06-08-2018, 10:56 AM
Are you using any form of low pass filter at the output? At 45MHz output from a 125MHz clock you're getting fairly close to the Nyquist limit and the waveform will probably be a bit horrible without a filter.
Your counter is likely to be more accurate than a low cost crystal on the DDS board. But you can't be sure unless you've got a GPS or other high quality reference. Or generate 198kHz and see how it beats against R4 long wave. Then you can build a calibration offset into the software. Strictly it's not an offset, it's a multiplier.
I know it's a trivial error but your denominator 4294967295 should read 4294967296. In case others are wondering about the strange looking number it's 2^32. Most digital frequency synthesis is done with a denominator that's a binary power. It's nice and easy. If you need an exact ratio that can't be expressed as x/(2^n) this is also possible by switching the numerator when the accumulator overflows.
Your counter is likely to be more accurate than a low cost crystal on the DDS board. But you can't be sure unless you've got a GPS or other high quality reference. Or generate 198kHz and see how it beats against R4 long wave. Then you can build a calibration offset into the software. Strictly it's not an offset, it's a multiplier.
I know it's a trivial error but your denominator 4294967295 should read 4294967296. In case others are wondering about the strange looking number it's 2^32. Most digital frequency synthesis is done with a denominator that's a binary power. It's nice and easy. If you need an exact ratio that can't be expressed as x/(2^n) this is also possible by switching the numerator when the accumulator overflows.
www.borinsky.co.uk Jeffrey Borinsky www.becg.tv
