This work is the result of the activity described in the preceding thread:
https://www.sdrplay.com/community/viewt ... f=5&t=3352
and gives methods and concepts for optimizing low noise reception with a long wire antenna using RSPduo and other receivers of SDRplay.
By Gianfranco Lovisolo Ex I1LOV with the precious help & assistance of Phil, VK7JJ
The RSP class of SDRplay receivers has unique characteristics which can be used to advantage to optimize low noise operation in the HF frequency bands 3.6 to 28 MHz. Every antenna will receive atmospheric noise along with the useful signals. This paper shows ways to minimize the negative influence of noise on receiver operation.
The RSP receiver provides the operator with a panoramic representation of a wide span of the frequency spectrum, while at the same time giving continuous indication of the receiver’s available dynamic range. Furthermore the RSP receiver affords a Hi Z (1000 Ω) balanced input: by properly exploiting all these features, it is possible to optimize low noise reception on the HF bands to a degree not previously possible with conventional equipment.
An example of what can be achieved is given in Figure 1 below.
The two screens have been taken minutes from one another. Conditions were exactly the same, the only change being the balun used on the High Z input. Screen to the left was taken on Saturday, June 30th at 18:02; screen to the right on the same day, at 18:05. Screen on the left shows a “noise floor” of – 105 dBm, while on the right we have a figure of – 120 dBm, with a gain on noise of 15 dB, with wanted signals on the same level or better. But first we have to introduce some useful concepts.
1. RECEIVER NOISE & ANTENNA NOISE
If we leave all input ports disconnected, the panoramic display will show the “noise floor” of the receiver itself over the selected frequency span. Table 1, posted at the beginning of this thread, shows that the typical noise floor is in the order of -130 to 140 dBm, equivalent, in a 50 Ω system, to 0.07 to 0.022 µV. We also know that a SSB signal starts becoming useful, i.e. efficiently readable, at a level of -121 dBm, or 0.2 µV, which corresponds to S1. On the other hand , even in a “quiet” area, an efficient receiving antenna, with optimal output matching to the receiver, can easily output a noise level of -100 dBm, with the worst situation generally happening at 3.6 MHz.
Here, we must state again, we are considering reception only, so we are not concerned with the maximum possible signal transfer, as we should do if we considered transmission as well. Therefore our main objective is to optimize signal to noise ratio, not matching and maximum power transfer.
2. DYNAMIC RANGE
The maximum level a RDS class receiver can withstand ( MW to HF) before overloading is at a level of -43 dBm (1,600 µV) as can be easily shown by tuning in a strong station in the MW broadcasting band (please note that the overload threshold is also determined by the PC system you are using: the bigger the capacity, the higher the threshold. Details of PC system used are given below).. This means that the higher the noise we bring in by optimally matching the antenna, the lower will be the receiver’s dynamic range. For instance with a noise floor at -105 dBm the receiver’s dynamic range will be in the order of 62 db, but with a noise floor of -130 dBm our receiver will be 87 dB far away from overload. This means that as long as a receiver is far from overload we shall be able to copy signals buried in the noise with adjacent signals being at S9+. So dynamic range optimization is another reason why we surely must NOT strive for “perfect” antenna matching.
PC: IBM/LENOVO X61. OS: WINDOWS 7 PRO RAM: 3 GB CPU: INTEL CORE 2 DUO T7300 – 2 GHz
- Figure 1 - Comparison showing noise /dynamic range improvement
- TOTAL 7 MHz.jpg (95.48 KiB) Viewed 7863 times
Reason: No reason
Old timers in the long past fifties always recommended to youngsters like me (then!) to beware of antenna ground, as a very efficient noise source. The only receiver of the past I can remember with a balanced input was the HRO with a special drawer, the HRO did not have band switches, you had a plug-in drawer to insert for every band and the HRO reputation, as a very good communications receiver, was universally recognized. In this respect the RSP receiver seems to come out of an old timer’s dream, because the Hi Z balanced input lets you ISOLATE the antenna ground, while at the same time feeding the front end. It is this hardware feature that provides means to easily improve noise operation.
4. THE ANTENNA
All I have written until now was secretly, but not consciously, lurking in my mind, when I, with great thrill, unpacked my brand new RSPduo: the idea was to start HF testing with a long wire antenna, or better, a Beverage antenna, which, simply stated, is a long wire, as long as your available land goes, with a terminating resistor at the end. While a long wire shows directivity in both directions along the directions of the wire, the Beverage, being non resonant and terminated at the far end, has only one prevalent direction, as signals in the opposite direction are dissipated in the termination. These antennas, even if they work best on a poor ground plane (to achieve a useful elevation angle) still require a signal ground as little resistive and inductive as possible. In my case I drove two zinc coated big steel stakes 1.7 m into the ground at both ends. The antenna is 135 m long and suspended at an average height of 6m (19.7’). On the receiver side the antenna lead goes indoors with a length of 2 m (6.6’) to the matching transformer. The other input of the matching transformer is connected to the earth with a 7 m (23’) lead.
On the far side the antenna has a vertical leg 5.5 m (18’) going to the terminating resistor box, connected to the far end earth with a 1.2 m (3.9’) long lead. The 560 Ω terminating resistor is made with ten 5.6 KΩ ¼ W resistors in parallel. The three contentious points in this antenna are: A) the length of 2 m from antenna to matching transformer; B) the length of the earth lead at the receiving end and C) the antenna height, which in the literature is notably lower at 5’ to 9’, e.g. 1.5 to 3 m. Not considering point C) which should only influence the characteristic impedance, the “kosher” solution would be to drop the receiver side of the antenna down to the earth stake, thus increasing this vertical length to 7 m (23’) place a box there with the matching transformer inside and then come to the RSPduo with a coaxial feed, or with a balanced line. I provisionally tested a similar arrangement (but no coaxial or balanced line feed) by bringing the RSP near the ground stake, as advised by VK7JJ and described in the previous posts and obtaining a good result (4/5 dB advantage) not verified for more than the one field day. I plan to remote the RSPduo in a box and use a long & amplified USB cable to complete this type of test & confirm possible improvements in operation.
The antenna characteristic impedance to be met, as calculated by standard formulas and as confirmed by VK7JJ’s modelling, is in the order of 550 Ω.
- First 15 m. signal ever received on July 1st at 9:06 local time from Athens, Greece
- First signal ever 21 MHz!.jpg (184.35 KiB) Viewed 7846 times
Reason: No reason
Having designed and built many ferrite wideband matching transformers in my professional life, my idea was to build wireline transformers for this job. As testified by my various posts in the referenced thread, I built many and was able to achieve near perfect match and therefore near zero insertion loss.
Figure 2 – Several wireline and two transformer baluns - Figure 3 – Trifilar balun schematic
A wireline balun is essentially an autotransformer and by definition it has a lower insertion loss than a transformer, wich has separate primary and secondary windings. Being an autotransformer, the wireline balun does NOT have galvanic separation between input and output, as shown in Figure 3 below.
This means that there is NO GALVANIC ISOLATION between antenna ground and RSP input, regardless of the input used. In order to minimize insertion loss and to bring power transfer to the highest level possible, the windings are made of as many copper wires tightly twisted together. This closest coupling increases the parasitic capacitance among windings which hinders the balance & symmetry of the autotransformer. In other words with the wireline balun we have the least possible insertion loss and the best possible match at the cost of balance and parasitic capacitance, which promote spurious noise transfer. With wireline we are trading what we do need (minimum noise transfer) for what we do not need, minimum insertion loss & maximum power transfer. All this came to my mind the day I tested the Beverage antenna with minimum lead-in, as suggested by Phil, VK7JJ.
It is now clear how the balun must be designed and constructed: it must be a TRANSFORMER type balun, affording galvanic isolation between primary and secondary and must exhibit the minimum possible capacitance between primary and secondary. Just to put the above into perspective, a trifilar wireline balun wound on a 25 mm (1”) dia. ferrite core will exhibit an inter-winding capacitance of between 50 to 150 pF, while a properly wound transformer type on the same 25 mm core will show a capacitance of 6 to 12 pF. “Properly wound” means that primary winding is on one side of the core and secondary winding as far as possible away on the other side. One winding above the other is simply a NO-NO, because of added parasitic capacitance.
The other very important and not to be overlooked feature of the transformer balun is that it keeps the antenna lead always galvanically connected to ground, while providing galvanic isolation to the (supposedly) delicate RSP receiver input. It could save the receiver's life (or yours?) should you forget to disconnect the aerial before a lightning storm. I am not talking of a direct hit, but even induced voltages could destroy the RSP receiver and/or give you a big shock.
ADDED AS OF 08/07/18: Concerning safety, there is another important aspect peculiar to the Beverage. We have a ground at the receiver's side and a ground at the termination side. Often there is a voltage differential between the two grounds. I once measured 60 V A.C./50 Hz. So the antenna lead , through the terminating resistor, can be several volts above ground at the receiver's side. ANOTHER REASON FOR MAKING SURE THERE ALWAYS IS A GALVANIC SHORT AT THE BALUN'S PRIMARY!
Enamelled copper wire I used in all windings is 0.5 mm dia.
Anybody can test the difference on the RSP receiver’s noise floor in a few minutes:
- Build a transformer type balun on a toroid of any reasonable material (43, 73, etc.). Depending on
the nominally matched impedances use:
- Ns/Np = SQR(Zs/Zp)
- Example: Zs=1000 Ω ; Zp=500 Ω; SQR(2)=1.41; Use Np=3 T & Ns=3*1.4 = 4 or 5 T.
- Connect secondary between P and N terminals of the Hi Z input.
- Connect primary between antenna terminal & ground.
- Do not glue or resin the windings, just let them free to move along the core.
- Switch receiver on, try the 40 m band and note the noise floor: let us suppose it is at – 125 dBm.
- Now take a jumper and temporarily connect terminals N & GND: noise floor should move up, by a minimum of 5 dB. If you are also receiving a signal you will note there is no change in signal intensity.
- Remove the jumper and carefully move the two windings towards each other as close as possible: again you will note an increase on the level of the noise floor, normally more than +5 dBm, but it depends on local conditions.
The above tests are the more effective and convincing the higher the level of atmospheric white noise present at the moment. In my location, Piedmont wine country, Northern Italy, at the moment I experience the strongest noise before sunset, e.g. 16 to 18 hours, local time, 14 to 16 hours GMT.
Here below another example in Figure 4: going from a trifilar balun (GND terminal connected) to a transformer balun with secondary to P & N terminals only.
- Figure 2 - General wireline balun schematic
- Balun schematic.jpg (22.81 KiB) Viewed 7858 times
- Figure 3 - Several baluns
- Several baluns.JPG (107.39 KiB) Viewed 7858 times
- Figure 4 - Comparison on 14 MHz
- TOTAL 14 MHz.jpg (116.3 KiB) Viewed 7858 times
Reason: No reason
But we must not forget the dynamic range issue. Depending on local conditions, HF frequency band and ferrite core material we can develop a transformer not only with minimum possible capacitance feed thru and ground isolation, but with an optimum attenuation (through mis-match) which will maximize the dynamic range, without affecting the receiver’s ultimate sensitivity. It is an old noise figure concept, that NF in higher frequency bands (VHF, UHF & up) is optimized by mis-match. In this respect the figures already given in Table 1 can be used as a yardstick: in general one should dimension the transformer in order to keep the noise floor between -135 and -120 dBm. Another smart trick, with excess antenna white noise, is to use two transformers in series: the first balun with a 1:1 Z ratio followed by another with the actual calculated turns ratio. In this way transformer parasitic capacitance can be reduced to a very low value, due to the two capacitances in series.
Figure 7 shows how to prepare transformers for quick testing, as any one can be changed in a matter of seconds by using two way plugs currently available.
As already stated, this paper is only concerned with the HF portion of the spectrum, which is a small part of the entire RSP receiver’s capability. And to tell the truth, the methods herein described have been tested on the 15 m band only once and never on frequencies above, due to the present remarkably bad propagation situation caused by the current low sunspot cycle. Higher HF spectrum could be the subject of a future post. This fact also points to the positive effect of placing the balun and the receiver very near the antenna / ground terminals, as suggested by VK7JJ: as a matter of fact no signal on 15 m. has ever been received with a longer ground / antenna leads in since then.
7. SELECTED BIBLIOGRAPHY
- The A.R.R.L. Antenna handbook.
- The Amateur Radio handbook – RSGB
- John D. Kraus – Antennas
- Figure 5 - Transformer balun at 3.6 MHz
- 3.6 MHZ Good balun.jpg (188.25 KiB) Viewed 7847 times
- Figure 6 - Wireline balun at 3.6 MHz
- 3.6 MHz wireline balun.jpg (186.71 KiB) Viewed 7847 times
- Figure 7 - Transformer type baluns with quick connect/disconnect provision
- DSC_1122 (Medium).JPG (75.79 KiB) Viewed 7847 times
Reason: No reason
I would very much like to commend you on an exceptionally high quality and informative thread. This level of technical information is both rare and invaluable to the whole community.
Similarly the contribution from both yourself and Phil (vk7jj) regarding noise measurements and antenna modelling is equally impressive and extremely helpful.
Thank you very much.
Reason: No reason
I am glad to be of little help in the general understanding of the important revolution in communications which SDRplay receivers have brought about. To an old timer like me, who could experience the cost, the shortcomings and the technical limitations of HF communication receivers of the past, the daily use of the RSPduo is an astonishing experience, to which it is difficult to ever get accustomed.
Thanks, kind regards & do try to keep us astonished again with more!
P.S. I have verified that in a thread every post can be amended, save for the first or introductory post. This feature is very annoying, because errors can creep into any post, even in the first....
Reason: No reason
thank You for your very classy article.
I will read it for several times and compare with my own observations.
Reason: No reason
Thanks for your comments. Your experience will also be very useful and appreciated by all of us.
Reason: No reason
- Near overload @ 7 MHz. The S meter was above 9+40, but the PC screen collecting software reduced it while taking the picture
- Example of system overload.jpg (201.03 KiB) Viewed 7422 times
Reason: No reason
Lower frequency HF bands are plagued in daytime by impulse type industrial noise which mainly propagates through ground(s). The problem is compounded with the Beverage antenna which needs two grounds: one at the receiving end and the other at the termination. Any noise in between grounds will get thru. In my location daytime noise is in the order of -100/105 dBm on the 80 m band and -115 dBm on the 40 m band.
Yesterday it came to my mind that the balanced antenna configuration would be ideal for the insertion of a high pass filter which could cut off the fundamental and higher harmonics of the industrial noise. The filter should have a good impulse response, otherwise it would be uneffective. This rules out higly selective Chebyshev or Cauer responses, the preferred being the Butterworth prototype. Anxious to test the idea, I quickly calculated an N=4 Butterworth highpass with a 3 MHz cutoff and an attenuation of 24 dB at 1.5 MHz. It works, just look at the PC screens below.
To test it of course you must have industrial noise on air. The first tests revealed an improvement of over 10 dB on 80 meters and over 5 dB on 40 meters. Lower improvement on 40 meters is probably due to (a) lower initial interference and (b) to noise feedthru jumping over the terminals, because of a very quick and bad quality contruction (see illustration).
The fiter used is by no means optimal, it is just an example. A properly designed filter would have to be balanced (e.g. capacitors in series on the ground lead as well) and with several more sections: an n=8 Butterworth highpass, constructed with proper shielding, would probably be ideal for this job.
I am convinced that this idea needs further testing and hope that other Forum members will try it and post the results. Additional illustrations in next post.
- Noise well reduced by filter insertion
- Filter suppressed.jpg (188.97 KiB) Viewed 7400 times
- High level industrial noise on 80 meters
- Direct connection.jpg (188.28 KiB) Viewed 7400 times
Reason: No reason