Got my RSP2 today! First impressions:

[13dka]

I got my RSP2 this afternoon, woohoo! I was about time! Gladly I had the latest API etc. installed, so I just had to plug it in and put it to a test. But first I had to check the calibration, so I tuned to RWM 14996kHz - just in time to see the predicted coronal hole high speed stream making RWM and the rest of the upper HF portion fade out for the rest of the day. I swear that this happens every single time I get a new radio, but this time I got to watch it with my eyes!

VHF problem zones, e.g. 300-400MHz

I'm very glad to report that the massive problems with FM, pager and TETRA images are completely gone here - without using the FM bandstop filter! This proves that the tales of improved input bandpass filters are true, the obvious increase of passive components on the PCB was not in vain. While I was waiting for the postman to bring the RSP2, I jotted down several image-infested frequencies received on the RSP1 between 30 and 500MHz and compared them all with the RSP2, and it turned out they're all gone. This is smooth sailing on the VHF and UHF waves now, except...

...it took me all of 15 minutes (including unboxing) to to find a strange glitch when quickly tuning across 400MHz:

RSP2-Glitch3.gif (140.98 KiB) Viewed 40720 times

When you go fast enough over the ~400MHz mark, you may end up with an odd 20dBm dent in the spectrum that stays put. Doesn't always happen though, and you can undo that by doing the same tuning motion with the mousewheel again, or just stopping and restarting the radio. It's as if something is not switching reliably when you tune quickly, or in big (10MHz) steps. Another thing is the tuner AGC, now called "IF AGC", which still appears strange to me as it makes the spectrum levels jump around in very irritating ways, as soon as any kind of sizeable amplitude-changing signal is within the IF passband. Whenever you play with the gain reduction (or gain) level, the AGC will react in unquantified (or at least non-obvious) amounts of additional gain changes (when the LNA gain goes up, the IF gain goes down and vice versa) and I can't tell what the ratio of LNA gain vs. IF gain is. Even if you turn the IF AGC off, you get a slider with a scale but no numbers, but at least the pumping stops. I should also mention that I noticed some minor crosstalk between the antenna ports, but I can't think of a scenario where this would cause an issue.

Anyway, on VHF/UHF the RSP2 seems to be what the RSP1 only promised in its specs, or in other words, the RSP2 actually is what I thought the RSP1 would be and I think the only thing that's keeping it from showing a certain <cough> competitor product who's the boss, is the software. The hardware now fully supports clean and happy wideband scanning (without the scanner stopping on a lot of muck channels), proper range monitoring, milsat spotting...awesome!

VLF

Reception below 500kHz was difficult with the RSP1 due to the images, but once I found a baseband (IF) setting that fitted my station of interest, the signals were good. On port A of the RSP2, attempts on receiving VLF are entirely pointless with my dipole - images are at least as bad as with the RSP1, but the wanted signals are attenuated to the point that even the strongest signals (e.g. 147kHz) are almost inaudible now, the BC bandstop filter does not help with this either, since many (if not most) of the images are coming from the HF side.

These images are also present in the AM BC band, which becomes very obvious when you turn on the AM bandstop filter, the stations are gone, leaving all the images clearly visible. Two hours ago I was listening to Absolute Radio 1197kHz with some nice CW overlay - multimedia! I think it wasn't that bad with the RSP1.

Using the Hi-Z input with the 50 Ohms coax (had to whip up some adapter quickly), the whole affair is not only back to the behavior (images) and performance of the RSP1 - signals are much better and images are a tad less obnoxious on both MW and LW, and I found a nice clean chunk of spectrum down between 0 and 10 kHz for some "natural radio" experiments I hoped the RSP2 would make possible. However, VLF and 160m are the remaining areas where images are still causing me a few headaches.

So while I can understand why everyone is saying that it's a pity that the BC bandstop filter can't be used with the Hi-Z input, I just learned that it wouldn't help that much as several of the images come from around 3Mhz and so an additional lowpass (or a narrowband loop/antenna for LW) Seems to be the only way to get a clean, easy-to-browse VLF spectrum.

HF

Due to today's solar influence kicking the MUF way down, I had no chance to make meaningful comparisons on HF yet. On 80 and 40m I could not spot any obvious difference in performance though, NY Radio/Gander VOLMET on 6604 turned into a true grass roots station while I was preparing for switching radios, by the time I had connected the other radio it disappeared increasingly into the noise, however I did not notice anything bad at least, for some moments I had the impression that the RSP2 is a tad less noisy, but I needs much more time to say anything conclusive about HF on the RSP2.

Preliminary verdict:

No buyer's remorse triggered so far. A few quirks remain (brilliant with a few quirks, isn't that the recipe for a true British product? ) but I'm glad that it seems to be not only a good (like the RSP1) but also an interference-free performer on VHF/UHF now, and I'm pretty sure it'll do at least just as fine on shortwave. The 3 antenna inputs and the bias-T are the icing on the cake, making it a good upgrade also for HF-only people. Once calibrated, the RSP1's oscillator had no tendency to wander around a lot, so I could've done without the TCXO, however the clock in/outputs are inviting me a lot to make some passive radar experiments, which of course means buying a second RSP2! Ka-ching!

[13dka]

Unfortunately it looks like I spoke too soon about HF, something seems to be amiss about the LNA and gain:

By accident I tuned to 19450kHz and wondered about signals up there, the upper bands are dead tonight. Turned out that was all images from CRI mixed with R.Romania (likely from 49m and/or 41m, outside of the 12-30MHz bandpass), these nasty stations can peak at quite incredible levels at night, on multiple bands at once*. So I increased 'RF Gain Reduction' to the max., thinking this will do the same as "turning off" the LNA on the RSP1 (which is only reducing its gain to the minimum). But apparently it didn't, even when I turned off IF AGC and increased the overall gain reduction to the 123dB maximum, the images only got weaker. Same effect on all antenna ports.

I quickly hooked up the RSP1 and sure enough, turning off the LNA removed the images completely and I could decrease GR to the 43dB maximum, allowing normal reception on the frequency. Needless to say that the IF scheme was LIF with 8MHz and decimation 4 in both cases, and switching to narrower basebands did not help the RSP2 either.

The levels dropped after some minutes but still some images were creeping up every now and then at 123dB GR. The image doesn't give it away much because the gain is turned down so ultimately but I can assure you that it was way more audible than it looks.

RSP2-123dBGR.jpg (19.99 KiB) Viewed 40529 times

Increasing IF gain to any reasonable level above that brought back the heavy stuff. I switched radios forth and back several times and If I had the RSP1 running with the LNA off and 43dB IF GR , no images ever came up unless I turned on the LNA, while the RSP2 was set to be as sensitive as an old cheese sandwich of similar size and it still produced images. What am I missing? Not sure what to think, this looks like the RSP2 is less immune against strong signals than its predecessor?

EDIT:

After further examination of the problem this late afternoon it looks much like this issue is confined to around 19MHz (worst) and, to a lesser degree, several places between 16 and 25MHz. To reiterate, the issue is not that there are images, the issue is that the RSP1 can always deal with them by turning the LNA off, and the RSP2 can't in some cases, and hence also exhibits imaging where the RSP1 does not (unless it's being forced by raising gain much).

Here's an example at 16,955 kHz, empty air at that time. You see the RSP1 with its LNA even on and typical 50dBGR - no unwanted signals, below the RSP2 with the LNA off (RF GR to the max) and IF GR to some level that creates comparable reception:

RSP1_LNAon-vs-RPS2_LNAoff-16955.jpg (97.9 KiB) Viewed 40529 times

Yesterday, I could not see a difference in behavior between the Hi-Z and the SMA coax inputs. Then I connected a "reverse balun" to the balanced input terminal, and noticed an improvement on VLF. Today I compared the overloading behavior between the SMA port A and the balun"d Hi-Z input, here's the surprising result:

RSP2-16955-hiZ-53dBGR-fixedIFgain-vs-portA-50dBGR-fixedIFgain.jpg (32.91 KiB) Viewed 40529 times

At the bottom of the waterfall I had the Hi-Z port fed, then I switched to Port A, both were set to IF AGC off, maintaining an IF gain (reduction) setting in the middle of the slider, then compensating with the RF GR slider to get roughly the same amount of gain. You can see that something is badly distorted behind the SMA port, but given the moderate gain settings and the ADC not messaging overload in the SDRuno main window I have no idea what that could be.

Considering that the the gain/GR range is quite different on the Hi-Z port than on the SMA ports (and much more similar to the RSP1's figures), I came to think that the internal gain structure/gain staging might be quite different for the latter too. I further suspect that the SMA 50 Ohms inputs are more vulnerable to impedance mismatches for some (maybe the same) reason, and that using the reverse balun on the Hi-Z input changed the impedance mismatch map for the entire spectrum favorably. I'm sure the SDRplay guys can shed some light on the question why the 50 Ohms inputs behave so differently from the one found on the RSP1.

Tl;dr bottom line: At the latest if you're experiencing HF imaging on the SMA port, try the Hi-Z port instead!

[13dka]

You know you're getting old when you get a new radio and don't pull an all-nighter

When reading Jon's writeup about the Hi-Z port, I remembered that I had another of those NooElec baluns (isn't that an unun?) floating around and tried the "reverse balun" trick - with the result that nothing showed up (at least for a while) on LF that wouldn't belong there this morning:

RSP2-LW-Reverse-Balun.jpg (29.47 KiB) Viewed 40677 times

The balun can be connected directly to the port with 2 short pieces of heavy gauge wire, which is much nicer than the improvised pigtail with the SMA jack dangling from the RSP. I'm looking forward to see what influence this has on the higher bands. I wish I had another of those green terminal plug doodahs, I think I might be literally screwing around a lot with that poor thing.

EDIT:

Since I can't post the original post (the old "poll" error of the forum) I continue my prattling here:

HF:

I did more comparisons on HF over the day, and I found that I can't find a difference in reception quality, levels, or SNR between the models, if there is any it's possibly not relevant, or decisive when it comes to hearing or not hearing a station buried in the noise. Apart from the issue mentioned in the previous post, there is generally less muck flying over the screen with the RSP2 when I use the Hi-Z port with the reverse balun. Since the SMA ports seem to be more vulnerable to overloading (whatever is being overloaded there) compared to the single port on the RSP1, I join the crowd recommending using the Hi-Z port on HF by all means. There is no sensitivity penalty, and there might be a considerable off-band rejection plus with the odd green thingamajig.

(BC)FM

Comparing relatively stable low-power grassroots stations, there are no sensitivity differences in the FM broadcast range between the two either. Since I'm very happy about the performance of the RSP1 on FM, I feared that the successor would be worse, and "just as good" is all I wanted - there's no features-for-performance trade-in or something going on. Excelllent!

[g1hbe]

I use the Hi-Z port exclusively for LF/MF and HF. It works brilliantly with my Wellbrook loop. I help things along with a homebrew filter bank which gets rid of the remaining images.
I use the two SMA ports for VHF and UHF. I've noticed an irritating glitch which may be connected to the other strange goings-on mentioned here. When listening in the 200 to 400 MHz range, it is often necessary to change the LO plan to get around the images. But after switching, the gain seems to be at rock bottom and there are no signals present. The gain slider looks to be in it's normal position, but a quick nudge of it (up or down) will restore operation.
I've also had it happen when changing sampling rate. It's not a great problem, but I'm sure a lot of novices may get thoroughly frustrated with their 'deaf' RSP2!
I've reported this to the support e-mail address, but heard nothing so far.

[Tech_Support]

This issue with the gain being incorrectly set after changing the 1st LO and some other parameters such as the sampling rate is a bug. It should be fixed in the next release of SDRuno. Many thanks for flagging this.

Sincerely

SDRplay Tech_Support

[K0OD]

Nice work, 13dka. I'd like to see more side by side performance comparisons. The VLF performance of my RSP2 was disappointing at first brief glance.

I'm also wondering what's in store for the RSP2 when it has so many more components than the RSP1 as seen in this photo.

[g1hbe]

Glad to hear it's known. Looking forward to a fix. Thanks.

[F1BJB]

however the clock in/outputs are inviting me a lot to make some passive radar experiments, which of course means buying a second RSP2! Ka-ching!

Hi
Thanks for this nice report.
On that subject is there a software possibility to tune 2 RSP2 exactly the same way ?
That could be useful.
Sharing the output of a single RSP between a master and a slave software would be nice too.
I found by accident that sound cards based SDR's can do that in Windows.

[mikael]

I'm also wondering what's in store for the RSP2 when it has so many more components than the RSP1 as seen in this photo.

Seems they had to add an extra micro controller, presumable due to the fact that the GPIO of the Mirics chipset wasn't enough for the additional preselector filters, new LNA etc. *guessing*

[13dka]

I edited and completed some of my posts above over the day. Today I tried to compare the two RSP's susceptibility to imaging. To do this, I used a wide 8MHz bandwidth and a common start frequency, then I used the mouse wheel to sweep the frequency upwards. Anything not having the same slant as the other signals is an image. Images come in different flavors though, and this method does not show the ultra-fast little whistlers or the "classic" images keeping their distance to the source (thus behaving like a normal station when you tune past them). Since the start frequencies are in the middle of a wide spectrum, they also don't show the mayhem you'd see when you're tuning across e.g. VLF.

The receivers were set to show comparable gain while avoiding overloading, I used the Hi-Z port on the RSP2. Let's start from the top:

25.5MHz

RSP1-vs-RSP2-25500k-8MHzBW.jpg (113.91 KiB) Viewed 40427 times

Only a few things run across the screen here. Since the band is empty, your reference is the line coming out of the 24MHz mark (the internal clock of the RSPx I suppose). Clear winner is the RSP2.

16MHz

RSP1-vs-RSP2-16000k-8MHzBW.jpg (134.79 KiB) Viewed 40427 times

A bunch of very weak looking streaks are showing up on the RSP2 but not on the RSP1. I wrote "weak looking" because the screenshots don't really correspond with the actual audibility, or potential to annoy the listener. So the winner is the RSP1.

3.5MHz

RSP1-vs-RSP2-3500k-8MHzBW.jpg (205.22 KiB) Viewed 40427 times

The RSP2 clearly has an edge over the RSP1 here. If you look at the signals, you may get the impression that the RSP2 shows more signals and less noise. Since the forum allows only 3 pics per posting, I'll elaborate on that in the next one.