RSP2 LO 3rd harmonic spurious response only -20dBC

Post information or questions regarding SDRplay products here
oh1nd
Posts: 2
Joined: Wed Jun 28, 2017 7:32 am

RSP2 LO 3rd harmonic spurious response only -20dBC

Post by oh1nd » Wed Jun 28, 2017 8:08 am

This is incredible poor result.

I heard stations that for sure are not on 1.8 or 3.5MHz band. This did not depend on signal levels and therefore is not an overloading or nonlinearity issue.
So I did search frequencies that produre spurious responses. I used a professional HP signal generator and found out that RSP2 converts signals also with local oscillator 3rd harmonic. Amazing is that this spurious conversion is very effective and only 20dB below the wanted conversion.

This makes RSP2 useless.

Spurios responses are not specified. Is there any way to improve this (without external filters) ? Maybe this is "normal" performance for RSP2 ?? If so SDR radios are as good as conventional radios in year 1910.
Last edited by oh1nd on Thu Jan 01, 1970 12:00 am, edited 0 times in total.
Reason: No reason

OH2BUA
Posts: 83
Joined: Mon Aug 29, 2016 8:58 am

Re: RSP2 LO 3rd harmonic spurious response only -20dBC

Post by OH2BUA » Thu Jun 29, 2017 1:30 pm

Hi, I have experienced the similar, but haven't taken such an analytical approach on that as you've done.

Could you possibly share some more details, Sir?

- Did I calculate right, that when you are tuned into let's say 1.8MHz (Zero IF), you are also receiving signals which actually are at around 7.2MHz?

- This may be irrelevant, but did you happen to compare if there's any difference if you are using 50-ohms or Hi-Z antenna inputs?

73, Jukka oh2bua

Reason: No reason

g1hbe
Posts: 494
Joined: Sat Jan 17, 2015 3:28 pm
Location: Cheshire, UK

Re: RSP2 LO 3rd harmonic spurious response only -20dBC

Post by g1hbe » Thu Jun 29, 2017 7:13 pm

I think I've seen this documented somewhere. There are responses based on odd harmonics of the sampling frequency/LO. Just like the 'image' response on a conventional superhet receiver, there is no way to get rid of them without using front-end filtering. The RSP's have filters, but the lowest one covers DC up to 12 MHz, so it includes the unwanted responses when you are tuning the lower bands.
This exposes the dilemma facing all SDR designers - leave the front end open so that they can brag about a 10 MHz wide spectrum display or narrow the filters to get rid of the unwanted responses and lose the big waterfall?
I find a huge spectrum display a novelty but ultimately I always go back down to a more useful 200 KHz as it makes it easier to see what's what. With that in mind, I use homebrew narrow filters for my bands of interest. ATM I have a VLF filter (0 to 100 KHz), an LF filter (0 to 500 KHz), a MW bandpass filter (500 KHz to 1.7 MHz) and another one for the 160m band (1.8 to 2.0 MHz).
All these bands are now clean and very nice to use. For freqs above this (up to 30 MHz) I have a basic HPF that starts roughly at 2 MHz.

It's not 'SDR' radios that are the problem, it's the current fashion for spectrum displays. As well as the RSP2 I have a 'conventional, knob-tuned' SDR receiver, but as it does not drive a spectrum display it does not need to use wide filtering so it does not suffer spurii. IIRC, it uses track-tuned filters for HF and another 18 bandpass filters for VHF and UHF.

Reason: No reason
Andy

oh1nd
Posts: 2
Joined: Wed Jun 28, 2017 7:32 am

Re: RSP2 LO 3rd harmonic spurious response only -20dBC

Post by oh1nd » Thu Jun 29, 2017 7:46 pm

It looks like the IQ mixers are not analog multipliers like in old Motorola MC1496 circuit but instead IQ mixers are working in switched mode. So the LO is like a square wave. The spectrum of that includes odd harmonics and the only way to get rid of spurious conversion products is to filter input signals at 3rd harmonic with a low pass filter. This can be done ofcourse with external filters. For 1.8 ... 2MHz filter attenuation at about 3x1.8MHz = 5.4MHz should be 50dB. This figure is based on 70dB dynamic range and the measured -20dBC spurious conversion for 3rd harmonic.

RSP2 price tag is more than 200€. In my opinion this kind poor performance should be mentioned in specifications. And including few LP filters inside RSP would have not costed more than 1to 2£.
Last edited by oh1nd on Thu Jan 01, 1970 12:00 am, edited 0 times in total.
Reason: No reason

g1hbe
Posts: 494
Joined: Sat Jan 17, 2015 3:28 pm
Location: Cheshire, UK

Re: RSP2 LO 3rd harmonic spurious response only -20dBC

Post by g1hbe » Thu Jun 29, 2017 9:20 pm

Yes but then the waterfall/spectrum display would not show up to 10 MHz wide! You can't have it both ways, this is the problem.

Reason: No reason
Andy

DanubeBCL
Posts: 203
Joined: Sat Jul 02, 2016 12:39 pm

Re: RSP2 LO 3rd harmonic spurious response only -20dBC

Post by DanubeBCL » Fri Jun 30, 2017 6:44 am

I think what Juha wanted to say is that they shoud have provided low pass filters behind the LO.
My opinion: When someone expects high standards of a spectrum analyzer or monitoring receiver
shouldn't he go better for a Rhode & Schwarz FSW8/FSW13
https://www.rohde-schwarz.com/us/produc ... 11793.html
or an EB500-HF
https://www.rohde-schwarz.com/us/produc ... 11225.html
or similar equipment from other specialized manufacturers and not buy an RSP for 200 bucks. I personally would not expect ITU monitoring quality from an RSP.
73, Heinrich

Reason: No reason

OH2BUA
Posts: 83
Joined: Mon Aug 29, 2016 8:58 am

Re: RSP2 LO 3rd harmonic spurious response only -20dBC

Post by OH2BUA » Fri Jun 30, 2017 7:47 am

Andy, as you seem to be experienced with external filtering, can you possibly recommend a solution for me:

For listening MW and 160m I have a Double-KAZ antenna, with about 1 kohm impedance, symmetric. It's fed to the RSP2's Hi-Z port by using 12 meters of 450 ohms ladder line.

I'd like to add an external low-pass filter for cutting at let's say 1.9MHz. Preferably by using some off-the-shelf solution. But all the available gear (like Mini-Circuits stuff) seem to be for 50 ohms unbalanced.

Shall I make a chain of 9:1 balun - LPF - 1:9 balun? Sounds hmmm, non-kosher.

73, Jukka oh2bua

Reason: No reason

g1hbe
Posts: 494
Joined: Sat Jan 17, 2015 3:28 pm
Location: Cheshire, UK

Re: RSP2 LO 3rd harmonic spurious response only -20dBC

Post by g1hbe » Fri Jun 30, 2017 8:14 am

Jukka - if you want to use off-the-shelf filters, you will need to transform to 50 ohms on the input side and then back up to 1k to feed the RSP. Not ideal. A better solution would be to download an RF package (something like RFSim99) and use it to design your own filter for a 1000 ohm system. All you do is enter the filter type (Chebyshev or Butterworth) HP or LP, roll-off frequency and impedance and press GO. It'll give you a circuit diagram and will also simulate the response for you. Even I can use it!
I've just thought - your system is balanced, so either you'll need to do a balanced filter or transform to unbalanced on the input to the filter (use a 1:1 balun) and back to balanced on the output (another balun).
RFSim is quite old, but it still works, even on my W8 laptop.

Just one more thought re the spurious responses. If you think about a conventional receiver, you don't know about all the unwanted signals until you happen to tune across one. So you tend to think the receiver is virtually free of such things.
When you can see a wide chunk of spectrum, all the nasty stuff is laid out before you and you think the receiver is a sprog box! Good luck with it.

Reason: No reason
Andy

OH2BUA
Posts: 83
Joined: Mon Aug 29, 2016 8:58 am

Re: RSP2 LO 3rd harmonic spurious response only -20dBC

Post by OH2BUA » Mon Jul 03, 2017 10:25 am

Thanks Andy for your tip for RFSim99.

It's easy to get learned and fun to play with.
Jukan_1.png
Jukan_1.png (6.99 KiB) Viewed 25664 times
73, Jukka

Reason: No reason

F1BJB
Posts: 118
Joined: Wed Jul 15, 2015 4:24 pm
Location: Beauvais France

Re: RSP2 LO 3rd harmonic spurious response only -20dBC

Post by F1BJB » Wed Jul 05, 2017 2:33 pm

Hi
On HF bands RSP is using up conversion so the problem of third harmonic LO responses should not occur.
Or am I wrong somewhere ? :)

Reason: No reason

Post Reply