Page 1 of 1
LC Filters and AVCC
Posted: 01 January 2008, 16:33 PM
by Don_Kirby
I've been looking at SMT LC filters, trying to find something equivalent to what is recommended in the Atmel Datasheet. My math gives me a center frequency of about 5 Mhz from the values shown in the datasheet for a Mega32. The values shown in the ZX24 schematic are slightly different, however, and give a center of around 1.6Mhz. Is this on purpose, or is my math faulty? What is the logic behind picking a particular frequency? I would have thought that it would be primarily based on the circuit/board/environment, but nowhere do I see anything suggesting that.
Math aside, is there anything wrong with using one of
these devices, or something similar?
-Don
Posted: 01 January 2008, 18:28 PM
by dkinzer
It is probably a good idea to start with Atmel's recommendation and then vary the parameters to fit the application. In the case of the ZX-24, we chose the largest inductor that we could find in the 0603 package.
The purpose of the inductor and capacitor is to filter transients on the supply rail in order to minimize their effect on ADC conversion results. If your application doesn't use the ADC, you can omit the inductor and capacitor (i.e. zero inductance and infinite capacitance). If you do use the ADC, I would recommend using an inductor as close to the size recommended by Atmel that will fit your physical constraints. You should be able to find a larger valued inductor in the 0805 size. It is important to note that all of the current that is sourced by pins on Port A plus the current drawn by the ADC itself must pass through the inductor. An inductor with a current rating of 200mA or more should be sufficient.
As for the device to which you referred, it appears that it offers minimal inductance. The equivalent circuit makes it appear more like a feedthrough capacitor.
Posted: 01 January 2008, 18:47 PM
by Don_Kirby
dkinzer wrote:It is important to note that all of the current that is sourced by pins on Port A plus the current drawn by the ADC itself must pass through the inductor.
Good point. This particular application does not use port A for anything but the ADC itself however.
dkinzer wrote:
As for the device to which you referred, it appears that it offers minimal inductance. The equivalent circuit makes it appear more like a feedthrough capacitor.
That's what it looked like to me as well. Would you agree that the Atmel specs should be considered absolute minimum values then; center frequency being unimportant?
Ideally, I would like to use the smallest possible package, 0402 or 0201, or even better yet, combine the inductor and the cap into a single package. I will not do that at the expense of proper operation however.
-Don
Posted: 01 January 2008, 19:03 PM
by dkinzer
Don_Kirby wrote:Would you agree that the Atmel specs should be considered absolute minimum values then; center frequency being unimportant?
I suspect that the center frequency is relatively unimportant. I would regard the Atmel recommendation as a desirable target and select a part that meets your other needs while getting as close to the recommendation as possible.
You can do some tests using chokes with different inductance and see what effect they have on the accuracy/repeatability of your ADC readings. If you can get the accuracy that you need with an inductor 1/10th the recommended size which also meets your physical requirements you're good to go.
Posted: 01 January 2008, 19:12 PM
by Don_Kirby
dkinzer wrote:If you can get the accuracy that you need with an inductor 1/10th the recommended size which also meets your physical requirements you're good to go.
Sounds good to me. I'll most likely end up with something on the order of what the '24 uses, as all of the other passives are 0603 on that side of the board anyway. Might as well be consistent.
Thanks
-Don