DDS AWG Square Rise Time 2x Faster Than Built In Square Rise TimeI was working with an arbitrary waveform that has 10 cycles per period so that I could try higher output square wave frequencies on my SDG2042X generator. In looking at the Arbitrary Wave square output I note that the rise time is much faster than when using the built in Square wave shape. See the attached scope captures.
Test setup is generator direct to scope with 50 Ohm BNC cable with the scope set to 50 Ohm termination. The scope is an old (circa 1992) Tek TDS-420A which is specified with 200 MHz/1.75 nSec rise time bandwidth. So, these measurements are a little slow compared to actual.
The first one shows the built in square wave at 10 MHz. Measured rise time is just under 9 nSec. Using RSS to remove the scope 1.75 nSec risetime and the actual is probably more like 8.7 nSec. My memory of testing with a different 1 GHz bandwidth scope was more like 8.4 nSec. All good and as expected. The instrument's data sheet lists 9 nSec max which matches well.
The second scope picture shows the arbitrary square wave shape running in DDS Arb Mode at the same 10 MHz. Now the rise time is 4.6 nSec and again using RSS to remove the scope 1.75 nSec rise the actual is probably 4.3 nSec or about 2x faster. Also note this waveform has a little ringing and slight undershoot before settling.
The third scope picture shows the arbitrary square wave shape running in TrueArb Arb Mode at 1.5 MHz. Now the rise time is 8.8 nSec and again using RSS to remove the scope 1.75 nSec rise the actual is probably 8.6 nSec. Also note this waveform has no ringing and settles in fine like the built in square. I would say it is the same as the built in square mode except it is consistently slightly faster (8.8 for True Arb versus 9.0 for built in square).
My theory on why these results are different is that the 300 MHz update A/D 4x linear interpolator is turned on for the built in square and TrueArb modes but is
not turned on for the DDS arbitrary wave. With 4x linear interpolation the DAC output will have a linear slope with duration of 4/1.2GHz = 1/300MHz = 3.33 nSec while without the linear interpolation method turned on the DAC will try to output a step at the edge.
The output of the DAC is followed by a high order filter. With a 3.33 nSec linear slope at the edge the output looks good with no ringing. With a perfect step you can better see the dynamics of the filter.
Unfortunately I can not explain why the built in square and the TrueArb square are not exactly identical. Perhaps the lower frequency of the TrueArb measurement?
The net, net of this observation is that the usable upper frequency for DDS arbitrary square waveforms is higher than for the built in one by about 2x in frequency.
In my opinion the DAC interpolator should be on for arbitrary waves like it is for built in waves. However, for the TrueArb and for built in square modes I would have thought it would be best to be off. Perhaps better yet, the instrument should have a setting to turn the DAC linear interpolation mode on and off??
What do others think??
Siglent, can you help explain this behavior?