47 nH inductor really needed?

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

I'm in the final stages of designing/laying-out a PCB with a ATMEGA324PA part. In Atmel's app note, "AVR042: AVR Hardware Design Considerations", figure 6-1 shows a 47 nH inductor for power supply filtering. Is this really needed? With a good ground plane and thick VCC traces, can I eliminate this part?

Thanks.

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

Building one for yourself, or a commercial application?

Are you using the ADC input on the micro?

If your answer is Yes to the above, then including the inductor is a good thing.

Otherwise ignore it.

JC

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

Include it, if you are using the ADC. I always do.

Leon Heller G1HSM

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

No plans for using the ADC and this is a prototype. So, I can drop it?

If this is going into production, I should keep it? Is this for reduced EMI? Or, better PS filtering?

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

Better PS filtering for the ADC as a rule. The digital power supply on most designs is quite noisy. If anything sensitive is being used on the AVR (such as an ADC), the inductor will remove a lot of the noise.

If there is a chance this will go to production I highly recommend you put it in. Then there will be less surprise when you add it later. :)

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

Maybe, I'll add it to the layout and then just jumper it for the proto. I can add it, if needed.

I'm having trouble sourcing a 47 nH TH inductor. Digikey only shows 100 nH in stock in small quantities. Is 100 nH okay?

I'll have to re-layout for production since this is a TH layout. For production, I'll switch to SMT. Thanks.

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

100nH would be just fine.

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

devils4ever wrote:
I'm in the final stages of designing/laying-out a PCB with a ATMEGA324PA part. In Atmel's app note, "AVR042: AVR Hardware Design Considerations", figure 6-1 shows a 47 nH inductor for power supply filtering. Is this really needed? With a good ground plane and thick VCC traces, can I eliminate this part?

Thanks.

Yes - you can eliminate it.

The only place where you might need an inductor is to filter AVCC on devices with a separate pin for the analog supply to the A/D. In that case the inductor is in the range of 10uH - it is also not necessary on most cases.
47nH is a very low value that will note have any effect on most PCBs.

Note that in the app note the purpose of the inductor is supposedly to filter noise from the power supply to prevent it getting to the processor. With the values shown this attenuation will be limited by the self-inductance of the capacitor (many nH) and the value of the capacitor. I've only used this or seen other designs use the LC filter for sensitive sections of the circuitry such as ADC or PLLs. For normal digital logic use the brute force approach and just put plenty of decoupling.

The self-resonant frequency of a through hole 0.1uF capacitor will be 10MHz or so - above that frequency it will act like an inductor.

kevin

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

Inductors in addition to capacitors make the decoupling filter one order higher so it will attenuate ripple and noise 6 dB more in the transition band compared to if you just use standard capacitor decoupling. It depends on your application, ie if you use the ADC or the same rail is shared with analog electronics.

An VCC pin of a microcontroller is not sensitive to ripple and noise and normally does not require a coil filter, except if you use the ADC.

I often use two coils and capacitor decoupling to separate AVCC and DVCC on a microcontroller when the ADC is in use. One coil helps to filter away the digital noise that comes from the MCU (DVCC), and the other filters the power supply noise.

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

The LC filter in the appnote has a cutoff frequency of:
1/(2pi*sqrt(L*C)) = 2,3 MHz so it must be a EMI filter.

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

Who needs to buy a 47 nH inductor. Four (4) turns of 20 SWG 4mm diameter over 4 mm gives 47 nH. http://www.vwlowen.co.uk/java/coil.htm A single wire turn on a ferrite bead woud do about the same.

Charles Darwin, Lord Kelvin & Murphy are always lurking about!
Lee -.-
Riddle me this...How did the serpent move around before the fall?

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

Quote:
Better PS filtering for the ADC as a rule. The digital power supply on most designs is quite noisy. If anything sensitive is being used on the AVR (such as an ADC), the inductor will remove a lot of the noise.

If there is a chance this will go to production I highly recommend you put it in. Then there will be less surprise when you add it later. Smile

On the issue of "filtering" Avcc with an inductor the surprise very well may be *worse* behavior with it, especially when powering the system with switching regulators. Step-function noise and/or ripple on Vcc may cause the inductor to ring, and *increase* the noise on Avcc. A well designed and decoupled board should not have more than a few mv of noise on the power rails, and this will not bother the adc. In my experience mixing sensitive analog circuitry with digital, opamps are more affected by power supply noise and each should have its own rc filter network on its power supply pins. The Avr's Avcc should not need filtering, but if for some reason it must be done, I suggest a rc network, not lc.

For anyone probing noise issues, I consider having a good broadband ac voltmeter essential, scopes are not sensitive enough. Ebay is loaded with good serviceable Ballantine meters, imho, the gold standard for this sort of thing. I would love to hear if anyone measurably improved their situation with an inductor on Avcc.

Tom Pappano
Tulsa, Oklahoma

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

Interestingly, Microchip recommends an RC filter for the AVcc supply on their chips. It seems to work just as well as the inductors I've used on AVRs.

Leon Heller G1HSM