NCO/PWM jitter

sjev

Hello,

I am using the following code to generate a frequency (copied from the counter application notes):

---

PUB StartDOD
  ctra := %00100_000 << 23 + 1 << 9 + dod_pin 
  frqa := $8_0000
  dira[noparse][[/noparse]dod_pin] := 1 
  repeat

---

The problem is that I get a pretty bad jitter (+-2% of the period).
What did I do wrong? Didn't I read the manual carefully? Or shouldn't I buy the DLP-PROP module? My oscilloscope is not fast enough to measure the clock frequency.

Funny thing is that a pulse generator at a different cog (using waitcnt) produces a pulse with a jitter sychronous to the NCO. So, does my system clock jitter???

CJ

I'm pretty sure the DLP-Prop is using the oscillator from the ftdi chip and that isn't super accurate, someone else had trouble using it with serial comms

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Parallax Forums - If you're ready to learn, we're ready to help.

sjev

So I have measured the 6MHz clock at the XI pin.
It jitters!!!! Damn FTDI!
Tomorrow I'll test another chip at work ( yes, it's more than a hobby .
Maybe a word of warning should be added to the DLP-PROP page. Unstable frequency in industrial control applications is a pretty serious issue.

Post Edited (sjev) : 4/3/2007 10:56:34 PM GMT

QuattroRS4

What industrial application are you using a DLP-Prop on ? Especially when its frequency generation ...

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
'Necessity is the mother of invention'

sjev

QuattroRS4 said...
What industrial application are you using a DLP-Prop on ? Especially when its frequency generation ...

I am building an inkjet printer prototype. It is just a simple research device, but the timings should be absolutely solid for a dozen of reasons.

sjev

Yes, the other chip is much better. With a good scope Ican measure the standard deviaton of the jitter. On this one it's 0.1 %. Not rock solid, but workable. I guess my next order will be a bunch of prop sticks instead of dlp-props