Post by Jim LuxPost by Jeff LiebermannPost by unknownI'm curious, did you do any measurement and testing with an analyzer, etc., or
are all the figures simply calculated estimates?
There are no affordable "analyzers" that work at 2.4Ghz. I also have
not owned, or plan to own an MFJ-1800 antenna. The physical
measurements were done by a friend that owns this antenna and was
wondering why it didn't provide anywhere near the gain promised (as
compared to a commercial panel antenna).
Hmmm.. while you might not be able to buy an analyzer, that doesn't mean
that you couldn't have done the measurement by other means.
Well, not having the antenna prevented me from doing much. If I had
the antenna, I would have connected it to a home made reflection
coefficient bridge and my Wiltron 610D sweeper. I also have some
directional couplers, but all my detectors are currently blown.
<http://802.11junk.com/jeffl/pics/home/slides/BL-shop6.html>
(Right now, there's a chain saw torn apart on the bench). Either
bridge or coupler would have shown high VSWR for the antenna.
Incidentally, I don't consider it my job to supply test data for MFJ.
MFJ should run the tests and post both the procedures and test
results.
While it's a good thing to verify models with actual measurements, in
this case, just creating the model burned all my spare time for about
a week. I don't think I would have had time to do anything more than
a rough sanity check type of test.
Post by Jim LuxA power
meter (perhaps an eval board for one of the analog devices parts)?
Maybe even using a WiFi card as a field strength meter at a little
distance away, and then using an adjustable stub tuner (e.g.
microstripline on a PC board. (or one of those "check your microwave
oven for leakage" meters from radio shack)
I have several microwave oven testers. They will barely detect RF
from an access point. Basically, they're just a diode detector, with
no RF gain. I wouldn't expect much from them:
<http://802.11junk.com/jeffl/pics/Microwave%20Leakage%20Detector/>
The detector diode is on the far left.
I do have an indoor antenna range of sorts. It's full of reflections,
but is good enough for measuring gain using a reference antenna. In
this case, I had a patch antenna and a biquad antenna tested in an
anechoic chamber and obtained fairly accurate gain curves. My guess
is that I can measure gain at 2.4Ghz to +/-2dB. Not great, but better
than guesswork.
Post by Jim LuxSure, it's not as nice as that VNA from Anritsu or Agilent, but people
have been making microwave impedance measurements for decades with
pretty primitive gear.
I still have several slotted lines buried somewhere.
Post by Jim LuxBuilding a directional coupler in microstripline is a cookbook endeavor.
An hour with some copper clad board and a razor blade would do. If
you happen to have some copper foil tape around, it's even easier.
Yep, I've done that. On a good day, I'll get 20dB directivity, which
isn't so great.
<http://www.qsl.net/n9zia/24swr/index.html>
I prefer a reflection coefficient bridge, as in:
<http://pe2er.nl/wifiswr/>
I've built several of these. It won't tell you whether the antenna is
inductive or cazapitive, but will supply the VSWR (actually, the
reflection coefficient).
Post by Jim LuxAn impedance measuring system along the lines of the "3 meter" technique
(where you have a known L or C in series) is a possibility.
Pardon my ignorance, but what's that?
Post by Jim LuxNot to mention that there are a lot of folks on this list who DO have
access to suitable analyzers.
Great. Let them borrow and MFJ-1800 antenna and run their own tests.
Post by Jim LuxSo Renee's question was pretty reasonable.
True. I should have tested my model to verify my allegations. Perhaps
I should also do indoor/outdoor tests, in various environmental
situations, in the presence of interference, and with a side by side
comparison with other antennas. Yeah, that would be great, but I
don't have that much time (or money). I threw together an NEC2 model
that suggested that the antenna is mis-designed and left it there.
Post by Jim LuxPost by Jeff LiebermannIncidentally, I recently purchased a pair of totally misdesigned
2.4Ghz yagi's on eBay for about $8/ea.
<http://802.11junk.com/jeffl/crud/wi-fi-yagi-that-sucks.jpg>
Note the crude driven element, exposed coax conductors, and lack of a
balun. What happened was that this was a cheap clone of another
similar antenna, which did provide all the required matching
circuitry. This vendor decided it was too much trouble to copy the
matching network and balun, so he just left it out. It has more gain
to the sides of the yagi, than in front.
Fascinating... especially since the matching network wouldn't normally
have any effect on the pattern. Balun/Choke having an effect, I could
see (coupling to the feedline screws up the pattern).
I didn't include the feedline in my model, which due to the lack of a
balun (I don't consider the ferrite bead to be a balun due to the high
frequency) will radiate and produce weird patterns. Never mind the
fairly large lengths of exposed conductors at both ends of the coax
cable. I'm not sure exactly what a 200 ohm antenna in a 50 ohm system
will do. 300 ohms in a 50 ohm system is a 6:1 VSWR and a 3.1 dB
mismatch loss. So what if we loose half the power in the mismatch.
Post by Jim LuxPost by Jeff LiebermannPost by unknownHas anyone confirmed what the piece is that connects the driven element with the
coax connector that looks to be a short piece of coax covered in heat shrink?
Has anything been found or measured hiding under the heat shrink?
There's a ferrite bead under the shrink tube. If you can find a
ferrite material that actually works at 2.4GHz, it might pretend to do
something useful. However, no way is it going to provide a 4:1
impedance transformation.
2.4 GHz is 12 cm wavelength
4-6cm of transmission line is a good fraction of a wavelength. How do
you know it's not 3/4 lambda of 75 or 92 ohm line, or something like
that. (1/4 wavelength of 75 ohms would transform 200 ohms to about 30
ohms, for instance)
The owner of the antenna measured the coax cable dimensions. From
that, we determined that it was 50 ohms coax. I have the numbers
buried in email correspondence somewhere. I'll see if I can find
them. Digging.... Nothing yet, but I did find one more photo:
<http://802.11junk.com/jeffl/MFJ-1800/>
While it's difficult to transfer dimensions from the photos, the coax
cable even looks like a 50 ohm cable. Not the thick center conductor,
which makes 92 ohms improbable, and 75 ohms unlikely. You can see the
ferrite beads through the stink tube.
Post by Jim LuxI've worked with enough UHF and microwave antennas with weird funky
feedpoint things to not trust in first impressions. There could be all
sorts of parasitics that help the situation out, and they're tough to
model accurately (you're certainly not going to do it with NEC, for
instance.. I'd trust NEC for the feedpoint impedance without
matching/balun/stubs..)
Well, that's what my model did. The characteristic impedance of the
model is set to 200 ohms at the folded dipole. No matching network
involved. See:
<http://802.11junk.com/jeffl/antennas/mfj1800/slides/VSWR.html>
and note the 200 ohms in the upper left of the VSWR graph.
Post by Jim LuxYeah, it's probably not a nice 1:1 match over the entire band, but then,
most antennas aren't.
Look again at the graphs.
<http://802.11junk.com/jeffl/antennas/mfj1800/slides/VSWR.html>
It's absolutely gorgeous across the band. Less than 1.3:1 from 2400
to 2483.5. I would also have expected a much more narrow band antenna
and was very surprised at the wide bandwidth. That's not normal for a
high gain yagi.
Post by Jim LuxIt just has to be "good enough" and in that
business, the loss in the feedline probably dominates anyway. Figure 3
dB of loss in the feedline, and you get a 6B improvement in return loss.
Yep. That's why I measure VSWR at the antenna, not the source. I
like to use LMR-400 with 6.6dB loss per 100ft. 3dB loss would be
about 50 ft of coax cable, which is MUCH longer than I would use at
2.4GHz. Also, I'm partial to putting the radio next to the antenna,
which reduces coax losses to nearly zero.
Post by Jim LuxThe model you have shows a beamwidth around 30 degrees (H) and 28
degrees (V).. that's a directivity around 16 dBi
Yep. For the benefit of those trying to follow all this, see the
numbers in red at the right:
<http://802.11junk.com/jeffl/antennas/mfj1800/slides/Horiz.html>
Incidentally, notice the asymmetrical pattern, caused by the off
center line position of the copper folded dipole driven element.
Post by Jim Lux4NEC2 is giving you 14dB gain.. that's moderately consistent, although 2
dB is something I'd want to figure why the difference.
The 2dB is from the -2dB gain in the reverse direction (180 deg).
14dB forward gain. -2dB reverse gain. 14 - (-2) = 16dB directivity.
I don't see a problem.
Post by Jim LuxI'd want to
figure out why 2dB different, especially since you didn't put resistive
loading into your model, which, at 2.4GHz *will* make the match better
Ummm... why would I need a load? I'm only interested in what the
antenna looks like to whatever I connect to the terminals. If I
included the generator source impedance in the model, I would have to
also include the generator output impedance characteristics,
connecting cables, and possibly some manner of 4:1 xformer.
Post by Jim LuxMFJ claims 15dBi, and that is also consistent with your model and the
back of the envelope from the pattern.
Close enough. When in doubt, always round up to make the numbers look
better.
Post by Jim LuxPost by Jeff LiebermannPost by unknownI realize we can all assume the antenna impedance is 50 ohms, but is it actually
stated in any of the specifications from MFJ? I didn't see any mention of
impedance specification at all for this antenna, but I could only find
information on the web page at the link you provided as well as in the pdf file
of the instructions for the MFJ-1800. The instructions are rather sparse.
Good point. One can assume that if it uses a 50 ohm coax pigtail, a
50 ohm N connector, and is designed to connect to a wireless device
that presents 50 ohms, it just might need to be a 50 ohm antenna.
Nope.. not if the feedline is a couple dB loss (which wouldn't be
unusual for a laptop pigtail using RG-174 or smaller).. All they really
care about in this kind of application is the directivity.
Maybe. Using coax loss to improve VSWR is an old trick. Whether it's
being done intentionally is debatable. I prefer to assume that if all
the major components are 50 ohms, then perhaps the antenna should also
be 50 ohms.
Post by Jim LuxAfter all, how many ham antennas for HF have a PL-259/SO-239 "UHF"
connector on them.
Too many. However, HF antennas are not as critical to construction
and component differences as microwave antennas. What hams get away
with building HF antennas, would be doom and disaster at microwave
frequencies. You also don't see UHF connectors and banana jacks at
UHF frequencies.
Post by Jim LuxI wouldn't take the fact that it has a 50 ohm
connector as indicating anything other than it's a common connector that
people are likely to have mating connectors for.
Ok, I'll conceded that there's a possibility that there's some
matching magic going on behind the curtains, that's not visible to
mere mortals. Lacking the necessary insight, I prefer to judge based
on what I can see, feel, touch, understand, and test.
Post by Jim Lux75 ohm BNC, Ns, etc exist, but how many people use them on dipoles
(which have 72 ohm impedance)... nahh. they just use a 50 ohm cable, UHF
connectors, and are done with it.
Most of the feed lines to my roof are 75 ohm CATV rigid cable. Also,
plenty of 75 ohm RG-6a/u. Worst case VSWR for doing this is 1.5:1.
The tiny mismatch loss for this arrangement is more than compensated
for the lower coax loss of 75 ohm coax, compared to similar sized 50
ohm coax.
<http://www.qsl.net/n9zia/wireless/75_ohm_hardline.html>
Post by Jim LuxIf they used 75 ohm coax on this antenna, and it's close to a multiple
of 1/4 wavelength, then the match at the N connector is probably not a
lot worse than using 50 ohm coax on a dipole.
The antenna impedance (by my model) is between 200-300 ohms. To use a
1/4 wave section to match that to 50 ohms, the coax would need to be:
Zo = (Zout * Zin)^0.5
Zo = (200 * 50)^0.5 = 100 ohms
93 ohm RG-62/u would probably be close enough. However, the coax
cable in the photo is certainly not 93 ohms (judging by the fat center
conductor).
--
Jeff Liebermann ***@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558