EMRFD Message Archive 9219

Message Date From Subject
9219 2013-10-06 07:23:09 Chris Howard w0ep questions about a DDS spectrum
Hi,

I recently came into possession of an HP 8566 spectrum analyzer
and I have been finding things here in the shack to
feed into it, both to try to figure out how to use it
and to get some idea of what signal sources I have and how
they work.

I have an FCC-2 signal generator which is a DDS-based
kit from NorCal (it is not the MKII model).

Naturally, I hooked up the FCC-2 to the SA and I've
been dialing it around.

I set up my span for about 5-6 MHz on the SA.
Down around 1 Mhz I can see the harmonics of the
main signal, about 50-60 db down. And they
march away to the right as I turn the tuning knob on
the FCC-1 to make the output freq go up.
They leave the scene at an exponential rate.

I continue to dial upward.

Then, when I'm tuning up around 11 MHz I start
to see a larger peak, nicely formed and about
45 Db down, larger than the harmonics,
which comes back from the right of the
display and then converge with the main
peak and pass on to the left.

As I go higher I see more of these, maybe 3 before
I hit the max freq output of my FCC-2, just above 21 MHz.

The delta between the main output frequency and these
peaks is not quite linear and it is inverse. I have one
converge at 12.40 MHz and another at 16.65MHz.
At the same time those larger peaks are coming
in from above, there is a cluster of smaller peaks
coming up from below at the same rate and they
all meet with the main output at those two frequencies
and maybe others higher on.

If that doesn't make sense, I will try to get a little
video of it.

I'd like to know what is creating these and how
to understand them.

Chris Howard
w0ep
9220 2013-10-06 10:14:28 kc7nj Re: questions about a DDS spectrum

Hello Chris, 

Welcome to the crazy world of DDS!

Dave

K2ZU



9221 2013-10-06 10:40:27 Thomas S. Knutsen Re: questions about a DDS spectrum
What you are seeing is the image products of the sampling process used in the DDS. 

Lets do some math together:
The DDS is clocked at Fclk 100MHz (this makes our calculation easy to do, your may have an different clock frequency.)
The DDS output signal Fout is at 20MHz.
The max frequency or Nyquist frequency of the DDS (Fclk / 2)  will for this purpose then be 50MHz.

Then since we are sampling, we have to have an product at Fclk - Fout or 100MHz - 20MHz = 80MHz.
The amplitude of this is given by the function Sinc (X) or in an different way: Sin(X) / x. That means that the clock frequency is in the first 0, so the amplitude will fall off towards 0 when we reaches that frequency.    
This is the first product that appear to be an inverse. 

Then, above the clock frequency Fclk there should be 2 products: Fclk + Fout and 2*Fclk - Fout, leading to the following products:  100MHz + 20MHz = 120MHz and 2*100MHz - 20MHz = 180MHz. 
The 2. zero of the sinc(X) function will then fall on 2*Fclk. 
Then there are products at 2*Fclk + Fout and 3*Fclk-Fout.


This math may seems a bit advanced at first, but work trough it and it should become clear.  This is also outlined in quite some more detail in the DDS datasheets, and Analog devices have an extencive set of articles avaible on their homepage explaining this. 

Good luck. 

73 de Thomas.





2013/10/6 Chris Howard w0ep <w0ep@w0ep.us>
 


Hi,

I recently came into possession of an HP 8566 spectrum analyzer
and I have been finding things here in the shack to
feed into it, both to try to figure out how to use it
and to get some idea of what signal sources I have and how
they work.

I have an FCC-2 signal generator which is a DDS-based
kit from NorCal (it is not the MKII model).

Naturally, I hooked up the FCC-2 to the SA and I've
been dialing it around.

I set up my span for about 5-6 MHz on the SA.
Down around 1 Mhz I can see the harmonics of the
main signal, about 50-60 db down. And they
march away to the right as I turn the tuning knob on
the FCC-1 to make the output freq go up.
They leave the scene at an exponential rate.

I continue to dial upward.

Then, when I'm tuning up around 11 MHz I start
to see a larger peak, nicely formed and about
45 Db down, larger than the harmonics,
which comes back from the right of the
display and then converge with the main
peak and pass on to the left.

As I go higher I see more of these, maybe 3 before
I hit the max freq output of my FCC-2, just above 21 MHz.

The delta between the main output frequency and these
peaks is not quite linear and it is inverse. I have one
converge at 12.40 MHz and another at 16.65MHz.
At the same time those larger peaks are coming
in from above, there is a cluster of smaller peaks
coming up from below at the same rate and they
all meet with the main output at those two frequencies
and maybe others higher on.

If that doesn't make sense, I will try to get a little
video of it.

I'd like to know what is creating these and how
to understand them.

Chris Howard
w0ep




--

 Please  avoid sending  me  Word  or  PowerPoint  attachments.
 See  <http://www.gnu.org/philosophy/no-word-attachments.html>
PDF is an better alternative and there are always LaTeX!
9223 2013-10-06 11:19:42 Chris Howard w0ep Re: questions about a DDS spectrum
Thanks Thomas and Dave!
I will work it through and see what I get.

It has been very interesting to watch this.

Chris




9224 2013-10-07 02:20:32 Nick Tsakonas Re: questions about a DDS spectrum
Hi Chris,

Thomas gave a perfect explanation.

If you are in the mood of reading a bit more pages without too many math, you can read the following paper from Analog Devices

http://www.analog.com/static/imported-files/tutorials/MT-085.pdf

I also suggest that you give a look
9225 2013-10-07 20:02:45 Chris Howard Re: questions about a DDS spectrum

I did some more careful measurements.

I think I figured out what I am looking at.

The FCC-2 has a 50MHz clock

Peak #1 F1 = F + (50- 4F)
Peak #2 F2 = F + (50 - 3F)

Peak #1 meets F at 12.5 MHz
Peak #2 meets F at 16.6 MHz

Chris

Chris Howard w0ep wrote:

 


Hi,

I recently came into possession of an HP 8566 spectrum analyzer
and I have been finding things here in the shack to
feed into it, both to try to figure out how to use it
and to get some idea of what signal sources I have and how
they work.

I have an FCC-2 signal generator which is a DDS-based
kit from NorCal (it is not the MKII model).

Naturally, I hooked up the FCC-2 to the SA and I've
been dialing it around.

I set up my span for about 5-6 MHz on the SA.
Down around 1 Mhz I can see the harmonics of the
main signal, about 50-60 db down. And they
march away to the right as I turn the tuning knob on
the FCC-1 to make the output freq go up.
They leave the scene at an exponential rate.

I continue to dial upward.

Then, when I'm tuning up around 11 MHz I start
to see a larger peak, nicely formed and about
45 Db down, larger than the harmonics,
which comes back from the right of the
display and then converge with the main
peak and pass on to the left.

As I go higher I see more of these, maybe 3 before
I hit the max freq output of my FCC-2, just above 21 MHz.

The delta between the main output frequency and these
peaks is not quite linear and it is inverse. I have one
converge at 12.40 MHz and another at 16.65MHz.
At the same time those larger peaks are coming
in from above, there is a cluster of smaller peaks
coming up from below at the same rate and they
all meet with the main output at those two frequencies
and maybe others higher on.

If that doesn't make sense, I will try to get a little
video of it.

I'd like to know what is creating these and how
to understand them.

Chris Howard
w0ep

9226 2013-10-08 04:58:40 f2dc2003 Re: questions about a DDS spectrum
9233 2013-10-11 07:25:43 ha5rxz Re: questions about a DDS spectrum
What you're seeing are spurs, the odd signals that used to be called 'birdies' in my youth. They're generated from combinations of your output frequency and the frequency of the clock driving the DDS chip. Rather than try and explain it all have a look at the following link:

http://designtools.analog.com/dtDDSWeb/dtDDSMain.aspx

Once the page has loaded look about half way down on the right, just underneath the display of your output waveform and you'll see another link, 'View Tabular Display of Spurs', click
9234 2013-10-11 08:06:24 Chris Howard w0ep Re: questions about a DDS spectrum
Peter,

Thanks for the link.

The peaks I was looking at appear as Fout is nearing Fclock/4
and Fclock/3. They are about -50 dBc.

My clock is 50MHz, device is AD9834.

I do see the same effect using that simulator.

Thanks for sending that!

Chris
w0ep