Integrating Software Defined Radio (SDR) and Standard Transceiver

Chavdar Levkov LZ1AQ  

 www.lz1aq.signacor.com      lz1aq@abv.bg    ,  Sofia, Bulgaria

 

SDR it should be integrated into a standard environment – transceiver (TRX), power amplifier (PA) and the antennas.  In this way the SDR will find much wider practical acceptance in ham radio community.

The schematics of one possible solution is shown here:  SDR_TRX_splitter.pdf 

 

Splitter box

The simplest solution is used: the antenna input/output of the transceiver (TRX) and the antenna input of the direct conversion receiver (DC RX) are tied together in rx mode.  In TX mode the DC RX input is grounded.  A small fast relay (FRT5 telecom relay; switch time < 2 ms) is controlled by a –SEND signal which is standard output control signal of most commercial transceivers for PA control.  Return to RX mode is delayed with 10 ms for safety reasons. The connection to T-connector of ANT TX input of the DC RX should be made very short, better without any cable in order to keep to minimum the stray capacitance from the connector to the relay contact. This capacitance plays role in TX mode when the receiver input is grounded. The splitter is housed in a small screened box made from PCB material with copper foil.

Special care was taken for the safety and noise immunity of the total system.  The IC 40106 consists of six Schmidt trigger input invertors. The splitter box is powered from the same TRX 12V power supply. Ground loops are avoided with common mode balun transformers in control and power cables of the box. Additionally the DC RX input is insulated with a broadband transformer.  There is an input attenuator 3.7 dB which smoothes to some extent the influence of the input impedances of both radios adding strong resistive component. This is important when one of the receivers is on another band and its input filters add reactive component which is not good for the other radio. (I have measured the DC RX input impedance in receiving bandwidth -  around 70 ohms mostly resistive). But more important is it’s  “fuse” function. If for some reason the relay is not switched to ground in TX mode, the total transmitter power is applied to the attenuator and the 0.125 w resistors simply blow away thus saving the DC RX input circuit. (It has happened 2 times since I forgot to plug the control cable!). Additionally the broadband transformer went into saturation and limits the voltage that can enter the DC RX. This system is foolproof.

Using this small box (Fig.1) there is no need to make any changes in existing station – just a coaxial T-connector to plug there ANT TX input of the splitter box.  There is no additional relay contact junction in TRX-PA-ANT path.

 Fig.1            Fig.2

Audio switch box

The audio switch box (Fig.2) permits to listen to main transceiver, to the SDR output or both (split into left and right earphones). I have headphone audio amplifier in DC RX so there is another switch which permits to listen to pure DC RX audio without any DSP processing.  There is also “RX mute” switch which disconnects the DC RX input from TRX antenna in the case where we do not want to load the main transceiver input.

 

   Using these boxes I do not have any problems with RF pickup or other noise. Moreover I am using only open wire feeders for all my antennas (this is my other hobby) with home made automatic antenna tuner so the level of the surrounding RF field was not so low. The T-connector is inserted between TRX and PA so a maximum of 100 V RF voltage is applied there. 

 

 

On the air tests

The SDR should have to compete with the mainstream equipment and to prove its virtues in real environment so I decided to make a heavy test  -  the CQ WW 160m CW DX Contest 2008, QTH= Europe (some people call it EU Zoo contest.)

Standard Equipment:

TRX: Icom 756pro, PA: 350 w, home made, Antenna: full sized offset center fed dipole for 160m at 15 m height, Automatic antenna tuner, home made.

SDR equipment

Home made DC RX is used with double balanced Tayloe mixer with CD4052 analogue multiplexer. The oscillator is 28MHz VFO divided by 16 and   D-trigger IQ splitter.  The audio pre-amplifier is TS462 with 4nV/Hz1/2 noise density.

I have measured this radio on 14 MHz band (with Audigy SE 24 bit card) with 16MHz low-pass input filter and obtained:

MDS -129 dBm @ 500Hz BW, IP2 = 62 dBm,  IP3 >= 15 dBm,  Sound card saturation @ -26 dBm. (59+50dB)

It is expected this radio to have similar or better performance on 1.8 MHz band.  It must be noted that the oscillator phase noise on this band will be very low. See the schematics of the radio SDR_sch.pdf. Additionally I added a band pass input filter (1.8MHz; 700 KHz bandwidth).

A description of this radio and how it is measured can be found on my home page www.lz1aq.signacor.com  but the articles are in Bulgarian language. I plan to translate them into English if the free time permits.

Computer

Notebook Dell 2 x 1.8GHz AMD core.   2 GB RAM. Sound card is SoundBlaster Live! 24 bit USB external. The computer has internal Sigmatel High definition audio sound which appears to be useless for SDR since it was not possible to reach any I/Q balance by unknown reasons.

Software

Windows XP.  Rockie 3.2 ( http://www.dxatlas.com/  ) and Winrad 1.3.( http://www.weaksignals.com/ ). ASIO drivers are not used. The sampling rate was 96 KHz @ 16 bit resolution since this USB card uses USB1.1 and the bandwidth is insufficient for 24bits/96KHz resolution. Running Rockie or Winrad occupied between 15-20% of the CPU resources at 96 KHz s/rate.

QTH is KN12KD near Sofia, Bulgaria in a small village in a deep valley (not quite good for DX contacts). See the shack and test setup in Fig.3.

 

  Fig.3

 

Tuning

I tuned the VFO at 1854 KHz and entered this value in the software. I had now 96 KHz bandwidth from 1806 to 1902 KHz. (at the band edges some digital aliasing occurs. I do not know the reason for that but these few KHZ at the band edges were not very usable). This bandwidth was suitable since most of the activity took place from 1810 to 1890 KHz.  I used the main TRX for basic work and simultaneously watched at the waterfall display for weak stations or  pile-ups. If I found a perspective spectral line I clicked on it and listened to the station. If I needed this station I tuned the TRX manually on this frequency (as in very old days). I took the exact frequency from the SDR display. Then I switched the audio box to TRX, listened there and made a call. This was not very convenient method but the idea was to test the equipment and SDR together, not to win the trophy.

 

Results

Most of the time the band was totally occupied. In 80 KHz bandwidth I counted almost 240 spectral lines on a frozen waterfall display. That means that on every 350 Hz and less there was a transmitting station!! I had continuously almost 30 to 40 booming stations with KW power (distance up to 300 Km) with signals sometimes above 59+30dB. In 20 km diameter I did not have a powerful station.

  1. I did not have even a single event of sound card saturation. (Any signal which exceeds saturation level -26 dBm (59+50dB) makes the SDR useless). This means that sum of the power of all signals on the band in any moment did not exceed the limit of -26 dBm.
  2. The wide band waterfall display was the main used mode and for CW it was very convenient.
  3. Extremely narrow bandwidth filters (20 to 100 Hz) with excellent shape factors can be used fast and easily due to mouse click tune. The IC756pro has also very good narrow digital filters down to 50 Hz but they are not easy to be used in conventional way during the contest. (I own this radio 4 years)
  4. The very close-in selectivity of the SDR subjectively is better than IC756pro (using both programs). I was able to pick up very weak stations at 100-200 Hz offset from a 59++ signals without problems.  On IC756pro the AGC was popping and clicking even on the narrow 50Hz bandwidth mode when there is a very strong near station. These were fair tests since I compare the radios in the same environment and on the same signal. I think that in this case the way the AGC is implemented plays a major role. Rockie 3.2 was the best in this respect. Switching off the AGC sometimes helps but additional tests must be performed. Winrad has a nice and smooth AGC control (which can not be switched off) but for very uttermost CW contest conditions it is not as effective as Rockie. For normal situations Winrad AGC is much more pleasant to listen. Both programs have excellent narrow band CW filters.
  5. Narrower waterfall displays were often used since most of the weak stations were simply not visible on the 96 KHz bandwidth between tremendous wide band clicks.  During the contest I found that the best strategy was to look approximately at 10 - 20 KHz bandwidth where even the weakest stations are visible. For the extreme cases highest resolution waterfall bandwidth of 5-6 KHz was invaluable.
  6. I saw clearly that most of time the real limitations were not the receiver dynamic range but the wide band signals of some stations with bad clicks which are perfectly visible on the display.
  7. I managed to work 53 countries and 1 province but my antenna and power are not what many will dream about. What is more important that I heard at least 20 more countries and states most of them with SDR. I am totally convinced that without SDR waterfall display combined with very narrow band filters I will not be able to find them in the crowd, working in the usual way.

 

Comments and Suggestions                             

  1. Most of the modern TRX have serial computer control and it will be very convenient to have some automatic way to pass the RX frequency of SDR to the standard TRX.  I think that in Winrad (Alberto, I2PHD) there is a DLL mechanism to do that. Rockie surely can also be improved moreover there is a new and wonderful SDR software called “CW Skimmer” from the same author (Alex, VE3NEA) which has already a CAT control (but not for the I/Q radio mode!?).
  2. The proposed audio switch box is the simplest one. It will be good to put some kind of automatic switching – for example to switch to TRX audio automatically when PTT is on. The online CW monitoring from SDR is confusing due to the time delay.
  3. The 16-bit mode of the USB sound card was not a limitation in the case of this contest. The best 24 bits cards have not more than18 effective bits and theoretically 12 dB better dynamic range compared to 16-true-bit card ( I suppose that  my SBLive! USB card has 16-true-bit resolution). Much more important is the maximal sampling rate since this gives immediately a practical improvement of enlarged bandwidth.
  4. The receiver sensitivity is reduced due to parallel connection of the two receiver inputs. Theoretically the input signal level will be reduced with 3.5 dB if both receiver inputs and antenna have equal 50 ohm impedances. The attenuator reduces additionally 3.7 dB and the signal at the DC RX input will be 7.2 dB less. Another source of losses is the increased SWR=2 of the feeder in rx mode. Calculations for 50 m RG213 antenna feeder show additional losses of 0.06 dB for 1.8MHz rising to 0.25 dB for 14MHz compared to SWR=1 case.  On the low bands (1.8-7MHz) this degradation of sensitivity is not significant. The atmospheric and man made noise are much higher than the rx noise level even in rural QTH. For higher bands this solution might be unacceptable and the input attenuator should be with lower attenuation or omitted at all. (but then the “fuse” function will be lost).
  5. When listening with transceiver there is a strong carrier at the center of the band from the DC RX oscillator. I measured the level of this carrier with IC756pro S-meter (756pro attenuator and preamplifier in off position ;  antenna was replaced with 50 ohm resistor). It was approximately S6, S8, S9 and S9+10dB on the respective bands (1.8, 3.5, 7, 14 MHz). The amplitude of this leakage carrier increases with frequency. These observed levels do not impair the receiving parameters of the transceiver. I do not assume this as a big problem since this is only one frequency and in the contest I even did not noticed it. With “RX mute” switch the DC RX is disconnected so this signal disappears.  Probably with proper  mixer design this leakage carrier can be reduced. I am using double balanced mixer which might be better in this respect but I have not compared it with a simple balanced mixer.
  6. Sometimes I observed weak phantom signals in the transceiver. They appear far outside of the amateur band and were obviously products from the DC RX mixer. One possible explanation is that the oscillator of DC RX and strong out of the amateur bands signals are mixed and mixing products appear back at the antenna terminals. If the DC RX has an input band-pass filter these spurs are removed. An HF pre-amplifier or transformer type splitter might be used for radical separation but the simplicity of the splitter box will be lost and I think this is not necessary until further experiments are performed. 

Remark:  The results from selectivity tests are valid only for CW mode.

 

Conclusions

A simple hardware is developed in order to use simultaneously SDR and conventional radios in fast and convenient way without any modification of the existing equipment. Using two senses: eyes and ears give certain advantages especially in the uttermost conditions. Extremely narrow band filters of the SDR can be used easily. The practical very close-in selectivity and dynamic range in CW mode is better compared to a hi-end transceiver. The SDR hardware and the sound card used in this experiment were very cheap and common. Minor additions to existing software which benefit simultaneous usage of SDR and standard transceiver are proposed.

 

February 2008,

73,  Chavdar LZ1AQ