Author Topic: Correctly calculating impedance of a biconical antenna and impedance matching  (Read 30632 times)

0 Members and 3 Guests are viewing this topic.

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Hi
I want 100R terminators for my 3-port balun balance tester.  For the tests above, I used standard 50R terminators for the X-Y inputs.  These reduce the input impedance seen by the balun to 25R.  Using 100R terminators will look like 50R to the balun. 

If I could buy 100R terminators, I expect they will be way too expensive.  I turned to the huge shop that Aliexpress is and found cheap sma attenuators.  The BNC equivalent are about 3x the price.  I figured I could buy the sma attenuators and modify them to make DIY 100R terminators.

Given the price, I had low expectations on quality and performance.  What I found inside surprised me.
The attenuator is one-piece brass body construction with a sleeve to cover the insides.  This sleeve is a press fit.

Inside is a ceramic substrate with laser trimmed resistors. Measurement of the resistors indicates they are accurate.  The substrate is soldered to the body.  Easy to remove.  Based on physical inspection, the 8GHz bandwidth is plausible. 

It is possible that these are rejects from a production line for a name-brand manufacturer.  If they are, I can't see or measure any obvious defect.

The photo of the attenuator in pieces was the result of destructive exploration.  Now that I know how to open these up, they will be ideal candidates for making 100R terminators.


Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 
The following users thanked this post: ch_scr

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Re: Correctly calculating impedance of a biconical antenna and impedance matching
« Reply #101 on: February 04, 2024, 02:53:00 am »
Another ant step forward today.
I received the  SMA attenuator to replace the butchered one.   I am back to a pair of these.
I made a tool to remove the sleeve from the attenuators without damaging them.
I replaced the attenuation part with a straight-through wire and a 100R resistor to ground.

I now have two 100R scope terminators.  My scope has 13pF inputs so I should match this with inductance in series with the 100R resistor.  I have chosen not to do this because it will be difficult to make the impedance of both DIY modded scope terminators. 

The photos show the two completed sma terminators.  The tool flexs to open around the sleeve.  When gently closed, the tool grips the edges of the sleeve.  The aluminum punch fits into the sma socket to push the body out.  The same tool is used to install the sleeve.   The tool works without damage or marks.
Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21842
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Re: Correctly calculating impedance of a biconical antenna and impedance matching
« Reply #102 on: February 04, 2024, 04:20:55 am »
Mind, it's probably not full capacitance. You'll want to do a reflection measurement to determine what the actual input impedance is at high frequencies.  It's rated xx pF so that, in combination with a probe of yy pF, you get a 10x divider, but both of those actually have ESR, and the equivalent circuit trends towards a low-Z probe at HF.

Of course, what they actually are, they don't tell you, and won't be consistent from instrument to instrument, so you'll have to measure.

In any case, you'll likely find inductance is not necessary, or that less is necessary to achieve whatever peaking is possible [than would be calculated from the mid-frequency pF value].

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Re: Correctly calculating impedance of a biconical antenna and impedance matching
« Reply #103 on: February 04, 2024, 09:15:04 am »
Mind, it's probably not full capacitance. You'll want to do a reflection measurement to determine what the actual input impedance is at high frequencies.  It's rated xx pF so that, in combination with a probe of yy pF, you get a 10x divider, but both of those actually have ESR, and the equivalent circuit trends towards a low-Z probe at HF.

Of course, what they actually are, they don't tell you, and won't be consistent from instrument to instrument, so you'll have to measure.

In any case, you'll likely find inductance is not necessary, or that less is necessary to achieve whatever peaking is possible [than would be calculated from the mid-frequency pF value].

Tim

For this application (making a DIY 3-port VNA) it won't matter if the 100R terminators are exactly impedance matched.  What does matter is that both x/y inputs to the scope are connected with terminators that have the same impedance so the scope can display any imbalance in the balun. 

To test the balun, I will drive it in reverse.  I will be driving the unbalanced balun output with the unbalanced signal generator output.  The balanced antenna balun inputs will be connected to the x/y scope inputs.  With 2x 50R scope terminators, the signal generator will see 25R.  With the new 100R terminators, the signal generator will see parallel 2x 100R = 50R load on the x/y scope inputs.  The higher 100R resistance will increase the displayed amplitude of the x/y signals.   This simple vna will  work as long as the impedance of the x and y scope inputs is the same. 

Now that I have the tooling to open up this type of sma style attenuator, it is relatively easy to make custom adapters.   They are cheap but good, even when left original as attenuators.  I should be able to make custom scope terminators impedance matched to a specific scope.  The adapters are cheap enough to make a set of matched terminators for each scope.  For now, I just need 2x 100R terminators with the same impedance.

Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Hi
I has been a while since I posted on this topic.
Made some progress on my balanced to unbalanced 3-port VNA.  I need a fixture to hold everything in place.  This is a piece of Aluminium Clad Composite. 

At the bent end, two sma connectors and 100R oscilloscope terminators are fitted.  These will connect straight into the scope without cables.
Along the length of the fixture are holes for a chassis SMA connector connected to the unbalanced output of a signal generator.  This is the RF ground reference point.

I am going to try toroid ferrites as a balun.  It is not an original idea.  I can't find any literature on this as a balun.  The hypothesis is that it should have a much wider bandwidth than threading a coax as a coil through a larger toroid. 
Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21842
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
Try "current balun" -- choking the current flow (raising CM impedance) to the point where balance is good enough.  It can never be perfect, you can't null the current this way -- but it can be good enough, where "good enough" might be 20 to 40dB.

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Try "current balun" -- choking the current flow (raising CM impedance) to the point where balance is good enough.  It can never be perfect, you can't null the current this way -- but it can be good enough, where "good enough" might be 20 to 40dB.

Tim

Hi
The coax coil through a toroid was he result of research on current baluns.   Tests showed the upper frequency performance was limited by the number of turns.  Not enough turns = not enough CM attenuation.  Too many turns = reduced upper frequency response.

So my logic is to unwind the coil to make it zero turns.  Balanced currents will completely cancel out and the ferrites won't see a magnetic field.  The ferrites won't be excited so only coax loss.    The upper frequency response of balanced currents should be defined by the coax cable, not the ferrites.

Any unbalanced CM current is going to be seen by the ferrites, and impeded.  Any unbalanced current coming from the sig-gen is going to have the choice of a 50ohm cable, or a CM impedance.  I think it will look like a RF voltage divider, and I expect slight imbalance. 

If I find the expected imbalance, I plan to experiment with the tiny Ethernet torroids I rescued from a MoBo.  I plan to pass the centre core conductor of the coax through one of those tiny ferrites.  The aim being to add just enough impedance to achieve balance.  If it works, the trade off will be increased insertion loss of balanced signals. 

I key question to answer is how many ferrites will it take to provide a reasonable lower frequency limit?

In many applications, the long skinny form factor would be an issue.  For the Biconic, it would fit nicely inside the tube supporting the antenna.

If will be a while before I run any tests because I am waiting for some connectors/adapters from China.
Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21842
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
So my logic is to unwind the coil to make it zero turns.  Balanced currents will completely cancel out and the ferrites won't see a magnetic field.  The ferrites won't be excited so only coax loss.    The upper frequency response of balanced currents should be defined by the coax cable, not the ferrites.

That's... not quite how that works :)

For one: as long as it's a current balun, there's no cancellation, regardless of number of turns, or cores, period.  At best there are nulls due to resonances, at specific frequencies, but not in general.  The simplified circuit is, putting an impedance in parallel with one diff port (the CM/shield impedance to GND), and not the other (the core/signal/positive port is untouched otherwise), and therefore it intrinsically and necessarily must be unbalanced, by a margin given by the ratio of port to CM impedance.  You could tweak that impedance by loading the positive port with a compensating impedance, but this is only correct when the unbalanced port is common ground; or add balancing transformers to better enforce it, but that would be a voltage balun, or hybrid.

For two: turns is the number of loops through the core aperture.  The fact that the wire is locally straight, near the core, is irrelevant: it still must close into a loop around the core; or if not, it must be an open circuit!  You can't have zero turns, or you don't have cores looped around the conductor at all. :)

(Note the core is itself a single turn of core around the winding!  In fact, there are some games that can be played with the winding number of a core, for material with very high permeability, but that doesn't apply here.)


Quote
Any unbalanced CM current is going to be seen by the ferrites, and impeded.  Any unbalanced current coming from the sig-gen is going to have the choice of a 50ohm cable, or a CM impedance.  I think it will look like a RF voltage divider, and I expect slight imbalance.

Yes -- keyword "impeded".  That impedance must be made infinite, for current imbalance to be reduced to zero.  Any finite impedance will necessarily result in some loading of the CM-connected (here, negative diff) port.

RF may be complex, but it's not black magic.  There's no point where we avert our eyes and pray to elder gods; circuit theory doesn't simply go out the window because waves bounce around sometimes!  We can -- and must! -- apply basic circuit theory, in situations where it remains applicable (which, here, will be for the chain of cores / TL stub being less than 1/4λ).   Draw the equivalent circuit, do the sums, and determine the required core count.  Check the datasheet for impedance(freq), beads always(?) provide this.

Offhand, I'm guessing seven cores about that size, might be around 100Ω each, and 700Ω out of 50Ω and loading one but not the other port is half the effect or 1/14th, and 14x (volts) is 23dB.  If you measure wildly different from 23dB balance at up to, oh, 200, 300MHz or so, I'll be pleasantly surprised. :)


Quote
If I find the expected imbalance, I plan to experiment with the tiny Ethernet torroids I rescued from a MoBo.  I plan to pass the centre core conductor of the coax through one of those tiny ferrites.  The aim being to add just enough impedance to achieve balance.  If it works, the trade off will be increased insertion loss of balanced signals. 

I key question to answer is how many ferrites will it take to provide a reasonable lower frequency limit?

Yes, smaller, thicker cores are handy -- you need to use smaller transmission line though, which may make problems not just in terms of impedance matching but losses (smaller wire = higher loss) as well, and we can't ignore the common mode through the ferrites themselves -- they aren't air, ferrite has a fairly modest dielectric constant actually, and MnZn ferrite has notable resistivity -- so we expect some antenna effect where the strip shown will eventually radiate as a 1/4 wave antenna, if a poor one*.  This also limits how much impedance we can put on, even with enough turns on very thick cores.

*Probably the effect will be down in the +/-0.1dB sort of range, and may not be visible as a result.  One way you could tell, is testing the system as-shown, then bringing a reflector near it -- a sheet of metal parallel with the existing ground plane, some distance above it, but still far enough that it's not affecting near-field too much.  (Up close, you can imagine it increasing CM capacitance, or reducing Z_CM as it begins to resemble a ferrite-loaded stripline geometry.)  Bonus points for demonstrating the reflection is, well, reflective -- say, pointing the reflector at different angles and observing no effect except for angles close to parallel; but so close in, and for lengths and distances comparable to a wavelength anyway, I suspect the effect would be small.  Or, at distances where the reflector is reasonably in the far field, the return gain will be so small (< -40dB say??) that the change in balance or port impedance or whatever will be imperceptible.

There's also skin effect in the core material itself; how much depends on material, but this is one reason MnZn isn't so suitable for VHF+ use for example (the skin depth is too shallow and you don't get much impedance, or it even looks capacitive).  NiZn ferrite may also be low enough loss that, rather than a skin effect as such, there is some wave propagation through it -- that is, if you consider the amplitude and phase of an incident wave versus depth through the material, you may find it goes through a few cycles, i.e. propagation is dominant, and it looks more like a lossy bulk material, than a poor conductor.  (Which makes some of these materials useful at high frequencies, like the application of magnetic bias to obtain relatively massive Faraday rotation, and construct a circulator or isolator.)

But some combination of these will likely get you a very good balun; I would imagine better than 40dB balance is possible.  You may need to use a stack of cores, like NiZn close to the line, MnZn outside those, and nanocrystalline outside that (allowing operation down to quite low frequencies, 10s kHz or below even).

It's doubtful that better than 20dB is really needed here, and the mast / feedline can be grounded to the test chamber at the base, further attenuating CM / feedline currents.  20dB balance error corresponds to about a 0.83dB amplitude error, readily calibrated out.

Cheers,
Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Hi
I agree with everything you have said.

The coax coil through a toroid proved to have almost good enough performance.
Basically I am aiming to improve on that performance by wrapping the toroid around the coax instead of wrapping the coax around the toroid. 
I don't expect perfection.  I am aiming for good enough.

This could be a complete failure, but that would be a useful thing to learn.  If it does fail, I will fall back to the coax coil through a toroid.
Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Hi
My theory works.  The concept of winding a toroid around a coax rather than winding a coax through a toroid has performed better than expected in my tests.  I will call the concept a linear coax balun because it is straight, not a coil. 

The test balun achieves a practically constant 2degree phase error and 5% amplitude error from about 5MHz to 300MHz.  Insertion loss is near zero. 
The upper frequency range is limited by my 100MHz scope.  I suspect the upper bandwidth will easily exceed 500MHz.

The simplicity of my 3-port VNA does not appear to affect the results of the test.  I made a balun substitute to prove the 3-port VNA fixture. I tested the simple 3-port VNA with an ideal balun substitute to prove it works.

The photo shows the balun substitute laying beside the 3-port VNA fixture with the linear coax balun installed.

The balun substitute consists of a coax with a sma connector at one end, and a sma Tee adapter at the other end. A signal from the sig-gen is split 50:50 at the Tee.  The branches of the Tee are connected to the X-Y inputs of the scope. Note that only the coax core is connected at the Tee.  The coax shield is open. The X-Y inputs see signals that are perfectly in-phase and of equal amplitude.  The unbalanced output of the sig-gen is converted to a balanced X-Y input.  The coax substitute was used to test the 3-port VNA fixture.  It showed that:
  • the displayed amplitude was attenuated above 100MHz (stating the obvious for a 100MHz scope)
  • At over 150MHz, the oscilloscope shows some imbalance and non-linearity.  (again no surprises there)

The linear coax balun consists of a length of coax threaded through a number of ferrite toroids. The X input is connected to the coax shield, and the Y input is connected to the coax core wire.  That is it.  The toroids impede any imbalanced current in the coax.  The single ended signal from the sig-gen appears at the other end as a balanced signal seen by the oscilloscope. 

I know toroids over cable have been used in emc forever.  What I haven't seen anywhere is toroids over coax used as a wide-band balun.  If someone else has come up with the same idea, my search of the literature hasn't found it.

The X-Y signals show a small imbalance that is frequency insensitive.  I did expect a little imbalance and I did try adding very small toroids to the balun coax core, but saw no significant improvement.  I suspect this balun needs trace imbalance to work properly.

There is no sign of any self resonance because there is no inter-winding capacitance.

Tests to date demonstrate the linear coax balun is easily the best balun tested. It is very simple to make with good performance.  It could be described as untra-broadband. 
« Last Edit: April 02, 2024, 07:22:02 pm by dazz1 »
Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 

Offline shabaz

  • Regular Contributor
  • *
  • Posts: 206
Glad its working! Regarding:
What I haven't seen anywhere is toroids over coax used as a wide-band balun.  If someone else has come up with the same idea, my search of the literature hasn't found it.
It's quite a common technique, e.g. this baluns article (I wrote it a couple of years ago) mentions it (in the section titled Real-World Baluns.



« Last Edit: March 30, 2024, 08:31:17 pm by shabaz »
 

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
...
It's quite a common technique, e.g. this baluns article (I wrote it a couple of years ago) mentions it (in the section titled Real-World Baluns.

Yes, I see that you mentioned a coax going through a tunnel of toroids.  That exactly describes what I have done.  Your comment is the only reference I have seen to this type of balun, so maybe it isn't so common.  However, if others have arrived at the same solution, then that is good because it confirms my thoughts and experimental findings.

If you have been following my posts, you will know that I have tried a number of different balun types.  They all looked good on paper but suffered when exposed to the real world. None of the theoretical papers I have reviewed discuss in detail, or measure, the effects of inter-winding capacitance.  For a 1:1 ratio balun, the coax in a ferrite tunnel easily has the best real-world performance.   In my case the measured performance appears to be limited by my available instruments.   I don't think I have reached the frequency limits of the test fixture or the balun.

The attached image shows the phase plot of the balun at 100MHz. The amplitude deviation is 5%.  The phase angle deviation is 2 degrees.  Insertion loss is practically zero.   This balun is not far from being ideal.

My target is 350MHz and I have been able to measure to 300MHz on my 100MHz oscilloscope.  I think the balun would be good to at least 500MHz but making measurements at that frequency would require Siglent to donate a free scope, and that will only ever happen in my dreams.   More realistically, I will ask a favour of a local company with the right gear.

At frequencies above 500MHz, the physical dimensions of the test fixture are likely to start affecting the accuracy of the measurements.  I would be able to quantify any errors with the balun substitute.   

In any case, my search for a suitable balun has ended.  This will allow me to complete my biconic antenna.  One of the nice features of combining a very wide band balun with interchangeable plug-in antenna elements is that I could easily make a range of element sets to change the frequency range of the antenna.  For example, I could double the length of the elements and still use the same balun and machined parts.  The elements are cheap and easy to make.  The machined parts are difficult and expensive (in time) to make.

One of the really nice features of this linear type balun is that it will conveniently fit inside the mounting tube that is commonly used on biconic antenna, including mine.

This project isn't finished yet.  The next phase is to make a final version of the balun and connect it to the antenna. I will then need to measure the performance and characteristics of the completed antenna. 
« Last Edit: April 02, 2024, 07:22:37 pm by dazz1 »
Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 

Offline T3sl4co1l

  • Super Contributor
  • ***
  • Posts: 21842
  • Country: us
  • Expert, Analog Electronics, PCB Layout, EMC
    • Seven Transistor Labs
I mean, it's nothing unusual, easily enough derived from principles; and, you mentioned EMC, but if you're kind of excluding EMC references, I dunno, EMC uses a lot of antennas, baluns and other cable-impedance-modifying and coupling structures so it's pretty much the go-to domain to find electromagnetic structures and tricks like this.

For example, a typical EMC cable clamp for measuring / injecting signals on cables/harnesses, uses a tall stack of ferrites, alongside a coupling strip that's grounded to plane at one end and couples to the cable at the other end.  Thus giving isolation along the cable, coupling to a BNC... it's a clip-on bias tee, pretty nice eh?

And I say "easily enough", but those principles aren't exactly the most common information, you can pick it up from an undergrad curriculum I would say, but you're more likely to find it in explicit detail in grad level EE, and if you're not officially EE at all, it can take quite a lot of study and insight to develop these ideas, and many working in electronics never get there at all, it's just not in their wheelhouse (say, those working with audio, equipment repair, digital/embedded, etc.).  It's fairly specialized knowledge, and so you have to search it out in particular, it won't come to you on its own, so to speak.

If you're confining yourself more to radio references, and ham stuff in particular, or online references even more narrowly, yeah there's a lot of junk information there unfortunately.  There are a lot of RF and EMC engineers that are hams, but there are few hams that are RF engineers.  It works one way but not the other.  So that selection slices the literature the wrong way and you...don't get very good results.

The tipoff is, such results -- postings, web pages, informal articles -- rarely cite anything; they should only be used as possibly correct information subject to further proof.  When they do make citations, whatever of those references you can find in turn, check that they support the claims, that the claims are reasonable given what theory you know or are able to test, etc.  You are searching for proof of correctness of the claims, and of the supporting material in turn, recursively, until you have reduced the problem all the way to well-proven claims and verifiable theory.  Burdensome at first, but as you map out available knowledge, you can pin more things as proven, or in what ways they claim things correctly and what others they miss and haven't yet proven as such.  This is the challenge of academic rigor, and the primary means of identifying, developing and curating well-proven science.

So, you see claims from, say, someone selling an antenna kit -- maybe they claim directionality/bandwidth/size beyond the theoretical (Chu–Harrington) limit, but if they provide no test results (feedline impedance or SWR or etc.; CM current; exact configuration/geometry tested; radiation pattern, at least along axes; calibration where possible; etc.), feel free to ignore their claims, doubt them fully, it's just a bunch of words it doesn't mean anything in relation to anything else until proven as such.  And indeed, you will find claims of impossible performance, where the configuration is basically an unbalanced dipole let's say, but it's far wider bandwidth and higher gain than should be possible, and the only possible resolution to that is ground effect and feedline acting as antenna element, and perhaps nearby structures acting as reflectors if it was actually measured for radiation pattern.  Or another common "trick", they're claiming SWR is antenna efficiency, or gain even, when this is only true when the antenna is lossless, and they might've simply cooked up a particularly shite configuration with a lot of equivalent resistance so the actual efficiency is awful but the bandwidth is wide and SWR is low.  Again, verifiable claims, consistency, does it match theory, are the methods correct, etc. etc.

Regarding balun design specifically, you probably don't find many references discussing interwinding capacitance, but the problem can be decomposed into the N-turn impedance of a ferrite bead or other core, which you sometimes see plotted in datasheets, and yes the peak goes to higher Z at lower F as turns goes up as one would expect, and, you can test this independently by rescaling the problem say to magnet wire on a smaller core, it doesn't have to be coax -- and then substitute solid wire for coax and the outside fields / shield/CM impedance don't know the difference -- it can't know that there's a wire inside the shield -- and that's your answer, that's the CM impedance across the balun from shield to shield, an impedance that isn't shared on the signal wire and therefore imbalances the result.  And by scaling, I mean the frequency response is proportional to length scale, give or take material invariants (paramagnetic resonance perhaps, but most applications are well below that and ferrite just looks like a bulk lossy permeable material as we usually model it, plus minor wave effects due to core size and shape).  So it's not a balun design problem, it's a transformer or inductor design problem, and you will likely find many more results in that more general direction, than for baluns alone.

Beyond that, you can do some hand-wavey thoughts about fields within and around the ferrites, because dielectric constant is fairly high so the tubular stack itself looks like a fat wire above a ground plane, and that impedance adds with the total bead impedance; fields and waves means wave mechanics and you could have cavity resonances or radiation or whatever depending on if the stack is in an open or closed configuration (open as shown, or closed in a box / surrounded with shield).  Pretty soon, you'd want to measure or simulate these, but likely these are also to a degree of precision unnecessary (e.g. maybe if you were doing a VNA with a precision directional bridge, but for an antenna, who cares), and so we can truncate our efforts depending on accuracy required.

So, applying these principles, let's say, this:
http://on5au.be/content/a10/trans/41balun-2.pdf
No references (or maybe inlined, I just skimmed), but lots of measurements of basically what you're doing, different orientation (1:4 instead of 1:1+1), lower frequency (but why stop so short, first resonance might be 100s MHz) but we have every expectation it works to much much higher, and the measurements being very flat in the range tested certainly checks out with expectations.

Ooh, this is a good one:
https://www.nonstopsystems.com/radio/frank_radio_baluns.htm
I wonder if this is kind of the holy grail that you've been looking for forever but just never found (probably because Google sucks these days); the references aren't necessarily high-tier journal results, but it looks like a reasonable selection of experimental data that can be checked in turn.  (And local copies are provided for easy viewing!)

https://www.nonstopsystems.com/radio/pdf-ant/BALUNS2006-ang.pdf
This is one of the references, it seems familiar, I might've seen it a long time ago, but anyway -- showing off cores and some basic geometry, and more or less confirming the manufacturers' curves on these (at least of the EMI bead parts where impedance is most likely to be provided -- you'll have to look them up and confirm), and showing testing methods, and a range of typical applications.  Good stuff.

...Maybe the formatting is a bit gaudy?  I mean, it's no \$\LaTeX\$ article. :-DD  But what do you expect, it's PowerPoint, that's easily excused -- don't let superficial aspects like aesthetics get in the way of some good data.  (Now, when it's poorly presented/arranged data, or misleadingly formatted, by all means critique; such aspects go a little deeper in meaning and impact!)

Tim
Seven Transistor Labs, LLC
Electronic design, from concept to prototype.
Bringing a project to life?  Send me a message!
 
The following users thanked this post: Kean

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Hi
My research did not find the documents you referenced, but I did find similar, which I soon discarded, or discarded after experimental results were unsatisfactory.  The main reasons I discarded papers were:
1.  They could not be readily adapted to provide a 1:1 impedance ratio. 
2.  My experimentation revealed that coils on toroids did not have the band width and performance I was seeking,
3.  They used technology unsuitable for the frequency range of interest, especially if they were working in the GHz range,
4.  No measured data or other evidence to backup the claims,

The critical information I was seeking to define the performance of a balun I never found.  I wanted to see measured data for:
1. The phase difference of the balanced lines (ideally 180 degrees),
2. Amplitude difference between the balanced lines ( ideally zero),
3. Insertion loss (ideally zero).

Those 3 parameters essentially define the quality and performance of a balun.

The most useful paper you referenced is the last one, the gaudy power point presentation.  It has useful info on measuring ferrite characteristics I have not seen elsewhere. 
Page 24 shows the exact balun configuration I ended with, but no measured data relevant to my application.
Page 53 also shows ferrites over coax used as a balun, but again without measured data.

I know there is a lot I don't know about RF, but I try to fill the knowledge void with facts and evidence. My old Professor used to say "It isn't true if it isn't measured."  I made my own truth by experimentation.  Part of that experimentation included designing, building, and validating custom fixtures to make the measurements to provide the data and evidence I was seeking.  So yes there are plenty of papers out there on baluns, but in my experience most would, and did,  end up in my discarded bin.   

Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 

Offline shabaz

  • Regular Contributor
  • *
  • Posts: 206
I think most EE undergrad courses won't cover any of this, it's too vocational, so unfortunately, unless one comes across it in their line of work, some extra study is needed by engineers and non-engineer hams alike, and that includes trying to read academic papers, reading the decent online information (granted a lot will be wrong or omit information, but recommendations from others helps to weed out the bad links) even if some of it is hard to initially follow (some content is extremely narrow-focussed so it's natural it may be hard to follow if one isn't an expert in that specific field), and finding books on the topic at hand.

For what it's worth, if you're an amateur radio enthusiast, there's a very popular book by Jerry Sevick called 'Transmission Line Transformers' that is easy to read; it's practical with examples. It is not worth the new price on Amazon; there are second-hand copies from time to time at low cost. But the original Guanella paper is easy to read, too (the book contains it from memory). Also, the Chris Bowick book 'RF Circuit Design' is super-easy to read (although it is brief, which could be considered a positive thing) and covers quite a few ferrite and inductor topics with clarity, and non-domain-experts will easily follow. Again a used copy is worth getting, since the new price has almost tripled over the past few years.
« Last Edit: March 31, 2024, 03:42:39 pm by shabaz »
 

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Hi
I am a pro EE and its been nearly 40 years since I last did an EE undergrad course. My career path has included a lot of RF stuff, but not R&D. Courses and books explain the theory, but little of the practical problems with application.  I am not a licensed ham radio guy.  When I started this project, I new what a balun is, how it worked and why I needed one. I had never made one before.  I started with a review of the literature and of commercial baluns.  Practically all of the papers ended up in being discarded.  I had to invent my own wheel.

At the risk of making a huge generalization, the main problem I have seen with amateur radio info is the typical lack of validated measured data. What they say "works" but no evidence of how well it works. Exceptions apply.   The commercial antenna manufacturers don't give any measured data on the baluns they use.  The modern academic papers tend to be undergrad or at GHz frequencies. 

The other problem I hit is instrumentation. One example, I needed a VNA to measure the performance of my balun.   The VNA needed to have a 50ohm unbalanced port and  a 100ohm balanced port.  I found a good HP reference on VNA measurements.  In order to measure balanced signals with a 50R VNA port, the manual said a calibrated balun is required (HP part numbers given).   Even if I had all of that, I still needed a fixture and a means to validate the test method and setup.  That lead to a mini-project to create a 3 port VNA and test fixture with a signal generator and an oscilloscope.  This may be a common method but again I don't know of any similar setup, pro or amateur.  There is no known course or paper on this topic that I could find.  My undergrad courses and later experience gave me an understanding of the theory, what I needed to do, and how to do it.   Most importantly, it worked. 

So for me this has been a self made journey of learning that I have recorded on this forum for the benefit of others who may be interested.  This added discussion is healthy because it draws out more relevant info.

Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 

Offline coppercone2

  • Super Contributor
  • ***
  • Posts: 9709
  • Country: us
  • $
Ehh for the balun I got 2x 50 ohm RF resistors with 3GHz performance and soldered them between banana jacks in a bakelite piece (it has irregular spacing on the HP adapter for 100 ohm).

I consider it good enough because IMO usually you are going for cable impedance, and that is probobly gonna depend on how you lay it out and stuff.

Since that standard is on banana jacks, it probobly is not that good. I was very annoyed by the price of the 100 ohm standards.

but 3Ghz 1% through hole resistors on the 10Mhz banana jack fixture is probobly gonna be pretty damn close. I got like 200 of them so I can pick the best values at DC. Still cheaper then the HP standard, and I have a bunch of cool through hole RF resistors left over.

I think the prices people are asking for differential stuff on ebay is unfair

You need another manufactures transformers to get past 10Mhz IIRC for the HP vna fixtures also.


Unless we are thinking of somethign different, HP did not appear to sell a calibrated balun, they sold a 100Ohm banana jack insert calibration standard to calibrate their balun or north hills brand transformers.
« Last Edit: April 01, 2024, 05:02:29 am by coppercone2 »
 

Offline uer166

  • Frequent Contributor
  • **
  • Posts: 907
  • Country: us
Did you get a chance to test the fixture with an actual biconical or at least a dipole? I have a similar interest (and reading the same book but 1st ed.)
 

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Did you get a chance to test the fixture with an actual biconical or at least a dipole? I have a similar interest (and reading the same book but 1st ed.)

Hi

Not yet.
If you read my posts wayyyy back, my aim was to separately develop and quantify the performance of the balun before mating it to an antenna.  I have arrived at the view that the balun is THE critical component of a Biconic antenna, but the least discussed.   Also, I have made two identical antenna specifically so I can accurately measure and calibrate their response. 

I now have a balun with a flat response across a very wide frequency band width.  I will be able to accurately characterise the antenna performance.

I think there is still potential to improve the linear ferrite over coax balun in two areas:
1.  Further increase the bandwidth by using ferrite beads of different materials along the length of the balun.  I have not tried this.
2.  Add components to reduce the phase deviation (2 degrees) and amplitude deviation (5%) of the balanced output.  I tried adding very small ferrite toroids over the centre core wire of the balanced output to little effect.  Further experimentation is required. 

What I have is "good enough" for my application, but I can see the potential to do better.  Any additional effort to improve the balun would qualify as over-engineering.





« Last Edit: April 02, 2024, 07:22:18 pm by dazz1 »
Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 
The following users thanked this post: uer166

Offline uer166

  • Frequent Contributor
  • **
  • Posts: 907
  • Country: us

2.  Add components to reduce the phase deviation (2 degrees) and amplitude deviation (5%) of the balanced output.  I tried adding very small ferrite toroids over the centre core wire of the balanced output to little effect.  Further experimentation is required. 


When you calibrate the setup against a standard, I'm sure this is easily cal-ed out. For my (similar) use case, these numbers are more than enough, though for a different band (500-1.5G).
 

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz

2.  Add components to reduce the phase deviation (2 degrees) and amplitude deviation (5%) of the balanced output.  I tried adding very small ferrite toroids over the centre core wire of the balanced output to little effect.  Further experimentation is required. 


When you calibrate the setup against a standard, I'm sure this is easily cal-ed out. For my (similar) use case, these numbers are more than enough, though for a different band (500-1.5G).

Hi
I made a balun with coax and a SMA Tee.  It outputs equal amplitude and 0 phase difference.  This is my cal standard as shown alongside the fixture (fitted with the actual linear ferrite balun) in the photo.
The other photo shows the A and B channels of the linear balun at 25MHz.    If you stand far enough from the computer screen, the two sine waves look identical.  Given that the waveform from the centre core is slightly greater amplitude, I think there is the possibility of reducing the phase and amplitude deviation and improving balance. 

My review of papers revealed that balun imbalance is a significant error, causing beam pattern distortion that degrades measurement accuracy.  The deviation measured in my balun is probably within the noise.  I am thinking of looking for a simple method of reducing deviation.  Over engineering is fun.

Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 
The following users thanked this post: ch_scr

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Hi
I have produced a video about what I call the linear ferrite balun, because no one else seems to have given it a name.  You can see the video here:  https://youtu.be/Bc4mT_oOxRQ

The video assumes limited RF knowledge, including those that will never own a proper vna.  Some of you might find the video a bit slow but I could have easily made it an hour long.  It does demonstrate the diy 3-port vna including the fixture, the calibration substitute balun  and the measured performance of the balun out to 500MHz using a 25year old 100MHz scope. 

I don't think the video will hit the top 10 on YouTube this week, but it may be of interest to some.  Remember to watch right to the end because not much happens.
Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 
The following users thanked this post: Kean, uer166, ch_scr

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Hi
I thought it might be useful and relevant to compare the linear ferrite balun with a more conventional toroid based balun.

I made a video here:  https://youtu.be/T0NOIWpfvf8 that sweeps the frequency from 1MHz to the maximum limit of the toroid balun.
The test results prove that the linear ferrite balun is significantly better than a more conventional coil over a toroid.   I think the key difference is that the toroid is affected by inter-winding capacitance that changes the relationship between the X and Y inputs.

The image shows a toroid based balun being tested a while ago.  Previous measurements recorded and bandwidth of 20MHz to about 300MHz.  That is confirmed to looking at the test data from the 3 port vna.

« Last Edit: April 05, 2024, 10:35:48 am by dazz1 »
Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 
The following users thanked this post: Kean, ch_scr

Offline dazz1

  • Frequent Contributor
  • **
  • Posts: 707
  • Country: nz
Hi
I have just found a video on a fixture to test common mode current on a balun.


The principle of operation is the same as the fixture I made for my spectrum analyser.   That is to divert the signal to the coax shield. 

This fixture has just been released as a commercial product here:  https://electronics.halibut.com/product/common-mode-current-choke-test-rig/
There are no specs but based on the video, the user manual, and observation of the internal photographs, it would appear the bandwidth is limited to the HF band.  That is fine if that meets the requirements. 

It does not surprise me that someone else has developed a common mode test fixture applying a similar design principle, because standard affordable test equipment doesn't do it.  It necessary to create something to go between the DUT and the test equipment. 

Just for comparison, my test fixture is shown attached.  The chart plots the insertion loss of the test fixture with the calibration link in place up to 500MHz.  This was done by comparing the fixture with calibration link fitted, referenced against a simple coax cable.  I think I tested it to about 1GHz but only recorded the plot to 500MHz.

There are probably more versions of common mode test setups and fixtures somewhere out there.    If you know of any other examples, post the links here.  It is interesting to see what others are doing.

Dazz

Over Engineering: Why make something simple when you can make it really complicated AND get it to work?
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf