Re: my clock test above, can anyone else confirm that the phase noise on the output is horrible? I.e., scope both an output and the internal Ref out (from the back BNC), set the output to 10MHz and trigger off the rear 10MHz reference. I would expect to see both the ref and output signals sync'ed.
Yes there is a problem and it is more likely a software one. The 10Mhz generated by the standard square function as 2 problems:
- it is not accurate ( my MSO reads 9.99986Mhz)
- the duty is varying (ghosts traces on falling edges)
I did the test using TTsource and the ARB function with a "clean" 10Mhz square signal:
- my MSO reads 10.0000Mhz
- no more ghosts falling edges
It is tricky with TTsource to create such accurate files because the number of samples has to be a multiple of 16 (TTsource rounds it itself otherwise). So for 10Mhz you can't directly draw a cycle of 25 samples as it will round it to 32... you would need to have 400 samples at least (25x16) and draw 16 cycles.
At the moment I use Agilent BenchLink Waveform Builder to draw signals (f.e. 10 seconds to enter you want 16 cycles of a square wave on 400 samples) and export it, run a macro that converts the file to a .hwf file that I import in TTsource and download to the Hdg. Yes that's wired but than you have very acurate signals out of the HDG. Plus Waveform Builder is much more powerfull than TTsource.
Another problem of TTsource with arbs is that it always tries to have around 6000 samples when you enter a frequency (I suppose to reduce downloading times) and its choice of divider is not always the best one. So it is often better to directly chose yourself the divider and the number of samples that meets your needs.
Could you try the same tests with the attach files? (just unzip and use TTsource to import and download to the HDG)