Tilt switch test rig

rq3

Senior Member
Looking for on-axis and off-axis magnetic position sesnor with analog output, best i can find is the MLX90371, actual part number being MLX90371GDC-BCC-300-SP maybe rq3 will have 5 mins to see if he thinks its suitable? It does seem to support both on-axis and off-axis planar detection and seems to be analog output, they do also do a development pcb with one attached which could be useful to play about with.

The analog output is 10%-90% vdd, they do an evaluation board with a chip pre mounted its part number is EVB90371-GDC-300-Rev1.0

https://www.melexis.com/en/product/evb90371/evaluation-board-mlx90371 is the webpage for the board and it gives a circuit digram from the board looks like it will go straight into a adc terminal on the picaxe?
@Rampz, my gut feeling is that the MLX90371 might really complicate things. It is capable of three axis sensing, including an analog (or digital) response to the field strength of an approaching magnet (or two), in addition to angle sensing. Tweaking magnetic field strengths, especially in anything approaching an outdoor environment, is something I strongly suggest not getting into unless absolutely necessary.

Sorting those reponses, when it appears that you only need a repeatable and reliable angle sensor, seems like jumping into the deep end while holding an engine block, while what you really need is a reliable life preserver.

I think I've told this story before. Many years ago I worked with a digital engineer who was reponsible for the test interface to a holographc airborne RADAR system. His ultimate design was a wire wrapped board, about 18 inches square, containing a mix of over 500 integrated circuits (CMOS, TTL, and ECL). The interface connector to the RADAR had 600 pins, required over 100 pounds of force to connect, and was on the end of a cable roughly the diameter of my leg.

As debugging progressed, the design became more reasonable, with more and more functions combined, chips deleted, and wire wraps removed. Eventually this monstrous board had one lonely resistor on it, until the day, about 3 years later, that the final Engineering Change Notice (ECN) was issued: Delete R47.
 

Rampz

Well-known member
Sorting those reponses, when it appears that you only need a repeatable and reliable angle sensor, seems like jumping into the deep end while holding an engine block, while what you really need is a reliable life preserver.
rq3 I am sure your right but need to start somewhere, it seems its got a reprogrammed output I can use, I don't have any experience with this sort of thing and nowhere to turn, off-axis planar is what I was looking for ideally says someone that doesn't know what he's talking about.

I will try and keep any unwanted magnet movement under control and hope for the best.

Thank you for your help.
 
Last edited:

hippy

Technical Support
Staff member
I just wonder if the very very slow speed that i am tilting will it will sense it or not, the motor on the chiming side of the clock don't move at all till the clock chimes i'm quite sure it would pick up movement very well, just the clock side that might struggle as may any kind of sensor,
I cannot see that a solid state sensor wouldn't track slow moving changes. If the sensor shows one angle of tilt when things start, and it progresses to an angle where the re-tension is triggered that's all which is needed. There may be some noise while things are changing slowly but it should be easy enough to average that out and it can possibly even be ignored.

The first thing to do when you get the sensors is to see how they perform away from the current test jig., check that you can detect tilts etc, see what sort of results you get, what the effects of varying mounting orientations are. Once the sensors have been characterised one can then decide how best to implement the tilt solution.

I really don't know about the MLX9037. It's not what I would have chosen given it seems to require some sort of additional magnet to work but I am not familiar with how it is meant to be used. My experience has been limited to a PICAXE reading orientation from a Wii Nunchuk which uses gravity as its reference. I can't remember what I actually did there, but I recall I could identify if that was held horizontal, tilted forwards, backwards, left or right.

Given it is pre-programmed, outputs a voltage, I suppose it will be simple enough to connect it to the tilt-detect part of your test jig and measure what the voltage is when it moves from start to re-tension trigger point.
 

papaof2

Senior Member
As Hippy said, test the sensor in the range you need. If vibration appears to be a problem, read the sensor several times and average the readings, then come back and read it a second (10 seconds? 20 seconds?) later to be "just past" some mechanical event that causes vibration. Perhaps borrow a double-bell wind-up alarm clock and see what the sensor does when attached to the case of that clock while it is alarming?

The original winder seems to have been designed as a "by the angle" device but is that the only way to do the job?

How often does the rewind occur? Measured in seconds, minutes or hours? Perhaps have the PICAXE read an RTC chip and run the winding motor every X minutes/hours? Could the rewind trigger be related to the movement of a gear which could be sensed by an LED plus photodetector, perhaps some small dabs of flat black paint every XX degrees on that gear to block the reflection of the LED into the photodetector? With either of those triggers, you could drive the winding motor until it hits stall current, then turn the motor off and wait for the next signal from the photodetector or the RTC - no angle measurement needed and less chance of vibration being a problem.
 

Rampz

Well-known member
How often does the rewind occur? Measured in seconds, minutes or hours?
Thank you papaof2 i was half worried about vibration with the test rig since the motor didn't run too smooth at its slowest speed, on a turret clock there will be some vibration as it ticks but not expecting it to be the problem thats causing the issues.

The clock re-triggers around every 15mins, but it all depends of the clock, and since these things gets sent out to be self fitted it really needs to be angle change rather than a pulse every so many minutes.

Dust could be an issue with opto pickup, could always use a disk with a hole going through a slot type optocoupler as an alternative, many options, i'm not saying the magentic version i have ordered will be any better in my hands

The first thing to do when you get the sensors is to see how they perform away from the current test jig., check that you can detect tilts etc, see what sort of results you get, what the effects of varying mounting orientations are. Once the sensors have been characterised one can then decide how best to implement the tilt solution.
Hippy yep if i can set an 08m2 up to read the voltage being between 0.5v - 4.5v and display that on the PE6 terminal then we can play with angles and see how it reacts, the additional magnet i would hope would be embedded in some sort of shaft hung under gravity, with the sensor detecting the orentation of the magnet.

Going to be trial and error, but solid state seems like being a good route to follow, rolling ball seems to have issues
 

PhilHornby

Senior Member
Dust could be an issue with opto pickup, could always use a disk with a hole going through a slot type optocoupler as an alternative, many options,
I thought about the DIY tilt-sensor approach and came up with the same idea. However, the issue is that it doesn't give that 'snap' effect. So the hole might be fully uncovered at 15°, but half-uncovered at 7.5°. At what point does it trigger? And is it the same every time?

Micro-switches have a nice ON/OFF snap effect :) ...
... if you end up having to build extra structures to hold magnets, then they might be worthy of consideration ...
 

Rampz

Well-known member
Micro-switches
Phil that's the problem I guess getting a snap action when the whole assembly rotates, only really have gravity to go against, the floor can be a couple of metres below, not supposed to attach anything else to the clock and the motor can be fitted at any angle and clocks drive in different directions
 

hippy

Technical Support
Staff member
Hippy yep if i can set an 08m2 up to read the voltage being between 0.5v - 4.5v and display that on the PE6 terminal then we can play with angles and see how it reacts
It shouldn't even need a PICAXE, just a power supply and a multi-meter to read the voltage.
 

Rampz

Well-known member
It shouldn't even need a PICAXE, just a power supply and a multi-meter to read the voltage.
Hippy yep i will try that in the first instance and see if i get anything either useful or repeatable.

I think regards the rig i will ask the manufacturer to make something much more like the clock movement if he can, so i have a better basis for testing, not to say the existing setup hasn't shown some of the downfalls of the existing tilt switch, its has been useful

I have designed a simple 08m2 board with relay output and input for ADC, PWM and the tilt switch so in first instance i can try and see whats happening with the test unit that fails every couple of weeks, will give me the chance to play with code to either go on first pulse, first pulse + delay, etc, will code it to start with so the relay follows the tilt switch and go from there, i have included an led so i can see the status of the tilt switch
 

Rampz

Well-known member
Just another thought, just been looking at rotary encoder incremental type, seem to give pluses out on 2 pins allowing direction to be worked, I'm not interested in direction, but maybe counting pulses in either direction and acting on a particular number is another way to do the job, bounce seems to be really small, output is digital, encoders are nice and a small weight could be attached to turn the encoder against gravity.
Any thoughts?

Looks like they don't give to many pulses per revolution, for 15 degrees I would need at least 24 per revolution to get 15 degrees

I suppose I can count A and B pulses would give me 48 pulses per revolution, count to say 3 and pulse motor for 400ms
 
Last edited:

PhilHornby

Senior Member
It's an interesting idea, but ...
Looks like they don't give too many pulses per revolution; for 15 degrees I would need at least 24 per revolution to get 15 degrees
That doesn't leave much margin for error! (as in none :cautious:) Setting it up in the first place might be entertaining too!

Also, from the datasheet I saw, it might need replacing on a fairly regular basis. (I don't know what exactly what counts as a "cycle", but you only seem to get ~15000 of 'em)
 

papaof2

Senior Member
A "cycle" of a rotary encoder is probably one revolution and perhaps related to the mechanical wear when the shaft turns - unless it has a very good metal support structure, wear will eventually affect the detents and possibly the alignment. There are probably better encoders - for a price - because things used in A-10s and 767s are built for a cycle life with at least a couple more zeroes. Have you seen any high-end electronics scrapped recently? Maybe a high end oscilloscope or other test equipment? This is an item where the surplus dealers might be the best source. In the US, that would include http://www.allelectronics.com/, https://www.goldmine-elec-products.com/, https://skycraftsurplus.com/, and http://www.surpluscenter.com/. http://www.halted.com/ has moved their inventory to another dealer but their URL has a link to that new dealer.

If you find something in surplus AND its spec sheet, buy as many as you think you can use and then double the order - they probably won't be available next year. I have a PICAXE water flow measuring device that uses the water flow sensor (magnetic switch with vane magnet turned by water flow) from a Westinghouse(?) Duet washer. My device was built to monitor the flow in a gravity-fed low pressure drip irrigation system - works great, using interrupts to count 815 pulses per gallon. If I ever needed to replace that sensor, it might be difficult to find now - I built the device probably 8-10 years ago and the number of spare sensors decreases each year.
 

David_Reynolds

Well-known member
Hi everyone . Just a thought, can you find a way to resolve 0.0012V on a PIC ? because 15 Deg is not a lot and even the best I can find is 10Bit resolution and this means for 15Deg we only have 170 steps of 1.2mV for a 0.to 5V output ( no matter where it is on this scale) so only 0.2V total.

Even if that can be done, it has been noted that it would take a lot of setting up, be subject to possible vibrations , mechanical movement, interference, and possible tinkering. Using anything that is that sensitive for such a menial task is always going to be a problem for almost anything., I suppose that a lot of reads and averaging, may smooth out the signal, but any mechanical damping, given that the movement is only 15 Deg is difficult since, to put the tilt angle in perspective, it only represents just 2 1/2 minutes on a clock face.

It also means that any kind of tilt switch is right on the edge of its resolution as well, indeed it is a wonder that any of them works given those constraints.

That in itself seems very mean for a spring winder, so is it correct Rampz?
Or does the wider move a bit more than that?


I see that some of the sensors can be programmed to resolve a total of 18 Deg, but I think they would be a nightmare to set up and a bitch to service.

I still keep drifting back to a simple On or Off resolution, for whatever, it is and even if it is using some position sensor making it resolve only an on or off situation.
A pendulum of 20 mm long will give a total of 5mm movement , this has got to be enough for a simple optical switch?
Consumption should be around 5mA, and a 25mm long pendulum will have 6.5mm of movement.
Yes, it will have some hysteresis but I think that if it took five degrees or so to resolve its state it would still be a useable device.
It could even be damped if, it were to be immersed in a low viscosity oil.

Anyone?

Dave.
 
Last edited:

Rampz

Well-known member
It also means that any kind of tilt switch is right on the edge of its resolution as well, indeed it is a wonder that any of them works given those constraints.
Yep just my feeling not much angle to work with, the manufacturer wanted the smaller angle the better he said then the motor can trigger more often and move less each time.

I still keep drifting back to a simple On or Off resolution, for whatever, it is
Yes simpler the better, i was wondering further to use a small pendulum attached to say a 50mm cog driving say a 25mm cog, the angle of movement would be greater on the second cog so attached the tilt switch to that?

Another thought was to say mount 3 tilt switchs 5 degrees appart and feed each switch to a picaxe pin and look for the first pulse on each tilt switch and act on that?

Regards set up of any switch the procedure is to pull the safety pin that disconnects the motor but still allows the led on the side of the control box to display what the switch is doing so its angle can be set with the motor in its top position and when happy put the safety pin back in and ready to go

Also, from the datasheet I saw, it might need replacing on a fairly regular basis.
Phil the reliability wasn't something i had considered, but a it would be possible to make the encoder only move in one direction so each time its activated it would move say 15 degrees in the same direction allowing it to use the whole 360 degrees increasing the 15000 to 360000, maybe still not enough, at 15mins a time could still be 10 years life.

I did see on ebay there are versions with upto 600 pules per revolution for around £16 so about 25 pulses for 15 degrees, bit more possible, the life could be awful at that price :)
 

David_Reynolds

Well-known member
While an encoder might be subject to the wear and tear on each detent an absolute positioner is field effect and is only subject to bearing wear and electrical stress. It would last for many years but will come at a price, but as it resolves 4096 steps it could give 170 for 15Deg. the main problem is that it is twelve bit and the PIC only does 10 bits, as far as I know. Hippy?
My own thoughts on this are that this will not be too bad 141 steps. unless the output is amplified first, then you can resolve any angle.
The pendulum only needs to be heavy enough to turn the post and bearing. as long as it stays out of the weather all should be OK.

That is if you want to go that way.
I always try to keep things as simple as they can be, but I also understand that some like to be more up to date and techno about it. I can see that if you only react to the first incoming pulse then set the time that you want the motor to run for ignore the input for say thirty seconds , you can adjust it to do all that is required.
The timer could be adjusted by a pot setting inside the control cover so it can take care of any situation, take care not to use any relay, since they consume a lot of power if the mains is not present., unless they are a mains relay.
When on site the switch can be setup so that it is just on and a pulse to the motor could be say a minimum of fifty millisecs. up to to a certain maximum. all derived from the amount of movement your motor gives for its period.
This although seems a bit primitive, should be all that is required. Only the initial input pulse is all that is needed to trigger, time , PWM and hence winding angle .
 
Last edited:

Rampz

Well-known member
The timer could be adjusted by a pot setting inside the control cover so it can take care of any situation
Thanks David, the prefered method is to pre set as much as possible leaving the client to set the angle onsite, we prefer to keep people out of the control box if possible, warranty is voided if they remove that lid, you can only imagine what happens if the average user is let loose with a screw driver.

This all=though seems a bit primitive should be all that is required
As long as its repeatable is fine

take care not to use any relay, since they consume a lot of power if the mains is not present
I use a n-channel mosfet, i have removed all relays from any part of the setup and power suppliers
The pendulum only needs to be heavy enough to turn the post and bearing. as long as it stays out of the weather all should be OK.
Yep that was what i was thinking, they are normally under cover so weather shouldn't be an issue just temp and possible dust
 

PhilHornby

Senior Member
It would last for many years but will come at a price, but as it resolves 4096 steps it could give 170 for 15Deg. the main problem is that it is twelve bit and the PIC only does 10 bits, as far as I know.
Presumably the 11th bit only comes into play at 90° ... (as in, 10bits = 1024. 1024 is 0.25 x 4096. 0.25 x 360 = 90)


On a different tack, would it be possible to measure the spring tension directly?

Using a simple momentary-contact pull-switch e.g. http://www.castelco.co.uk/pull-cord/

or

some kind of electronic spring-balance style arrangement: e.g. https://www.amazon.co.uk/Freyla-Weighing-Tool-WH-A22L-Electronic-Measuring/dp/B097F156GT

In the case of the pull-cord (especially the one shown) , it would be highly unlikely to be able to take the full 'power' of the spring, so it would have to be initially slack and connected 'in parallel' with the spring attachment point (maybe connected to the 'motor end' of the spring. (So the switch could be attached to the same surface as the spring, with its cord threaded through the spring to its far end).

The 'electronic spring balance' approach, would involve figuring out how to interface with whatever is in there. (My googling has revealed nothing, so a bit of dismantling would be required to find out.)
 
Last edited:

David_Reynolds

Well-known member
Rampz I know no more than the the devices give 4096 steps simply because they are a field effect system te hat has its own outputs resolved to twellve bits and given out according to the output that it was made for, such 0-5V , 1-5V, 0.5-4.5V , 4 -20mA , 0-10V. from three solder Pins.
Since , and this is only my opinion I see these as too costly and too delicate and not really needed.

The opto limit switch would seem a better way and if two can be fitted the pulse from each can be used to start and stop the tensioner. they normally have an indicator to tell you that they are switched on, bonus!
You have peaked my interest in this, I might even try to see how small a box can be used to work the switches with the limits described + and - with a large margin of adjustment possibly even a clear lid so it can be seen working.
It has to be noted that these are so fast at switching on an off that there could be a lot of dither, but your picking up the first pulse so will not be a problem. the only wear will be on the bearing and that can be replaced in the future, but I would imagine the drive will have collapsed before this happens.

Are you going to produce a "universal" already set up system that sets its own fixed motor drive pulse length?

So that anyone setting up one of these spring tensioners and the switch angle is purely down to their skills.
Taking off the weights and cables etc. setting the device in the right area of operation, setting the spring to its minimum tension to work the clock around the correct speed
It can I suppose be pre-set to low and high limits of 0 to 15 Deg. as well as a speed that still ensures correct function.
 
Last edited:

Rampz

Well-known member
Are you going to produce a "universal" already set up system that sets its own fixed motor drive pulse length?
David yep thats normally how its done, certain details are obtained about the clock in question, i don't know what is asked, but the 15 degrees is always the same

So that anyone setting up one of these spring tensioners and the switch angle is purely down to their skills.
yes, full instructions are provided i understand with pictures

The manufacturer does try to get the yearly clock service if possible, so he can keep an eye on the units and their condition along with maintaining the clock

If going optical then a couple of gears and a pendulum type set up, to try and get more movement on the second gear, rather than a hole to trigger, go the other way have a slot that is = to 15 degrees and what for the first off pulse and triger the motor, the user then sets up the gear position to suit just being ON by the led indicator at the motor tentioned position.
 

PhilHornby

Senior Member
Using an old RS microswitch from my bits box - obsolete, but apparently replaced by this (159-4562), I claim to have cobbled together the beginnings of a tilt switch :) .

The "ZF SP-CO Roller Lever Microswitch" claims to offer "a high level of repeat accuracy "

25199 25201

2xAA cells are about 50g, but wouldn't operate my old switch. 3xAA did (using the weighing scales I got an answer of about 70g too - why do scales measure in grams, rather than Newtons?)

Comparing my cobbled together 'demonstrator' against an on-screen 15° angle, it operated at very close to that. Gluing on an extension bar and adding a more dense mass, would improve things no end. You probably wouldn't want the model with the 'roller' end in this application.

But just to reiterate - the claim for that switch is "a high level of repeat accuracy" - the datasheet indicates that the operating point of the switch is ±0.5mm - you'd have to do some trigonometry, to convert that to degrees - I think it's about 2.5°. (Curiously, the angle of the actuator bar to the body of the switch, is ~ 15° ...)
 

PhilHornby

Senior Member
Version 2 is a bit more presentable. I dremel'd the roller off and epoxied the bolt on (after filing a flat on one side).

I note there is appreciable hysteresis - probably so much, you could just wire it straight to the motor! I think you would need to sense it resetting, rather than just sending a timed pulse.


25202
 
Last edited:

Rampz

Well-known member
Version 2 is a bit more presentable
Looking interesting Phil, the concept has some merit, maybe the actual switch mechanisim needs to but a pcb push button with almost no travel? I'm meeting the manufacturer tomorrow so hope to make some progress there or at least get him to start building something
 

PhilHornby

Senior Member
maybe the actual switch mechanisim needs to but a pcb push button with almost no travel?
Interestingly, I've used PCB push switches as 'limit switches' on my CNC Router, because I decided they'd be easier to fit than the microswitches that conventional wisdom dictates. The arduino-based controller manages to sense them and stop the motor, before it crushes them to death. Either that, or there's just enough flex in the PCB to save 'em :)

25204
 

Rampz

Well-known member
I met with the manufacturer today, he's not keen currently on making the tilt mech are more involved as it currently is, so i left his the test rig as it was coded for him to play with, while i was there we tested a tiny mercury version and it had a kind of sticky effect inside the glass tube, so with the test rig set currently to detect on going off, that had alot less contact bounce, and at slow speed switched off really well, i did try uploading a phone clip of how the motor triggers but seems i can't
 
Last edited:

PhilHornby

Senior Member
Something you said, early on:
The issue I have had previously was the tilt switch would function correctly for a week or so, then for some reason it just stops, when giving the drive a flicking the tilt switch would correct itself just as if it hadn't activated the tilt switch correctly

and something I said, recently:

I note there is appreciable hysteresis - probably so much, you could just wire it straight to the motor! I think you would need to sense it resetting, rather than just sending a timed pulse.
Is that the crux of the problem - as Hippy put it, "it's like a brick sticking to a plank" (maybe not those exact words!)

So the tilt switch operates - the motor is run for 400mS and everything moves - but the switch is sometimes still "ON". So when everything moves back - as the spring tension is lost - there is no new signal from the switch and the clock stops.

In which case, continuing to run the motor, until the switch goes "OFF" (or some sanity-timer kicks in) is the answer.

If you can get the PWM working, so the motor runs slowly and gently, at some point, the switch must surely go "OFF".

(PS. I wouldn't let you try and wind my grandfather clock up in 400mS :eek: )
 

Rampz

Well-known member
(PS. I wouldn't let you try and wind my grandfather clock up in 400mS :eek: )
We are only moving 15 degrees in 400ms not winding anything any further than that, no speed to it. I did video it for you all but not sure how you can see it, i couldn't upload it, file issue

So the tilt switch operates - the motor is run for 400mS and everything moves - but the switch is sometimes still "ON". So when everything moves back - as the spring tension is lost - there is no new signal from the switch and the clock stops.
Could well be something like that, the 555 on the motor that looks after the chimes runs differently, if the switched stayed on it would carry on driving because for the chimes it takes much more power to run them, on the clock side its set as one shot and only resets when it gets a off then an on, these were the issues we were having when using a 555 timer, with the picaxe we should have a lot more room to adjust, we can do starting on or off etc.
 

PhilHornby

Senior Member
I did video it for you all but not sure how you can see it, I couldn't upload it, file issue
The "... Insert Media" option lets you link to a Youtube video. You can upload your file to Youtube, set it private and then get a link to it, which can be posted into the forum software.
 

Rampz

Well-known member
This is a clip of a vibration issue on an auto winder from last week, it was operating the tilt switch on each "tick"

Vibration issue

This had recently been upgraded to a 555 timer board, i guess shortly it will get a picaxe board, where i think i can check for a longer on pulse and disregard the pulses shown in the clip
 

PhilHornby

Senior Member
OK - can see them now.

Out of interest, is there enough energy stored in the spring to strike "12" ... so there could be just a single 'wind', rather than many? Would the connection angle of the spring allow it?

In the second clip - operating on the "tick" - presumably, if the motor had run for a smidge longer, or the tilt sensor had been slightly repositioned, it wouldn't happen? As in, it's still positioned right on the 'edge' of detecting a tilt?
 

Rampz

Well-known member
Out of interest, is there enough energy stored in the spring to strike "12"
Yes because the 555 is setup to keep resetting and then give a 400ms delay after the last pulse, so the motor will run longer for striking 12

presumably, if the motor had run for a smidge longer, or the tilt sensor had been slightly repositioned,
Not in this case, with the puling effect to motor was moving 400ms worth on each pulse and keeps pulling its safety pluh out, altering the postion of the motor doesn't help because you then have to adjust the tilt switch to suit and the vibration has enough force to activate the tilt switch regardless, the clip is from the manufacturer, i would expect if it could be allowed for then he would have done, he had to dampen the vibration to make it work, this auto drive was an existing unit that started having problems with the previous set up and was upgraded to the 555 version and tilt switch, i'm sure the picaxe version will be the way forward which i think could deal with the issues better?
 

David_Reynolds

Well-known member
This is a clip of a vibration issue on an auto winder from last week, it was operating the tilt switch on each "tick"

Vibration issue

This had recently been upgraded to a 555 timer board, i guess shortly it will get a picaxe board, where i think i can check for a longer on pulse and disregard the pulses shown in the clip
Hi, Rampz , I see the vibration issue and it is exactly what you would expect from a tilt switch that is only just off. Since you are going to reverse the situation to a opening switch this should stop happening,
I am guessing this was the thought when the timer was originally fitted, but in itself it only added to the problems .

As long as you make the motor move the minimum amount needed to stop it from bouncing back against the contacts then no more problem.
When I suggested it I was under the impression that the timer was to pulse the motor on for a fixed period, not that it was used to stop contact bounce.
It also reveals the snubber you mentioned as being some kind of pulse extender.

I was pleased to see that the contact clearing was a sticky motion it just means that the minor magnetic field holds on a little a bit , like the gas solenoid being held on whilst the hot junction is still in the pilot flame, just a few microamps from it is enough. Do not put any snubber across it for this purpose since you need the contact to hang on as much as it can.

The one thing I see from all this so far is, the switch is better working break to go, "BUT" that in itself requires the motor to have one timed pulse to set the distance it needs to travel to give the switch a clean run. (or a stop switch) . Without that timed pulse the contacts may roll back and keep bouncing every few seconds.
The fifteen degrees from the tilt switch means the glob moving from one end of the switch to the other this is not happening to your switch because it is not at the centre of the axis, it is just giggling about on the outer edge, as such it will never run out a full cycle, but that does not matter.
Since we can see that it can be done with a simple timer that only pulses once on an open contact and no other purpose than to set a fixed winding angle each time, spring pressure then applies the return cycle.

As for using a picaxe , yes you are getting the benefit of bring to alter the outcome and control the action of the winder to greater precision have alarms, etc.
As a last note , I see from the winding pulse that you are trying to achieve a very quick pulse at regular times so that it makes the clock easier to regulate, though without seconds sweep I think it would not make a damn of difference if you pulsed it every half hour with a two second burst it might be a few seconds faster for the first qua,rter then slower for the last fifteen, easy enough to compensate on the regulation.
I know that from what the winder is doing, it momentarily stops the clock as the winder moves and this is why such a short pulse is presently used.
 
Last edited:

rq3

Senior Member
Yes because the 555 is setup to keep resetting and then give a 400ms delay after the last pulse, so the motor will run longer for striking 12



Not in this case, with the puling effect to motor was moving 400ms worth on each pulse and keeps pulling its safety pluh out, altering the postion of the motor doesn't help because you then have to adjust the tilt switch to suit and the vibration has enough force to activate the tilt switch regardless, the clip is from the manufacturer, i would expect if it could be allowed for then he would have done, he had to dampen the vibration to make it work, this auto drive was an existing unit that started having problems with the previous set up and was upgraded to the 555 version and tilt switch, i'm sure the picaxe version will be the way forward which i think could deal with the issues better?
Rampz, is there no "sustaining" or "maintaining" mechanism on a tower clock? Just curious.
 

PhilHornby

Senior Member
is there no "sustaining" or "maintaining" mechanism on a tower clock?
The video I linked to in Post 109 shows one ... @ 2"12secs


Incidentally, the guy winding manually takes about 3 secs per revolution. This works out only125mS for 15° :unsure:
 
Last edited:

David_Reynolds

Well-known member
Yes Rampz, I get that it would be a real problem if the drive kept engaging and driving itself further and further away but it was not moving with a signal that should have made it move, It was being masked, that seems IMHO not to the answer .
The company chose to use a bang bang control, but then chooses to alter or ignore it, the switch must never be feathering and the drive should move enough to ensure that it does not. but if and when the switch state occurs it should move, it should never be able to pull its own plug if the switch is not asking for a pulse, barring a motor driver or switch failure.
That said, greater logical process can be used even given such a mean trigger arrangement.
 
Last edited:

David_Reynolds

Well-known member
Phil. the weights are kept in the very expensive electric winder albeit that they are smaller because of the larger pulley drum.
The format that is chosen by the company Rampz is negotiating with, is to do away with them and just pulses against a spring to give a few minutes of power to the clock.
I note the the hand winder has to set a small weight on a lever arm to keep the clock moving as he winds up the counterweight.
I wondered how it was done, now I know.
It does not happen with the pulsed winder so I suppose they want to keep it as short as they can, still I can't help thinking one second pulse every half hour is nothing in the grand scheme.
 
Last edited:
Top