Kite Aerial Photography
Picavet Testing

Many KAP enthusiasts have commented on their method of attaching the Camera Rig to the kite line. Whilst the pendulum method has generally been superseded by the Picavet much of the material as to its stability is anecdotal to say the least. Why doesn't size matter, when as an engineer I would expect it to? Why does my picavet keep moving about? What is the relationship between the other variables involved? In an attempt to answer some of these questions, and improve the stability of my rig, I decided to look at the picavet in much more detail. There is a lot of material here, so feel free to start at the conclusions, they are not what I expected!

This page assumes you know the basics of the picavet. I suggest you read the very informative pages by David Hunt on KAPER first if this is a new concept to you.

There are many variables to consider, from the wind conditions to the kite, the picavet itself and the rig, so experience built up in the field may come up with a good solution for your combination of variables, but it may be somewhat suspect to try to draw too many generalisations. So to start with I removed the wind, the kite and the rig from the equation. The objective was to build an experimental rig and perform a number of controlled experiments with a view to drawing conclusions about the dos and don'ts of picavet stability.

Dummy picavet and Rig weights

The Picavet Experiment used the unlikely equipment on the left. In the background three weights of 1.6kg, 1kg and 600g made of concrete filled plant pots. Each has a plastic pipe embedded in its center which is used to attach to the Picavet cross ensuring the center of gravity (CG) passes through the cross center. 

On the left of the picture is an extension piece to allow you to lower the effective CG of the weight from the picavet cross. In the foreground is a 300mm (12") metal rule to give scale.

Dummy picavet fixing details

The picavet cross detail on the right shows the simple but effective "nail" used to attach the weights, allowing quick changeover between weights. This was repeated to attach the extension tube between the cross and the mass.

Also shown are the screw-hooks  that allow three settings for the Pecabe 515 pulleys that the picavet hangs by. Normally eyes rather than hooks are recommended for security of your precious camera, but in the interests of quick changes, and the low cost of concrete, hooks were used. 

Axes and Conventions

Others before me have defined naming conventions for the picavet. Rather than re-invent the wheel the image on the left shows this based on the definitive article in Aerial Eye 1:4. It also shows the standard threading with the attachment points marked in red (this was the threading used for the experiment). 

In addition I have added an axis convention in blue:

Finally, I have called the angle subtended by A1,A2 - the center of the cross - B1,B2 the hanging-angle. 

If you do not have access to Aerial Eye, there is also information about the Picavet and its variants on the KAPER site.

Definitions:

The introduction above gives a rough idea what we're going to do, but lets be thorough, the set up was as follows:

• The picavet was made of wood as shown above, and was strung the standard way (the + Picavet: A1-1-B1-Ri-4-A2-3-B2-2-Ri-A1, see Aerial Eye 1:4) with a single ring, using 11m (36ft) of 50lb line. The picavet anchor points on the kite line were 400mm, 1.2m, and 1.8m (16in, 4ft, 6ft) apart. Some additional measurements were taken at 650 and 900mm. The kite line was at approximately 60º,  attached between a telegraph pole and a heavy weight on the ground.
• Pekabe 515 pulleys were used at the cross, but at the kite line end simple 10mm rings were used.
• Mounting points for the pulleys were 10mm diameter hooks at diameters of 100mm, 200mm and 300mm (4", 8", 12"). In all cases the length of the picavet arms remains 310mm, so any apparent movement could be easily compared.
• The load was balanced and fixed to the center of the cross so it became part of the mass of the cross. This is how the rig behaves when the pan servo is not in motion. Masses of 600g, 1kg and 1.6kg were used. Each mass was located with it's CG 200mm below the cross. An extension piece was available to double this distance.

The following set-up variables were considered:

• The weight of the rig,
• The size / shape of the picavet,
• The hanging angle (Which for a fixed kite line angle, and a fixed picavet line length is proportional to the distance between fixing points A1,A2 and B1,B2)
• The distance between the cross and the CG / load
  (this was tested but not as thoroughly as other variables)
• The use of a Brooks Leffler Horizon Helper^TM (Aerial Eye 4:1)

Under the following external conditions:

• Artificially induce a fixed twist in the picavet. This simulates the force that needs to be resisted when the rigs pan servo operates.
• Slow, general "random" movement of the kite line, and observe how this affects the stability of the picavet cross. The period of these movements was about one second, and the amplitude was ~250mm. This is an attempt to simulate the effect the kite's motion has on the kite line.
• Change the angle of the kite line (simulating kite rise and fall)
• Lateral (sideways) movement of the kite line (simulating kite movement)

These final two were considered, but not performed:

• Changes in kite line angle (simulating a climbing kite)
• Kite line moving along its own plane (simulating kite movement)
  (because of the way the kite line was attached to the telegraph pole these were not possible to perform)

Results:

The following tables detail the experiments and their results. By their nature the results are subject to observation, I've tried to be as objective as possible. The results were taken over several different days, and not necessarily in the order they appear in the tables. This is why some values are missing, but it helps because I couldn't see some of the trends while collecting the data.

cross (300mm)	large
medium	small
wide	narrow

First of all I spent some time just playing with the set-up trying to get a feel for it, this determined the sample points and combinations that were to be used in the experiments. This led to the picavet cross being tested in five ways:

• large, each arm is 150mm (6") 300mm total,
• medium, each arm is 100mm (4") 200mm total,
• small, each arm is 50mm (2") 100mm total,
• wide. The arms in direction k (kite line) were 50mm (2") each, and the arms in direction s (sideways) were 150mm (6") each.
• narrow. The arms in direction k (kite line) were 150mm (6") each, and the arms in direction s (sideways) were 50mm (2") each.
  

TEST I 

Fixing points 1.2m (4ft) apart:
a. Induce a 90º twist in the picavet and count the number of swings to come back to rest
b: Induce slow random kite line vibration of ~250mm just below the picavet and monitor the size of the cross rotation

I		a. 90º TWIST			b. Random kite line movement
	rig mass	1.6kg	1kg	600g	1.6kg	1kg	600g
	wide	2	1-2	1	20º	15º	15º
	large	3-4	3	2	30º	30º	15º
	medium	4		2	45º	30º	30º
	small	5-7	5	3-4	30º	30º	30º
	narrow	8		4-5	45º	45º	45º
So what's best?		LESS MASS Large or Wide cross			Wide cross

TEST II

Using the 1.6kg weight:
a. Induce a 90º twist in the picavet and count the number of swings to come back to rest
b: Induce slow random kite line vibration of ~250mm just below the picavet and monitor the size of the cross rotation

II

a. 90º TWIST

b. Random kite line movement

line dist. hanging angle

400mm 15º

650mm 20º

900mm 25º

1.2m 32º

1.8m 46º

400mm 15º

650mm 20º

900mm 25º

1.2m 32º

1.8m 46º

wide

4

3

2

1-2

1-2

15º

15º

15º

20º

15º

large

6

4

3-4

3-4

3

30º

30º

30º

30º

15º

medium

8

6

4-5

4

30º

45º

30º

45º

30º

small

10-12

6-8

4-5

5-7

5

30º

30º

30º

30º

15º

narrow

10-12

9

8

30º

45º

45º

45º

45º

So what's best?

MORE DISTANCE Large or Wide cross

Little difference in hanging angle Little difference between symmetrical crosses, Wide best

TEST III

Using the 600g weight: 
a. Induce a 90º twist in the picavet and count the number of swings to come back to rest
b: Induce slow random kite line vibration of ~250mm just below the picavet and monitor the size of the cross rotation

III

a. 90º TWIST

b. Random kite line movement

line dist. hanging angle

400mm 15º

650mm 20º

900mm 25º

1.2m 32º

1.8m 46º

400mm 15º

650mm 20º

900mm 25º

1.2m 32º

1.8m 46º

wide

2

2

1-2

1

1*

15º

15º

20º

15º

15º

large

4

2-3

2-3

2

1

30º

30º

30º

15º

15º

medium

4

3

3

2

2

15º

30º

45º

30º

30º

small

5-6

4-5

4

3-4

3

15º

30º

45º

30º

30º

narrow

7-8

8

5-6

4-5

4

30º

45º

45º

45º

30º

So what's best?

MORE DISTANCE Large or Wide cross

Little difference in hanging angle Little difference between symmetrical crosses, Wide best

Observations: 

1. At combination marked * the cross did not come back to rest in the correct position, gentle shaking of the kite line was required for the correct position to be achieved. I assume that resistances in the picavet line and pulleys are becoming significant factors leading to what I call "over damping". 

2. Be careful when interpreting the 90^o twist, and random movement tests, they are trying to look at different things. The 90^o twist test is indirectly measuring the picavet's resistance to rotation in the z axis. The random movement test is trying to assess how much movement is induced in the picavet from the kite line. Observation indicates that this motion appears in the picavet as predominantly z axis rotation.

3. Some of the above points were also repeated with the extension piece to lower the CG by 200mm. They results were not included above so as not to confuse an already complicated picture. There was no perceivable difference in the results. This may be because the extension of 200mm was small compared with the length of the picavet cord.  

Conclusions:

The large and wide picavet are the most resistant to twisting. The wide picavet is least affected by random line movement. symmetrical (or standard) crosses perform similarly, within the measurement error. So, how much of the instability comes from the rig and how much comes from the kite line? The wide performed best all round. The distance between the fixing points is problematic: Too large and kite line movement is easily induced in the picavet, too small and the picavet tends to become an undamped pendulum. Choose a compromise kite line distance (~1.2m) between the extremes.

TEST IV

Using the 600g weight, & 1.5m (4½ft) less picavet cord:
a. Induce a 90º twist in the picavet and count the number of swings to come back to rest
b: Induce slow random kite line vibration of ~250mm just below the picavet and monitor the size of the cross rotation

IV		a. 90º TWIST			b. Random kite line movement
	line dist. hanging angle	400mm 16º	1.2m 36º	1.8m 55º	400mm 16º	1.2m 36º	1.8m 55º
	wide	2	1*	1*	15º	20º	15º
	large	3	1*	1*#	30º	30º	#
	medium	4	2	1#	15º	45º	#
	small	6	3	3	15º	30º	30º
	narrow	16	8	6	30º	45º	45º
So what's best?		MORE DISTANCE Large or Wide cross			Hanging angle inconclusive Little difference between crosses, Wide best

Observations:

1. At combination marked * the cross did not come back to rest in the correct position, gentle shaking of the kite line was required for the correct position to be achieved. I assume that resistances in the picavet line and pulleys are becoming significant factors leading to what I call "over damping". 

2. The points marked # resulted in slack cord when the cross was turned through 90º.

3. Observation 2 for Tests I thru III also applies here.

Conclusions: 
Although a heavier mass may have helped, these results are all similar if a little worse than for the standard length of picavet cord. So the conclusion from I,II,III above holds, choose a compromise distance. With hindsight, I think we should be measuring the angle subtended by the picavet cords at the picavet cross, not the distance between the fixing points. Angles are included in the table for your information.

TEST V

Fixing points 1.2m (4ft) apart:
The kite line was lifted and dropped 300mm in under 1 second to simulate the kite line angle changing suddenly. This introduced a swing in the picavet along the k axis, the number of swings till the system came to rest is measured.

V	rig mass
		1.6kg	1kg	600g
	large	10	8-9	7-8
	wide	12	10-12	8-9
	small	12	10-12	8-9
So what's best?		MARGINAL EFFECT: LESS MASS & Large cross

Conclusion: 
Larger picavets cope better but this is not as marked a change as in I thru III. I suspect because of the nature of the picavet this is a mode of movement that is difficult to dampen out. With the subject matter far away this change is not as noticeable in the camera as twisting or other swinging. This effect is not significant.

TEST VI

Fixing points 1.2m (4ft) apart: 
The kite line was moved sideways 300mm back and forth in approximately 1 second. 
This introduced a swing in the s axis and a twist around the z axis in the picavet:

VI	rig mass
		1.6kg	1kg	600g
	wide	2 rotation low swing	-	1-2 rotation low swing
	large	3 rotation mid swing	-	3 rotation mid swing
	small	4 rotation high swing	-	3 rotation mid swing
So what's best?		MARGINAL EFFECT: LESS MASS & Wide

Conclusion: 
By introducing this sideways movement in the kite line two things happen. The rig tends to sway in the S axis. This is inevitable, and somewhat dependent on the mass. Secondly a twisting was introduced from the kite line to the picavet, in the same way as in I thru III. This is because only one end of the kite line was moved sideways not both. In this case the smaller the picavet in the k axis (small or wide) the better. If we look at how energy is transferred from the kite line to the cross, because the picavet pulleys all run in the k direction (on all arms) then the k arms are more efficient at transmitting a twist in the kite line to the picavet than the s arms. This test was similar to the random test detailed in I thru III, and yields similar results. 

TEST VII

Using the Brooks Leffler Horizon Helper^TM

This is a device that stops the movement of the picavet line through the upper rings. These experiments confirm that the operation of the helper does not effect any of the above experiments, however it can be used to ensure the rig is horizontal, or brace the system against windage. The helper was tested at various points, however the picavet line movement was also closely monitored, as to whether it was moving through the attachment rings. Since no movement was observed this confirms Brooks' theory as to the helper operation. 

In addition the helper has a positive effect on the introduction of a 45º tilt upwards to one side of the picavet cross. This tilt is impossible to achieve with the helper unless some of the picavet lines go slack.  

Conclusion

1. Under some circumstances Large and Small picavets perform similarly, this could lead you to conclude that size didn't matter.
2. The Wide picavet performed best all round, the Narrow was worst.
3. The distance between kite line fixing points is a compromise. 

What does it mean?

My initial way of thinking about the picavet is that it dampens the rotation of the rig. The easiest way to think about this is to consider the pan of the rig and how to overcome that. However the picavet must also resist twisting induced from the kite line. Different KAP systems may have either of these two factors as dominant depending on the external factors. We should choose a picavet that minimises both of them. 

For crosses with four equal arms, as the cross size is reduced the induced twist in the cross from the kite line also reduces, hence the observation that small crosses can work well. However the tests also show that larger crosses recover better from the 90º test, and are more stable in some configurations. The  wide picavet takes advantage of the fact that the picavet does not need to be the same size in both the K and S axis. Small arms in the k direction reduces energy transfer from kite line to picavet, whilst larger arms in the s direction maintain the higher stability and quick recovery of the large picavet.

This also begs the question- "What if the pulleys were mounted differently, and not fixed in the k axis?" This will of course introduce a whole new set of dynamics! 

I suspect some of the issues here are to do with the resonant frequency of the kite line (its sway frequency, not its "singing" frequency) the resonance of the picavet line (pendulum effect), and the size of the cross. Unfortunately these may conspire against you, hence some systems work and some don't. Having arms of different length should give you more chance of success.

As for the distance between fixing points. I would recommend a distance that gives a hanging angle of around 30º. Although not thorough, the experiment with shorter picavet cord confirms that larger hanging angles are not optimum. So the user must choose the picavet line length as a compromise between ease of handling and hanging angle. Intuitively the ring holding the centre picavet lines together should be positioned in the open, with the lines free to pass through. The ring itself may be another variable in the system worthy of closer examination.

Flight Tests

Wide Picavet as built June 2003

The first wide picavet was built of brass square section tube 6.3mm (¼") across. Total length 305mm by 110mm (12x4"), lap jointed in the middle with aluminium reinforcing pieces top and bottom held in place with 4 x M2.5 screws. 

Tested on the 24th June I noted immediately that the rig was more stable and less liable to swing around the Z axis. The test was in light but steady wind and in my opinion performance was equal to the best performance I had previously seen (only once) with my previous 80mm cross. In my opinion subsequent testing in the following months confirms my confidence that this is a good solution.

At KAPiFRA05 I was very pleased to meet Nico Chorier who has also experimented with unequal armed picavets, his research shows that the same conclusions apply to picavets strung for 'x' rather than '+' orientation. With the 'x' picavet the arms are of the same length but the angle between the arms is not 90degrees, the footprint of the resulting picavet is the same however: oblong instead of square. The results are also similar.

Since this page was published a number of flyers have experimented and been pleased by the performance of the "wide". 

On the left, an example expertly fabricated by Gerhard Zitzmann from Austria, demonstrating another technique for fabricating the cross.

In 2006 Brooks Leffler started selling the "wide", calling it the "gent-X" (B) to compliment his existing picavet cross (A), both are shown right.

 There are other tests and variables that could be considered, but for the moment I'm going to take pictures!

Thanks to Peter Bults, David Hunt, Nicolas Chorier, and Tony Cowlin for their critique of this work.