Attached is the schematic of my implementation of DDS - AD9951
I am using AD9951 - DDS and I have programmed it through SPI Interface successfully and I have been able to generate frequency of the range of 1MHz to 100MHz. I am observing the output using DSO - 100MHz Aplab with 1GSPS.
What I find is instead of getting 0.5Vp-p output from DDS, I am getting 0.25Vp-p output at lower frequencies say upto 10MHz. Beyond this this voltage keeps on reducing drastically and drops to around 20mVp-p. For no frequencies I get 0.5Vp-p. I have removed R17 for this implementation and I am observing the output at the transformer output U4.6. All the readings are pertaining to U4.6.
I have configured ASF (Amplitude Scale Factor) = 3FFF. I am using RSET = 3.9K.
Please suggest me where are we going wrong.
Thank you all.
Thanks for the reply and your application notes. I have worked through the excel sheet and found that If I make Load Resistance = 220 ohm and the pull up resistors = 110 ohm each then using my 1:1 coil craft transformer, I can get 1Vp-p output from the transformer end. May be since the load value has changed, so I may need to change my Cauer Filter Design in this case (I have to!!!).
Thanks again a lot and I am still to test it on the hardware. Will come to you in case some discrepancy is there.
Thanks for letting me know these design basics in a great more detail to know my fault.
Not being funny, are you already allowing for the de-tuning efffect of the scope probe on your filters ? - to be rolling off at 10MHz sounds quite badly wrong. A standard scope probe looks like a capacitance and the inductance of the grounding tail in series, and is the worng thing to be using round a filter with small Ls and Cs in it.
Also what impedances is your filter designed for - there seem to be too many 50 ohm resistors in your design.
If you can make your scope present a 50 ohm input, and remove one of your series resistors, and solder a bit of coax to your test board such that centre is on the signal and outer is on the ground, then the 'scope is taking the place of the 50 ohm load. This needs thin coax ~ 1mm dia. and the shortest possible length of transition wire, as each mm in the path to the 'scope that is not true coax is a nanohenry or two of extra inductance.
Apologies if you have already been doing this.