Spurs in Rigol RSA3000 and RSA5000 series in realtime mode

[Strix]

We got a Rigol RSA3030N with the 40 MHz RTSA option last November.  Pretty soon after, I noticed a ~4dB spur at CF-12MHz visible in RTSA mode, regardless of frequency range and span (wide enough to include it, of course), even with a 50Ω load terminating the input.  It's not there in GPSA mode.

I sent screenshots to the dealer, who eventually came back and said that Rigol would exchange the unit and throw in our option of choice for the bother.  After some time and occasional prompting from me, the dealer came back and said that Rigol couldn't find a unit that didn't have this spur, but that they they had a unit with the extra option if we wanted it.

That's nice of them and I can work around it, but I'd rather a unit that worked as designed, so I suggested that they replace it with the equivalent from the RSA5000 series instead.  Dealer came back to me and said they likely have the same problem.

What gives?  After rummaging through EEVblog, I understand that Rigol and others in this space are known to have their wrinkles, but they're subtle ones like the DHO series' PLL issue.  This one stands out bold as brass because it's always there and, even if it's no R&S, £6.5k worth of kit (scope+option) should do better than this.

I've found no mention of such spurs either in the 2020 thread (other than after upgrading PSU components) or anywhere else online.  Has anyone with an RSA scope and either the B20 or B40 option seen anything like this?  Can it be the case that it is a systematic problem with the platform, is it something that crept in after it was first released or is it just my unit (or batch of units)?

Hardware v00.01.03, FW CPU v00.01.00, SPU v00.01.06 and WPU v00.01.02, SW "firmware" v00.03.06.

Cheers for any light shed.

[G0HZU]

The same -12MHz spur is just visible in the youtube video below.
The RBW is higher at 200kHz and this masks it to some degree. But it is still there.




The  noise level across the 40MHz span shows quite a positive slope with increasing frequency, both in your screenshot and in the video. It appears to be about 4dB. Is this there all the time as well?

Is it possible to play with the ADC settings in the menus to try and reduce the spur term? Maybe a different ADC gain setting will help here?

As a comparison, I've got an old Tek RTSA here that can do up to a 110MHz real time bandwidth. I've set it to 40MHz and copied your  RBW setting of 100kHz and used 10dB attenuation and put it into Advanced DPX mode. The response isn't that flat but it is spur free. I also tried turning the RBW down to 10kHz and it is still spur free. However, the datasheet suggests that it won't always be spur free like this. I could probably hunt around and find some.

Does the Rigol let you play around with the ADC gain settings?

[Strix]

The same -12MHz spur is just visible in the youtube video below.
The RBW is higher at 200kHz and this masks it to some degree. But it is still there.

So it is.  I don't think it would have occurred to me that that was a spur rather than some random bit of external noise.  Not at a casual glance, anyway.  Well, that rather does suggest it's a systematic thing.

The  noise level across the 40MHz span shows quite a positive slope with increasing frequency, both in your screenshot and in the video. It appears to be about 4dB. Is this there all the time as well?

Is it possible to play with the ADC settings in the menus to try and reduce the spur term? Maybe a different ADC gain setting will help here?

Yes, there's a fairly consistent ~2dB from left to right edge (tested at 140 MHz, 1.5 GHz and 1.9 GHz).  Turning on the power amp reduces that to about 1dB, but there's still a small positive slope.

I haven't found a way to mess with ADC parameters.  The closest I've seen is the maximum mixer level, but that's a bit different.

Thanks for having a look with your Tek RTSA!  Alas, in RTSA mode, the minimum RBW of this instrument is a function of span.  It can do down to 1Hz RBW, but only in GPSA mode.

[scottapotamas]

I see the same general slope across the span, and the -12MHz off-center spur.

RSA3015N, about 2 years old.
HW00.01.03
FW CPU 00.01.00 SPU 00.01.06 WPU 00.01.02
SW Firmware 00.03.03

I've attached some screenshots from my unit. I terminated the RF input with 50R after a N-SMA adapter.
One screenshot tried to be as similar to your screenshot as possible, then some others with 500M and 1G center frequencies, all 40M span and 100kRBW

[Strix]

Thank you so much for taking the time to confirm/replicate the condition, and apologies for the delay in getting back to you.  Things got a bit frantic after our week at the EMC lab, as you might imagine.

I see the same general slope across the span, and the -12MHz off-center spur.

I've attached some screenshots from my unit. I terminated the RF input with 50R after a N-SMA adapter.
One screenshot tried to be as similar to your screenshot as possible, then some others with 500M and 1G center frequencies, all 40M span and 100kRBW

Well, that's, um, disappointing and not just a little surprising.  Between your efforts and what I gather from the dealer, it does, indeed, appear to be a systematic problem with the design.  What were Rigol thinking, and how did it take over six years before the problem was noticed (or, at least, commented on it publicly)?

I can work around it, now that I know it's there, but it's a big gotchya that could lead to some seriously misleading results if one weren't aware.  It's a pretty solid swept analyser, but my impression is that the realtime mode is a bit gimmicky, other than in very specific other circumstances. 

Matters are not helped by the limited range of choices of RBW.  100kHz is the lowest it goes at 40 MHz span, which is only just good enough to approximate the CISPR/FCC-stipulated RBW of 120kHz for emissions > 30 MHz.  To be fair, though, RTSA does help in EMI mode with the aid of software like EMCview Pro, at least in conducted emissions (which is the only thing I've made any serious attempt at replicating being that I don't have a TEM cell nor anything remotely like an RF-quiet room!)

What's your experience, have you found RTSA mode to be particularly useful?

[scottapotamas]

Heh, I noticed it pretty much instantly when I installed the BW40 option, I just assumed it was "normal Rigol behaviour" and was thankful the option was included.

My most common use for RTSA has been getting a general feel of a signal's behaviour in time-domain, or quickly checking the spectrum for unexpected noise.
It did get used a bit in the public writeup for some tests I was doing around the start of the year
, but if I'm measuring anything I'm normally in a swept mode.

I believe/hope the VNA and EMI modes are exclusively swept and therefore unaffected. I suspect I'll buy a SDR if I need to do anything meaningful with RTSA, most have wider spans and better software. 

How is EMCview with this hardware? I have a dual-channel LISN, and a slowly accumulating pile of Tekbox gear (near field probes and TEM cell) but have just used integrated EMI tooling and limit lines so far...
There's barely any information online about these kinds of precompliance setups once you filter out sales material.