[time-nuts] DMTD for different input frequencies

Magnus Danielson magnus at rubidium.dyndns.org
Sun Nov 14 20:01:33 UTC 2010


On 11/14/2010 08:28 PM, jimlux wrote:
>
>>> Cheers,
>>> Magnus
>>>
>> One still has to ensure that close in spurs generated by the DDS
>> sources aren't an issue.
>> If the zero crossings of the beat frequency signals at the outputs of
>> the 2 final mixers arent closely aligned then such spurs can be
>> problematic.
>>
>
> How close in for the spurs?
>
> For most DDSes the spurs come from phase truncation errors, and to a
> lesser extent from harmonics aliased into the passband. There's a couple
> ways to reduce this for basically fixed frequencies.. You can drive the
> DDS clock from a variable frequency source (e.g. another PLL or DDS) so
> that the frequency you're generating doesn't have spurs in "bad" places.
> Another approach is to do your DDS with a FPGA/DAC, and make the phase
> accumulator and sin/cos LUT not a power of 2 long.

The simplicity of power of 2 DDSes and it's effects on spurs seems to be 
not fully understood. I have just not taken then effort to write up a 
DDS approach with variable length, but it is definitively on my list of 
things to do.

> And finally, there's all kinds of clever ways to suppress close in spurs
> with things like error feedback.

Feed forward corrections have also been seen.

But all this is really a complete side-track to what I was trying to 
discuss, a method for dual frequency DMTD.

Cheers,
Magnus



More information about the time-nuts mailing list