AVR32 PLL Details - numeric limits on m/n

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

The AVR data is sparse here, they give a formula, but no numeric ranges.

If I look at the register mapping, I infer that just 4 bits are used for M,N, but that hardly seems right ?
ie if your PLL goes to 240Mhz, then /16 stops at 15Mhz, missing a whole range of crystals.
Is there any hard info, on the numeric ranges that the PLL can actually work with ?

The whole idea of a PLL, is to allow more choices from a given crystal, and on a 32 bit chip, allocating very small numbers to PLL ranges, does not make sense.
We need to cover 3 target frequencies, and with good precision.

I see the LPC17xx has a useful PLL spreadsheet, that takes numbers, and reality-checks them - is there a AVR32 PLL-Calc equivalent ?

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

Hello,
the four bits are right. Atmel didn't implement the best PLL but for division you have the bits PM.CKSEL too. I don't know a PLL-calc tool but for that limited range you won't need it.

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

Thanks.
Bummer; looks like it's the LPC175x, which is a larger package, and more $$, than we otherwise need.

Does anyone know if any of the newer AVR32's have better PLLs ?

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

What's your need?

-sma

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

sma wrote:
What's your need?

Using the LPC175x PLL calculator, we can hit our 6 possible target frequencies, within 10ppm. (we should need only 3 in the final project)

Here, that's a VCO of 275-550MHz (post Div2), and then a XtalDiv of 5 bits (even tho the data suggests more) and a Freq Multiplier of 9 bits.

The RefFreq needed is from ~780KHZ to ~2-3MHz

So, only a slight improve from the very limited AVR32 design of 4 bits is needed : On the Xtal Div: 5 or 6 bits, and on the VCO Divider : ~9 bits. - and the PLL needs to cope with Ref Frequencies of >= 750KHz.

The LPC family can work down to below 32KHz Fref, and has some modes for 32.768 locking; nice features, but not needed on this design.

They also have a nice CLKOUT divider, that is separate from the CPU Divider, so we can get our targets on a pin, without imposing too many caveats on the core.

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

Hello,
if you want some help please ask some proper question or answer the right way to questions from the community.
What's mis understandable on

Quote:
What's your need?
?
You won't get an answer if you talk for a page about LPC175x but didn't tell us what's your need.

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

uC-Harry wrote:
Hello,
if you want some help please ask some proper question or answer the right way to questions from the community.
What's mis understandable on
Quote:
What's your need?
?

If you read carefully, you will see I gave an example of what I need.

uC-Harry wrote:

You won't get an answer if you talk for a page about LPC175x but didn't tell us what's your need.

The LPC is simply an example, and again, if you read carefully, this is what an _AVR32_ needs, in order to meet what we need :

["So, only a slight improve from the very limited _AVR32_ design of 4 bits is needed : On the Xtal Div: 5 or 6 bits, and on the VCO Divider : ~9 bits. - and the PLL needs to cope with Ref Frequencies of >= 750KHz."]

There, I have given actual bit-counts on the Xtal Div, the VCO div, and the RefFreq, which is as complete info as most need, to design a PLL.

What I asked above, still unanswered, was if any of the upcoming AVRs improved the shortcomings on the PLL ?

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

Hello,
sorry I didn't realize that you were talking about improvements of AVR32 PLL. We all know that the PLL could be better.
I thought you want some community support to find a good solution for your needs.

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

as a followup, a just unveiled alternative opens using a 'proper' external Synthesizer - yes, it pushes up the parts count, (and adds a supply risk), but does allow a MUCH cheaper uC to be considered.

This news from IDT
http://www.idt.com/?id=5627
[" 20-pin and 24-pin QFN, sampling, $0.80 to $1.25/10K" ]

VCO is 100..475MHz, and XtalDiv is 8 bits, FeedBackDiv is 11-12 bits, and PostDiv is 8 bits.
Up to 120MHz out, 1..40MHz Xtal IN, and with EEPROM and an impressive 128 bytes of Config array.