AD9914 Spurs

Dear Gents at ADI,

Kind reminder.

This query is open from end of last year with no feedback.

Thanks a lot

Freddy (ex DFAE)

Hi gil_he ,

Some thread worth reading: 

1. AD9914 outputting more than single frequency (unknown spurs?) - Q&A - Direct Digital Synthesis (DDS) - EngineerZone

All the best,

Jules

hello Jules,

Thank you for a detailed response.

We are aware of the ADIsimDDS web tool.

It shows only PPT (Primary Phase Truncation) spurs.

I calculated the two PPT spurs for our setup (using an-1396, which discusses it).

Fppt1=~1.972GHz and Fppt2=~1.027GHz.

These are not the spurs we have.

I am looking for a way to predict the 2nd and 3rd order phase truncation spurs.

According to an-1396 they might be what we are looking for.

And another thing:

The 1M offset spur we see, changes power after every DAC-CAL (sometimes, go below -100dbm).

How can that be explained?

https://www.analog.com/en/resources/app-notes/an-1396.html

Thanks,

Gil

hello Jules,

Adding to my above reply:

Phase truncation occurs due to the truncation of the lower 15 bits of the phase accumulator in the AD9914.

This results in phase truncation spurs.

In my original message you can see the SA plot with Fo=563MHz.

Then we zeroed the lower 15 bits. Fo slightly changed.

However, the 8M and 1M spurs remain with same power and slightly different offset.

See SA picture:

Zeroing bit 16 also had no effect on the spurs.

According to phase truncation theory, the ~8M and ~1M spurs are NOT phase truncated spurs.

How do we proceed with finding the origin/cause of the spurs we see?

thanks,

Gil

hello Jules,

Adding to the above:

We identified the spurs also in Profile mode (no modulus).

Please advise.

thanks,

Gil

Hi gil_he ,

Apologies for the late response. Thank you for providing me your results.  Let me compile this and I'll find a way to give you some answers. 

All the best,

Jules

Hi gil_he ,

Apologies for the late response. I will try to make this on time as possible in the upcoming queries. 

To summarize your concerns above:

1. Asking how to calculate higher order truncation spurs (2nd and 3rd order).

- This was explained in the AN-1396 as you discovered. 

2. Why a 1 MHz spur changes strength every time the DAC is calibrated?

- This can happen because calibration adjusts internal DAC gain and matching. Small changes there can cause certain small spurs to move up or down in level each time calibration runs. 

I believe the 552 MHz is a 1st order PPT spur (Fout-1 MHz) and the 571 MHz is a higher order PPT spur (Fout+8 MHz). 

I have other suggestions on things might be improving these:

1. Have you tried to turn on Phase Dither? This adds a little randomness, so the error does not repeat, and the spur disappears into a noise. 

2. Can you try to use not rounded up number? Change the frequency by even 1 Hz? 

All the best,

Jules

Hi Jules,

Thank you for your answers.

My comments:

1. AN-1396 explains how to calculate the largest first order spur, which is the PPT spur.

It does not describe a method to calculate 2nd order PT spurs (result of the DAC harmonics) and the 3rd order PT spurs (result of the quantization spurs associated with the angle to amplitude converter and the DAC).

2. For Fo=563MHz calculation and ADIsimDDS calculates them to be Fppt1=~1.972GHz and Fppt2=~1.027GHz.

I wrote that I performed a test to confirm the spurs we see (~8M, ~1M) are indeed PT spurs, by loading FTW that has its lower 15 bits equal to zero. According to theory this should eliminate any PT spurs, since the 17bits the the angle to amplitude converter has not error with regard to the phase accumulator output. The spurs are still there thus we assume these are not PT spurs.

3. I am not able to find Phase Dither in AD9914.

Is it a hidden option in one of the control registers?

4. I can create a frequency with FTW (profile mode) and increase the LSB by one, thus, with our 3GHz Fs, it will increase Fo by 0.7Hz.

Do you mean input a not rounded up FTW (not a power of 2)?

Wouldn't that introduce more spurs into the output? (since the phase accumulator will zero at different times).

thanks,

Gil

Hi gil_he ,

Just wanted to address your concerns:

1. I was trying to look for some old files relating to finding how to predict 2nd and 3rd order PT spurs but I can't seem to find one. I will continue looking and will update on this thread incase I find one.

2. I tried to zero out the right most bits as shown below to get 562.5 MHz: 

And based on the spectrum I got, it seems like it is gone as shown:

3. Apologies I may have overlooked the Phase Dithering part. I was referring to AD9954 not AD9914. Apologies for that and yes, you're right AD9914 does not have Phase Dither. 

Just let me know if this works for you or we can try other things. 

All the best,

Jules