After
@nctnico's review of the Tektronix AFG31022, I just had to have a more detailed observation of my "DG811+++'s" inter-channel phase accuracy behavior. So I hooked up both channels of the AWG to my 53310A modulation domain analyzer, configured as a high-speed timer. Channel 1 would start the measurement, Channel 2 would stop it.
Since I wanted to analyze a possible "phase drift" at whole-number multiples of frequencies between the two channels, I always configured channel 2 to output a whole number multiple frequency of channel 1. I experimented in "couple mode" with both frequency and phase coupling enabled or disabled with little observable effect.
The result was disappointing: depending on the frequency pairs, I can find a phase drift, regardless of the coupling parameter, at factors sometimes as low as ten between the frequencies. I didn't dial in ultra-odd combinations. 60kHz and 1.2MHz for example walks away at approx. 100ps per minute. Other combinations are drifting even faster.
Seems like Rigol didn't get the phase accumulator math exactly right. I'ld expect that the "phase coupling" has some real effect, even if the frequencies are not identical but whole number multiples. Otherwise, the "frequency copupling" should work accurately to multiples of at least 1000.
IMO, the DG811 "hacker's delight"
is still a good AWG for its price, but never ever pay the price for a higher model in the range, especially not the DG900 or DG2000 series. For that kind of money, the performance will be disappointing, and there are better choices available.
I'm quite curious if we'll eventually see a firmware update that takes care of this... IMO it would be sufficient to reset (re-align) the phase automatically each time when the multiple of periods has passed. The additional phase noise in the faster channel should be neglible. It may not be necessary to modify the math routines that probably, due to rounding inaccuracies, cause the phase accumulators to drift.
Home
Help
Search
Login
Register
