In all cases, the DDS has a 16 bit phase accumulator and a 5 bit phase truncation.
The spectrogram x-axis shows frequency scaled by clock rate (i.e. 0 to 0.5, because the simulation is real only)
The y axis is the DDS set frequency, also scaled by clock rate. The spectra were computed with a Hamming window, and a 4096 point transform.
First, for frequencies near zero
Then, for frequencies near fclock/15
![](https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEhRcMdnu7Ma5f171zy9MYsO0n-3Hf18j_KFhkb7DzVLSKS0KEMUQDv6PTVtDbWcgkI6VaHTAMNNuZfECvXXunuJXB1THbYjNwwKP8PY_-eDNQrArE5cYRimkarYBE6coPzgPjX7BFwL5Vc/s640/dds2.png)
Finally, for frequencies near fclock/4
You can see the "big spurs" move in nice lines, but the small spurs are essentially random (little speckles throughout the image)