I have built a number of antennas over the years - first using yagi designs from the handbooks, then using the NBS designs, then a log yagi from ARRL, then a design by DL6WU.  The latest for 4m is a DK7ZB 6 ele yagi, a nominal 28ohm design, very slightly modified to take account of half inch elements instead of 12mm OD. The dimensions are as follows:
 

Element	Length	Position
Reflector	2.094	0.0
DE	2.024	0.53
D1	1.95	1.135
D2	1.896	2.305
D3	1.883	3.79
D4	1.86	5.06

The analysis of its performance using MMANA is shown below.
 



My yagi was kindly built for me by Phil G3TCU who has an article describing (amongst other things) its mechanical construction here (pdf, 1.5MB)

Building an antenna

For suppliers of aluminium tubing and booms in UK, try Aluminium Warehouse or in south east England there's Rudgwick Metals.
You can also order parts including tubing and element clamps from Nuxcom in Germany.  Nuxcom are also selling kits to make up 4m and 6m antennas to DK7ZB designs.
 

Stacking 4m and 6m antennas

Wanting to have both 4 and 6m antennas on the same mast, I was interested in the possible interaction between them.  The antennas are the 6 ele DK7ZB yagi described above, and 5ele 6m M squared yagi.  I used MMANA v1.0.0.60 to model them, and to find out what deterioration there might be in forward gain and front to back ratio.

The results of the modelling are in the table below. I've modelled the antennas with the reflectors aligned (not very practical) and the centres aligned (more practical) for comparison.  The 4m yagi is mounted above the 6m yagi.  Gain comparisons are modelled in free space to avoid any height effects due to ground reflections.

MHz	Height separation	Gain dBd	f/b dB	Z
50.1	6m yagi alone	9.27	19.8	16.9+j11
50.1	1m, reflectors aligned	9.24	18.03	15.9+j10
50.1	1.5m, centres aligned	9.24	18.1	16.1+j11
70.2	4m yagi alone	10.14	20	27.8-j0
70.2	1m, reflectors aligned	9.57	13.26	19.8-j4
70.2	2m, reflectors aligned	10.09	18.7	26.2+j0.2
70.2	1m, centres aligned	9.66	14.1	23-j6.6
70.2	1.5m, centres aligned	9.94	16.5	25-j1.9
70.2	2m, centres aligned	10.09	18.5	26.5-j0.5

 
The modelling shows the effect of a 1.5m separation as

I have done some measurements using the nearest beacon on 4m (G4JNT/P) and the GB3MCB beacon on 6m (not close, but the best for this purpose).
The best f/b ratio on 4m before stacking was 22dB, and 20dB after.  There was no detectable difference in forward gain (less than 0.5dB change).

On 6m, the f/b ratio after stacking was again 20dB.  These ratios are both better than the modelling predicts.

A cautionary note - when running 400W on 6m with this layout, there is about 0.4W returned on the 4m feeder, equating to about 4.5V.  Similar results apply when transmitting on 4m (160W out -> 0.1W returned on 6m feeder ~2.2V).  Some protection of both receiver front ends is advisable, although I have found no problem so far.

Conclusion

A 1.5m separation of antennas is satisfactory with no discernable degradation of performance.

Operating cross band - a 6m bandstop filter 

In trying to operate cross band I found an excessive amount of noise entering the 6m receiver from the 4m transmitter.  To overcome this I designed a bandstop filter to go in the 4m transmitter output feedline.  Details are on the equipment page here.

Adding a 6m vertical

There are several ways of adding a vertical antenna -

1. a ground plane mounted on an additional 3m pole above the existing antennas
2. a vertical quarter wave element using the 4m beam as its ground plane
3. a vertical dipole mounted ahead of the support pole

Modelling these with MMANA showed that (1) had a gain of -0.2dBd, a low impedance ~19ohms, and an elevated polar diagram with a peak response at about 28 degrees above horizontal.  Slightly better was (2) with a gain of 1.3dBd, but too high an impedance (153ohms), and too much horizontally polarised component.

In (3) I used a driven element 0.9m ahead of the support pole and length 2.76m. A half reflector length 1.54m is added above the support pole - this gives a gain of 2.6dBd, front to back of 4dB, and impedance of 67ohms. Horizontal radiation is more than 16dB down.    The polar diagram is shown below.  The vertically polarised pattern is shown in red, horizontal in blue.  
 
With an impedance of about 68ohms there seems little point trying to match it any better since direct use of 50ohm feeder will give a VSWR of 1.34:1.  It is however desirable to use a balun, and a quarter wave balun was chosen - see eg ARRL antenna book page 26-26. RG58CU has a velocity factor of 0.66 so for 50.16MHz the length of balun is 0.987m.

For a photo, see the diversity page.  The interaction between the 4 and 6m yagis and the vertical was analysed using MMANA. There was no change in gain, front to back ratio, or impedance of either yagi.  

After installation, the VSWR measured 1.1:1 at 50.0 and 50.16.  I measured the power being received on the vertical when transmitting on each yagi. There was 0.1W received when transmitting 160W on the 4m yagi, and less than 0.1W when transmitting 400W on the 6m yagi. These values indicate good orthogonality and a good measure of protection to a receiver connected to the vertical.