AD9910: output circuit clarification

Document created by analog-archivist Employee on Feb 23, 2016Last modified by analog-archivist Employee on Feb 23, 2016
Version 2Show Document
  • View in full screen mode

About the output circuit of the AD9910 eval board , after a deeper analysis,
there are several inconsistencies and items that you may need to be aware of
and I thought you'd find this discussion interesting and useful. I've
summarized in the attached file.


1) 75 ohm transformer in a 50 ohm circuit: Here's the word from When you operate a 75 Ohm transformer in a 50 Ohm
system you should expect the operational bandwidth to be slightly shifted in
frequency. So if you are using the transformer in the specified central
frequency region, there should be no practical difference. We picked this part
because of its wide bandwidth (800MHz). Because it is not operated in a 75 ohm
system, we expect this BW to decrease somewhat (according to Mini-Circuits),
but it will still be far larger than all the Mini-Circuits 50 ohm transformers
we've found.

2) If you look at the transformer's datasheet (attached here), the primary has
no center-tap (i.e., pin 5 is not used) while the secondary does. Since we need
the DDS outputs to be biased, we operate the transformer "backwards" so that
the center-tap is on the DDS side of the circuit to provide this bias. That is,
we connect the transformer such that it's secondary windings act as the primary
and vica-versa. Because this is a 1:1 transformer, this can be done without any
strange effects. We grounded pin 5 because it shorts out a potential parasitic
signal input path and was easy to do.

3) With both sides of the DDS terminated in 50 ohms, the equivalent resistance
seen at the transformer input is 100 ohms. Since this is a 1:1 transformer,
this resistance is reflected to the output of the transformer. The
reconstruction filter sees a 100 ohm source resistance. However, the design of
filter assumed a 50 ohm source and load. Because of this, the frequency
response of the filter will not be what was intended. However, if you look at
the filter's simulated response (see attached Spice simulation), it hasn't
changed much. Compared to a 50 ohm source, the 100 ohm source allows a higher
output level (because the DDS is driving a higher impedance) but also has
slightly more ripple. Note that transformer bandwidth effects have not been
included nor has the ultimate attenuation limitations of PCB stray signal paths
& non-ideal grounding.
2 people found this helpful