EMRFD Message Archive 3485

Message	Date	From	Subject
3485	2009-08-18 23:50:28	ashhar_farhan	Si570
Since someone discussed Si570 on the list, I have been pursuing some literature about it. The chip is quite fascinating. I was wondering if someone could explain how it works. specifically : 1. how do you get a low noise vco that spans such a huge frequency range. 2. what kind of PLL/DSP is being used in this design? the step size seems to be too fine for it to be a PLL, the spurs seem too low for it to be DDS. thanks, - farhan
3486	2009-08-19 00:34:26	ehydra	Re: Si570
ashhar_farhan schrieb: > Since someone discussed Si570 on the list, I have been pursuing some literature about it. The chip is quite fascinating. I was wondering if someone could explain how it works. specifically : > > 1. how do you get a low noise vco that spans such a huge frequency range. > > 2. what kind of PLL/DSP is being used in this design? the step size seems to be too fine for it to be a PLL, the spurs seem too low for it to be DDS. > It's is simply a form of noise-shaping. Read the Si570 patents. The technique behind is at least 20 years old. I think it was a Racal synthesizer. If you like to implement it yourself, go here: http://www.google.de/search?hl=de&safe=off&q=%22an+all+digital+fractional-n+synthesizer&btnG=Suche&meta= http://t03dsp.skydan.in.ua/Download.htm regards - Henry -- ehydra.dyndns.info
3487	2009-08-19 02:08:24	leon Heller	Re: Si570
----- Original Message -----
3488	2009-08-19 03:32:26	dixonglennb	Re: Si570
> 1. how do you get a low noise vco that spans such a huge frequency range. > Hi Farhan, Since it is a digital clock, The VCO only has to cover some portion of the highest frequency range. The lower frequencies are obtained by the divide-by-n as shown in Figure 2 of the data sheet. > 2. what kind of PLL/DSP is being used in this design? the step size seems to be too fine for it to be a PLL, the spurs seem too low for it to be DDS. > I think the method uses DSP-based phase correction...calculating and correcting phase detector output on the fly. This allows a higher frequency phase reference than normally allowed for the small step size. Henry mentioned the RACAL technique. National Semi produces some similar synthesizer chips that require an external VCO, but that go to, I think, as high as 5 or 6 GHz. What is nice is the faster/better integrated circuits available nowadays make error sources lower and the technology cheap. 73, Glenn AC7ZN