Any AVRs with > 10-bit ADCs?

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

I searched through all the devices and they all seem to have only 10-bit ADC's. I was just wondering if I was missing anything... Do any exist with resolution greater than 10-bits?

Thanks,
James

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

In general, the answer is "AVRs have 10-bit A/Ds".

Now, after the Feb. 2007 promised announcement of the "XMega" (which still hasn't happened AFAIK) we can expect 12-bit A/D:
https://www.avrfreaks.net/index.p...
https://www.avrfreaks.net/index.p...
https://www.avrfreaks.net/index.p...

At least one "specialty" AVR has 12-bit: Mega406.

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

If you are able to oversample the signal, you might be able to increase the resolution. This is documented in application note AVR121.

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

Depending on what you are doing, you might implement a slooow dual slop approach using capaictor, timer & the builtin comparator. While this might be a headache to get high accuaracy easily, it could easily give high resolution. But probably only a few conversions a second.

hoyt

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

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

If that dual slope is only going to net a few conversions per second you might as well go with kms' sudgestion and do the mulitple conversions needed to get better resolution (4 conversions for +1bit res, 16 for +2, 64 for +3 etc). Remember, some noise is good in this situation.

Edward

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

If you could either sample 16 times with 13 us duration conversions and average the result (taking a total of 208 us for the 16 samples) or sample one time with a 260 us conversion time, which method would give you a more accurate answer?

James

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

The signal needs to sit still for 16 samples for oversampling to work right?

Imagecraft compiler user

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

Yesish, it actually helps if you have about 1 lsb of random noise . But yes, there should be no drift.

Edward

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

Quote:
The signal needs to sit still for 16 samples for oversampling to work right?

But what I meant was... if you could either do several short 10-bit samples & average them... or one long 10-bit sample of equal length to all the short samples... assuming the signal's value held still... which way would give you the more accurate answer?

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

A 10 bit sample is always going to have the same resolution and accuracy no matter how long you take to do it, it's still 10 bits. The oversampling has the same effect as a floating point average but it really isn't and shouldn't be confused with averaging which is to reduce noise.

Edward

PS. If your asking if 4x oversampled 10bit ADC (giving 11bits) is more resolved or accurate than an 11bit ADC that takes 4 times as long as the 10bit ADC. Resolution will be the same (both are 11bits) and accuracy will entirely depend on the accuracy of the ADCs as specified in their datasheets, niether is inherently more accurate I believe.

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

Quote:
A 10 bit sample is always going to have the same resolution and accuracy no matter how long you take to do it

I just assumed that since you can set the ADC prescaler to get conversion times ranging from 13 - 260 us that the longer conversion must have higher accuracy... or why would anyone select that? And then I wondered if you'd be better off doing several 13 us conversions and averaging or just one long 260 us conversion.

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

Yup, not much reason to go for the slow end. I would much rather go with a 65us (the fastest conversion on my chip the 644) and over sample it 4x for 11bit resolution, or average it for better noise handling, than going with the 260us conversion.

Edward

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

Quote:

Quote:
A 10 bit sample is always going to have the same resolution and accuracy no matter how long you take to do it

Quote:

Yup, not much reason to go for the slow end.

Well, there IS one reason for plodding along like us old bit pushers. At a slower ADC clock rate, the sample time for the S/H is (seems to be?) increased, which can help with relatively high-impedance signals such as 30k thermistors.

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.