Which decoupling caps on Atmega32U4?

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

Hi! I have a question about decoupling caps in a minimum circuit with Atmega32U4. How much and which value? The datasheet keeps silent about it. I have 0.1µF caps on almost all VCC lines. Some guy told me it is common sense to have these values.
Alright. But then I would assume having 0.1µF on each VCC line. Why is there a "1µF" (not: 0.1µF) on UCVV? I copied this value from Teensy2.0. But that doesn't explain me why "1µF" there?

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

Why is there a "1µF" (not: 0.1µF) on UCVV? I copied this value from Teensy2.0. But that doesn't explain me why "1µF" there?

It's not on UVCC so much as it's on VBUS:

 

https://www.pjrc.com/teensy/schematic2.gif

 

 

And the datasheet recommendation is 10 uF:

 

 

It wouldn't be the first time someone didn't follow the manufacturer's recommendations:

https://www.avrfreaks.net/forum/caps-pin

 

Also, everyone who works with AVR should familiarise themselves with (at least) AVR042: AVR Hardware Design Considerations.

 

"Experience is what enables you to recognise a mistake the second time you make it."

"Good judgement comes from experience.  Experience comes from bad judgement."

"Wisdom is always wont to arrive late, and to be a little approximate on first possession."

"When you hear hoofbeats, think horses, not unicorns."

"Fast.  Cheap.  Good.  Pick two."

"We see a lot of arses on handlebars around here." - [J Ekdahl]

 

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

There are three rues of thumb:

 

1.  100nF (0.1uF) ceramic for decoupling integrated circuits.   Especially digital switching.   Short leads have lowest self-inductance

2.  10uF - 470uF electrolytic or Tantalum for "current reservoir" in power supplies.   Leads are less critical.   High currents may need special quality.

3.  22pF for crystal oscillator circuits.

 

Yes,   your UVCC pin needs (2).   But any other chips supplied by UVCC will have their own local (1).

An Analog circuit might use (2) and (1) close to the chip.

 

You have been very wise.   e.g. inspect the schematic and pcb layout of a Teensy.   Copy the proven design.

Most importantly,  do not omit any of the capacitors.  If a commercial board uses them,   it is for a good reason.

 

David.

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

@joeymorin I see. The teensy2.0 makers used "1µF" instead of the recommended 10µF, because there is another 1µF in the circuit (on UCAP), and by this they reduced the amount of different components to a minimum and spare production costs, right?

 

@david.prentice Thanks for these rules of thumb. This part of your post is still not clear to me

Yes,   your UVCC pin needs (2).

Wouldn't UVCC need a decoupling 0.1µ too?

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

right?

I think you'd have to ask them.

"Experience is what enables you to recognise a mistake the second time you make it."

"Good judgement comes from experience.  Experience comes from bad judgement."

"Wisdom is always wont to arrive late, and to be a little approximate on first possession."

"When you hear hoofbeats, think horses, not unicorns."

"Fast.  Cheap.  Good.  Pick two."

"We see a lot of arses on handlebars around here." - [J Ekdahl]

 

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

Look at Joey's post.   And the Teensy schematic.   You have 1uF as a "reservoir" whether you select 3.3V or 5V.    10uF would be equally appropriate.

 

Anything that eventually uses the "VCC" derived from UVCC / UCAP will have their own 100nF capacitors.

 

There is no need to know the maths or the physics.   Just copy the proven design.

 

If you want a layman's explanation.   100nF provides short term current to fast switching circuits.   10uF or 470uF provide longer term current.

Think of a sprinter versus a marathon runner.    (just realise that instead of a 10 second sprint you have a 0.000000001 second switching time,  instead of 3 hour run,  you have a 0.01 second mains ripple).   The sprinter probably eats glucose.   The marathon runner eats slower-release carbohydrates.

 

David.

Last Edited: Sat. Jul 16, 2016 - 04:53 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ye, good layman explanation. I get a glimpse of the logic behind that. I'm using 4 "sprinters" with 0.1µF to compensate leads, and 2 "marathon runners" with 1µF to compensate gaps. Recommended is, to have 0.1µF on every power line, so I rather should have 6 of these sprinters, instead of 4, right?

Last Edited: Sat. Jul 16, 2016 - 07:10 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Krümelmonster wrote:
so I rather should have 6
6 is the recommended number. Depending on how the circuit is constructed (breadboard, single layer circuit board, presence of ground planes, etc.), how (electrically) quiet the environment, and how harsh the temperature range, you may be able to fudge that a bit.

David (aka frog_jr)

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

frog_jr: Got it. Since my power source is pretty clean (pc USB hub) I probably can get away with 4. Well, then ...

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

Since my power source is pretty clean (pc USB hub)

Maybe.

 

Is it a stand-alone hub with its own wall-wart power supply, or is it internal to a PC?

 

The stand-alone unit might be pretty clean.

 

If the hub is part of a PC then it might well be very noisy.

 

Are you building 1,000,000 PCB's?

 

If you are only building one or several, then just install the "extra" two caps!

 

JC