Friday, 22 December 2017

Using the KiwiSDR to provide receive audio to WSJT-X

Listening to modes such as CW, SSB and AM using the KiwiSDR is straightforward as the default output of the HTML5 sound from the browser goes to the default soundcard, which typically powers speakers or headphones.

A challenge in using digital modes is that the output from the browser needs to be sent to a third party application, such as WSJT-X, using a virtual audio cable. A limitation of the common browsers is that they do not provide any options to re-route the sound to other devices. This post pertains to Windows - Linux provides various other options.

After some research, a solution was found in the form of a Firefox Add-on:


The following steps were followed:

1) Downgrade Firefox to release 56 (the Add-on is not yet compatible with Firefox Quantum).


2) Once installed, disable automatic Firefox updates. As you now are not running the most recent release, there are security considerations. In my configuration these concerns are minimized as the Firefox instance is running on a separate machine and used only for accessing the KiwiSDR (no general browsing).

3) Install the Add-on.


4) Configure the Add-on. Select your virtual audio cable in 'Options'.


5) In WSJT-X audio settings, select the correct virtual audio cable for audio input.


------------------------------------------------------------------------------------------------------------------------------------


As a test, a QSO was attempted in JT65 mode on 20m using WSJT-X.

The KiwiSDR was started in Firefox and set to the correct frequency.


After verifying that the receive audio from the KiwiSDR was being decoded in WSJT-X correctly, I chose to reply to the CQ of S53PM to test the configuration.


Observations

1) There was a significant disparity in sent and received signal reports. This was anticipated due to the large difference in effectiveness between the respective transmit and receive antennas.

2) The internet adds delay, which is manageable within the generous DT (Difference in Time) tolerances of JT65/JT9, but it made FT8 a challenge. The receive audio delay was typically between 1 and 2 seconds. In FT8 you have to be really fast to double click the callsign of interest to start the QSO (as otherwise your first transmission will start too late and not be decoded) - however if you can do this the auto sequence works normally from that point onwards.

KiwiSDR (over internet) on the left, local (IC7100) on the right

Saturday, 16 December 2017

AM Bandstop evaluation

I had my first opportunity today to assess the AM Bandstop, which should reduce the strength of signals on the Medium Wave band and hopefully clean up the receiver.


378-1880 kHz - 3 dB attenuation

526-1710 kHz - 9 dB attenuation

KiwiSDR1 on the left, KiwiSDR2 with AM bandstop on the right.

Unfortunately, the addition of the bandstop was not enough to stop ADC overload on the KiwiSDR when the attenuator was removed. As this was the primary purpose of its procurement its continued retention in circuit will have to be justified by improvements in other areas, which will require further observations.

Other strong signals outside of the Medium Wave band could be the cause of the ADC overload or it could be because of the aggregate strength of all the signals being fed in from the loop across the LF/MF/HF range.

Reducing the circumference of the loop would give the KiwiSDR an easier time, however this would affect the Red Pitayas which can tolerate larger signal inputs. A compromise is therefore needed.

Further observations were made in the early morning of Sunday 17th December. A snapshot taken at 0610z of the 80m band shows a real improvement with the bandstop fitted. The improvement is so marked that continued retention of the AM bandstop is justified.

KiwSDR1 on the left, KiwiSDR2 with AM bandstop on the right.

Friday, 15 December 2017

Lightning protection installed / AM Bandstop / Second KiwiSDR

Protection from lightning discharge is an important consideration at a remote installation. This morning I installed an earthing rod and a Nagoya lightning arrestor near to the antenna feed point. No degradation in performance was noted.



A second KiwiSDR was installed onsite, for additional user capacity and side-by-side experimentation.

Currently, the second KiwiSDR has a newly purchased AM bandstop in circuit to determine its effectiveness. Currently, both KiwiSDRs have a 10 dB attenuator on their inputs to avoid constant ADC overload. I would like to find a way to remove these as this solution is not optimised. Users should be aware that this second KiwiSDR will be used for experiments and will have a lower quality of service.



Thursday, 14 December 2017

Adjusting the KiwiSDR display for optimum spectrum presentation

New users of the KiwiSDR often comment that the spectrum display seems noisy. It is necessary to change the display settings to get optimum results once you have set your frequency and display bandwidth.

Suggested steps:

1) Select 'Spectrum' if you have not already done so.

2) On the upper display, note the noise floor.

3) Adjust 'WF min' until the noise floor just disappears below the display.

4) Adjust 'WF max' to raise the peaks of the visible signals.

A KiwiSDR session optimised for good presentation


WSPR Reception

One of my primary interests recently has been WSPR reception. I have found that WSPR is an excellent mode for receiver and antenna optimisation.

The input from the antenna is distributed to the receivers by an Elad ASA-15 antenna splitter. Two Red Pitaya boards act as WSPR receivers running Apline linux software by Pavel Demin. Each Red Pitaya can decode 8 bands simultaneously.

Presently, no impedance transformation solution is in circuit for the Red Pitayas and they are directly connected to the Elad. The Red Pitaya's antenna ports are high impedance inputs. The current receive performance across the frequency range seems satisfactory in the present arrangement.


Wednesday, 13 December 2017

Noise found and mitigated on 20m, source - Dell PC

A Dell Core-I5 PC was installed near to the receiver location for running CW Skimmer Server and other decoders. It was evident that the 20m band was affected by noise, which increased when the computer was switched on. The PC and its PSU were the cause, the effects went away when the PC was moved some distance away from the receiver. As CW Skimmer Server will be running over ethernet, sharing the same switch, moving the PC away and using shielded ethernet cable back to the switch is a satisfactory solution for the moment.

Some isolated ethernet spikes remain - I will try and tackle these with ferrites at a later time.
 



Interference from Dell PC located next to the receiver.


Dell PC switched off, PSU on. Interference reduced but still visible.


Dell PC and PSU switched off. No PC interference.


Dell PC on, but relocated 30m away. No PC interference.



Tuesday, 12 December 2017

Noise reduction and receive antenna relocation

After some prompting, I've been persuaded to document some of my amateur radio activities in a blog. Over the next few weeks, I will provide further detail of the steps taken to reduce received noise and the composition of the developing receiver and demodulation system.

Much useful information about noise reduction can be gained from the entries in Jim's blog:

GM4FVM's radio world

After taking the following steps at the home location:

1) Using ferrites (Fair-Rite 2631801202 FT114D-31)
2) Replacing switched mode power supplies (SMPS) with linear supplies
3) Using shielded ethernet cable
4) Removing powerline adapters
5) Trying a QRM eliminator
6) Employing mains filters

It became evident that, even though the above steps had been helpful, I was fighting a losing battle as noise sources seemed to be increasing in type and variety.

When ever-increasing vDSL signals in the 8.5 - 12 MHz range could not be mitigated successfully, a plan for antenna and receiver relocation was initiated.

The concept was straightforward. Move the receive antenna away from suburbia into the countryside, away from man-made noise sources. Provide access to the antenna over the internet using a KiwiSDR and use additional devices such as Red Pitayas to process data onsite and upload.

Wellbrook Loop ALA100LN, 50m wire circumference (a bit hard to see!)

Over the weekend of 9/10 December 2017, a test antenna was constructed and erected at the remote site. 50m of RG213 feeds the signal from the head unit back to the building which contains the receivers. The internet connection from the receivers is a mixture of WiFi/ADSL and somewhat fragile - the connection drops sometimes. This is an area that needs improvement.

Testing and evaluation is now in progress. The Wellbrook ALA100LN Loop is primarily designed for low band reception, but can receive up to 30 MHz. First impressions are that the project has delivered significant improvements and was worthwhile. Further investigation of remaining noise and an assessment of receive performance across the HF spectrum will be needed.



80m SSB in the afternoon

The KiwiSDR is at the following link:

G0LUJ KiwiSDR