Testing RSPduo receiver sensitivity with noise generator

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

Re: Testing RSPduo receiver sensitivity with noise generator

Post by glovisol » Tue Jun 26, 2018 7:48 am

Thanks Tech Support, very informative and timely!

glovisol

Reason: No reason

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

Re: Testing RSPduo receiver sensitivity with noise generator

Post by vk7jj » Wed Jun 27, 2018 12:52 am

Decimation will have no discernible affect on sensitivity or noise figure.
"descernable" ... depending on whether you watch the small snr reading on the screen stay constant or you watch the visible "noise floor" drop in level as you increase the decimation!

Thank you very much Tech Support, having forum input from you is one of the many good reasons for choosing and enjoying your products. I sure wanted to believe that article though I did want to see it actually measured.

--

Hi again glaviscol!

My first transistor was an OC70 as an add on to a crystal set. After that I scraped the black paint off and used it as a photo-transistor, after which I killed it, cost me a whole weeks work in school holidays.

Re. signal leakage, I wondered how much signal got around your unshielded (series as opposed to shunt?) attenuators though it doesn't matter because your signal measurements were relative as you measured the differences.

Have had that problem here, I use an AA-1400 (0.1-1400 MHz) antenna analyser as a sig gen and struggled with ad-hock attenuators until I built one along these lines in a die cast box:

https://diodetech.blogspot.com/2013/07/ ... uator.html

Fine on the Beverage, love to hear how it goes. Never built one myself mostly because of it's directionality but their noise performance has achieved legendary status

In the end after a lot of work leading to an appreciation of how RF travels along mains wiring and from there along your coax braid radiating all the way, then radiates some more as an integral part of your antenna unless it's a loop with a balanced feed line or you take a lot of care with common mode chokes in the shack as well as at the antenna, I've ended up with a wide band low noise omni alternative with inexpensive wire loops. /end soapbox.

Regards, Phil

Reason: No reason

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

Re: Testing RSPduo receiver sensitivity with noise generator

Post by glovisol » Wed Jun 27, 2018 8:52 am

Hi Phil,

Glad that Tech Support put the decimation issue to bed. As I wrote you yesterday, I was initially very curious about it, and tried to discover differences in sensitivity, while with very weak signals from a short antenna, but did not discover any.

Now I understand...in order to kill leakage I took care to use not one, but two attenuators in series, each providing only 46 dB, the attenuators being "H" type, not "T" type: e.g. one shunt resistor on each side and one series resistor in between. Of course the RSPduo shield is very effective and as soon as you remove a connection from the input, nothing gets thru. Yesterday I discovered the detailed specs on the RSP duo and was gratified to see a very good correlation with Table 1. Also the Hi Z sensitivity advantage at lower frequencies is confirmed in the specs.

For measurements I use a Picolog 2204A PC scope good up to 10 MHz. This unit, thanks to its excellent software, works also as a spectrum analyser and I am using it now while developing baluns for the Beverage. It is very useful to have a spectrum analyser using Hi Z, compensated scope probes, it gives you a great deal of freedom. There is a lot of material published on baluns, but not very convincing, as many claims are made, but no measurements shown. I am active on the Picotec site and below is the last of many published works of mine.

https://www.picotech.com/support/topic36891.html

For the Beverage at present I am just concerned with listening, so I have no problem with RF travelling around...but I am in the process of designing parts for a "third method" SSB generator, of which the audio filter is the first part. I have also taken care in providing a good ground for the far end antenna termination and it is amazing to see the noise go away on 3.6 MHz when you connect the termination there...

Kind regards,

Gianfranco

Reason: No reason

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

Re: Testing RSPduo receiver sensitivity with noise generator

Post by vk7jj » Thu Jun 28, 2018 1:01 am

Good morning Gianfranco,

You have certainly been busy! Well done, lots of effort and good insights. And baluns are part of the black art of amateur radio, a mystery to all but the cognoscenti. Every Amateur Must Have One!

Well, it is true that they are very useful things and also that we need to make and measure our own; that was part of the self-justification for buying that antenna analyser because it's let me build and test and compare everything. If you are contemplating anything that looks like any of these, which are all different, I've been able to measure and compare them very accurately from 1KHz to 1.4GHz. It can even measure the difference between 1 and 2 turns of that red wire on the snap on.

If you want to know which of the winding techniques offers more impedance, what the typical impedances are at any given frequency etc, I could put you in the ballpark.

Regard, Phil.
chokes-overview.jpg
chokes-overview.jpg (284.46 KiB) Viewed 24650 times
Testing arrangement for basic quick evaluation. It does also have a companion Windows app for PC control and large screen display.
test.jpg
test.jpg (111.99 KiB) Viewed 24650 times
Information available:
AA-1400.jpg
AA-1400.jpg (50.43 KiB) Viewed 24650 times

Reason: No reason

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

Re: Testing RSPduo receiver sensitivity with noise generator

Post by glovisol » Thu Jun 28, 2018 12:49 pm

Dear Phil,

Thanks a lot about your offer of assistance, which I surely need. I can see from your photos that you must have accumulated a big experience!

The necessary complement to a Beverage antenna is a properly working balun. To feed an RSP type receiver we need a nominal 600 Ohm antenna input to a nominal 1000 Ohm output for the Hi Z port and a 600 Ohm antenna input to a nominal 50 Ohm output for the 50 Ohm ports.

I am waiting for fully rated 73 material, toroids and binocular cores, but in the meantime I have characterized and used toroids I already had in the lab. I am enclosing the results for your comments & guidance. From the time I was professionally designing solid state RF amplifiers for HF SSB (1970 to 1980) my experience is with twisted wireline, this is why I have used wireline, although I do not know if common windings are better here.

Beverage impedance calculations give a value very near to 560 Ohm, therefore:

50 to 560 Ohm - impedance ratio: 11 - turns ratio: 3.3
560 to 1000 Ohm - impedance ratio: 1.78 - turns ratio: 1.33 very near to 1.41
SEE SCHEMATIC BELOW

With a trifilar balun we are spot on with terminals 1-3 for 50 Ohm and 1 - 6 for 560 Ohm.
With a quadrifilar balun, we have a turn ratio of 2 with terminals 1 - 4 for 560 Ohm and 1 - 8 for 1000 Ohm, hence impedance ratio is 1.41, very near to 1.33.

At present the quadrifilar gives best (and extremely remarkable) practical results on 80 and 40 meters, but I still have to do the measurements. Here are the measurements on trifilar units so far:

Experimental baluns Frequency range 1 – 10 MHz

All baluns on 25 mm diameter toroids. Trifilar or quadrifilar twisted wireline.
Core data: 20 turns produce 26.5 µH, hence Al = 26.5*1000/400 = 66.3 nH/T
Noise generator output is 350 mV RMS on 50 Ω load – P = 2.45 mW

1. Trifilar balun 6T test with 560 Ω load:
Input reading: 520 mV RMS (input impedance far from 50 Ω)
Theorical output reading: 1560 mV RMS
Actual output reading: 820 mV RMS
Loss: 20 LOG (1560/820) = 5.6 dB
Test screen #2
Coupling too low, power not transferred

2. Trifilar balun 28T test with 560 Ω load:
Input reading: 500 mV RMS (input impedance far from 50 Ω)
Theorical output reading: 1500 mV RMS
Actual output reading: 640 mV RMS
Loss: 20 LOG (1500/640) = 7.4 dB
Test screen #3
Too many turns, core eats away lots of power

3. Trifilar balun 11T test with 560 Ω load:
Input reading: 330 mV RMS (input impedance very near 50 Ω)
Theorical output reading: 990 mV RMS
Actual output reading: 970 mV RMS
Loss: 20 LOG (990/970) = 0.177 dB
Test screen #4
Measured total inductance: 70 µH, hence at 7 MHz we have Xl = 3077 Ω, thus 3077/560 = 5.5 core does not contribute to the load.
Conclusion with 11 turns we are in the ball park

4. Quadrifilar balun 11 Turns to be measured.

More PICS on next upload.


Kind regards,

Gianfranco
Attachments
#1_IN-OUT.jpg
In-out noise voltages as read by scope - true RMS
#1_IN-OUT.jpg (68.98 KiB) Viewed 24503 times
11T Balun (Medium).JPG
Trifilar balun
11T Balun (Medium).JPG (77.75 KiB) Viewed 24503 times
Balun schematic.jpg
General schematic
Balun schematic.jpg (24.14 KiB) Viewed 24503 times

Reason: No reason

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

Re: Testing RSPduo receiver sensitivity with noise generator

Post by glovisol » Thu Jun 28, 2018 12:52 pm

More PC screens enclosed
Attachments
#4_Balun 11t.jpg
11 Turns balun performance - excellent
#4_Balun 11t.jpg (71.88 KiB) Viewed 24396 times
#3_Balun 26T.jpg
26 Turns balun performance - not acceptable
#3_Balun 26T.jpg (71.12 KiB) Viewed 24396 times
#2_balun 6t.jpg
6 Turns balun performance - not acceptable
#2_balun 6t.jpg (71.63 KiB) Viewed 24396 times

Reason: No reason

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

Re: Testing RSPduo receiver sensitivity with noise generator

Post by glovisol » Thu Jun 28, 2018 12:57 pm

More PC screens enclosed.

Use of a correctly matched balun lets you use all the sensitivity advantage of the Hi Z input from broacast band up to the 40 m band. By using normal windings on the cores (not wireline) it is possible to keep the Hi Z input balanced as well (terminal "N" floating) and this arrangement should provide a noise rejection advantage. I shall try this as soon as I have the rated 73 material cores.
Attachments
ScreenHunter_71 Jun. 27 22.11.jpg
Same good performance on 80 m with quadrifilar.
ScreenHunter_71 Jun. 27 22.11.jpg (190.86 KiB) Viewed 24286 times
ScreenHunter_69 Jun. 27 22.06.jpg
Full 40 m band with the quadrifilar balun, amazing to me!
ScreenHunter_69 Jun. 27 22.06.jpg (209.94 KiB) Viewed 24286 times
#5_Scope Voltages- 11 T Balun.jpg
In/out voltages as een on scope for 11T balun
#5_Scope Voltages- 11 T Balun.jpg (96.42 KiB) Viewed 24286 times

Reason: No reason

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

Re: Testing RSPduo receiver sensitivity with noise generator

Post by vk7jj » Fri Jun 29, 2018 12:40 am

Wow Gianfranco! You've swept me away! I wonder how can I best contribute to your efforts! I've never modelled a Beverage before so it's a great time to try.

Fundamental frequency of operation: 7.05MHz
Construction:
#1 vertical wire 0.78m high, feed point at the bottom of this wire, driven against ground
#2 horizontal wire 90.6m long, 0.78m above ground, connected to the top of wire #1 and the top of wire #2
#3 vertical wire 0.78m high, connected to ground via a 500 Ohm non-inductive load

The table shows it is remarkably flat across the bands from 40m up, with a radiation resistance around 400 Ohms and little bothersome reactance.
The next two pics show radiation pattern and angle of take off at 7.05MHz and 51.0MHz. As you would expect the other frequencies range in between.
beverage-calcs.jpg
beverage-calcs.jpg (103.7 KiB) Viewed 24151 times
radiation-7.05.jpg
radiation-7.05.jpg (119.93 KiB) Viewed 24151 times
beverage-51.jpg
beverage-51.jpg (121.24 KiB) Viewed 24151 times
I would simply feed it with 450 ohm ladder line. If you have a problem bringing ladder line into your shack because of bad close contact with metal parts of the building, then you could use twin lengths of 70 ohm TV coax as a shielded balanced line.

At the radio, if it is an RSP2, I would connect it straight to the HiZ input through a common mode choke. A simple twisted pair on a toroid. In the picture of my chokes from my previous post the toroid with the orange twisted pair was built explicitly for 450 Ohm ladder line.

Here is a point that has always seemed to me to be important; please find fault or I'll keep on being wrong!

If you use a toroid to transform impedances then your frequency range and efficiency will be limited by the ferrite, which of course is why you're chasing 73 material. That limits your received signal bandwidth, eg. no single ferrite material goes from DC to 70cm.

If you don't go that way and instead use the inductance of the core to provide as high an impedance as possible just so as to reduce common mode signals over the frequencies of interest, then you are *not* limiting your received signal bandwidth. What you are doing instead is limiting the effectiveness of the common mode rejection and balanced to unbalanced performance, but only outside the ferrite's range. Neither of those characteristics matters nearly as much as losing all that frequency response!

Also (imho of course) I'm willing to suffer a non-perfect impedance match. If you use a balanced feed line of simple 300 ohm, 450 ohm or twisted pair (where necessary I use home made twisted pairs or home made ladder lines so I can create an impedance to suit my antenna) then there is negligible power loss caused by the poor SWR in your feed line, especially at HF, because as you know the losses caused by standing waves are because of additive feed line loss, principally in coax. Not only that but a balanced line and antenna inherently helps reject the common mode signals that plague coax.

So I've come to choose and design antennas and feed lines so as to require as little as possible impedance transformation: impedance transformation whether it uses L and C, coax stubs, etc must always be frequency dependent because we are dealing with reactances. Ferrites are not as bad but I've found I really need three different mixes to effectively cover from 100KHz to the top of 6m

Your Beverage seems like a great candidate for that approach with it's broad flat response.

It's great to compare notes Gianfranco, I used to be shy about putting myself out there to be knocked down but as I got old and didn't care as much I found it was the fastest way to learn from others and be presented with things I'd never seen or been able to do as you have done with your work, thank you.

Best regards, Phil VK7JJ

Reason: No reason

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

Re: Testing RSPduo receiver sensitivity with noise generator

Post by vk7jj » Fri Jun 29, 2018 4:01 am

General comment, not to you Gianfranco! (see prev post for reply to you) ...

As an afterthought, because the Beverage has such a flat response it is an exceptional antenna and hides the following problem with trying to match impedances. Here is a more usual antenna:

A 20m dipole at the ideal practical elevation of half a wavelength over perfectly conducting ground:
20m-dipole.jpg
20m-dipole.jpg (84 KiB) Viewed 24027 times
Let's say we want to feed it with 300 ohm ribbon and build a 4:1 "balun".
No worries, 4 times it's 79.87 ohm radiation resistance at 14.15MHz gives us a nice neat 320 ohms.

But on every other band it doesn't.

I guess most amateurs understand that because they try to transmit on them, but RSP users almost by definition are going to want to use the antenna on every frequency that tickles their fancy. Yet they are encouraged by the pressure of countless articles and countless people with baluns for sale to believe they've just gotta have a (say) 4:1 balun or their antenna won't work.

"End fed" wires are even worse of course.

And then we have the effect of their random length of coax feed line when it's connected to the ultra high off-resonance antenna impedance. How can anyone guess what impedance transformation has occurred at the RSP end of that coax run?

What to do? Shut up and let everyone get on with enjoying life or discuss the issues? I'm not sure!!

But nobody discusses the trade off of the effect of "baluns" limiting frequencies, presumably there is a frequency response chart of the Nooelec device somewhere I've not seen but I've built and tested enough small binocular ferrites to know already what it's likely performance is.

So I'd rather encourage people to use balanced antennas and balanced feed lines, home made or otherwise, along with a simple home made twisted pair common mode choke to assist with noise as mentioned in the previous post.

I've toyed with the idea of trying to build a small compact balanced loop using something like 3 x metal slinkys to post here for RSP users; that might offer a better alternative to all those end-fed-wires. I've got several and have used them configured for various antennas but time is a problem!

All a bit of fun and eminently do-able by anyone at home with an RSP and a great way to encourage more appreciation of the issues and get into the hobby a bit. :-)

Sincerely, Phil VK7JJ

Reason: No reason

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

Re: Testing RSPduo receiver sensitivity with noise generator

Post by glovisol » Fri Jun 29, 2018 4:21 am

Dear Phil,

It is six in the morning here and I have just seen your fantastic response. Unfortunately today I cannot even read it properly, as I have to go to sea, on the Riviera, to bring help to a friend of mine who has burned out the diesel on his sailing boat and I have to go there to see the work t be done for a rebuild.

Very many, many, many thanks for assisting me. Shall resume the good work tomorrow.

Kind regards,

Gianfranco

Reason: No reason

Post Reply