On Jun 4, 11:25�am, julius <juli...@gmail.com> wrote:> Weren't you saying in some other post that you go to Boston > occasionally (from VT or NH where you live?), if you do please holla > the next time you come down.well, i'm in Boston (more precisely Waltham) now and will be here for a couple weeks. i'll click on the ellipses in your email (i'm using Google groups) and send you an email. you *do* take mail at your gmail account, no? L8r, r b-j
Computing complex phase rotation in fixed-point
Started by ●June 2, 2009
Reply by ●June 4, 20092009-06-04
Reply by ●June 4, 20092009-06-04
julius wrote:> On Jun 3, 10:06 pm, robert bristow-johnson <r...@audioimagination.com> > wrote: >> i thought it was you getting outa college a short decade ago. wasn't >> it? i even seem to remember a phone call from you (outa the blue) >> when i lived in New Jersey. maybe it was someone else. >> >> you dunno (yet) what getting old is. :-) i'm gaining a glimpse of >> it. s'pose Jerry sees it pretty well. >> >> r b-j > > I moved from Berkeley to MIT in 2001, and I remembered that you > lived in the northeast. So I thought I'd give you a call to try and > meet > up. Long story short, I got done with my PhD in 2006, then went to > Houston for 2 years, and now I'm back in Boston. > > Weren't you saying in some other post that you go to Boston > occasionally (from VT or NH where you live?), if you do please holla > the next time you come down. > > My in-laws are in south NJ and I've been trying to meet with Jerry > but it's been hard with all the family hijacking of your time when you > visit them .... Sorry, Jerry!Last time we tried, It was I who couldn't make it. Some day ... Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
Reply by ●June 4, 20092009-06-04
On Jun 4, 11:37�am, bellda2...@cox.net wrote:> > Are you going to repeatedly rotate this vector? �If so, amplitude > error may accumulate; I have seen the result go unstable in floating- > point. It may not remain a unit vector. A phase accumulator with a > lookup table can solve this, at the price of a few multiples. �You can > save memory on the LUT by having one LUT for 360 degrees, and one LUT > for much smaller steps between the 0 and the next point in the first > table. Use cos(sum of angles) and sin (sum of angles) to get the > result. Again at a cost of a few multiplies. > > DirkHi Dirk, Not quite repeatedly. This is for phase tracking. So I'll get a sequence of symbols, run each through a CORDIC to get the angle error, track it, and then do corrections. Good warning on amplitude error accumulation, but I think it does not apply in my case. Thanks, Julius
Reply by ●June 4, 20092009-06-04
julius wrote:> On Jun 4, 11:37 am, bellda2...@cox.net wrote: >> Are you going to repeatedly rotate this vector? If so, amplitude >> error may accumulate; I have seen the result go unstable in floating- >> point. It may not remain a unit vector. A phase accumulator with a >> lookup table can solve this, at the price of a few multiples. You can >> save memory on the LUT by having one LUT for 360 degrees, and one LUT >> for much smaller steps between the 0 and the next point in the first >> table. Use cos(sum of angles) and sin (sum of angles) to get the >> result. Again at a cost of a few multiplies. >> >> Dirk > > Hi Dirk, > > Not quite repeatedly. This is for phase tracking. So I'll get a > sequence of symbols, run each through a CORDIC to get the angle > error, > track it, and then do corrections. > > Good warning on amplitude error accumulation, but I think it does > not apply in my case.Read Ray Andraka's CORDIC writeup. http://www.andraka.com/cordic.htm Jerry -- Engineering is the art of making what you want from things you can get. �����������������������������������������������������������������������
