The Top Mount is the structure on which the wire is fixed, where the magnet and the Hall sensors sit and where some adjustments can be done.
The top mount must be fixed very ridgidly to the structure of the building.
In the early days of my experiments I have used a number of alternatives which simply did not work or did not last long enough.
The (bending part of the) wire I use now is in almost continuous operation for several years.
Overview
Cable clamp
Magnet mount
Hall PCB
OptoSolution
Other Opto Solution
Note that the wire is flexing several mm below the clamp. The first few mm stay almost straight.
Overview
Fig 1. Fig 2.
The Top plate allows vertical fine-adjusting the assembly in East-West and in North-South direction, and it can be rotated to point North. See Adjustments.
The second plate carries the Proxxon Clamp which holds the wire. You may sharply bend the wire above it to prevent slipping out.
Then we have the PCB with the Hall sensors.
The bottom plate keeps things together and has a hole in which the blue collar fits tightly for a calibration purpose.
The bottom plate also functions as a safety trap for when the wire should break.
Threaded rods and nuts are made of brass.
Fig 3. Magnet carrier.
The black part is the magnet, harvested from a defect loudspeaker. But nice magnets are for sale in many metal shops. Find one with a through hole, do not try to drill a hole in a magnet yourself, it is almost unmachineable material.
The magnet has its South pole on the bottom side.
The blue collar fits tightly in the hole of the lower plate for calibration purposes. Normally it sits low so that it does not touch anything.
Fig 4. Ridgid Mount.
This is one of the ways to mount your pendulum ridgidly to a structure in the building. It allows rotation in the horizontal plane to precisely adjust it in the North - South direction. Fine adjustment of verticality is done with the nuts on the upper plate. See Adjustments.
Cable clamp
The cable clamp is made from a piece of M10 threaded rod and spare parts for a Proxxon PCB drilling machine. .
Fig 5. Drawing Fig 6. Proxxon Mount.
The proxxon cap has threading M8x0.75, which is not standard. I was able to cut the threading with my lathe.
Magnet mount
Fig 7. Magnet mount. Fig 8. Magnet Clamp.
Also see fig 3.
The wire is clamped on both ends of an aluminium tube of 6mm outer diameter with brass nose pieces and modified M6 bolts sitting in a long M6 coupling nut. The aluminium tube is threaded M6 on both ends.
The manufacturing of the noses is not evident, so I give the order of working on the lathe.
Fig 9. Drawing of nose Fig 10. Order of working.
1/ Start with a brass rod of 6 mm diameter and flatten the face.
2/ Drill a very shallow centerpit with a centerdrill.
3/ Drill the hole of (in my case) 1mm diam.
4/ Taper the end.
5/ Cut it off slightly longer than required with a handsaw.
6/ Flatten the other face.
7/ Bring to the required diameter to fit tightly into the aluminium tube.
8/ Center, and drill the larger hole.
9/ Cut M6 threading on it.
10/ Cut the slit in the nose.
For 6, 7, 8 the piece has to be clamped in the lathe with the nose pointing inwards. for 9 and 10 it has to be clamped on the diameter which fits into the aluminium rod.
Fig 11.
Slit cutting with a 0.5 mm thick circular saw. Image from the USB microscope.
Do not use a diamond saw, it wil produce a much wider slit.
Fig 12. Slit cutting setup
The 0.5 mm thick saw was mounted on the Proxxon machine, clamped to the lathe support.
To accurately position the saw I used a USB microscope looking downward.
Hall PCB
The Hall PCB contains the 4 Hall sensors and a temperature sensor.
Fig 13. Schema of the Hall PCB.
The position of the Hall sensors is quite critical. The temperature sensor appears to have non-magnetic wires. I managed to find non magnetic capacitors too. You might know that many electronic components have ferromagnetic wires. In the vicinity of the magnet and the sensors it is probably wise to avoid that.
I did not use a RJ45 connector here, I just soldered the cable to the board. Do have a good tensile relief.
Fig 14.
Top side of the board.
Fig 15. Bottom side with hand wiring.
Early 2024 I replaced this board with a new one, a real PCB with the Hall sensors kept in place with a mask made with a laser cutter.
Fig 16. The new PCB with the sensors held in place with a mask.
The sensors are now positioned correctly by design and the RJ45 connector allows the use of a standard network cable.
Note the use of nylon screws and nuts, we do not want magnetic materials in the vicinity of the sensors.
Fig 17. Mask for the Hall sensors.
The mask consists of one plate 2mm thick with the free spaces to adapt the sensors (left), and 2 times the cover, one below the sensors and one on top, cut from 1 mm thick material.
I used black acrylic material because I used a visible blue light solid state laser for this job. (SculpFun S10) If you have acccess to a CO2 laser you can use transparent materials.
Opto Solution.
Later on I I've been thinking of a solution for position measurement without a magnet. It resulted in these sketches:
Fig 18, side view Fig 19, top view
It works with 4 stationary Leds + photocell and 4 shutters connected to the cable.
The edges of the shutters should be on the height of the pivoting point to minimize cross talk between X and Y.
The photocells should be terminated with a low impedance, such that the current is measured. The open circuit voltage is very non-linear with the ilumination, the current is quite proprtional.
The data processing goes much the same as with the Hall sensors.
I thiink the frame for Leds and Photocells and the carrier for the shutters can be made with a 3-D printer, for the shutters I'd prefer metal.
And yes, the whole thing must be placed in the dark and there is some internal light shielding required.
Other Opto Solution
Use 4 cheap (a few €, $, £ each) analog distance sensor units like the GP2Y0A51SK0F (download PDF)
Mount a cylinder of ca. 6 cm diameter and a few cm length on the cable, at a height where it deviates ca. 1 cm at maximum swing.
Paint it matte white, there should be no specular (mirror like) reflection.
Mount 4 of these sensors vertically oriented at that same height, and with the face ca. 3 cm from the cylinder when the bob is in rest.
In the horizontal plane the orientation should be N-S-E-W.
Connect them the same as the Hall sensors in my design, and apply the same adjustement / calibration procedures.
Yes, these sensors have quite some non-linearity. But by using two of them in opposite positions the non-linearities and X-Y crosstalk will cancel out to a large extend.
November 2024 I had constructed a top mount with these sensors and after testing decided not to use them.
The problem was that the output signal is analog, but not continuous. I used the detector in the range of 27 to 43 mm, and there it showed a discrete behaviour I had not noticed before, and which is also not mentioned in the datasheet.
In this range only 13 discrete distances were reported.
Below the transfer and the setup is shown.
Yes, I could have placed the sensors much further from the top to get a larger range, but stil it would have a limited amount of levels.
And in the mean time I had sucessfully experimented with a capacitive measuring system and I decided to go for that. I'll report later....
Fig 20. Tranfer of the above mentioned optical distance sensors over the range 27 to 43 mm.
Red, Yellow: outputs of two opposite sensors
Brown, green: outputs of the other pair.
Blue: displacement.
Fig 20. The arrangement of the opto sensors.
On the wire, shown in the extreme position, a cylinder of 50 mm diameter was attached.
A piece of white paper was wrapped around the cylinder to give it a matt reflection, needed by these sensors
The sensors were at 120 mm distance from each other.