It was nice not only to test and do some DXing at Muhos during the weekend 4th - 5th December but also to celebrate 2 x 50 years anniversary of two PUDXK-members (the two Jaris). We started again (like two other PUDXK-meetings) at "Sokeri-Jussin Kievari" at downtown Oulu (at Pikisaari, a small island just near the City Centre) - http://www.sokerijussi.net/.
At the listening QTH (JPRs DX-QTH Muhos) we had nice weather, some sunshine during the few daylight hours between 10 and 14 local time(sic) and temps around -10 degrees. Only about 5 cm snow, whereas in Southern Finland we got more and more during the week after this weekend, now accumulating to about 1/2 a meter at time of this writing!
Testing the netSDRTesting was done only by "ear" without any equipments. I made some simultaneous files with Perseus & netSDR for later review. The Perseus was hooked up to an AcerLaptop (dual-core @ about 2 GHz)while the netSDR was connected to a bit older Dell (single core @about 2 Ghz tabletop pc). So the limitation of processing power limited us just to record everything on netSDR at FFT of 2048FFT/BLK and all the analyzing has to be made aftwerwards. The listening has to bedone using the spectrum display of the total span of 1300 kHz without measuring the exact carrier frequencies. The software used was SpectraVue 3.09. I did make some listening and freq. measurements also using a small span to achieve very good bandwidth resolution (picture) - and comparing the sensitivity with Perseus at the same time.
Performance of the netSDR as expected ...
was very good with a low noise floor and oscillator noise as stated in the specs. You have to notice of course that the unit tested was a prototype and as far as I know the real units (coming Q1-2011 ?) will be a bit better and there will be an option (?) to have even an oscillator of even lower noise soon. So by ear the output of netSDR sounded to be better than that of Perseus and/or IQ we had working at the same time. I do not know if some noise of Perseus was coming from the Perseus PSU as it was using the original power supply unit. The NetSDR was using an "old separate linear (?) PSU" originally used with SDR14 - not the original of 14 or netSDR. It seems that also the netSDR is equipped with a small switching PSU which should be replaced. An issue I fully do not understand - that was the case also with the original PSU of SDR14. Anyway the low signal performance of netSDR was good as well as filtering with the provided software (SpectraVue 3.09). I made a small comparison with very low signals on HF (the Papuan later on on 3905 kHz) by listening it directly on both the netSDR and Perseus. In genaral the reception was quite similar but the sound of netSDR sounded "better" and cleaner (the bandwidth of both rx´s was about 3 kHz). But the difference was very small. What we are missing still with the SpectraVue is the notch which is good on Perseus and on some situation wins the competition, when there is a strong het or signal nearby. As reported earlier the measurement of exact carriers was easier with netSDR/ SpectraVue as SVue tells you "automatically" 4 most powerfull carriers of the viewed span - and you can also see the spectrum (and carriers) at small spans when LSB/USB is used while on Perseus you have to use AM when looking at the spectrum or waterfall on small spans (usually we use 100 or 200 Hz spans). With Svue rather exact measurement can be done like on the picture: using the FFT of about 1Million and demod span of 100 kHz the resolution is about 0.1 Hz. Well, this is not often suitable and when you record 1300 kHz span you have to do measurements with the resolution of about 0.7Hz - which is, however usually enough.
As the unit was an prototype one, there was some decrease of spectrum (maybe the sensitivity as well) on the higher end of the mediumwave band (1500...1800 kHz) which could easily be seen on the spectrum. This issue has been reported also by RFSpace, however.
Recording the Spectrum ...
of 1300 kHz was good, and the 5 minute files were about 2 Gb each. While recording it was possible not only to listen but also to transfer file to external HDs, connect them etc which was very good as the netSDR was SUPER stable - it was running about two days without any crashes. This could be because it was connected by Ethernet and not by the USB. The USB-connected Perseus did suffer of some instability - an issue detected earlier also - you have to be careful when connecting other usb-devices when you have Perseus recording - we had some issues when the Perseus stopped working during these situations. The connecting of netSDR via Ethernet worked really good - this was reported by TuA earlier also when he was using the device at ENOX in November. The 1300 kHz span was good as this is enough for the entire MW-band to be captured. The size of files is so about 12 x 2 Gb = 24 Gb/hour when Perseus record files (using the 2 MS/s/1600 kHz span) of about 2,3 Gb/5 minutes which makes about 28 Gb/hour. OK, not a very big issue, but during the 10 hour session you can save about 40 Gb of HardDisk. We did not tried to record other than 1300 kHz SPANs duringthe tests. It seems we did not do the calibration very well, as the both NetSDR-files and Perseus-files seems not to show correct frequencies, the NetSDR files seems to have an error of about 1 Hz (too much) while Perseus shows about 2-3 Hz too much. Shame on us.
Evaluation of files later ...at home I finally made some afterwards listening ("Post DX-session evaluation") of the files of total of about 300 Gb. Again there was no big difference between the signals I listened to, so in general both hardwares worked ok. Maybe I can do some real "DX-listening" with those files later (I transferred only some files of TOH´s of the two nights with me) in a few days. I know some files have already been analyzed by JVA who has also a powerful "automatic split frequency capture&analyze software" - but he can tell about it later on. The listening was done using my Pentium dual-core @3GHz (4 Gb memory) PC so it was easy and took maximum 10-20% of the processing power. The recommended PC/Laptop for the NetSDR is at least about 2 GHz/dualcore, but as we noted the recording and listening (without exact carrier measurements) can be done easily with older pcs/laptops of about 2 Ghz/single-core. Then you have to use the full spectrum view and set the FFT low (like 2048 FFT/BLK) and possible set the screen update speed (n=10-20). And furthermore - you do need a BIG HD (preferably an internal, but the external works ok and there is no problem with multible USB-connections as you have the RX hooked up to the Ethernet port).
The reception conditions...towards "TA" were rather average to good especially on 5 December (coming better later that week) and some early reception of common NAs were noted starting at around 19 utc. Also some Latin American were noted at night on both 4 and 5 December. The conditions were rather wide spread however with stations from Venezuela to West Cost.The antennas used during the tests were beverages (700-900 meters)usually at around 290 ... 305 degrees.
Last but not least...I have to point out that the NetSDR is priced at about 1300...1400 euros - and you can get a Perseus at about 700...800 euros, so if the money talks you can get "about the same" with 500 euros less. However, if you want to see the exact carriers easily, prefer using the SpectraVue (or SDRRadio) software and like the "look and feel" of NetSDR and last but not least - like to use the Ethernet connection (and easily connect the device used remotedly via net) the choise is NetSDR. And there will be some software development going on with the SpectraVue (I believe), SDRRadio and what is the most interesting with the totally new "CuteSDR", which seems to be an open source platform to develop softwares available also for Mac OS - an issue most wanted by the increasing number of Mac-users among radio enthusiasts!
Well, which one? for myself - I´d like to have them both.
Thank You very much for RF-Space letting us to test this proto of the great new receiver, let´s hope we will see the final products soon!
A testfile of 5 minutes covering 500-1800 kHz will be published later on for those wanting to test with the SpectraVue 3.09 or later.
73 TK
