12/29/10

Temporary page

Eventually this information will be revised for presentation in the style common for the Primer.

The bad news is that the new 4228HD and DB-8 are both wrecks. I haven’t yet found any good news.

4228 Gain:

A old Channel Master 4228

B Raw gain of 4228HD (new) with harness and balun removed

C 4228HD as delivered (as manufactured by Channel Master)

D 4228HD with feed point bent (pushed rearward) away from dipoles

E 4228HD with improved harness (and with the Channel Master 4:1 balun that came with the antenna)

F 4228HD with improved harness and simple 4:1 balun

G 4228HD with improved harness and simple 2.5:1 balun

All of these except plot B are net gains (the raw gain minus the coupling loss). B is like the total raw gain of two 4-bays placed sided by side. The goal of the feed system is a net gain as close to B as possible.

Channel Master 4228HD

This is developed from the 4221HD, which is an excellent 4-bay. But the 4228HD engineer was some clown who knows a lot less about antennas than he thinks he does. The “phasing harness” is awful. It looks like it was designed by a plumber. It is so close to the dipoles that it can touch them if anything is slightly bent. The balun box tends to sit almost in the plane of the dipoles. If you just push it back so that it is two inches behind the dipoles, the gain improves by about 1 dB (plot D above). You will have to tie it into that position, perhaps with a tie-wrap around the mast.

Plot E show the gain with a redesigned feed harness. This harness more closely resembles an ideal twinlead transmission line. The huge gap between plots C and E shows how badly they bungled the feed harness.

Plot F replaces the Channel Master balun with a simple 4:1 balun made from 150-ohm quarter-wave segments, an idealized version of the Channel Master balun. Plot F roughly matches plot E, indicating that the Channel Master balun is reasonably efficient.

Plot G uses a 2.5:1 balun made from two 120-ohm quarter-wave segments. The improvement is considerable. This is no surprise. A 4228 is essentially two 300-ohm antennas connected in parallel, which would be 150 ohms. One would guess that the 4228 requires a 150-ohm to 75-ohm transformer, also called a 2:1 balun.

As delivered, the 4228HD outperforms a 4-bay by only a very small amount, which is pretty sad. The main problem is that each half of the harness is made of two wires of different lengths. Part of the longer one is un-cancelled and thus radiates. Although this seems like a small and inconsequential segment, keep in mind that it carries four times the current of the other dipoles and thus radiates four times as effectively. The harness radiation disrupts the radiation pattern of the other eight dipoles, lowering the raw gain of the antenna.

(Some of the above plots have dips at channels 23, 30, and 60. These are caused by resonances in the loops of the reflector. Also there are resonances at channel 47 and 39 caused by vertical currents that include the mast and which disappear if the mast is an insulator. Some of these resonances are excited by the feed harness and are made worse by pushing the feed harness back towards the screen. If Channel Master knew about these, it might explain why they put the harness so close to the dipoles. But the correct fix is to make the harness out of closely spaced, smaller gauge wires.)

In theory, fixing the 4228HD is not hard: You just replace the phasing harness with two 4:1 baluns and a 2:1 combiner. A nut-driver is the only tool you need. If the loss in the baluns and combiner is 0.5 dB then the performance will be 0.5 dB less than plot B. Unfortunately finding low-loss devices seems to be impossible. Such devices are becoming common, but they are not yet sold as free-standing units.

Until I think of something better, if you want an 8-bay, you should probably buy two Channel Master 4-bays instead. I believe that antenna comes with a low-loss balun, 0.2 dB or better. (Tom, at http://www.antennahacks.com has tested splitters, and the Perfect Vision PV22-233 was the best he found, averaging a roughly 0.5 dB loss (0.5 dB beyond what is expected for an ideal device). I have not tested it. I suspect it is also sold under other names.)

VHF gain:

A old Channel Master 4228 net gain, balun loss not included

B 4228HD (new, as delivered) net gain

C 4228HD raw gain

D 4228HD net gain with the four screen mounts replaced by insulators

E Rabbit ears, 40” 45º, net gain

Plot A is without a balun. A typical ferrite transformer balun has a loss of 0.5 to 1.0 dB for these frequencies.

Comparing Plot B to C shows that the UHF balun is causing about a 3 dB loss at VHF. In effect the antenna is direct-connected to the 75-ohm cable with no impedance matching. Also, the lack of a proper VHF balun means there could be additional losses due to radiation from the coaxial shield.

In plot D, the four 4-inch mounts that connect the radiator assembly to the screen are replaced by insulating bars. (Insulating washers would not be good enough. At 200 MHz their capacitance is almost a direct connection. It is best if the whole 4-inch rod is replaced with plastic.)

I had High hopes for this antenna for VHF. I figured that with fewer vertical wires, it would not have the vertical currents of its predecessor. But the vertical rods it retains are enough to cause trouble.

The dip at channel 7 is caused by the antenna radiating greatly straight up and down, which diminishes its forward gain.

The dip at channel 9 is caused by the antenna radiating more out the back than out the front, which diminishes its forward gain. (The harness and dipoles induce a high current in the vertical front ½” support bar, which travels to the screen where it goes to the top and bottom and then turns horizontal. One complete path runs from the upper left corner to the bottom left corner. The other path, in the right side, has the opposite phase. Each path is about 1 wavelength, so there is a current null at the middle of each vertical bar where the current reverses direction. At channel 9 the path is slightly less than a wavelength, and so the screen acts as a director. In plot D, this path is cut, eliminating the dip.)

(The dip in channel 7 starts with a current null coinciding with the feedpoint in the shorter harness wire. For channel 11, it is the longer wire. The null blocks the current but doesn’t stop it from flowing in from the opposite side of the antenna. The result is a massively unbalanced current flow in the harness. That unbalanced current in the vertical feed bus induces a large current in the vertical front ½” support bars. This current flows into the screen, disrupting the currents there.)

All of these problems would be eliminated by a symmetric feed and an insulated screen.

NEC simulation programs are usually a little off. Although the graph shows dips in channels 7 and 9, the graph is not trustworthy as to how close the dips are to channels 8 and 10.

To use this antenna for VHF, I recommend replacing the feed harness with two baluns and a splitter.

The 4228HD balun

The balun is in a black box that is glued shut.

The above diagram comes from the ARRL Handbook. The 4228 balun is roughly the same. Some differences are:

1. In theory, if the two twinlead lines are long then this balun has a very wide bandwidth. But if they are only a quarter-wave long then they have less bandwidth but are quarter-wave transformers that can match anything to anything. If the quarter-wave segments have a characteristic impedance of 150 ohms then this is a 4:1 balun. If they are 106 ohms then it is a 2:1 balun. (To match Z1 to Z2, the quarter-wave segment must be sqrt(Z1*Z2).)

2. The 4228 balun segments are a quarter-wave long for UHF. The two coil structures are not magnetically linked significantly. When I first saw this circuit I assumed it was a 2:1 balun, which would be a good match for the 4228.

3. According to the ATLC program, the circuit board trace is 129 ohms, 0.62 velocity factor. (A 3:1 balun?)

4. Apart from being a quarter-wave transformer, each coil is also a step-up transformer with a Vout/Vin of roughly 1.2. Considering these two effects together, this is clearly a 4:1 balun. A NEC-4 simulation verified this.

5. The two segments are not equal in length: One is 12% longer than the other. I cannot figure out why they did this. Generally, asymmetries are always bad in antennas. If anybody has additional insight into this balun, I would like to hear from him. Email me at kq6qv@aol.com.

With slightly different artwork this could have been a 2:1 balun, and the antenna would have worked better. This is a UHF balun. For VHF it will pass the signal but will not properly transform the impedance or block unbalanced currents.

A properly designed 8-Bay

I have taken a stab at redesigning the 4228. Any antenna twice as big should be 3 dB better. Anything less than 2.5 dB is bad engineering. The feed harness is a transmission line and follows rules that every electrical engineer is supposed to know. If you depart from ideal, you must determine what you are giving up.

This try disregards the VHF performance. Someday I hope to take another look at this for VHF.

The feed harness will serve as an impedance transformer. Two lossless types are common: A quarter-wave segment, and a half wave tapered segment. I chose the quarter-wave segment since it can match extreme impedances without itself ever having to be those impedances. I divided each harness-half into two 5-inch transformers. I simulated the antenna with the harness missing, then used Mathcad to search for the best impedances for the transformers. The 4-bay seems to average about 425 ohms. A 316-ohm quarter-wave will transform that to 235 ohms, and a 188-ohm quarter-wave will transform that to 150 ohms. The two 150-ohm halves in parallel make 75 ohms, which needs a 1:1 balun. Such baluns can be lossless and very wideband. Transforming to 300 ohms also works. The 300-ohm version seems slightly lossier, even before adding in the 4:1 balun loss.

The harness geometry:

The length of the B segment is 6” plus whatever wraps around the standoff screw terminal. Both wires of each B segment must be exactly the same length. Each of the four standoffs is a 1.75” F/F threaded ¼” hex or round aluminum spacer. I recommend 0.1-inch diameter aluminum wire. Such wire is not stiff enough to support itself, so a few plastic spacers will be necessary to maintain the required wire spacing. (Warning: Too much plastic will change the velocity factor. Using larger gauge wire would require a larger spacing, increasing crosstalk with both the screen and the dipoles. Building the feed harness out of commercially available 300-ohm twinlead does not work because of the velocity factor.  I looked for a good compromise based on that but did not find one.)

Plot D shows the expected net gain:

(I have never actually built this antenna, so I have no photos of it. Antennas are mathematical devices, and any proposed design must first work mathematically.

This modification is a little difficult, just tricky enough that I am leery of encouraging anybody to try it. The few details on this page are just enough for an engineer to work from, but someone less serious should probably avoid this project.

I had two reasons for creating and publishing this proposal: First, to encourage manufacturers to move in the right direction, and second, so that buyers would recognize a good 8-bay should one ever appear on the market.)

Tom Ballister has constructed this modification. He has good test equipment for measuring the results. His efforts are described at http://www.antennahacks.com/AntennaHacks.htm . He reports that the improvement over the unmodified antenna is very significant at UHF, but the VHF performance is worse below channel 10. I hope someday to look into fixing its VHF performance. –Ken 12/29/10

AntennasDirect DB-8

All the 8-bay makers seem to be copying each other’s mistakes. The DB-8 has some of the same harness errors as the 4228HD. Replacing the harness with two baluns and a combiner would make it the same as the old DB-8. But the DB-8 is not fixable. The DB-8 dipole elements are only 6.2 inches long, compared to 8.0 inches for the 4228. This biases the DB-8 toward the higher channels. The DB-8 was always a bit weak below channel 40. Now that channels above 51 are gone, it is no longer a reasonable antenna, even with the harness fixed. Some day AntennasDirect will figure out that they have to rescale this antenna.

The DB8 balun

The diagram comes from the ARRL Antenna Book. The DB8 balun is the same except the U-section has been divided into two quarter-wave transformers having characteristic impedances of 54 and 75 ohms, velocity factor 0.60. It will match 200 ohms with 75 ohms, making it a 2.66:1 balun.

This balun is a good impedance transformer but not a very good balun. A balun is supposed to block unbalanced currents. This one creates them. Also it does not prevent radiation from the coaxial shield. A balun of the 4228HD general type would have been a little better.

This balun will mostly filter out VHF, so don’t even think of using this antenna for VHF.

Some new simulations: net gain

A old 4228, not including balun loss

B new 4228

C old DB8, not including balun and combiner losses

D new DB8

E ClearStream 2

F ClearStream 4

G Winegard HD-7698P, not including balun loss

H AntennaCraft HBU44, not including balun loss

I DB2, not including balun loss

J Channel Master 2020

A Rabbit ears, 40” 45º

G Winegard HD-7698P, not including balun loss

H AntennaCraft HBU44, not including balun loss

J Channel Master 2020

Winegard HD-7698P

The Winegard literature is correct: The antenna is for channels 7-69. So this antenna will be obsolete on June 12. In every other respect this seems like a good antenna. But throwing away 2 dB for a short-lived market advantage seems silly to me.

The antenna has a combiner/balun that employs a ferrite transformer balun. Just from looking at it I would say it might be especially low loss for a transformer balun. But presently I have no way to measure the loss in such devices.

This antenna has no useful reception for channel 2-6. The UHF part of the antenna is 86 inches long.

AntennaCraft HBU44

This seems like a partial redesign to me. The antenna still goes up to channel 69, but the placement of VHF directors hurts the upper channels some. Also I get the impression that the corner plane rods are too sparse.

This antenna has no useful reception for channel 2-6. For channels 7-13 it matches the Winegard 7698, but the larger 7698 is much better for UHF. The UHF part of the HBU44 is 56 inches long.

Channel Master 2020

Announced long ago, this antenna is finally available. As a channels 7-69 antenna first delivered after the digital transition, this antenna was obsolete on day one.

Its VHF portion is a bit weak. It is a good choice for close suburban locations with obstructions, where a strong UHF antenna is needed but the VHF part can be medium strength.

The UHF part is 60 inches long. The poor performance on channels 14-30 is due to poor isolation between the VHF and UHF parts. (The quarter-wave stub is meant to serve as isolation: it is a short circuit for UHF. But it is perfect for only one frequency, roughly channel 45, and gets less effective further away. The UHF picked up by the VHF antenna sometimes adds, sometimes subtracts with what the UHF antenna receives. This gives the plot the roller coaster appearance on the ends of the band.)

AntennasDirect ClearStream 2

This is an excellent antenna. It is roughly the same size and performance as the DB2. Since the DB2 is the reigning champion of indoor antennas, and since the “C2” is a little better, the C2 is the new champ. (The DB2 has more bandwidth, but with the new channel lineup this is no longer important. The DB2 would be stronger than the C2 if made a little bigger, but in relation to size the C2 would still win.)

AntennasDirect thinks this is an outdoor antenna, and ships it without a stand. If you buy this antenna for indoor use, you will have to devise something to hold it up. If you are in a poor-signal area and are forced to use an indoor antenna, the C2, DB2, and 4220 are your best choices. The C2 is the only one of these without a poke problem around little children, so you should buy the C2.

This author continues to believe that a 4-bay is generally a better choice for an outdoor antenna.

The antenna elements are too small to have any response to VHF. And in case a really strong VHF signal sneaks in anyway, the UHF-only balun will filter it out. The manufacturer’s web site used to specify this antenna for channel 7-69. But that was wrong, as this graph shows:

Forward horizontal net gain:

Do not buy this antenna for VHF.

The C2 balun works about the same as the DB8 balun, and has the same imbalance problem. A simulation suggests the balun will match 345 ohms to 75 ohms, making it a 4.6:1 balun. (The delay line is a half-wave segment tapered from 115 ohms to 134 ohms, velocity factor 0.66. There is a 4.5 nH inductance 1.0 inches from the split. Although the delay line is a tapered segment, it is easier to understand as two quarter-wave transformers.)

AntennasDirect ClearStream 4

Antennas that look the same work the same, true? Maybe.

The ClearStream4 is the most directional medium gain antenna available. The ClearStream2 is the least directional medium gain antenna available. The difference is profound. In an urban or close-in suburban setting, one of these antennas will be a good choice, and the other probably a very bad choice.

Beam widths (to the -3 dB points, for channel 30):

ClearStream 2 72º

Silver Sensor 64º

DB2 56º

CM4221 55º

ClearStream 4 33º

DB8 25º

4228HD 25º

Within 25 miles, if you see ghosts on your analog channels then the “C4” is your best choice among medium gain antennas. The 4228 and DB8 are still more directional, but those are much bigger antennas. If there are no ghosts then the C2 is your best choice for avoiding a rotor when stations are in multiple directions.

For channel 13, the ClearStream 4 outperforms rabbit ears. It might pick up channels 11 and 12. Do not buy it for channels 2-10. (The C4 works on channel 13 because of the two cross braces. But below channel 12 the balun betrays it.)

The C4 employs the same balun and roughly the same harness as the DB8. That harness, which works so badly on the DB8, works quite well on the C4. Apparently there is less harness crosstalk from loops than from straight dipoles.

This page is part of “An HDTV Primer”, which starts at www.hdtvprimer.com