Making splitters/combiners
with asymmetric cables

The following discussions about making power splitters/combiners with asymmetric cables took place in the Moon-Net reflector. I found the subject so interesting that I decided to put together all the messages in this page.

On 24-Sep-2001 M0BPQ wrote:
Hi All,

I have a brief question wrt using 70 ohm coax as a phasing harness for combining two 50 ohm yagis. I appreciate the theory of using 70 ohm coax as published in many places. However, I thought that one should use equal lengths of co-ax (in odd multiples of lambda/4) for each "arm". I see from the UKSMG pages:

http://www.uksmg.org/split.htm

that one arm can be 1.75 wl with the other 0.75 wl. Does this mean that I could have one arm as 2.25 and the other at 0.25 wl (for example). I think I understand why you can do ths, but I am far from sure. Can anyone explain ?

I already have some 10.3 mm 70 ohm co-ax around (UR57?), and some N-type T peices, so this seems like the easiest way to go (for 2m). If there are any serious reasons why I sould ditch this and build a quater wave tubular power splitter, I would like to know. A rough calculation says that my proposed 70 ohm phasing harness will be 0.15 db worse off than with a tubular splitter and Westflex 103 coax. I can live with that so are there any other reasons ?

I look forward to hearing your thoughts

On 24-Sep-2001 G3SEK wrote:
You could build a divider with asymmetric legs, but it's not a good idea. The often overlooked problem with dividers having legs more than 0.25 wl long is that the sensitivity to small errors in physical length and velocity factor becomes progressively greater and soon gets out of hand. Sensitivity to changes in antenna impedance also becomes very high, especially if one antenna changes differently from the other. It becomes very difficult to get the expected SWR, and as for the phase error and its effect on beam tilt... well, you just don't know.

(If a certain person is reading this, he can tell you himself about the problems with 1.75+1.75wl divider on 2m!)

If the legs are asymmetrical as well, the chances of these errors are again increased compared with a symmetrical divider - and all that is in addition to the unavoidable amplitude error caused by the difference in losses in each leg.

The errors are basically caused by the extra and/or unequal lengths of coax that are operating at an SWR >1. It's far better to build the rest of the phasing harness in 50 ohm coax operating under matched conditions, and use the shortest possible divider, either 0.25 wl or 0.25+0.25wl.

The 0.25 wl and 0.25+0.25 configurations have different pros and cons. A 0.25wl divider requires a 35 ohm section, which you'd probably have to make in tubing but a zero phase error is guaranteed. The 0.25+0.25 type can be made in either tubing or 70/75 ohm coax, but it takes a bit more effort keep the phase error low.

>I already have some 10.3 mm 70 ohm co-ax around (UR57?), and some N-type 
>T peices, so this seems like the easiest way to go (for 2m). If there 
>are any serious reasons why I sould ditch this and build a quater wave 
>tubular power splitter, I would like to know. A rough calculation says 
>that my proposed 70 ohm phasing harness will be 0.15 db worse off 
>than with a tubular splitter and Westflex 103 coax. I can live with that 
>so are there any other reasons ? 

Only the one you mentioned in the follow-up post: your coax is actually 75 ohm, so there will be small additional SWR error on top of all the others.

If you don't have access to a vector network analyser, it's always better to choose the method that is more error-tolerant.

On 24-Sep-2001 G8MBI wrote:
>that one arm can be 1.75 wl with the other 0.75 wl. Does this mean that 
>I could have one arm as 2.25 and the other at 0.25 wl (for example). I 
>think I understand why you can do ths, but I am far from sure. Can 
>anyone explain ?

uhmmmmmm....maybe

this appears to combine two "in theories"

'in theory' ....1....you can have unequal lengths in a splitting system as long as you maintain phase....this works because the bandwidth in use is significant, so there is no need to maintain perfect phase+delay, only phase.

'in theory'....2....impedance repeats at half wave multiples as you move away from the load, so if you have a half wave cable and put that on an antenna, then the impedance (R+/-j) that you see at the end of that cable is the same as that of the antenna feed itself.

SO........

In this case the situation is complex as you are combining two 'in theory' techniques, the first is the desire to maintain phase, I am one hundred % sure that with care this is ok, even at 144, myself and many others do it....the second is a desire to repeat the impedance that appears at the end of the first 1/4 wave 'transforming' section, for that I am _much_ less confident.

I'll share a good quote (unattributed) that showed up on a microcontroller list today:

"The difference between theory and practice is bigger in practice than in theory."

I'll be very interested in the result for my files if you do it........

On 24-Sep-2001 K2TXB wrote:
Stephen, you can view a phasing line of .75, 1.25, 1.75, etc. wavelengths as a line consisting on a 1/4 wave section followed by some integral number of 1/2 wave sections. Since the impedance and phase repeats at each half wavelength along the line, it doesn't matter how many are present for purposes of having equal impedance and phase at the end of the lines. The longer line will have slightly more loss however.

Hope this helps,

On 24-Sep-2001 PE1OGF wrote:
On the other hand you have to be very precise in cutting the cable when using different lengths like 0.75 , 1.25, 1.75 e.g. WL of cable. The longer cable you are using the more precise you have to be. As 2 mm error on 0.25wl will give a 7x lager fault on 1.75wl of cable. this will result in a phase AND impedance error. Think it is wise to stick to equal lengths of cable. In this way you will have the same fault on both cables. Or just use a 0.25 piece of cable for impedance transformer and continue with 50 ohm cable. Or build a simple power splitter just a 0.25wl splitter with 35 ohm impedance.

On 24-Sep-2001 K2TXB wrote:
Well, yes. Of course any time you are building phasing lines it is important to make them the right length. Even when making phasing lines for an all 50 ohm system, I stick to an integral number of half wavelengths. But in this case the more important part is to make the phasing lines all equal in length. Thus any measurement errors will be equal and present an equal match and phase to all the antennas.

However, if the phase lines are carefully measured, to a multiple of 1/2 wave, with the right equipment, there is no reason why unequal length lines will not work just as well - whether they are connected to an impedance matching transformer or not.

On 24-Sep-2001 SM7GVF wrote:
Hi From my view, the only problem with unequal lengths is that it may be hard to cut them with exactly n x Lambda difference by just knowing the velocity factor and a folding rule. The difference has to be one wave length to keep antennas in phase, unless every second antenna is mounted up side down (trickier).

I use a VSWR bridge, or similar, to measure relative standing wave, and I cut every line to the same response, with the feeder either open (n *L/2) or shorted (L/4 +n* L/2) at the end, with the set up below.

          50 ohm 
             | 
TX --- VSWR -+--------feeder-------

I tune across the band in question and notice the indicated VSWR and cut till the lines behave similar.

I have successfully used unequal lenghts tuned this way for EME on 144 MHz.


On 27-Apr-2003 KE7NR wrote:
I have a dumb phasing line question....

I have always used equal length odd 1/2 wavelengths for phasing lines. Easy to do with 4 antennas!

I'm working on a larger array and, for mechanical reasons, it would be very advantages to use unequal lengths of odd 1/2 wavelengths for phasing lines.

Is there a huge penalty for using different length 1/2 wavelength sections for phasing lines? I'm talking a difference of maybe 3 1/2 wavelengths. I would think at some point the difference would become a problem, but, where is that?

On 27-Apr-2003 K1FO wrote:
Don:

Phasing lines that are an odd multiple of 1/2 wv does not accomplish anything different than phasing lines that are an even multiple of 1/2 wv.

There are 3 basic thoughts on phasing lines.

1. Phasing lines cut to multiples of 1/2 vw Does not matter if they are odd or even multiples. The reason for 1/2 wv multiples is that impedance repeats on multiples of a 1/2 wv transmission line (odd or even). So the theory is that multiple 1/2 vw length phasing lines will repeat the impedance (SWR) of the individual yagis, that is the phasing lines will not introduce any additional mismatch in the array.

2. Phasing lines cut to odd multiples of 1/4 wv The reason is, that 1/4 wv transmission lines, or odd multiples of 1/4 wv transmission lines (Phasing lines) act as a natural choke which prevents RF current from flowing on the outside of the shield on the phasing lines. (Same reason why you would put a balun on your 75 M dipole or 2m Yagi). By preventing RF current from flowing on the outside of the shields of your phasing lines theoretically you would have a quieter array (no noise pickup and pattern degradation from unwanted phasing line radiation). Although you may have baluns on your yagis, when you stack them in an array, mutual impedance effects from the other yagis can cause some imbalance.

The down side of using odd 1/4 wv multiple phasing lines, is that a 1/4 wv transmission line causes an impedance transformation. That is if your individual Yagis do not have a perfect match, odd 1/4 wv multiple phasing lines could make your array SWR worse than the individual yagis.

I have used odd 1/4 wv multiple phasing lines for years and belive that if done properly it can improve the performance of your array, especially on RX. However there is more opportunity to make a mistake with this system.

3. Phasing lines that are any length as long as all are equal in electrical length. The reasoning for this system, is having the phasing lines the same electrical length is the most important factor, so don't bother to worry about any of the stuff in items 1 & 2. This systems also lets you pick any convenient length for the size of your array.

Only downside is you could accidentally end up with system 1 or 2.

Now as far as un-equal length phasing lines go, there is no problem in doing this, but the difference in electrical length must be multiples of 1 electrical wavelength. Unequal phasing lines can be used with any of phasing line systems #1, #2 or #3 as I described above.

I had a 8 Yagi 432 EME array many years ago that used a single 8x power divider in the middle with inner phasing lines that were 2 wv shorter than the outer 4 lines. They system originally used 1/2 wv multiple lines The inner lines were 3.0 wv and the outer lines were 5.0 wv. The system worked very well this way. Total phasing line losses were also lower than using 2 - 4x dividers and 1 - 2x divider. This system got rid of 2 power dividers, 2 phasing lines and 4 connectors.

I then re-cut the lines to an odd 1/4 wv system, the inner lines were now 2.75 wv and the outer lines became 4.75 wv. With the odd 1/4 phasing lines, reports indicated no noticeable TX change, but RX performance appeared slightly better. The pattern was a bit cleaner, that is minor lobes in the region of 60 degrees to 120 degrees off of the front of the array were reduced by at least 5 dB, although all this stuff was pretty far down to start with. Sun noise appeared to improve by about 0.3 dB and I just felt like I heard better, especially at low elevation angles. So my conclusion was that the odd 1/4 wv system can result in a small RX improvement.

However since sky noise is so much higher at 144 MHz, I don't think that you would see much of an improvement with the odd 1/4 wv system unless local man made noise was your limiting factor.

There are other phasing line systems that use intentionally unequal line lengths in order to beam steering and to control the pattern. For example by unequal lengths you can intentionally steer your main lobe up in elevation which can reduce the side lobes that point down into the earth to reduce the noise pickup on the array, especially at low elevations (but the sidelobes on the other side of the main lobe will be stronger. This can be a help if you have lots of man made noise on 144 or also can be used on 432 to lower Earth thermal noise at low array elevations.

Hope this helps you,

On 27-Apr-2003 K2TXB wrote:
Don. Using an unequal number of 1/2 wavelengths for phasing lines will not materially affect the performance of your array. The only change will be in the increased loss due to the longer coaxes. That is negligible for the lengths you are likely to use.

Also, there is no requirement that the phasing lines be made up of an ODD number of 1/2 wave lengths. The impedance and phase of the signal in the line repeats at every 1/2 wavelength so you could, for example, use 8 half waves on one side and 11 on the other side with no ill effects.

The requirement for an odd number of sections comes in when you are using 1/4 wave sections for impedance transformation. In that case it has to be an odd number or else it would become some multiple of half wavelengths and would no longer transform the impedance.

Hope this helps,

On 27-Apr-2003 OZ5IQ wrote:
The SWR is NOT affected at all whatever how nany odd 1/2L you use BUT Seen from the point of view that we are " a wave" leaving the cable at different times - "travelling through" different delay zones , it must affect the pattern ( in theory) . I do NOT have any experience in practice . But it will naturally work. So within short eqalities I would say go for it.

On 27-Apr-2003 K2TXB wrote:
Correcting myself. I said:

> Also, there is no requirement that the phasing lines be made up of an ODD 
> number of 1/2 wave lengths. The impedance and phase of the signal in the 
> line repeats at every 1/2 wavelength so you could, for example, use 8 half 
> waves on one side and 11 on the other side with no ill effects.

This is not correct. The phase does not repeat at every 1/2 wavelength, but at every wavelenth. So, you could use 7 on one side and 11 on the other side with no ill effects. Or you could use 8 on one side and 10 wavelengths on the other side. But mixing and even and an odd number of 1/2 wavelengths in the same system will not work.

Sorry for the confusion, I realized my error just after I hit the send button (:

On 27-Apr-2003 DL1EJA  wrote:
you can use any cable length in a system if the are all the same lenght. The advantage of using multiple of Lamda/2 is that you won't create any transfomation in the cable itselve. So input resistor and output resistor (complex) of the cable is the same. Only in the power divider will be a transformation.

In time of failiar conditions it is easyer to locate the problem...and the cable is used in the specifications (50 Ohm).

Good luck

On 27-Apr-2003 K2TXB wrote:
> ...The SWR is NOT affected at all whatever how nany odd 1/2L you use BUT 
> Seen from the point of view that we are " a wave" leaving the cable at 
> different times - "travelling through" different delay zones , 
> it must affect the pattern ( in theory) .

I believe that the only thing that affects the pattern is a difference in phase of the antennas in relation to each other. For example, if the line to one antenna is 1/4 wavelength longer than the other, then the antennas will be 90 degrees out of phase and the pattern will be affected. but the delay caused by the longer line does not enter into antenna 'steering'. Someone correct me if I am wrong about this.

I calculate that it would take about 1500 wavelengths (at 2 meters) difference in phasing line length before any effect would be noticeable at the receiver. That difference would then be seen as a slight echo, or multipath.

On 28-Apr-2003 SM5BSZ wrote:
If you have correctly working baluns it does not matter at all what cable lengths you use as long as the antennas are fed in the same phase. You can save a lot of cable in a big array by deviating from the conventional arrangement like SM5FRH has done in his big array.

If the shortest cable length is X, all other cables have to be X + N*wl/2. VERY important, if N is odd, the antenna has to be turned upside down to add one more 180 degree phase shift on top of the 180 degrees you get from an odd number of half waves.

You may also route a cable to the feedpoint of one antenna, then continue from the feedpoint N*wl/2 with another cable to the next antenna. You then get Z/2 at the first feedpoint but you may save a lot of cable in a big system. Z/2 is easily transformed to Z*2 with a quarter wave line and then connecting two in parallel will restore Z.

There are many more perfectly correct ways to feed a yagi array:) You have to be able to measure the phase shift which is very easy. A T-connector and some means for measuring attenuation.

If the phase shift is correct at 144.000, it will have an error at other frequencies. If you really want the antenna to work all the way up to 144.4MHz (0.3%) you have to keep the difference in cable length below about 10wl for a phase error of less than 10 degrees. You might see a small gain loss if the phase error is above 30 degrees or so.

Extra cable in the antenna will always add attenuation, I suggest you try to keep cable lengths at the absolute minimum.

On 28-Apr-2003 K5GW wrote:
In support of Steve's (K1FO) comments about array performance using 1/4 wavelength phasing lines here is an additional comment:

One of the interesting properties of the odd 1/4 wavelength line is that when two of them are fed from a common source (such as one end of a power divider for example) the currents through the two lines are forced to be equal, even if the impedances on the far end are not equal. Our low band friends use this principle to force equal currents to phased vertical antennas when the base imedance are vastly different due to mutual coupling influences. This property is the actual basis of how impedances are transformed in a 1/4 wavelength transmission line.

In our application, if some of the yagis in the array have slightly different impedances (and they will in a large array where some of them are adjacent to other yagis and some are adjacent to air) the equal current forcing can correct small patern distortions.


(Note: I'm not showing any E-Mail address here in order to avoid them from being collected by SpamBots. You can possibly find the E-Mail addresses of the above OM at QRZ.COM.)

This discussion is still not closed. If you have any opinions or additional related information you would like to be published here, just send me an E-Mail

Read other Moon-Net discussions

 
Share: Twitter   Facebook   Home page Site map Radio Sherlock search engine Terms of service  Privacy  Cookies