RSP2 frequency calibration

General discussions
Post Reply
PeterB
Posts: 3
Joined: Sun Nov 27, 2016 12:51 pm

RSP2 frequency calibration

Post by PeterB » Sun Nov 27, 2016 1:17 pm

Newbie SDRplay and SDRuno user here - although I do have a fair amount of SDR experience with dongles, Airspy, SDR-IQ etc.

A question with preamble ...

When using SDRuno CAL function to calibrate the TCXO to a reference frequency, it seems that there are large discrepancies depending on what sample rate and decimation I choose

For example, tuning with a sample rate of 2 MHz and decimation of 4 gave me a final CAL adjustment of -0.072 ppm
but tuning to the same marker with rate of 4 MHz and decimation 16 gave a CAL adjustment figure of +25.875

So, the question ( in 2 parts )

a) is this normal and/or expected operation for either RSP2 or SDRuno ( or both ) ?
b) assuming that the answer to a) is Yes ... is there any method of setting the correct correction automatically based on sample rate and decimation
or do I have to do an adjustment every time for every combination ?

I fear that the 2nd option could get pretty ordinary pretty quickly.

Thanks for reading, look forward to any responses.
Pete.
Last edited by PeterB on Thu Jan 01, 1970 12:00 am, edited 0 times in total.
Reason: No reason

Tech_Support
Posts: 499
Joined: Mon Jun 01, 2015 7:00 pm

Re: RSP2 frequency calibration

Post by Tech_Support » Sun Nov 27, 2016 5:01 pm

Hello Pete,
What are you using as a reference source?
Can you confirm that the SAM demodulator has correctly locked to your reference signal before you attempt the auto calibration? The signal needs to be within a few kHz of your VFO frequency and when locked, you will see a clean tone centred on 0 in your SP2 window.

Sincerely

SDRplay Tech_Support

Reason: No reason

PeterB
Posts: 3
Joined: Sun Nov 27, 2016 12:51 pm

Re: RSP2 frequency calibration

Post by PeterB » Sun Nov 27, 2016 6:45 pm

Hi T_S,

My original testing was done using WWV .

I've repeated the tests using a GPSDO ( Leo Bodnar unit ) :

test frequency 12.250 MHz :

Sample Rate 4 MHz Decimation 16 ( Final SR 250000 ) : + 19.650 ppm
Sample Rate 2 MHz Decimation 8 ( Final SR 250000 ) : - 0.090 ppm
Sample Rate 8 MHz Decimation 32 ( Final SR 250000 ) : - 0.088 ppm

test frequency 125.000 MHz :

Sample Rate 4 MHz Decimation 16 ( Final SR 250000 ) : + 1.828 ppm
Sample Rate 2 MHz Decimation 8 ( Final SR 250000 ) : - 0.113 ppm
Sample Rate 8 MHz Decimation 32 ( Final SR 250000 ) : - 0.110 ppm

test frequency 450.000 MHz :

Sample Rate 4 MHz Decimation 16 ( Final SR 250000 ) : + 0.427 ppm
Sample Rate 2 MHz Decimation 8 ( Final SR 250000 ) : - 0.111 ppm
Sample Rate 8 MHz Decimation 32 ( Final SR 250000 ) : - 0.109 ppm

So, it seems ( at least in these tests ) that the combination of 4 MHz SR and decimation of 16 is the one that gives the biggest discrepancy.
Obviously, this isn't an exhaustive test of every combination of SR and Decimation, but it looks indicative of something odd going on.

Other than that, the ppm corrections all look reasonable - at least to my untutored eye.

Comments Welcome.

Regards,
Pete.
Last edited by PeterB on Thu Jan 01, 1970 12:00 am, edited 0 times in total.
Reason: No reason

SDRplay_Tech_Support
Posts: 1
Joined: Tue Nov 29, 2016 11:35 am

Re: RSP2 frequency calibration

Post by SDRplay_Tech_Support » Tue Nov 29, 2016 11:54 am

Hi Pete,

We have taken a look at this but been unable to replicate your problem. Our findings are shown below

test frequency 12.250 MHz :

Sample Rate 4 MHz Decimation 16 ( Final SR 250000 ) : -0.656 ppm
Sample Rate 2 MHz Decimation 8 ( Final SR 250000 ) : - 0.661 ppm
Sample Rate 8 MHz Decimation 32 ( Final SR 250000 ) : -0.584 ppm

test frequency 125.000 MHz :

Sample Rate 4 MHz Decimation 16 ( Final SR 250000 ) : -0.646 ppm
Sample Rate 2 MHz Decimation 8 ( Final SR 250000 ) : - -0.54 ppm
Sample Rate 8 MHz Decimation 32 ( Final SR 250000 ) : - -0.537 ppm

test frequency 450.000 MHz :

Sample Rate 4 MHz Decimation 16 ( Final SR 250000 ) : -0.544 ppm
Sample Rate 2 MHz Decimation 8 ( Final SR 250000 ) : -0.531 ppm
Sample Rate 8 MHz Decimation 32 ( Final SR 250000 ) : -0.538 ppm

These results were taken using a signal generator with its external reference connected to a GPS precision reference. In our case the VLO = VFO and the gain set accordingly in each case so as not to overload. The signal was set to be within the capture window of the SAM demodulator, which is +/-500Hz (this means the frequency error in the Rx control window needs to be less than 500Hz). it will give a clean tone when viewed in the SP2 window. Can you give us a bit more information about your settings so we can see if we can replicate.
Last edited by SDRplay_Tech_Support on Thu Jan 01, 1970 12:00 am, edited 0 times in total.
Reason: No reason

PeterB
Posts: 3
Joined: Sun Nov 27, 2016 12:51 pm

Re: RSP2 frequency calibration

Post by PeterB » Wed Nov 30, 2016 11:15 am

Hi,

My methodology was similar to yours with one exception -

it had not occurred to me that the correct method of calibration would be to place the reference frequency directly on top of the LO -
I've always tried to AVOID the central 'spike' when calibrating, or indeed tuning in general, so I was calibrating with a LO shift of 50kHz.

When applying your testing method, I agree that there is little discrepancy shown between spans/decimations/calibration frequencies and ppm results.

However, I maintain that the 4MHz sample rate / decimation 16 does seem to behave differently when calibration was attempted away from the LO frequency, as I was able to generate sensible values except for that one combination.

It may be that this is no big deal, as I now know that calibration can be achieved as per your method.
I don't recall seeing any note about calibrating on or off LO frequency in the manual, so maybe that needs clarifying somewhere.

Sorry if I threw you a red herring :-)

Regards,
Pete.
Last edited by PeterB on Thu Jan 01, 1970 12:00 am, edited 0 times in total.
Reason: No reason

PG3
Posts: 29
Joined: Mon Nov 14, 2016 6:04 am

Re: RSP2 frequency calibration

Post by PG3 » Fri Dec 02, 2016 10:01 am

Yes, there seems to be a tracking problem. If you set the SR to 8, 9, or 10 MHz with a decimation of 1 this gives you an 8 MHz IF bandwidth which means that you can tune up to + or - 4 MHz from the LO. If you set the LO 4 MHz higher than WWV and tune to WWV (far left of display) you get a completely different calibration from setting the LO 4 MHz lower than WWV and tuning to it (far right of display). As you might guess, if you set the LO near (but not exactly on) the WWV frequency and to to WWV the calibration is midway between that of the previous two calibrations. Using RTL-SDR dongles with an up-converter I found that it was always necessary to have TWO calibrations to get spot on calibration across a large frequency span. One calibration for the dongle's frequency error, and another added or subtracted to the converter's offset amount to account for the converter's actual LO rather than it's claimed offset. Not sure if this would apply to the SDRplay hardware or not.
Last edited by PG3 on Thu Jan 01, 1970 12:00 am, edited 0 times in total.
Reason: No reason

Post Reply