As the builder of the redwood variant, perhaps I should comment.
Length was originally a bit longer and I had made a few of differing lengths, mounted them and did some rather crude but bottom line effective tests using a simple field strength meter (diode type!). It was the length shown which is the same as the martybugs variety that gave the best result. Admittedly we tend to use the lower channels, 1,6 rather than the higher end at 11 at our locations.
The coax, point is made....Cheap and available. In New Zealand, one does not have quite the abilities to get small lengths of more suitable coax. It was a bit of a gamble but the best we could readily find locally.
Like Jeff, I have been involved with RF as an amateur and professionally off and on for over 45 years so am very aware of all the optimum theory and parts but when you are working to a very limited budget in a place well removed from mainstream and without your usual facilities (I base in Melbourne, Australia), you tend to work by the seat of your pants. The real trick is that it is reasonably reproducible to a point that works. The biquad is comparatively wideband, more so than a stub in a can! so has a bit more tolerance anyway.
I leave it open for the average person to produce something so simply effective using only handyman tools and easily obtained items. Hopefully some will take the idea and experiment further and produce something even better, such is most of the experimental nature of handling RF and antennas et al.
One for you, Jeff, it certainly beats tuning a 10Ghz etched filter with silver paint and a scalpel under a microscope while watching a spec ani.! That was in the '80s and culminated in our setting the inaugural 10 Ghz VK DX two way record! The eyesight was a lot better then too.
Actually, you already have but don't realize it. You need to match the antenna feed to the dish and that's rough with a USB dongle.
I have some not very nice comments to make about using a reflector with a USB dongle feed. The basic problem is that the original PIFA or circuit board antenna has roughly a hemispherical antenna pattern. That means that transmissions from the dongle will go in all directions and only some of which will hit the dish, reflect toward the receiver, and provide useful gain. That also means that you will observe different apparent gains in transmit and receiver. Receive will be largely unaffected by the dongle feed pattern and will yield reasonable results. Transmit gain will suck as much of the power goes in useless directions.
See: |
formatting link
details. I ran the numbers for an 18" pizza dish in: |
formatting link
's probably a worst case example as the deep dish rice dryer seems a be a better shaped reflector.
Some not so light reading: |
formatting link
Part 4 on dish and feed basics. |
formatting link
look at the figures and read the captions if you don't want to dive to deep.
My suggestion is to replace the built in PIFA or whatever antenna in the USB dongle with a proper coax connector (SMA) or coax pigtail and use a dish that has a properly matched feed.
Incidentally, the same problem with feed "overspray" or "spillover" happens with foil reflectors hung on vertical omnidirectional antennas as in
formatting link
However, the effect is much smaller than the 15dB loss in the pizza dish reflector because reflector captures most of the RF from the vertical and redirects it in a useful direction. For example, if the reflector covered a 120 degree azimuth wedge (looking down on the antenna), two thirds of the transmitted RF will go to places unknown, while one third will hit the reflector. That's about 5dB loss which isn't bad considering that even the best dish antennas and feeds have about a 3dB illumination loss.
Also note that different parts of the dish provide different amounts of gain. This is best demonstrated by blocking parts of the dish and seeing how it affects the gain. Try this with an absorbent material (black foam or wet towel) and not with a reflector like aluminum foil.
The flat part of the dish, near the middle contributes the most to the antenna gain. Unfortunately, that tends to be blocked by the feed (or USB dongle) so offset feeds are used (as in DBS dish antennas) to avoid blocking the center. Gain derived from the "sides" of the dish is much less, especially on a deep dish (f/D < 0.35). As the dish gets deeper, the edges of the dish approach being inline with the antenna direction, and therefore capture very little signal relative to the surface area. It's also more difficult to design a proper feed for a deep dish.
This begs the question "Why are all these people building their own reflector antennas"? Easy. Even with all the problems mentioned, a suitable reflector type antenna is a major improvement over the stock antennas. Even the worst implementation will give about 6dB of gain which is the equivalent of 4 times the range. However, squeezing the maximum gain out of the system does require a proper design and testing.
No. That antennas are turned on only one at a time. That means you only get the gain of one antenna and cannot combine the signals to get more gain. Just use one antenna port.
LMR-400 coax is a good coax for long runs. 6.8dB/100ft loss at
2.4GHz. You will also lose about 0.5dB per connector pair so even a zero length coax extension will have 1dB of loss. Add a pigtail and you add 2 more connectors. How far you can go requires that I calculate the fade margin. I've done this about 20 times. Use Google Groups advanced search to look for "fade margin".
For a rule of thumb, 6dB loss will cut your range in half. 12dB will cut it to 1/4th.
I could never get consistent results using a field strength meter (diode hanging on my DVM with two resistors and a cap to filter the rectified DC). I now have a very old but useful spectrum analyzer. However, it only goes to 1.2GHz so I use an MMDS downconverter for a test receiver. My signal source is a 2.4GHz WISP transmitter located on a mountain top about 3 miles away. Line of sight and minimal local reflections. Also saves on the cost of a signal source.
formatting link
Note that the VSWR curve is quite flat over the entire band. That allows considerable creativity and variation in construction. I'm tempted to compromise this broad bandwidth by using a smaller diameter wire. That will make it much easier to build.
The problem is that the above model is directly from the original Trevor Marshall implementation without any of the fine details, such as the coax cable feed diameter thrown into the puzzle. I'll try to add these in the next few days and see what breaks. I really don't know (yet) if it will have a significant effect. I'm also going to change the quad antennas to a loop as putting corners on the antenna is a waste of effort and actually quite difficult to do accurately. The only critical dimension is the length of each loop, not the shape. But first, I gotta learn how to use the 4NEC2 optimizer.
Well, it's not so much the choice of coax that I'm complaining about. It's the effects of the transitions. You could have used a smaller diameter coax for the center feed if you had a better way to support the biquad elements. Methinks I can do that by using smaller diameter copper wire for the elements. A piece of RG-174 should work if stiffened with some solder slopped on the outside braid. Make your own semi-rigid coax. That should also solve the problem with the rather sharp right angle turn required where it exits the reflector. In addition, it will eliminate one coax diameter transition and eliminate the long exposed center conductor where it hits the radio circuit board.
Yep. Compare the can and biquad VSWR plots at the band ends: |
formatting link
|
formatting link
've only built two coffee can feeds. I wasn't thrilled. Biquad is much better.
I won't say anything nice about my attempt at building a biquad inside a 6" outdoor box: |
formatting link
mistake was using nylon insulators. They're hygroscopic and caused the antenna to screw up badly after about 3 days.
Yep. Learn by Destroying. One does not understand antennas until one builds a few that don't work as expected.
Idea for you. Go to stationary store and find some foam board. It comes in various thicknesses but most commonly about 1/4" thick. 1/2 wave aluminum foil reflector on one side of the board. Copper tape a biquad or patch antenna on the other side. Maybe 4 folded dipoles playing vertical collinear. Pile additional layers to get the correct thickness. Shove coax through foam and make connection with glue or aluminum duct tape.
Congrats and nicely done. I've built filters out of copper pipe fittings but never tried to do anything at 10GHz (other than use converted commercial equipment). I did machine an interdigital filter at X-band but I never could get it to work right.
No. A properly designed dish and feed have the same gain in both xmit and receive. In improperly designed system, with large amounts of overspray from the feed, will cause the dish to less than the theoretical maximum gain in transmit and about the same as the theoretical gain in receive.
This is rather muddled. A client radio, access point, repeater, and just about any type of wireless device works best when the transmit and receive gains are about the same. More specifically, they work best in a symmetrical system, where the fade margin on both directions is about the same. If you radically increase the gain in one direction, you obtain no additional range because the signal in the other direction hasn't improved at all. Whether you create such a situation by adding a tx power amplifier or by throwing together a poorly designed antenna system is not important.
Groan...
Chuckle. I just love the polarity marking in the 10pf capacitor, especially since ceramic caps are not polarized.
The reactance of 10pf at 2.4Ghz is: Xc = 1 / (2 * PI * freq * C) Xc = 1 / (2 * 3.14 * 2.4*10^9 * 10*10^-12) Xc = 6.6 ohms Compared to the 50 ohm load, that will result in about a 10% loss or about 1dB. Tolerable. Looks like the output matching network is trying to compensate for some lead inductance in the flip chip. The
10pf is therefore probably part of this matching network, in which case the loss is probably much less. I'm too lazy to model it to be sure.
If you're thinking of making your own amplifier, there are better chips that put out more power and in easier packages. Flip chips were made for automatic vapor reflow soldering and are not really suitable for do it thyself breadboards.
You might wanna dig through the collection of 802.11a/b/g app notes at: |
formatting link
anything interesting. Maxim also sells prototype boards with the chip and all supporting circuitry ready to test and clone.
Reading between the lines of your question, I detect that what you really want to know is where to cut the trace on your USB adapter and attach either a pigtail, connector, or antenna. Yes, it's on the antenna side of the capacitor to avoid having DC appear on the antenna. No, I don't have a clue where you should cut because I don't have the board layout, photo, schematic, or the time.
See above.
Always used in the better radios. It's almost impossible to build a perfect 50 ohm match. The designers take their best shots and compensate for any mismatches with a matching network.
Probably because he had some 150pf caps handy or couldn't handle the tiny 10pf or so capacitors. Self resonance is a big problem and needs to be avoided. 150pf is way too big. He added them because Netgear apparently has some DC on the ungrounded antenna elements and diversity PIN diode switch. External antennas should be AC coupled to prevent accidentally shorting parts of the radio to ground.
I also mentioned that it's on a neighbors roof and I can't find the photos. The PIFA antenna has 3 soldered connections. The middle connection is the 50 ohm point and is where you connect the coax center conductor. The nearest solder blob is ground. That's the shield point. The other solder blob is just for mounting. I used a PCB mount SMA connector and as much ground as I could solder.
Note that the Compaq WL215 is an Orinoco Gold classic card with an external coax connector. He could have done it without cramming the whole mess into the tube. I personally don't find anything wonderful about the arrangement.
That's been known to happen. That reminds me. My calculator batteries need replacement. 3ea N batteries. Ouch.
Nope. We have a small apples and oranges problem. The USB feed for a pizza dish is a 3 dimensional problem while the EZ-12 and such are mostly 2 dimensional. The USB source is treated almost like an isotropic point source (incandescent light bulb), while the omni antenna driving the EZ-12 is mostly a linear source (fluorescent light bulb). The illuminated areas are calculated somewhat differently.
For USB, you take the illuminated surface area of the dish, as viewed from the USB dongle. The ratio of the illuminated area of the dish, as a fraction of the spherical surface area of a sphere with the same radius as the feed to dish radius, is the illumination loss (in dB). That's because anything radiated by the USB dongle that does NOT hit the dish, is treated as a loss.
For the EZ-12 type reflector, the problem is mostly 2 dimensional. The illumination loss is simply the ratio of the wedge angle that hits the reflector, as divided by a circle with a radius equal to the feed to reflector distance. So, 120 degrees would be 1/3 of a circle or: dB = 10 log (0.3333) = -4.77 dB.
Obviously both my models are crude assumptions. The USB dongle is not really an isotropic radiator. The vertical dipole in the EZ-12 is not a perfectly linear radiator. However, the approximations are good enough for a first approximation and are certainly good enough for a comparison.
I'll throw together a 3D plot and demonstrate the difference between a biquad and a corner reflector. I'll also try to determine if the EZ-12 gain calc uses the transmit performance or the receive. As I mentioned, there is some loss in using a non-optimized feed for the EZ-12 but it's not as huge as using a pizza dish with a USB dongle feed.
I would call both of them 10dB based on all the various accuracies, assumptions, and significant figures. Most of my tests in trying to measure antenna gains and patterns have resulted in at *BEST* +/- 3dB accuracy. Usually, somewhat worse.
No. It transmits equal "forward" power toward the target and "incident power toward the reflector. The "incident" signal that hits the reflector bounces off the reflector and back towards the target. The position of the reflector is adjusted so that the "incident" signal and the "forward" signal are in phase and combine to yield additional gain.
One could build a biquad without a reflector, but the gain would be at least 3dB less. There would also be no physical isolation between the antenna and the RF circuitry, which can create all kinds of complex interactions.
Thanks for your reply. If I understand you correctly, that would mean that a dish type reflector is more effective on a receiver than a transmitter. Picturing the radio waves as rays of light, the rays would be nearly parallel at a receiver very far from a transmitter. Thus, nearly all the captured rays would go to the focus of the reflector. For a reflector a few inches from the transmitter, however, many rays would "spill out" and wouldn't contribute to the transmitted beam. So it seems to make more sense to put a reflector on a client antenna rather than an access point antenna. Is this verified by experiment? (I know, why don't I do it - not fully set up yet.)
Another question: I did some searching for IC RF amps starting from the page you gave in an earlier post. I found this Maxim data sheet:
formatting link
shows an application circuit at the top. In the circuit, the signal is shown ac coupled to the antenna through a 10 pF cap. Is that typical? Would I connect an external antenna at the RF out end of the cap? Do I cut any traces to an existing antenna? Are compensation circuits (not sure these apply to RF) sometimes used?
Also, at this page:
formatting link
author describes a way to add an external antenna to a particular wifi card by removing an inductor and adding 2 150 pF caps. Why is he doing that?
You gave some pictures of a dongle that you modified for an external antenna. Where did you connect to the circuit?
I'm thinking now that the best approach might be a something like this:
formatting link
a proper feed is made into the waveguide without using RF cables.
Sorry for asking so many questions but I find this topic so interesting. And thanks for the antenna theory references.
Cheers, Bruce
formatting link
for details. I ran the numbers for an 18" pizza dish in:
formatting link
That's probably a worst case example as the deep dish rice dryer seems
Good man - you show no fear of opening and improving these cards and dongles. Thanks for the good pics and tips on your web pages.
Well, could the cable be eliminated completely if you soldered a wire directly to the RF outpoint point on the circuit board and duct taped the whole card to the outside of the cantenna with the wire poking through? I wouldn't know what to do about the ground braid. Maybe I'm all wet.
But there is an RF cable, it's from the Compaq WL110 card via a small pigtail to an N type connector and then via a bit of RG195 coax which is then stripped back to make the feed into the cardboard can.
I'm not sure I'd go so far as to call it a "proper" feed but a quick and dirty bodge. Scraping the laquer off the foil liner and soldering to self adhesive copper tape really is a bodge! :)
I don't think your math is accurate. If you consider the 3dB beamwidth to be 50 degrees, then the 50 degrees of "overspray" is headed in the right direction. The EZ-12 looks like 120 degrees, so the effect should be 50 % of the radiated power going in the right direction.
Comparing your NEC plot of a BiQuad to the NEC plot of the EZ-12 looks like a minimal difference. The EZ-12 has a 3dB beamwidth of 50 degrees, your biquad is 60. The charts are based differently, but it looks like you have
11.1 dB of gain, the Ez-12 11.41.
And the EZ-12 is, as the name implies, easy to build.
Nope nothing wonderful. The reason it was all in the tube was so that there was only one cable, the USB one coming out. Plenty of space in the rear of the pipe so loads of space for the radio. Easier to do that than to find another box to piggy back on to and the point of using the USB was to have as little RF loss in 15' of thin coax.
I didn't say anything about the USB in a reflector.
120 degrees of reflector capturing the energy that was going the wrong way. There is also the original energy headed in the right direction to begin with.
The EZ-12 is not the corner reflector. The corner reflector on freeantennas has a NEC model that looks even more like your biquad.
I would acll it 12 dB, because that's what I see with NetStumbler.
Yeah, but it's not headed directly to the target receiver. Pretend the other end is a patch antenna about 1/2 wave across. At about 100 meters, that's: tan (angle) = (6.5cm / 10,000cm) angle = 0.05 degrees. That's very small part of the RF that's "illuminating" the target receiver. The rest of the signal from the vertical omni (dipole) is going in all kinds of other directions, none of which end up at the target receiver. Might as well ignore the contribution from the "forward" wave directly from the omni.
Sorry (again). I was too lazy to look at the web page.
Also, it's not my biquad. It's by Trevor Marshall. I just posted his NEC2 model.
12dB over isotropic or over some reference antenna? Also, are you sure you're not having reflection issues? I can move the antennas around in my crude indoor test range to maximize the signal and end up with about 3dB more gain than the calculations predict. Invariably, it's the result of reflections. When I do the testing in a real range or out in the open, the results tend to be more realistic.
Incidentally, remember that web page where someone compared the gain of various types of home made antennas? They ended up with some rather inconsistent results. Well, that was because they did it on the pavement and were getting reflections. Foundit: |
formatting link
This looks interesting. |
formatting link
'll read it when I'm done fixing the washing machine.
Well, that's fair. At first glance, it looked like the PCMCIA card antenna was being used to feed the can antenna directly. I missed the seperate driven element. Still not wonderful but good enough.
I pryed apart the cover on one of my Zonet ZEW2501 dongles. Surprisingly, it didn't break but opened easily. Even the plastic tabs stayed on so it looks like it may go back together again. I took some pictures and posted them at:
formatting link
's a metal can on the front side (RF amp under it?) and a Zydas chip on the back. I'm not sure where the antenna is. There's a short C shaped trace on the front on the end opposite the usb connector. Is that the antenna? There's an L shaped trace on the back right under the C. Is that a ground plane? If you have time look at the photos, I'd sure appreciate your best estimate on where to connect an external antenna. Where should the center conductor go? Where should the braid go? Should I cut any traces?
Here's a story to prove how new this RF stuff is to me: In some of your earlier posts you referred to the antenna in a usb dongle as a PIFA antenna. Each time I read that, I thought you meant "Pain in the ..." antenna. Some recent googling informs me that it really means Planar Inverted F Type antenna :)
Thanks to all who are posting in this thread. I'm learning a lot.
You've hit the nail precisely there, "good enough" was the objective. Quick to set up, easy to aim and with performance that was fit for purpose, in this case the Travelodge to the internet connection. :)
Cabling-Design.com Forums website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.