Diversity Reception with a Modified SDR Receiver and Small Active Antennas
Chavdar Levkov , LZ1AQ
Last revision 1.1 : 28 Sept 2015 ©LZ1AQ
Diversity reception is a method of combining two signals from different antennas to improve the S/N ratio . Most often the diversity reception is performed with two antennas (with the same or different types) placed at some distance between each other. The fading of the signals from these two antennas usually do not have the same phase and if we combine in some way these two signals there will be improvement in reception. This technique is usually associated with big antennas and large spaces, but it can be applied anywhere, even on the balcony of an apartment using small antennas and very simple additional hardware. The answer is to use two antennas on the same place but with different polarizations (polarization diversity ). The effect will be maximal if the antennas have mutually orthogonal polarization. It is possible to use two very small active antennas placed on the same mast each with different polarization. More about horizontally (HP) and vertically polarized (VP) small antennas can be found in another publication of the author .
Generally two receivers are needed to implement the diversity reception. Simple addition of the signals from the antennas does not work - the effect will be that a new antenna with different polarization and directivity will be created - e.g. phased array as an example, but the effect of the fading will be not diminished. Another approach is to use adaptive linear combination of the two signals with complex number coefficients (amplitude and phase) to maximize S/N ratio [26, 21]. Linrad SDR software  is doing that but fast real-time processing following the signal fading is not possible for the moment.
The simplest way is to send the signal from one antenna to the left ear and the other – to the right ear. This diversity reception technique is well described in [3, 4, 22]. The effect is a stereo audio signal moving in the space. This technique is successful for CW signals as well as for SSB (AM is still under question). Two phased locked but independent receivers are needed to implement this method.
A modified direct conversion (DC) receiver originally intended for SDR reception can be used for experimenting with diversity. These receivers are very popular, can be built easily, or obtained as a kit for minimal price . Each SDR receiver has a double channel mixer with common input for a single antenna. The modification is to split the input mixer into two separate channels (Fig.1, 2) for two antennas. Thus we will have two separate receivers which have common oscillator (there is a fixed phase difference of 90 deg. in between the local oscillator (LO) of the SDR mixers but this is not important in our case). Each mixer input is fed with its own antenna. The drawback of this approach is the existence of the image channel as in any simple direct conversion receiver.
Fig.1 Modification for diversity reception of the input of an SDR direct conversion receiver with a double balanced mixer. The RX has two separate channels with common LO with separate antennas.
Fig.2 Modification for diversity reception of the input of the Softrock SDR direct conversion receiver. Its original function as a single channel I/Q receiver can be restored by paralleling the Ant1 and Ant2 inputs. The local oscillator of the receiver must be either VFO or DDS/PLL device (e.g. SI570).
Fig.3 Experimental setup. This is a dual channel direct conversion SDR  with a modified mixer. At the right side are seen the control boards of two active antennas  with horizontal and vertical polarizations. SI570 PLL local oscillator board is seen at the left side. There is also a possibility to tune the receiver with VFO (the big multi-turn potentiometer which controls a varicap is seen at the left side of the board). This receiver has an additional headphone amplifier on its PCB which is similar to that given on Fig. 4.
Additional gain must be added to the SDR audio in order to drive the headphones. There must be a preamplifier and a headphone driver (Fig.4 ).
Fig.4 High dynamic range headphone audio amplifier for direct analogue monitoring of the input signals from any SDR direct conversion receiver. The audio gain of these receivers is not sufficient to drive headphones and additional amplifier is needed to convert them into classic direct conversion analogue receiver. The amplifier is connected to the SDR analogue outputs. There is a possibility to adjust the channel balance to equalize the antenna gains (+- 10 dB) . There is a 2-nd order low-pass RC filter at 3.2 KHz. TLV2464 is op. amp with high slew rate, high output drive and with very low distortions at 3 to 5 V supply . Diode limiter with Schottky diodes can be used to preserve the headphones and ears from strong signals. This amplifier converts the SDR receiver into a very good analogue direct conversion receiver. The audio quality, which is very important for these receivers, depends mostly from the headphone driver stage.
Fig.5 Fast headphone switch. Two micro-switches (SW1 and SW2in Fig.4 ) are mounted horizontally on a PCB plate and are activated by a lever made also from PCB. The phones are in diversity mode at neutral position. If left or right switch is activated, horizontally or vertically polarized antenna signals are fed to phones. This switch is needed to compare quickly the signals in order to choose the best mode.
I need a fast and ergonomic switch to choose between diversity and single channel reception in order to estimate whether there is an improvement in S/N ratio between these modes. The result is shown on Fig.5.
There are also some very good designs of DC receivers , for example KK7B “Binaural I-Q receiver”  can be used also for diversity reception with simple modifications.
2. Using HDSDR (Winrad)
Another software approach is to use HDSDR (or Winrad) [7, 8] without additional external audio amplifier. The direct conversion receiver is connected to the computer sound card in the usual SDR way. But in the software we must send the I and Q signals directly to the left and right audio channels in order to use the soundcard amplifiers.
HDSDR Settings :
- In the “Options - > “Output channel mode for RX” menu we must choose “IF as I(left)/Q(right)” or “IF as Q(left)/I(right)” lines. In this special mode the HDSDR signal processing is not active and we have unprocessed I and Q signals at the output of the sound card.
- The LO and Tune frequencies must be the same. The frequency change must be performed by changing only the LO frequency.
- If we have software controlled LO (SI570 PLL chip or some other kind of DDS ) we can use the software to tune on specific frequency but only by changing LO frequency. The “click on frequency” can be applied only if we set: “Options” - > “Misc Options” -> “Tune fixed to LO<-> Tune Offset”. The tune offset must be set to 0 .
The other possibility is to use VFO in the receiver with manual tuning, but keeping all the time the settings of LO and Tune to show on one frequency.
- Set the mode to USB or LSB - (in CW mode there will be frequency offset between left and right channel)
- Set the output sampling rate to the lowest possible value (4000 or 8000 Hz). Thus we can use the output DAC anti-alias filters to limit the audio bandwidth. For example using the minimal output sampling rate (4KHz) we have a very good 2 KHz software LP filter at the output. Some sound cards have also software equalizers which are actually very good filters. If ASIO4ALL audio driver is used, the input and output sampling rates of HDSDR must be set to equal values, otherwise the output stereo effect disappear.
- You can use sound card mixer (L/R slider) to equalize the antenna gain of both channels.
3. The antennas
The most pronounced effect will be when the signals from the antennas are not strongly correlated. – i.e. we must use either antennas with different polarization or widely spaced. The gain from both antennas must be almost equal (not very critical) otherwise the space effect will be diminished. This gain equalization might be performed to some extent in audio chain but it is not advisable to have considerable difference in antenna gains between both channels.
We do not need big or widely spaced antennas to experiment with diversity reception. I have good diversity reception from the balcony of my city apartment using two small wideband loops [9, 14] with different polarization on the same mast (Fig.6). The example of dual polarization small antennas are given in another article of the author .
Fig.6 Two active CC loops (horizontal and vertical) each with diameter of 0.7m on the balcony of a city apartment are used for diversity reception. Two AAA-1 amplifiers are used for each antenna.
Fig.7 Universal dual polarization antenna. The loops are with 1.27 m diam. Loops act also as dipole arms in dipole mode. Any combinations of HP-VP pairs can be chosen. The height is 7m.
The best pair is the vertical small dipole and CC (crossed coplanar) horizontal loop [2, 9] . They are pure vertically and horizontally polarized antennas even above real ground. For urban environment I will recommend the pair vertical loop & horizontal loop . Bear in mind that the horizontally polarized antennas need height (at least 0.1 wavelength) in order to have sufficient gain .
In  is described an universal dual polarization antenna which I have build in my country house (Fig. 7). Two AAA-1 amplifiers are used and any combination of HP and VP pair can be chosen.
4. The psycho-acoustic effect
The human brain is well trained to distinguish the direction of the sound . There are evidence that diversity stereo signal combined with brain processing will enable to receive very weak signals compared to single antenna reception [3, 11, 24]. One definite benefit is that the noise coming from the two antennas is weakly correlated and thus we must expect some increase (theoretically 3dB) of the S/N ratio. The psycho-acoustic effect is that the background band noise becomes somewhat transparent and space separated from the signal – thus the signal can be recognized easier by the brain.
For the cases where the change of the signal is slow, the diversity reception definitely improves the reception. This is not always true when there is a fast (period up to 1- 2 sec ) signal rotation in space. In these cases the mind is distracted by the rotation and probably there must be some training that will permit the mind to concentrate to information processing rather than to the fading effects.
One additional effect is that the signals are somewhat in different space locations and much wider filter bandwidths can be used. With some training, even in the presence of many stations in the receiving band, remarkable examples of brain processing can be demonstrated and listener can distinguish the desired signal easily. The “space effect” is stronger if wider the bandwidths are used – in very narrow bandwidths (below 500 Hz ) the space effect is not so pronounced by unknown reasons.
The human brain acts also as linear “bandpass filter” with bandwidth as narrow as 50 Hz [11, 27]. With training, even with this quite wide audio band of the receiver, the user can receive very weak CW signals which can be recognized with difficulty even using a good narrow band single channel RX . The exception is when there is another very strong undesired signal in this audio band – then the non-linear psycho-acoustic effect of “auditory masking”  takes place and the desired weak stations disappear.
I think that there is a difference between polarization diversity and space diversity cases. If any of the polarizations is prevailing (and this happens quite often) then there is no benefit to use diversity – with the better antenna you will have better reception since the other channel, where the signal level is low, increases only the overall noise. For space diversity reception (especially if you are using the same antenna types for both channels) this might be not true and the diversity scheme probably will be always better than a single antenna approach. So the best solution for polarization diversity is to have always the possibilities to switch to single channel mode with the better antenna. The optimal strategy is to listen in diversity mode and then to test the single channel mode for better reception (see the “fast” switch on Fig.5 which is very important).
For me the signals to which I listen with this setup are the “true” band signals. This direct conversion receiver has an excellent linearity, minimal processing stages, no AGC and two orthogonal antenna sensors. You can feel the band as it is - the fading, the polarization, the atmospheric crashes as they are. There are no additional filtering and processing – just your natural sensors (the ears) and the brain. This is a “ direct binaural electromagnetic receiver” which connects our brain to the RF electromagnetic space. It is very refreshing to listen to such a radio. Sometimes switching to conventional single channel receiver is almost the same as to put a choke in one of your ears J As a matter of fact, recently I prefer to listen to my direct conversion diversity receiver rather than to my Perseus or ICOM 756pro. I like wide bandwidths (but not in heavy contests) and the diversity reception gives me the additional selectivity without influencing the signal quality. Even the folded spectrum (image) of these simple receivers gives the opportunity to double the receiving bandwidth with useable tone pitch in CW mode. So the coin always has two sides – wideband diversity and narrow band single channel. Both are useful but do not miss the first opportunity.
5. Audio records with diversity reception
To demonstrate the diversity reception with wideband radio I have made several audio records with a pair of vertical (left channel) and horizontal polarization (right channel) small active wideband loops or dipoles. The loop antenna has two CC loops each with 1m diameter. Vertical loop is placed 2 m above the ground and horizontal loop is placed 7 m above the ground. The loops are used also as arms of a small dipole in dipole mode with AAA-1 active antenna amplifiers. These antennas and modes are described I details elsewhere [2, 9, 14]. These antennas are placed very near to each other in order to have a pure polarization diversity. Most of the records were made with H-loop and V-dipole pair but some of them are with H-loop and V-loop pair. The records were made with a simple modified DC receiver as described earlier (Fig.3). The audio signal for the recordings was taken from the usual SDR output before the final headphone amplifier. The DC receiver has at its antenna inputs simple LP filters at 16 MHz to avoid overload from nearby FM transmitters and no additional AF filters. These wideband records must be listen with good stereo headphones - using loudspeakers will not do the job. The output level in the soundcard mixer must be set near to the maximum since the records are a bit weak. The H and V antenna gains are with similar levels but for some records the listener might equalize the gain with the balance slider.
The records were made with PottyRecorder  at 11 KHz sampling rate so AF bandwidth of the records is somewhere around 4 KHz. All records are with 1 min. duration and compressed (*mp3 ; size 470 Kbytes). The file names consists of date and local time (UTC+2) automatically stamped by the recorder. The location is in semi-urban environment with moderate noise pollution.
5.1 CW mode:
- 5R1 on 20m, weak with fast polarization rotation, usual for Africa 2015-03-07 2212.mp3
- 5H3 on 20m.There is a strong pulse noise in vertical polarization channel. 2015-03-21 2254.mp3
- Europe on 80m, slow polarization changes 2015-03-07 2237.mp3
- AH6 on 20m, via the North pole with flutter 2015-03-08 1921.mp3
- 2 stations on 20m, they are with different H and V polarizations 2015-03-08 0619.mp3
- 40 m early evening, Eu, JA and pileup 2015-03-07 2305.mp3
- Usual USA 40m east coast propagation 2015-03-08 0625.mp3
- Antarctica on 20 m 2015-03-08 1917.mp3
- VK9NT on 20m very weak with QRM 2015-05-02 0833-4.mp3
- CW contest 2015-03-21 1944-f.mp3
5.2 SSB mode
- AL7 on 20m 2015-05-02 0845.mp3
- K8 on 40m in the morning 2015-03-08 0642.mp3
-TF3 on 20m pileup 2015-04-04 1740.mp3
- VE7 on 20m via the North pole with flutter 2015-03-08 1907.mp3
- W3LPL on 80 m ARRL contest 2015-03-08 0649.mp3
- Weak W6 on 20 m with excellent readability 2015-05-09 0801-25dB.mp3
- ZL2 on 20m 2015-05-09 0831-20dB.mp3
5.3 Records made from an urban environment
The loop antenna consists of 2CC loops each with 0.7m diam (Fig.6) 5 m above the ground level , placed on the balcony of my apartment in Sofia. The electromagnetic pollution is quite high but the diversity reception helps to minimize the man-made noise. Two active antenna AAA-1 amplifiers are used. The diversity pair used was horizontal and vertical loops, the dipole mode is very noisy in this environment. Below 6 MHz the signal level from the horizontal loop becomes too small due to its low height and the vertical loop gives always better signal.
- LU on 20m/CW 2015-06-18 2316.mp3
- VK5 on 20m/CW. Extremely weak signal with QRM. Here the diversity helps a lot. 2015-06-18 0830.mp3. Here is the same record but filtered with 400Hz band pass filter (post-processing with Audacity software) 2015-06-18 0830-10dB-Filt.mp3 . I can equally well get the VK5 callsign from both records irrespective of the fact that the unprocessed record has 4 KHz bandwidth.
5.4 AM mode
The AM records were made with fixed LO tuned exactly to the carrier frequency with resolution of 1 Hz (SI570). The DC receiver works in DSB mode. But there is always some small difference between these frequencies which is exhibited as a small beating phase difference. In binaural listening the brain is sensitive to phase difference between left and right ears. The effect is additional “space rotation” of the signal so the benefit of binaural diversity reception in this mode with simple hardware and software is under question. When the difference is very small the reception is quite acceptable. In AM reception with this setup one never knows whether the “space” effect is due to polarization and fading changes or to very small difference between LO and carrier frequency. These records were made in urban environment.
- Broadcasting station on 11.7 MHz. The carrier and LO are very close. 2015-06-17 1859.mp3
- The same station. The LO is +1 HZ from previous setting. 2015-06-17 1904.mp3
Dual synchronized receivers with two separate antennas are now available on the market in some high class transceivers [15, 16, 18] and also in some SDR DDC RX designs  but I do not have any experience with them . There are also open software programs [17, 20, 21 ] which will enable to use the diversity reception with SDR processing but for the moment I do not have appropriate hardware to test them.
There are some very nice diversity records in the net [ 3, 22] made with dual commercial receivers so the reader can get an idea what to expect from the audio diversity reception with higher class radios.
 Wikipedia, Antenna diversity, http://en.wikipedia.org/wiki/Antenna_diversity
 Levkov Ch., Horizontally mounted small active receiving antennas, 2015, http://www.lz1aq.signacor.com
 Tom Rauch, W8JI, Diversity receiver and transmission, http://www.w8ji.com/polarization_and_diversity.htm
 Tom Rauch, W8JI, Rotation Polarization Fades , http://www.w8ji.com/HF%20circular%20polarization.htm
 WB5RVZ Software Defined Radio Homepage http://wb5rvz.com/sdr/
 SD-RX manual http://www.lz1aq.signacor.com/docs/sdrx/SD-RX_manual_214.htm (in Bulgarian language)
 HDSDR home page , http://www.hdsdr.de/
 Winrad home page, http://www.weaksignals.com/
 Levkov.Ch. LZ1AQ, Wideband Active Small Magnetic Loop Antenna, 2011, http://www.lz1aq.signacor.com/docs/wsml/wideband-active-sm-loop-antenna.htm
 Barry D. Jacobson HUMAN EAR http://web.mit.edu/2.972/www/reports/ear/ear.html
 Leif Asbrink, SM5BSZ - Receiving Weak CW Signals 2006 http://www.sm5bsz.com/weakcom.htm
 Wikipedia , Auditory masking, http://en.wikipedia.org/wiki/Auditory_masking
 Ray Soifer W2RS The Weak Signal Capability of the Human Ear, 2002 , http://www.g1ogy.com/www.n1bug.net/tech/w2rs/The%20Human%20Ear.pdf
 Joe Martin K5SO, Dual-Mercury Diversity Operation, 2011, http://www.k5so.com/Dual-Mercury%20Diversity%20Program.html
 Leif Asbrink, SM5BSZ, Linrad , http://www.sm5bsz.com/linuxdsp/linrad.htm
 Piotr Hewelt, SP2BPD, video https://www.youtube.com/watch?v=sFpSQQ_dBE4
 Barry Gross, N1EU, Diversity reception, http://n1eu.com/K3/diversity.htm
 Rick Campbell, KK7B, A Binaural I-Q Receiver, 1999, http://www.arrl.org/files/file/History/History%20of%20QST%20Volume%201%20-%20Technology/QS03-99-Campbell.pdf
 Flex radio forum messages on diversity reception. https://community.flexradio.com/flexradio/topics/6700_vs_k3_still_no_contest?topic-reply-list[settings][filter_by]=all
 LZ1ABC Audio recorder http://www.lz1aq.signacor.com/PottyRecorder.rar
 Leif Asbrink, SM5BSZ, Adaptive polarization, http://www.sm5bsz.com/polarity/poltheor.htm http://www.sm5bsz.com/polarity/cpuadapt.htm
 Jan Simons PA0SIM, http://www.pa0sim.nl/Results%20and%20Audio%20Samples.htm
1.0 25 June 2015 Initial publication.
1.1 28 Sep 2015 Some additional links.