Hi there,
wow, if I'd only had a look at this thread earlier ... I'm an EE student from Austria in his hopefully last year
To be independant from my universities lab, I made myself a present in february, a DS2202. My father wanted to have his old Tek 468 back one day anyway ...
Not knowing too much about the internal construction of the scopes, I got the 200MHz variant, since I started working on my master thesis involving deserialization of a 125Mbit/s stream (nope, not ethernet) travelling over 100m of a 75Ohm coax cable. I simply wanted to have a chance to look at the signals I work on in the time domain to see why it's not working
Oh, and by the way, the DS2202 does a pretty good job in this setup (using a BNC T-adaptor with 75Ohm termination on one side), just in case this helps someone.
If I'd only knew about the real bandwidth of the DS2000 series, I'd probably chosen the DS2072 and saved a few bucks. Though having the 2ns timebase 'natively' is not the worst that can happen. Since I got the DS2202 from Batronix in Germany, they were also the ones that supplied me with the 00.00.01.00.05 firmware update (mine came with 00.00.01.00.02), and a second trial license code after my "keen attempt to run the self cal" to minimize the offset on channel one during low amplitude measurements.
When I found out about the existence of the amazing RUU software, I was really overwhelmed, not just because I'm quite concerned about the durability of the USB port over time, it adds quite some value to the scope itself. And even if it takes longer to transfer the screen data over SCPI, I tend to network everything via Ethernet, and let scripts do the rest. Sooner or later it just makes sense.
Ok, but enough of that, I do have a question about the FFT under the math menu. I was playing around with an AES3 stream (very similar to S/PDIF, a digital audio stream) and tried to do some spectral analysis as I had EMI issues with a driver interface I built a few weeks ago. While debugging the interface using a DIY h-field probe, the FFT function was really useful. Of course it's no comparison to a real spectrum analyser, but if manage to get decent levels into the scope it seems to work pretty well below 500MHz when some averaging is applied before running the math.
The only thing that puzzles me, how on earth does Rigol calculate the spectrum between one and ten GHz from a 2GS/s stream? I thought the math function would derive its output only from the existing samples and I don't think anything in the DS2202 is going faster than 2GS/s. Since I have no possibilities to generate known spectra in the GHz range, I can't verify whether they just claim to be able to do so. Has anyone with the appropriate equipment ever verified whether the spectrum above 1GHz in FFT mode is just a mathematical interpolation residual? If so, why do they do it?
Greetings from Graz,
Daniel
PS: Let me second that comment of Jean about the noise coming from the cooling fan. Honestly, I can't understand why a system below 50 watts of thermal power dissipation needs a fan that loud. Of course, 'loud' is relative, as marmad already stated, but I believe they could have built something at least 10dB softer. At my desk the Rigol is by far the loudest thing, so probably the first modification will be a different fan