[gta02-core] Audio ECN update

[Rene Harder]

Werner Almesberger wrote:
> Rene Harder wrote:
>   
>> This kind of EMI suppression filter is what i think we are looking for:
>>
>> TDK MEM2012P101R
>> http://www.tdk.co.jp/tefe02/e9617_mem_p.pdf
>>     
>
> And they even have them at Digi-Key :-) So that one would be good
> for data and audio ?
>   

That was the only suitable one i could find on Digi-Key. I'll check
Farnell/Newark today, maybe I'll find more.
Btw, are these UART/SIM lines terminated, and if what's the termination
resistance?

> They're a little bulk, though. Also, when I compare the attenuation
> plots of the MEM2012 (pi) and the Murata NFL18ST (tee), then they
> basically look the same.
>
>   

That depends quite a bit where you looking at. Unfortunately TDK only
recorded the attenuation till 2GHz, but if we assume the attenuation
profile is similar to MEM2012P75R0 and MEM2012P50R0 then the attenuation
is decreasing above 2GHz. (probably around 20-30dB @3GHz). However the
murata filter have a slightly higher attenuation in the area around 3GHz.

> We've been over EMI filters a lot in the Dash/HXD8 project, because
> of the high demands on GPS, and we had to switch from one type to
> the other (on UART lines) because of sourcing, apparently without
> ill effects.
>   

Well, that's not necessarily something you will see right a way, a
device might behave normally but once you fail the final EMC test
because your device doesn't meet the required standard you'll see the
difference between those filters.

> Back then, I didn't understand what the real difference between the
> two was. Maybe now I found someone who can explain ? ;-) What is
> there that the data sheet doesn't tell us ? You mention impedance,
> but all the data sheets show is attenuation for 50 Ohm.
>
>   

Yes that's exactly the point, the attenuation profile is recorded with a
50 Ohm load and source. Each filter type T and Pi has several pros and
cons and they perform differently with high or low impedance
load/source. So if in your application load/source is exactly 50 Ohm you
will get this attenuation profile but if not you'll probably end up with
a slightly different one.

Pi filter for example have a low input/output impedance at high
frequencies so in combination with a high source and load impedance they
offer a better performance against RF than T filters where the input and
output impedance is increasing with higher frequencies. [1], [2]



[1] http://www.pacaero.com/products/emi-filters-general-information.htm
[2] http://www.syfer.com/category_docs/filtsmisl.pdf