; in the case of electronics a common reference is
> 1mV, or 0 dBmV. Since power in a signal is proportional to V^2, when you
> compute dBmV from a voltage level it's 20*log(v) due to the power of 2.
> 
> But when you just want to use dB to represent a unitless ratio, then I
> agree with you it's simply 10*log(ratio).
> 
> Brian


  For purposes of humour, the larger magnitude number is always better.  Hence
I counted 6dB per doubling.  The fact that it was technically correct is a
mere beneficent side-effect!

    cheers,
      DaveK
-- 
Can't think of a witty .sigline today....


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]