Receive Antenna & Impedance "Matching"

Useful information regarding antennas for SDR products.
Post Reply
glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: Receive Antenna & Impedance "Matching"

Post by glovisol » Sat Jun 15, 2019 9:34 am

Hi sdrom33, you always persecute me with your keen eye, but you are right: that pic was taken with the RSP-1A overloaded in Spectrum Analyser service. I shall post the correct pic later on.

Hi Phil,

You lucky guy, you are still young enough to dance, not for me, very near to 80 in September!

You are really making me curious with this WSPR business and I shall take time to get the information on the site you uploaded a few posts back. Until now my attention has been devoted to the realisation of inductances with suitable Qu and ways to make pre-selector prototype setting up and testing easier and quicker.

By the way, I have looked up commercial pre-selectors on Internet and they look uselessly wideband, only good for guarding hundreds of KHz of bandwith against dangers sited MHz away: this cannot be helped if you resort to low Qu components.

Now is time, like you note, to devote attention to coupling. There is a misunderstanding here. I couple using binocular transformers when testing on the Spectrum Analyser, because both the Noise Generator and the RSP-1A have 50 Ohm terminations, but the pre-selector should work directly between Beverage antenna (450 Ohm) and HI Z input of the RSPduo, 1 KOhm.

So I have prepared ad hoc data: schematic diagram of the variable coupling pre-selector and Excel spreadsheet for quick calculation of number of turns for input/output links and relevant QL and insertion loss. These are uploaded below with relevant formulas, so that different values can be calculated. For 50 Ohm terminations I am afraid one will have to resort to higher inductance,in order to have at least 1 turn link, OR to use capacitive taps, OR to use taps on the coil itself, OR use ferrite transformers.

ERRATA CORRIGE: The Insertion loss formula shown in all calculations lacks a factor of 1.41 due to the pre-selector being double tuned, e.g.:

IL=20*1.41*LOG(1-(Ql/Qu)


However the spreadsheet calculated value is correct, the error being only in the formula, but not in the calculation.

This work shows that narrow 3 dB 20KHz BW can be achieved with links N1=1.5 turns and N4=2.5 turns and the insertion loss is over 15 dB with 25 mm dia. unspaced coils.

Wide 3db 180 KHz bandwith requires links N1=3 turns and N2=5 turns and the insertion loss goes down to 2 dB with 25 mm dia. unspaced coils.

These calculations explain the problem of high insertion loss in recent practical reception testing, as the turns for the all links were preliminarily set at N=1.

Will test over the weekend.

Kind regards,

Gianfranco
Attachments
Narrowband link. calc.jpg
Narrowband link. calc.jpg (82.54 KiB) Viewed 1799 times
Wideband link calc.jpg
Wideband link calc.jpg (81.22 KiB) Viewed 1799 times
Antenna double tuned Pre-selector 7 MHz (2).jpg
Antenna double tuned Pre-selector 7 MHz (2).jpg (94.06 KiB) Viewed 1799 times

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: Receive Antenna & Impedance "Matching"

Post by glovisol » Sat Jun 15, 2019 12:26 pm

Phil,

For narrowband work between 50 Ohm terminations, links must be half turn, so the simplest solution is to use half turn taps, as shown in the calculation. Otherwise the most practical solution is to use 2 turn links and match with wideband ferrite transformers: in this case the link impedance is 800 Ohm and 800 to 50 Ohm ferrite binocular transformers must be used.

ERRATA CORRIGE: The Insertion loss formula shown in all calculations lacks a factor of 1.41 due to the pre-selector being double tuned, e.g.:

IL=20*1.41*LOG(1-(Ql/Qu)

However the spreadsheet calculated value is correct, the error being only in the formula, but not in the calculation.
Attachments
narrowband 50 Ohm link calc..jpg
narrowband 50 Ohm link calc..jpg (88.47 KiB) Viewed 1798 times
Narrowband 800 Ohm link calc..jpg
Narrowband 800 Ohm link calc..jpg (86.2 KiB) Viewed 1798 times

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: Receive Antenna & Impedance "Matching"

Post by glovisol » Sat Jun 15, 2019 2:45 pm

First test this afternoon with calculated & appropriate links in place. Spectrum range has been altered to clearly see pre-selector response. This is the wideband 3 dB BW=160 KHz. Insertion loss negligible, as predicted. In these conditions strong Broadcasters above 7,200 KHz should be well attenuated. We shall see later this evening.
Attachments
WB operation 2.png
WB operation 2.png (328.47 KiB) Viewed 1787 times

Reason: No reason

vk7jj
Posts: 206
Joined: Thu Mar 02, 2017 9:56 am
Location: Tasmania

Re: Receive Antenna & Impedance "Matching"

Post by vk7jj » Sun Jun 16, 2019 5:57 am

Your test pic is looking really good, brilliant in fact!

That bandwidth and slope would likely make a lot of difference for many listeners plagued with strong unwanted signals.

Dancing is great, we've been members of a local rock n roll dance club for some years.

If you want a look at the scope of WSPR stations globally on a realtime map plus data tables and charts you could try my offering. It's rather piecemeal as I've added to it on an ad-hoc basis for my own needs, basically. http://wspr.vk7jj.com/

Back to the subject in hand which is almost a return to the thread title.

If you're playing round with coupling loops, perhaps it's worth while trying to resolve the fractional problem by starting out with the loop at the centre of the coil and moving it outwards to find the sweet spot. Originally I was planning to ground one end of the tank coil thinking I may need to for stability and so I put my coupling link at what was going to be the earthy end which ended up being 50 ohms, but your work has resulted in a balanced system. Maybe placing the coupling at the centre even if it requires more turns may be better from a loading point of view.

Using a coupling loop has the inherent advantages of galvanic isolation and a DC short circuit for antenna static bleed, as well as being able to work directly with coax or a balanced line on the input or coax to the RSPs coax A and B inputs or straight to the RSP2's balanced input so all of that is handy.

In the past I've built boxed helical resonators on higher frequencies but not much in the way of what we are doing now and have little real-world experience with tank coils since the old valve days apart from that squid pole, so I'm not much help.

I did get time to build a slider as per your layout and play with it, a long plastic mesh filter from the usual mains water cheapo irrigation system turned out to be the right diameter. Finding a foolproof way of sturdily mounting my beehive trimmers has temporarily stopped me at the moment.

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: Receive Antenna & Impedance "Matching"

Post by glovisol » Sun Jun 16, 2019 6:14 am

Impossible to do practical testing yesterday night, due to very stormy weather. Phil thanks for your suggestions, I shall reply in next post, as I am busy with relatives today. In the meantime I am uploading Spectrum Analyser frequency responses for the "wideband" solution. After practical tests with the wideband prototype, I shall change the links to test the "narrowband" solution & post relevant data. Numbers 3-5 in the pictures' captions are number of turns for input/output links.
Attachments
1. Wideband 3-5.jpg
1. Wideband 3-5.jpg (209.64 KiB) Viewed 1743 times
2. Wideband 3-5.jpg
2. Wideband 3-5.jpg (209.83 KiB) Viewed 1743 times
3. Wideband 3-5.jpg
3. Wideband 3-5.jpg (209.15 KiB) Viewed 1743 times

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: Receive Antenna & Impedance "Matching"

Post by glovisol » Sun Jun 16, 2019 6:15 am

More "wideband" responses.
Attachments
4. Wideband  3.5 critically coupled.jpg
4. Wideband 3.5 critically coupled.jpg (142.97 KiB) Viewed 1742 times

Reason: No reason

vk7jj
Posts: 206
Joined: Thu Mar 02, 2017 9:56 am
Location: Tasmania

Re: Receive Antenna & Impedance "Matching"

Post by vk7jj » Sun Jun 16, 2019 8:14 am

Nice pics again Gianfranco.

Looking at them shows another potential benefit of this preselector, the out of band attenuation is so good it would go a long way to protecting an RSP's front end from "other band" transmissions.

Annoying for you that the stormy weather got in the way, best hopes the relatives were of the enjoyable sort that made up for it :-)

Regards, Phil

Reason: No reason

vk7jj
Posts: 206
Joined: Thu Mar 02, 2017 9:56 am
Location: Tasmania

Re: Receive Antenna & Impedance "Matching"

Post by vk7jj » Mon Jun 17, 2019 4:18 am

Hi Gianfranco.

Have you given any thought to how SDRuno's calculation of SNR might be influenced by a narrow preselector?

If SDRuno calculates SNR as a ratio of wanted signal to passband power, if the SNR algorithm relies on samples or averages of spectrum power in the current bandwidth and assumes a flat passband then … ?

I did a quick search in the sdrplay docs but couldn't see anything that shed any light on the subject. If Tech Support happens to read this I wonder if they might like to comment, I'd love to know more.

Phil

Reason: No reason

glovisol
Posts: 662
Joined: Thu May 10, 2018 6:42 pm
Location: Piedmont, Italy

Re: Receive Antenna & Impedance "Matching"

Post by glovisol » Mon Jun 17, 2019 6:47 am

The "wide" pre-selector tests confirmed the advantage of attenuating very strong broadcast signals above 7,200 KHz. With a limited number (2) of limited Qu bandpass elements, increased selectivity is obtained at the price of a higher insertion loss, as shown in the uploaded screens below. The double tuned bandpass "wide" pre-selector provides an average attenuation of the broadcast signals of 10 dB to 30 db, depending on frequency difference, when at maximum coupling (no insertion loss). This means that the RSP can work at maximum dynamic range (unless interference exceeds pre-selector attenuation) and the RF gain can be set at maximum, with no overload alarm. Unfortunately, reducing the coupling, it is true we increase the selectivity, but at the same time we reduce system sensitivity, so the net gain is zero or thereabouts.This device has the disadvantage of being bulky and causing shielding to be problematic.

However potentially equal or better performance could be obtained with a fixed tuned Cauer type bandpass filter passing the wanted bandwith only (7020 - 7180 KHz) and with very sharp transitions due to near band attenuation poles. For example an Elliptic bandpass filter N=3, 0.5 dB ripple, should provide >46 dB attenuation at 6.9 & 7.3 MHz with low insertion loss and low Qu toroidal coils. A quick calculation showed that filter components for this filter are realizable, at least at 7 MHz. All this to be proven by practical testing, of course.

A different situation is presented by the "narrow" pre-selector, which we shall test & analyse next.

Phil, I do not think the SNR measurement is an issue here. In fact it would be a dream if we had the technology to limit the received bandwith to such an extent as to be able to alter the S/N ratio measurement. We shall look at this issue in detail soon.
Attachments
Minimum couplimg.png
Minimum couplimg.png (342.04 KiB) Viewed 1651 times
Maximum coupling.png
Maximum coupling.png (342.51 KiB) Viewed 1651 times

Reason: No reason

vk7jj
Posts: 206
Joined: Thu Mar 02, 2017 9:56 am
Location: Tasmania

Re: Receive Antenna & Impedance "Matching"

Post by vk7jj » Mon Jun 17, 2019 10:19 am

Hi Gianfranco, sounds like the Pandora box of filters is open!

My own needs are well satisfied with a sharper filter despite it's loss.

One of the reasons for that is the difference in our antennas and another is the difference in what we each are trying to listen to. Your antenna is directional and its design goal is lower noise at the expense of gain, mine is omnidirectional and its design goal is to maximise gain across as many ham bands as possible, so I'm presuming we are working with significant differences in the number of signals received and their strength.

I'm trying to exclude as many unwanted signals as possible so I can decode them as low down into the noise as -33dB and only within 200Hz of bandwidth while I'm presuming you are enjoying listening to a wide range of signals within any given "band" as comfortably above the noise as possible.

More people want what you want than what I want so it certainly makes sense to pursue as many other solutions as present themselves and it sounds like you're right on the money with your suggestions :-)

The SNR issue still bothers me though. After a bit of searching I found this article by Jon which highlights the bandwidth issue. Clearly people who know much more than I do are wanting to log SNR and are choosing to do it over specified bandwidths.

https://www.rs-online.com/designspark/u ... asurements

That begs the question: if the bandwidth is not specified and SDRuno derives it's SNR by measuring the total power across whatever bandwidth the user has set by choosing to view (say) a 2MHz swathe of spectrum, it would also be assuming that 2MHz of bandwidth was flat.

Looking at your pics even with the wide filter the bandwidth is very unflat and the band edges are very many dB down on band centre, the result of a very successful Glovisol build :-)

I would love to be corrected, I'd like to be able to confidently measure my SNR, but I can't see how the (default?) SNR will not be significantly altered by that.

So here's the thing, taking Jon's link above into account the inference seems to be the above problem may possibly be overcome by a user specifying a particular bandwidth for SDRuno's measurement, ie. we can measure SNR over (say) the -3dB points.

Regards, Phil

Reason: No reason

Post Reply