MASH Noise shaping

Nov 05, 2005 3 Replies
A

I'm interested in fractional-N synthesizers, I've built a couple for hobby projects, and I'm playing with software simulations to evaluate the Fourier series of (a) the MASH output, and (b) the waveform at the PFD output. Noise at the MASH output rises at 80 dB per decade for a 4-stage MASH. My simulation of the PFD output for a synthesizer using the same MASH shows a



20 dB per decade slope. I don't know if this is right. My question is: what should the noise at the PFD output look like, and why? Since the output of the MASH controls frequency via the PLL divide-by-N, and the PFD output is measuring phase, I was expecting them to be seperated by an integration i.e. 20 dB per decade, but it seems that either my simulation is wrong, or it's more complicated than that. Any pointers would be appreciated.

TIA Andrew.



MASH:

formatting link
PFD output:
formatting link
(*)



  • this is a 16-bit MASH, but I get the same slope for 24-bits.


Vote
Twitter Facebook LinkedIn
J

Isn't it just the fractional component of the quotient? E.g., if you are generating 1.1 MHz from a 10/9 divider, you'd expect there to be a

100 kHz sawtooth artifact in the PFD output.

Have you looked at the latest ADIsimPLL tool at the Analog Devices site? I know they're selling some fractional-N chips these days. Good chance that ADIsimPLL will simulate them.

Also: I recently pulled the service manual for the HP 8657A down from agilent.com, and noticed that it has a nice tutorial on fractional-N loops. You might check out that service manual.

I'm not at all familiar with MASH technology, but that looks like some seriously-nice work. Bookmarked for later perusal.

-- jm

------------------------------------------------------

formatting link
Note: My E-mail address has been altered to avoid spam

------------------------------------------------------

Vote
Twitter Facebook LinkedIn
L

...Since the output of the MASH controls frequency via the PLL divide-by-N, and the PFD output is measuring phase, I was expecting them to be seperated by an integration i.e. 20 dB per decade...

Why were you expecting them to be seperated by an integration i.e. 20 dB per decade?

Vote
Twitter Facebook LinkedIn
A

Because phase is the integral of frequency.

Vote
Twitter Facebook LinkedIn

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required