Canadian TV, Computing and Home Theatre Forums banner

PCB 4:1 balun perfomance evaluation

13K views 18 replies 6 participants last post by  Lassar 
#1 ·
#4 ·
For top perfomance (squize every possible bit) it's quite easy to DIY practically lossless (air gap) microstrip balun, optimized for specific band (e.g. minimize SWR in 470-690, or get SWR=1 for specific channel). Insulators can be used only as supports (most feedline air only), metal box can be used (with 150 Ohm strip or round wire pairs, soldered to the ends of this box, acting as a ground/coax shield wire itself).

So far, I learn exact dimensions for more generalized geometry: https://ypylypenko.livejournal.com/26409.html to get SWR=1.0 @ 600 MHz.
Found L for X=0, working on R=75

Also it's possible to develop antenna/balun as a whole. Antenna is not Z=300 +j0, it has arbitrary R+jX in wide band.
With out-of-the-box baluns (PCB or ferrite) output SWR is unknown, since transformation ratio is not 4:1 as claimed for simplification.

With custom microstrip balun one can predict and optimize SWR @ 75 Ohm coax load.
 
#5 · (Edited)
Probably doesn't make much of a difference, but I looks like HFSS model geometry is slightly offset from actual photo (see overlay in your link above).....

What is Insertion Loss vs Frequency???? For completeness, can you measure (or find a measurement and/or spec) for the actual Balun???

Although different design and somewhat different Freq Range, here is what ADTech measured for two of A-D's Back-to-Back PCB Baluns....about 0.6 to 0.9 dB each:
http://www.digitalhome.ca/forum/81-...igns-losses-diy-loops-etc-52.html#post2075514

FYI: 4:1 Balun measurements summarized for Ferrite and other PCB Baluns:
http://www.digitalhome.ca/forum/81-...igns-losses-diy-loops-etc-52.html#post2075514
I can't see "3rd Party Hosted" charts.....also found towards bottom of my fol. webpage:
https://imageevent.com/holl_ands/files/ota

Updated link to Johnrmckee's measurements for low-loss Philips Outdoor Ferrite Balun: [Unfortunately no p/n or photo]
http://www.avsforum.com/attachments/15540
UHF: 0.5 to 0.9 dB (470-698 MHz)
Hi-VHF: Under 0.6 dB
Lo-VHF: Under 0.7 dB

Note that Philips Outdoor Ferrite Balun and A-D's PCB Balun had about the SAME LOSS across the entire [N.A.] UHF Band....and that MOST Ferrite Baluns are significantly worse, esp on higher Freqs....
 
#6 ·
Probably doesn't make much of a difference, but I looks like HFSS model geometry is slightly offset from actual photo (see overlay in your link above).....
I'm not interested in the very exact graphs for this very exact sample. There are dozens of similar (but not identical) PCB-baluns around.
I bought this one for test, as little as $0.6
Whether this particular piece has resonance @770 or @670 MHz is not important to me.
It has quite poor tolerances (build quality), PCB traces width/gaps are very coarse/inconsistent (not 150 Ohm, not homogeneous).

12 years ago I did similar baluns for CDMA-850 Yagi's, this is from "Sprint Layout" PCB tool (now I understand that this drawing has a lot of minor faults):


Now, with HFSS I'm able to simulate and understand what perfomance consists of.

What is Insertion Loss vs Frequency???? For completeness, can you measure (or find a measurement and/or spec) for the actual Balun???
Loss in this balun is of 3 kinds:
1) PCB-heat loss (depend highly on material tangent loss, for e=4.4, h=1.5mm, tan=0.02 - heat loss is ~0.15 dB)
2) radiation loss, radiation is about -22...-25 dB, which corresponds to -0.015...0.025 dB loss
3) mismatch loss. for SWR=1.5 it's -0.177, for SWR=2 it's -0.511)

If good matched, I expect @ SWR=1.5 loss -0.3 dB, half of which comes from epoxy quality, half from mismatch.

With special HF PCB materials (Rogers, Neltec, Izola) or other insulators (air, teflon) heat loss can drop from 3.4% to 1-1.5% (-0.04...-0.07 dB)

-0.075 dB mismatch-loss corresponds to SWR=1.3

I expect that top perfomance -0.15 dB (0.06 + 0.07 + 0.02) is achievable, but not measurable, since tolerances for measuring equipment cables are worse.

In real world applications, combined mismatch loss (antenna + balun) has most influence on total insertion loss. Since antenna is not R=300 +j0, actual SWR after PCB balun can be both better or worse than on R=300 ohm ideal load. Specially designed balun can compensate antenna own reactances (mostly by traces Length, choosing appropriate Freq for L/4 length) and R<>300 (choosing trace width/gap other than 150 to compensate input R<>300)
 
#8 · (Edited)
Continue improving: https://ypylypenko.livejournal.com/26409.html





Line #1 (0.75 mm strip, 8 mm gap) narrowed to Zo=300 Ω, so it become matched and can go further.

Due to narrowing, jumper #2 shortened to 4.77 mm (ideally it's length should be zero)





Animation @ f=1000 MHz


Further optimization needed at low-impedance side:
- narrow gap to Zo=150 Ω
- rounded turns
- verify 75Ω pin location exactly at half/way (including Ω-jumper)
- draw 75Ω line, so 75Ω feed is outside balun and doesn't interfere with balun
 
#12 ·
Where is the evaluation?
Any bottom line / at the end of the day, real world measurements/comparison of one vs another in actual operation?
SNR / margin achieved one vs another? At the end of the day is all that matters.
How did we get from PC Board baluns to an LNA produced in Moscow?

Confused as to what the point of all this is?
Am I supposed to design artwork and etch this PC board balun in my basement and try it myself?
Guessing that would be a train wreck.
 
#13 · (Edited)
How did we get from PC Board baluns to an LNA produced in Moscow?
Produce your own. This is novel technique I've never heard before, it seems interesting to me, so I posted it in this balun topic.
It's controversial but sounds promising. You amplify differential signal, so any balun loss after pre-Amp doesn't influence SNR. I didn't find any real world measurements for that, only allegations at some russian DX forums.

Where is the evaluation?
in 1st post. I need to know wideband properties of PCB baluns (several similar types are sold here for $0.6-$1). I've heard controversial claims from local TV/SAT enthusiasts that these PCB baluns work better than ferrite @ 700..860 MHz, but worse at 470-650 MHz. I've also seen many type of DIY baluns at different forums (both PCB and bifilar-wire air type) with claims that they are better than those selling for $0.6.
There are also claims that @ 50 Ohm cable/load they work better than @ 75 etc

Also I wanted to add balun into my HFSS antenna models. Antenna is not Z=300 +j0, it has R different from 300 and X different from 0. I wanna see combined output Zo and optimize antenna/balun as a system, not to assume they are 300 +j0.
If antenna has SWR=2 and balun has SWR=1.5 at given frequency (@ ideal 300 Ohm load), combined SWR can be both >2 and <2 depending how balun R/X combines with antenna R/X

So I simulated one of most popular PCB (made in poland) in HFSS, to see:
1) where are frequency cut-offs, VHF-Hi DVB-T2 might be introduced in my area soon, so people are curios whether these existing baluns will work or not)
2) Wideband impedance curve
3) what is heat insertion loss for FR4. when people DIY PCB balun from better materials (Rogers, Izola, Neltec, foamed PVC) - is it worth or not?
4) is trace/topology from industrial design adequate, or poor designed

Most IL comes from ML (SWR), due to reactances and poor topology (trade-off for miniaturization)

I'm looking for balun+LNA for my 2-bay UHF bowtie project
http://www.digitalhome.ca/forum/186-antenna-research-development/282321-cm4251-h2-hybrid.html

Existing (at my local market) options are:
1) LNA with ferrite 4:1 balun
2) PCB or ferrite baluns
3) 75 ohm coax-through LNA

Polish PCB balun suits well Uda-Yagi folded dipole, but doesn't suite well bowties.

Polish company Anprel/Dipol modify their bowties line shape to suite their PCB:


it works extremely poor (SWR>5, work poor with LNA, doesn't work with passive balun), you can not acomodate coax cable - you must align it along between TR line

I hope to accomodate my bottle-shaped PCB design with 2bay CM4251-like bowtie
Also want to DIY 1R5D Uda-Yagi from holl_ands album, but since it is extremely high-Q type Yagi, want to verify model with balun.
 
#14 ·
roger that, I am with you now...
I agree, it looks like something that COULD work well but like you would be skeptical until trying it myself.
 
#15 ·
Yurii just a question , can the HFSS sim, do a coil wrapped balun, made out of 75 ohm copper coax cable.
If so.
I be interested to know my 4:1 home made coil balun how it stands up.
I start off with a 24 inch piece of coax
I take a 3/4 inch piece of electrical conduit, drill a hole big enough and 1/2 inch down from the end
so the coax will insert from the outside then pull it out 4 inches
Then I rap it 4 3/4 turns around the conduit
Drill another hole and push the coax in then out to the lenght , to match the starting 4 inch cut to match .
then I take a 50 foot piece of coax and run it up the middle of the coil to match the two 4 inch pieces.
Then wire it up as a 4:1 balun.
Tested and works from 170 MHz to 700 MHz
And out performs any store bought balun. That I have used.
I know it's not a pcb balun , I'm just wondering if you get bored. Or have time
Thank you
 
#16 · (Edited)
This task can be acomplished 2 very different ways using Ansys Electronics Desktop (former Ansoft HFSS):
- using FEM solver for arbitrary geometry/surface structures
- using "Circuit design" project type

FEM solver has no limits on surface type. It's very easy to draw coils, coaxial cylinders, coaxial coils etc. Bad news is extreme computational complexity.
All surfaces (including dielectric except vacuum) get meshed with tetraedras.
Any close/narrow space surfaces with high E/H field density (all transmission lines) require very dense mesh all over their length. Coarse mesh on TR lines influence Zo impedance a lot. With flat non-curve surfaces (rectangle) work is quite easy - tetraedra is flat, flat rectangle easily meshed with flat tetraedras.
With curve surfaces, especially coaxial, work is extremely (prohibitively) hard: you need 100-500x times more tetraedras.

Complexity can be illustrated on ATLC2 tool:


2000 pixel slice takes 1.4 minutes to calculate impedance. Even with 2000 pixels surface (both wire & shield) are quite coarse, not circle.
With 3000 pixels (fine image quantization) calculation time is 13.5 minutes (10-fold increase)

I suspect that your coax balun will nead >1 TB (>1000 GB) RAM and a week of computations on super-cluster using FEM solver.

Good news - there is no actual need for FEM solver if structure properties is well known (datasheet or measured).
Why mesh each kilometer of coax cable with billions of tetraedra if it's characteristic impedance and velocity factor is well known from datasheet/measurements.



Using circuit solver, any S, Z, Y reports are genereted instantly (4 seconds for 1000 points sweep)
 
#17 ·
Hello...as per my knowledge Half-wavelength Coax Balun has highest loss at the lowest and highest Band frequency and lowest loss mid-band. UHF will have higher loss due to trying to cover a much wider bandwidth and may also have somewhat higher parasitic losses higher loss at UHF Freqs is roughly offset by shorter coax length than for FM or Hi-VHF.
 
#18 ·
L/2 balun is extremely narrowband (single channel) design.
Only at 1 given frequency it lags signal 180 degree. At any other length phase lag is not 180, at L/1 it completely short-circuit input terminals.

Bandwidth is also dependent on L/2 line impedance.
Best L/2 impedance is Z/2 (150 Ohm line for 300->75 transformer). When line length is not L/2, it operates as transformer.
TR Ratio is proportional to difference if source impedance and line characteristic impedance. To minimize unwanted transformation, L/2 line should have Z/2 impedance
 
#19 ·
Balun link is bad

Updated link to Johnrmckee's measurements for low-loss Philips Outdoor Ferrite Balun: [Unfortunately no p/n or photo]
http://www.avsforum.com/attachments/15540
UHF: 0.5 to 0.9 dB (470-698 MHz)
Hi-VHF: Under 0.6 dB
Lo-VHF: Under 0.7 dB

Note that Philips Outdoor Ferrite Balun and A-D's PCB Balun had about the SAME LOSS across the entire [N.A.] UHF Band....and that MOST Ferrite Baluns are significantly worse, esp on hig
The link to http://www.avsforum.com/attachments/15540 is bad.

Would some one fix this?
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top