LP filter or decoupling capacitor

Go To Last Post
7 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hello,

This is the statement of AVcc in the datasheet of AT90CAN128:

Quote:
"AVcc is the supply voltage pin for the A/D Converter on Port F. It should be externally connected to Vcc, even if the ADC is not used. If the ADC is used, it should be connected to Vcc through a low-pass filter."

On the evaluation board DVK90CAN1 for AT90CAN128 there is only a 100nF capacitor near AVcc. In my case, a DC/DC converter is used to supply power as Vcc. Can I replace the LP filter with just a capacitor just like this? If really using a LP filter, which one is better, RC or LC?

Thank you
Senmeis

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

LC is better than RC. Are you actually using the ADC to measure critical signals? Because that is where the requirement for a clean analog supply will benefit you the most. Feed a fixed, filtered/bypassed voltage to the ADC and see how much the data changes from sample to sample, and if it is unacceptable, use a better filtered power supply.

Don't forget the PCB design guidelines. Ground and supply traces should be separate for the digital and analog sections. Separate ground planes too.

If you think education is expensive, try ignorance.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
LC is better than RC.

I would suggest caution here. If Vcc comes from a switcher, there probably is at least a small amount of high frequency ripple on the Vcc bus, and maybe even a lot. This can cause the inductor to ring and actually *increase* the amount of noise on AVcc. I would use a RC filter, if any, in this case. My experience has been that with a good layout and solid decoupling no special treatment is needed for AVcc. A nice wideband AC voltmeter such as a Ballantine can really be helpful in discovering where noise really comes from, and can read signals scopes can only dream about 8-)

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I guess that I'll throw in my two cents, based on experience and pretty much common sense:

If you want rock-steady results from the AVR A/D converter, you need rock-steady AVcc and reference voltage and ground.

If you want rock-steady, consistent A/D counts from converting your app signal, then the signal needs to be rock-steady.

There have been quite a few threads on this, and when all is said and done there ends up to be ripple on one or more of the above.

The solution will depend on the app and its requirements. For "general" work that can be "close enough"--e.g., a non-critical temperature reading via a thermistor that is picking up some mains frequency ripple--a compromise solution such as a simple averaging of a number of readins may work quite well. In other apps--e.g., reading an amplified mV signal from a bridge configuration such as a pressure sensor with best accuracy and precision--great care will need to be taken throughout the entire analog circuitry.

Lee

You can put lipstick on a pig, but it is still a pig.

I've never met a pig I didn't like, as long as you have some salt and pepper.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:

, and can read signals scopes can only dream about Cool

Them be fighting words, but luckily I'm not in a fighting mood. The multimeter might be better at quantifying the measurements, but for the ability to visualize the signal and identify problems that way the oscilloscope is unrivalled.

The use of a small ferrite bead is also something to be considered. Technically it is still LC, but the value of the inductance is small, albeit with cost and size advantages.

If you think education is expensive, try ignorance.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Them be fighting words, but luckily I'm not in a fighting mood. The multimeter might be better at quantifying the measurements, but for the ability to visualize the signal and identify problems that way the oscilloscope is unrivalled.

Wide band AC voltmeters (not multimeters) such as the Ballantine 323, etc. have ranges down to 1mv full scale. When seriously trying to observe the effects of various filtering and shielding tactics on noise, accurate quantification is exactly what you need to do to evaluate a design. The scope does let you visualize the noise and recognize perhaps where it is coming from if the amplitude is high enough, but 5mv/cm or even 1mv/cm sensitivity is not much help when you are trying to reliably observe the difference between .3mv and .1mv (imho).

Tom Pappano
Tulsa, Oklahoma

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you don't use much current from AVCC you can use an RC filter. If the current is more than about 10 mA, I would suggste the LC filter or maybe a comination of both, having L and R in parallel. This will give low DC drop and damping of possible resonances. Sometimes it's also usefull to have a filter in the normal VCC line as well.

Having about 100 nF at AVcc / AGND is allways suggestet no matter if the ADC is used or not.