0
\$\begingroup\$

I am wanting to use the AD9850 for a particular application to generate sine wave signals in the 100Hz to 5kHz range. I've purchased a AD9850 module from ebay which just looks like the standard ones that are floating around at the moment. I noticed that I get quite significant harmonics even at these low frequencies. I'm not sure how to design a filter for this because ideally I would just filter off everything above 5kHz but I've seen that when I set the AD9850 to 1KHz I get harmonics at 2 and 3KHz which are noticeable. So it seems like I would need to have a variable filter that gets rid of harmonics above whatever frequency the AD9850 is set at. From what I've read about this particular DDS chip I would not have expected harmonics at these frequencies but probably would expect it as the frequency gets into the MHz range.

Can any one offer any advice on this problem or suggest any simple modifications that could be made to the AD9850 module board to reduce, or ideally, get rid of these harmonics?

Thanks! Magnitude plot of AD9850 output dB plot of AD9850 output

\$\endgroup\$
3
  • \$\begingroup\$ What specification does the ebay bought item adhere to or, as I suspect, is my question a waste of time? Is it also a waste of time to expect a schematic for that particular design? Here's my tip when buying electronic bits - don't (ever) buy something that does not have a decently written data sheet or comes from a supplier that has no guarantee of quality. It's a simple rule but actually works out to be more cost effective than (say) just buying some unspecified pile of s**t from ebay. \$\endgroup\$
    – Andy aka
    Commented May 23, 2017 at 12:53
  • 2
    \$\begingroup\$ How many dB down are the 2k and 3K energy levels? \$\endgroup\$
    – analogsystemsrf
    Commented May 23, 2017 at 13:06
  • \$\begingroup\$ I've added some plots showing the harmonic content of the signal when set to 1kHz and the link to the schematic is qrp.hu/wp-content/uploads/2015/01/ad9850modulsch.jpg \$\endgroup\$
    – CakeOrDeath
    Commented May 23, 2017 at 14:20

1 Answer 1

Reset to default
0
\$\begingroup\$

By the looks of the datasheet from ADI, the harmonics are intrinsic to the part (and not the board). Mulling over the datasheet it looks like this DDS was never intended to be used for low frequency applications, and you use a high pass filter and only use the upper frequencies above 5kHz.

You might be able to knock some of this out if you used one or more digital notch filters to the input signal.

\$\endgroup\$
3
  • \$\begingroup\$ Thanks for your comment. I had knew that there were some issues with harmonics and distortions with this DDS but I thought they would be at much higher frequencies (getting close to the clock frequency). Can you recommend any webpages or sources for reading more into your suggestion of the digital notch filters? \$\endgroup\$
    – CakeOrDeath
    Commented May 24, 2017 at 15:39
  • \$\begingroup\$ Maybe this one. FIRs are good, you can generate the constants from online calculators or other code and then implement them on a processor (keep in mind the processor needs to be able to handle the additional multiplies and adds of the filter to run real time). I use matlab to generate my constants but that costs money. \$\endgroup\$
    – Voltage Spike ♦
    Commented May 24, 2017 at 16:30
  • \$\begingroup\$ Thanks again for the suggestions. I had come across FIR filters but wasn't sure how I would implement them. I'll have a look over the link you sent and see if I can get an idea figured out. I have access to Matlab so I should be able to generate constants that way. Are there any processors that you would recommend? \$\endgroup\$
    – CakeOrDeath
    Commented May 26, 2017 at 12:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.