1
\$\begingroup\$

The below lines are from AVR® Microcontroller Hardware Design Considerations (AN2519).

"In noisy environments the oscillator can be crucially affected. If the noise is strong enough, the oscillator can “lock up” and stop oscillating. To reduce the sensitivity of the oscillator to noise, the size of the capacitor at the high-impedance input of the oscillator circuit, XTAL1, can be slightly increased. Increasing only one of the capacitors does not affect the total capacitive load much, but unbalanced capacitors can affect the resonant frequency to a higher degree than the change of the total capacitive load. However, unbalanced capacitive loads will affect the duty cycle of the oscillation and should not be used. This is especially critical if the AVR device is utilized close to its maximum speed limit."

I have some questions regarding this.

  1. How do unbalanced capacitors reduce the sensitivity to noise?

  2. They also say "However, unbalanced capacitive loads will affect the duty cycle of the oscillation and should not be used". Does that mean: never use unbalanced load capacitors?

  3. Which one is better, balanced or unbalanced load capacitors?

\$\endgroup\$
4
  • 1
    \$\begingroup\$ Many crystals are embedded inside a metal case, or have a metal lid. Noise can couple to the crystal via this case. I'm surprised that grounding the case is not mentioned as a tactic to improve noise immunity. Trace routing on a printed-circuit board is critical too, if noise sources are nearby. \$\endgroup\$
    – glen_geek
    Aug 12, 2022 at 12:20
  • \$\begingroup\$ @glen_geek ,thnak you. but how unbalanced capacitors reduce the sensitivity to noise. \$\endgroup\$
    – Confused
    Aug 12, 2022 at 12:24
  • 1
    \$\begingroup\$ Hari, I don't have a good model of AVR oscillator's MOSfets. Have a model of 74HCU04 inverter, and looked at input Z and output Z, which were both in the same ballpark at high frequency (~10 MHz). Impedance goes WAY up at low frequency so this may be the region that AVR addresses... I can imagine that input Z can become much higher than output Z. If this is so, then increasing \$C_{in}\$ can lower the impedance at input node. Hardly any change in duty cycle with unbalanced capacitors at high frequency - I don't know the source of duty cycle changes that this APnote mentions. \$\endgroup\$
    – glen_geek
    Aug 12, 2022 at 14:42
  • 1
    \$\begingroup\$ Just a note: avoiding the noise altogether may be a good idea. Putting the entire device into a metal case, or putting a metal can over the processor, or making sure that the crystal, capacitors, and processor are laid out carefully and closely to one another, or using a crystal with a grounded case, would all be good things to consider. Changing cap values like this sounds like a desperation move, not a first choice. \$\endgroup\$
    – TimWescott
    Aug 12, 2022 at 15:24

1 Answer 1

1
\$\begingroup\$
  1. It's not the fact that you create imbalance that will remove sensitivity to noise. Like it says in the text you quote, simply adding capacitance to oscillator input does reduce sensitivity to noise. Which means you might not be able to increase value of both capacitors, so it might require to increase the other and decrease the other in order for the same crystal and same oscillator to still operate within same specs.

  2. No, it does not mean you should not use imbalanced capacitors. It just means that if you do then you should be aware of the consequences how it will affect the oscillation. Then you can estimate how well the circuit will work or if there is a risk it does not and how to design or re-design it so that it will work. If the option is to scrap 10000 already made blank PCBs and redesign it, or to just change the value of capacitors that need to be mounted, it may make a difference what is best solution in different situations.

  3. There is no general answer which one is better. It will depend on all the other components including the PCB layout. Some chips may simply require imbalanced capacitances. Always check the datasheets and application notes of the components you are using.

\$\endgroup\$
3
  • \$\begingroup\$ ,Thank you.I did not understand the answer to my 1st question. \$\endgroup\$
    – Confused
    Aug 12, 2022 at 9:19
  • 1
    \$\begingroup\$ @Hari OK. How should I put it better then? Increasing the capacitor value is the reason for better noise rejection. Changing value of only one capacitor creates the imbalance. So it is not the imbalance which improves noise rejectio, they both are results of changing one capacitor to larger value. \$\endgroup\$
    – Justme
    Aug 12, 2022 at 18:25
  • \$\begingroup\$ @ Justme thank you.I got it \$\endgroup\$
    – Confused
    Aug 13, 2022 at 13:21

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.