Wooden gear clock, sync to RTC ?

binary1248

Senior Member
I need a suggestion on how to sync my pendulum clock to a precision (RTC) 1 sec pulse.
The clocks are home built wooden gear clocks and the pendulum is given a kick from a coil burid in the base.
The pendulum has a magnet in it and as it passes the base, a tiny reed switch senses the bottom position. Then the 08m2 gives the pendulum a tiny repulsive kick after passing BDC (bottom Dead Center).
Now the pendulum still has to be carefully adjusted for swing period and I can get 10 sec over 10 hrs accuracy.
Since I already build in a 08m2 to time the pendulum kick pulse, if I could add a RTC chip with the 08m2 I should be able to get many months of accuracy.
I have built two of these clocks and of course people are amazed and watch them since all the gears are open and exposed.
You can see one of mine in action at:
https://www.youtube.com/watch?v=WtaijXGS4Gs
.
Yes, it's an overkill, but it's fun to apply advanced technology to old technology.
.
This happens twice a second, once for right swing, 2nd for left swing. But I was thinking to use only one direction swing and create a 1 sec timing pulse for the RTC.
I have an extra input to the 08m2 for the 1 sec timing pulse from the RTC.
I don't need code, just suggestions on how to generally do this. Attached is an waveform capture of the control sense and drive pulse.
The lower is the reed switch, the center pulse is dead center over the reed switch, I only use the leading edge of the reed switch pulse for delay timing
.
Click on image to expand.

NewFile1 - Copy.jpg
.
Here is current code, pretty simple.
Code:
;April 30 2015   PEH
;   *************************
;   *Pendulam Timer         *
;   * Flie = PendulamtmrXX  * 
;   *************************
;
;
; Changed timing on dly & push 5/8/2015
; Added test points (red, grn, yellow) 8/31/2016
; Added startup seq to get pendulam going 9/14/2016 (not working)
;
;
#picaxe	08M2
#com 4	'specify com port
;
#no_data   'save some load time
'#no_table
'#terminal 4800

;
symbol coilpwr=C.2 ;pin 5 on chip

	
symbol inputx=pin3   	;input pin 4 on chip
				'symbol outblck=pin4  ;input pin 3 on chip  not used
symbol dly = 20		;was 20
symbol pushx = 55		;was 50, then 55
W7=0

Start:
	input 3,4         ;make ports 3 & 4 inputs, physical pins 4,3 
  ;Auto startup 	
	;for b1 = 1 to 20
	;high coilpwr ;coil pwr on
	;pause 50   ;red led on, coil pwr
	;low coilpwr  ;coil pwr off
	;pause 400
	;next b1  
	   
		
	
	low coilpwr   ;Coil pwr off
	'

	'serout c.0,T2400_4,(0XFE,0X51) 'Clr LCD Display;
	'pause 100
	'serout c.0,T2400_4,("READY ...........") 'Display value follwed by Hom
	'Pause 100
	'serout c.0,T2400_4,("Clearing LCD ",0XFE,0X51,"READY") 'clear lcd command
	
	;
	Main:
	'
'	serout c.0,T2400_4,("Clearing LCD ",0XFE,0X51) 'clear lcd command
	
 Do
	;	Find entry switch tripped
	if inputx=0 then  ;physical pin 4, no bounce detection needed
		'  GREEN TEST LEAD
	  pause	dly    'let pend swing past center
	  high coilpwr 'Apply coil power
	  high c.1    'strobe led test 'YELLOW TEST LEAD
	  low c.1
	  
	  pause pushx     'keep pwr on for delay time ...RED TEST LEAD
	  low coilpwr  'remove coil pwr 
	 pause 400 'stay off in case pendulam stops and reed sw stays closed
	 		'prevent coil overheat
 	 W7=W7+1   'for some timing testing only send to LCD
  'serout c.0,T2400_4,("Clearing LCD ",0XFE,0X51) 'clear lcd 
  'serout c.0,T2400_4,("CTS = ",#W7,0XFE,0X46) 'Display value follwed by Home Cursor
	 
	 endif
	
	
	  low coilpwr	'
	 
  Loop

Paul
 
Last edited:

hippy

Technical Support
Staff member
[ Moved from Finished Projects ]

I would expect you need to track real time and pendulum time and adjust pendulum pulses to bring the two back into sync.

Which is easier to do; speed the pendulum up or slow it down ? I would guess the later, by simply not giving it a pulse, so it may be appropriate to design an algorithm which always has the clock run fast but corrects that to run on time. If it ever actually falls behind it will naturally catch up because it's running fast. If it's too ahead, don't give it a pulse until it's on time or running slightly behind.

If you want it accurate all the time you probably need to vary the start and length of pulse to cause it to speed up or slow down. That would probably require a PID or similar type of algorithm.

I would guess you need to characterise what effects changes in pulse start or length has on swing / timing of the pendulum to determine what a PID-like algorithm should be.
 

BESQUEUT

Senior Member
I don't need code, just suggestions on how to generally do this. Attached is an waveform capture of the control sense and drive pulse.
l
1) Start the clock with the reed switch as today, assuming it is too slow or too fast, but absolutely not accurate.
2) wait for the RTC pulse and the reed pulse are in phase (IE : pulses are very closes in time)
3) At this point, only use the RTC pulse to activate coil.

I think that this kind of clock can start and work with only step 3, IE only RTC pulses.
 

Steve2381

Senior Member
Would this not be a case of setting up the clock to always run very slightly fast if its pulsed every swing.
Then count the inputs from the RTC in one variable, and count the pulses from the reed switch on every swing in another variable.
If the pulses of the swing are more than the RTC count, then don't pulse the pendulum until they are equal again.

I might be way off the mark with that suggestion however
 

Buzby

Senior Member
If I remember my school lessons correctly, the period of a pendulum has a fixed relationship with it's length, no matter how wide the swing is or the weight of the bob.

This would seem to indicate that you would need to use the coil to slow down or speed up the swing, depending on whether the reed was seen too early or late relative to the tick from the RTC.

If your coil is mounted at the bottom dead centre of the swing it would look like you may even just get away with pulsing the coil with the RTC, and totally ignore the feedback from the reed. You would need to experiment with the length of the coil pulse, so it may take a few hours to work out.

It just an idea, but very easy to try.

Your clocks are cool !!!.

Cheers,

Buzby
 

fernando_g

Senior Member
First of all, I wish I had 1/10 your woodworking skills. The clocks are amazing!

Buzby is correct. A pendulum's period is related to its length, length meaning the distance from the pivot point to its center of mass.

So...how does one affect the mass center distance? That would be the key to precisely adjusting the period.
I'm no mechanical engineer, but perhaps a tiny compensating weight that may moved up and down the pendulum's bar via a wormgear. The wormgear driven by a small motor controlled by a Picaxe of course.

This may sound way too Rube-Goldberg like contraption. But perhaps someone has a far better idea.
 

Buzby

Senior Member
How about suspending the bob with 'muscle wires', and adjusting the current to control the length ?

Or a thin motor shaft at the top that winds a thread up/down as needed ?
( This one sounds better, as the thread could be hidden in a telescopic bar. Low power, and no modern tech on show. )

There must be plenty other ways, what else can you think of ?
 

BESQUEUT

Senior Member
This would seem to indicate that you would need to use the coil to slow down or speed up the swing, depending on whether the reed was seen too early or late relative to the tick from the RTC.

If your coil is mounted at the bottom dead centre of the swing it would look like you may even just get away with pulsing the coil with the RTC, and totally ignore the feedback from the reed.
I agree.

You would need to experiment with the length of the coil pulse, so it may take a few hours to work out.
Not the length of the pulse : only the phase !

- If the pulse is to early , the clock is too late, so the pulse have to push the pendulum,
- if the pulse is to late, , the clock is early, so the pulse have to pull the pendulum
- if the pulse is in phase, pulse is axial, so no effect.

All that is stricly electro-mechanical : no need for other stuff...
 

techElder

Well-known member
Oh, come on.

The good part of mechanical clocks is revisiting them occasionally.

This whole idea is intriguing but disappointing at the same time.
 

cypherf0x

New Member
Seems like an ideal application for a PID using the time period differential. Let it figure out the best pulse length and timing.
 

BESQUEUT

Senior Member
Seems like an ideal application for a PID using the time period differential. Let it figure out the best pulse length and timing.
Of course why make it simple, when you can make it complicated ?

if not :
Pendulum.jpg
Blue = pendulum at the start of pulse
Red = pendulum at the end of pulse
black = force during pulse

1) force will help pendulum
2) force will help, then slow down pendulum, so no effect
3) force will slow down pendulum

==> pendulum will auto-synchronize to pulses...
 

erco

Senior Member
That is an AWESOME clock. I love it all, clocks, wood, phhysics, microcontrollers. Simply fabulous.

Now that's a mighty long pendulum swing. Very atypical for a clock, since your pendulum is driving the gear instead of vice-versa. So you're not getting true simple harmonic motion, which is only approximate for small pendulum displacements. I imagine you are putting in considerable electrical energy to keep that pendulum moving so fiercely, through mechanical and air resistance. I would bet that the resonant frequency varies with displacement/travel, and you might be able to send stronger/weaker pulses to very the travel and thus make it run slightly faster or slower. But of course the ratchet may start skipping teeth if the swing travel gets too small, so there's a limit to that end.

I would suggest that you start by making the clock run fast (as others have suggested) then find a way to periodically slow it down as needed by purposely skipping ratchet teeth occasionally. Let the pendulum swing, but disconnect it from the geartrain so the hands don't move. You could add a servo or clutch up at the pendulum ratchet or hidden in the geartrain to let a few swings go by. Or just stop the pendulum pulses for a few seconds so the oscillation travel dies down and a few teeth are intentionally skipped.

If you go this route, you may need to add a feedback sensor somewhere, maybe a magnet on the ratchet wheel triggers a hall effect sensor to let the micro know when a full revolution has been completed to compare displayed time to the RTC.
 

AllyCat

Senior Member
Hi,

Yes erco has just rasied the issues that concerned me. The pulses are required not just to synchronise the pendulum but to energise it. Also, the period of the pendum does increase for larger amplitudes. So if it is "fast" you need to increase the energisation. Wikipedia puts a figure on that:

"at an amplitude of θ0 = 23° it [the period] is 1% larger than given by (1) [a small angle of swing]"

Also, the energisation of the pendulum will depend on the distance between the (permanent) magnet and the coil (inverse square law) so "phase" control might not work as expected. It may be possible to synchronise the pendulum by arranging narrow 1 second periodic pulses to pull the pendulum. Then (my theory suggests) if the pendulum arrives a little early it will get a bigger "pull", increasing its amplitude of swing and slowing the period. Converesly, "Pushing" a "late" swing (i.e. with the pendulum nearer) would increase the amplitude, making the next swing even later.

Of course a closed loop system (e.g. changing the length of the pendulum) probably does improve the chances that a "paper" design will work first time, but I think I'd try KISS first.

Cheers, Alan.
 

BESQUEUT

Senior Member
It may be possible to synchronise the pendulum by arranging narrow 1 second periodic pulses to pull the pendulum. Then (my theory suggests) if the pendulum arrives a little early it will get a bigger "pull", increasing its amplitude of swing and slowing the period. Converesly, "Pushing" a "late" swing (i.e. with the pendulum nearer) would increase the amplitude, making the next swing even later.
If so, just revert the magnet OR the coil...
Also note that energise or synchronize the pendulum is the same thing !
==> if it lack energy, the synchro will tend to energize it
if to much, synchro will tend to reduce energy.

Also note that period of a free pendulum only depend from his length, not amplitude (if small enough).
 
Last edited:

Buzby

Senior Member
Here is a calculator for pendulums, both large and small amplitude swings.

http://hyperphysics.phy-astr.gsu.edu/hbase/pendl.html#c1

I did not remember from my school days if we ever did large angles, that's why I thought only the length was involved.

I'm now tending to agree with BESQUET, that if the natural period of your swing is very close to 1 sec, then just pulsing the coil is all you will need.
As the pulse will advance/retard the pendulum by a very small amount per swing around BDC, I would expect that it will not strain the escapement too much.

This is a really interesting project, please keep us informed.

Cheers,

Buzby
 

binary1248

Senior Member
WOW, many thanks for all the suggestions. I will try many of them, working from the simplest to the more difficult. The pendulum timing is set by the center of the mass to the length to the swing point. I will try first to simply strobe the kick magnet and ignore the reed sensor. My RTC chip should be in shortly. Could have used a crystal controlled PIC without theRTC, but wanted to stay with the 08M2 chip.
Right now I am dealing with an incoming hurricane so I have been working on getting son-in-laws generator ready. He let it set to long (couple of years) without proper storage prep. My gen set is working and always started every few weeks.
.
Thanks Hippy for moving this to the correct area.
Paul
 

AllyCat

Senior Member
Hi Paul,

Make sure you try both polarities of magnetic pulse and with the natural frequency of the pendulum intentionally set both higher and lower than the required period. The frequency of some oscillators can only be "pulled" (or pushed) in one direction and the pendulum operation is rather non-intuitive (as shown by some of the suggestions on the first page*). Also you need to achieve the opposite of "normal" pendulum operation, i.e. to over-ride its natural frequency to correspond with an external source (instead of aiming for the best accuracy). In particular, look at "The Airy condition" about half-way down the Wikipedia Article.

* If you reduce the drive to the pendulum (by reducing the pulse duration, or current, or omitting pulses) then the amplitude will reduce and clock (frequency) will actually go faster (at least until the amplitude is so small that the escapement fails to operate correctly).

So Good luck with your expereiments and the weather. :)

Cheers, Alan.
 

mikeyBoo

Senior Member
1 Hz Clock

my internet is slow (don't seem like something as simple as a hurricane would effect it)
there is a subroutine called ElapsedTime_Get in my kayak control project (misc. projects) that could be adapted to do what you're asking
I use it to to update a "time on water" display (the Picaxe "time" is not accurate enough for real time display)
You would have to change the SREQ2 assignment to whatever pin you want to use (this project uses a 20M2)
Below I chopped out the parts I think you will need:

Code:
; SREQ2 will have to be set to the pin you want to use <-- LOOK
symbol SREQ2             = pinB.2  ; (Service Request2 via DS1307 (Used for 1 Hz clock input)
symbol lastClockLevel    = bit27   ; holds last sampled value (0 or 1) of 1 Hz clock (elapsed time clock 1 sec tick)
symbol currentClockLevel = bit28   ; holds "just now" sampled value of 1 Hz clock (elapsed time clock 1 sec tick)
; XW = variable X word value, XL = lo-byte of variable X, XH = hi-byte of variable X
symbol XW = w2           ; alias register for w2 word value
symbol XL = b4           ; alias register for w2 lo-byte value
symbol XH = b5           ; alias register for w2 hi-byte value


; assign I2C addresses
; Picaxe I2C ADDRESSING:
;     does not use actual I2C value, but rather the I2C address must be multiplied
;     by 2 (e.g. actual I2C = 33 (21h) will have to use 66 (42h)
;     bits 7...1 hold the 7-bit address, bit 0 is always 0
symbol DS1307 = $D0       ; actual I2C address for clock is 104(68h)

gosub init  ; initialize hardware & program variables

; VERY IMPORTANT: the main loop must not exceed 500mS or elapsed time errors will occur
main:
    gosub ElapsedTime_Get           ; elapsed time check
    gosub ??                        ; your other stuff
    gosub ??                        ; more stuff
    goto main


; ElapsedTime_Get --  (v1.00a by M. Ballew 02-27-2015)
;       updates elapsed time registers for a Picaxe app
;       requires an accurate 1 Hz square wave to maintain correct elapsed time
;       to maintain accuracy, this proc must be called at least once per second
;       at this time the clock is a DS1307 Fout (pull hi 100k) connected to the SREQ2 pin
;
; Example:
;       gosub ElapsedTime_Get  ; elapsed time update
; local variables: XW YL ZW
ElapsedTime_Get:
    let currentClockLevel = SREQ2               ; snapshot clock level
    ; toggle b.2                                ; simulate clock (toggles SREQ2) TEST ONLY
    if lastClockLevel = currentClockLevel then  ; has clock level transitioned?
        return
    endif
    let lastClockLevel = currentClockLevel      ; clock has transitioned, so save level (0 or 1)
    if currentClockLevel = 0 then               ; count only on positive transition (i.e. once per second)
        return
    endif
    ; INSERT YOUR PULSE OUT CODE HERE
    return

init:                                                   ; DS1307 only used for 1 Hz output at this time
    ; gosub RTClock.timeDate_Set                        ; skip this line if clock does not need setting
                                                        ; may have to do above call first time DS1307 is used
    hi2csetup i2cmaster, DS1307, i2cslow, i2cbyte       ; plug in Real-Time Clock I2C address
    let XL = 0 : hi2cout $0,(XL)                        ; write DS1307 seconds register to enable clock
    let XL = $10 : hi2cout $7,(XL)                      ; write DS1307 frequency output register (enable 1 Hz Output)
    return
 

hippy

Technical Support
Staff member
Seems like an ideal application for a PID using the time period differential. Let it figure out the best pulse length and timing.
Of course why make it simple, when you can make it complicated ?
I suggested similar. I have concerns over using a constant every second pulse -

Does that adversely affect the natural motion of the swing ?

How does it synchronise ? If the one second pulse fires when the pendulum is nowhere near the coil it will have little effect, if it fires just before it swings over the coil it may actually slow the pendulum down, if it fires just after, then isn't it just putting more and more energy into the swing ?

Perhaps these aren't problems.
 

Steve2381

Senior Member
Could you not leave the pendulum swing entirely alone (adding anything mechanical to the actual swinging arm would be a nightmare). Simply count its passes at the bottom of the swing.
Set the pendulum to swing slightly slower than you actually need.
Then, affect the speed by mounting an opposing electromagnet at a point near the end of its swing instead (thus shortening the swing on that pass). I imagine the smallest increase in magnetism would affect the timing quite a lot. Would not take much of a pulse however,
Trouble with all these ideas, is they will mean adding something not particularly cosmetically pleasing to your lovely masterpiece.

It all seems a bit complicated. There must be a simpler solution.

How about not altering the swing at all? Set the clock swing arm so that it always runs very slightly fast, and keep the swing count in a variable.
Then slow the clock by applying a brake to the first geared wheel (before it gets geared right down). A tiny electromagnetic solenoid applying a soft brake to the wheel would slow it for a tiny amount it when started to get too far ahead of the RTC count.
Stop the brake when the timing is back inline

Camera shutter solenoids are tiny. Or a cheap Ebay job...

View attachment 20276
 

BESQUEUT

Senior Member
Then slow the clock by applying a brake to the first geared wheel (before it gets geared right down). A tiny electromagnetic solenoid applying a soft brake to the wheel would slow it for a tiny amount it when started to get too far ahead of the RTC count.
Stop the brake when the timing is back inline
Unfortunately, this will not slow down the clock, but can only stop it or do nothing...
 

BESQUEUT

Senior Member
Look at #15.
There are only 3 possibilities to change period :
- theta : by construction, this is not an option...
- L : difficult, but a fun option to experiment ! (For example a little servo with a mass at the arm end, anywhere on the pendulum...)
- g : ? not possible ? ==> a magnetic field is not very different from a gravity field...
 
Last edited:

Steve2381

Senior Member
How about a tiny fan (think tail rotor on those tiny helicopters) that blows at the end of the pendulum to slow it down or behind it to speed it up.
I think I will shut up.
 

AllyCat

Senior Member
Hi,

IMHO there's nothing wrong with using magnetic pulses to drive the pendulum (they're probably the easiest method), but there are TWO independent conditions which must be met, regardless of whether you use an open loop (fixed period pulses) or a more complex closed loop system such as PID:

The pulse needs to be applied in such a way that it can control the period of swing to give the desired accuracy AND it must apply a force in the correct direction to maintain the swing. Since these conditions are independent and each has a required direction (polarity), there are FOUR possibilites and three of them wont' work as desired (two will stop the clock and one will adjust the timing in the wrong direction). On top of this you need to consider "exceptional" conditions such as starting (or recovering from an out-of-sync condition), or if the swing is too large or too small (in particular if it doesn't operate the escapement correctly).

As I said in #17, the system is counter-intuitive and there's also the Airy Condition to contend with, but I believe that the required conditions can be met by arranging the one second (or half-second) period pulses to PULL the pendulum just before it reaches the bdc position. This should be easy with a PICaxe because it can time the delay from the previous bdc position (if necessary) with sufficient accuracy to ensure that the pulse doesn't "kill" the swinging (by pulling after bdc).

This could be self-stabilising in normal operation: If the bob arrives a little "early" then it will be nearer to the magnet and receive a larger pulse than previously (inverse square law). Therefore, the next swing amplitude will be larger and take slightly longer, thus slowing the timing as required. Like any control system the pulse energy will need to be scaled to produce a "useful" timing control range, but not so large that it causes instability.

Cheers, Alan.
 

cypherf0x

New Member
If you use a large piece of copper or aluminum as the weight on the pendulum then it would be feasible and rather simple to use eddy currents induced by a pwm controlled electromagnet to act as a brake. Set the time period of the pendulum a bit fast and using something like a hall effect sensor at a different position to detect passing. The braking force could be adjusted to match the period you want.
 

BESQUEUT

Senior Member
How about a tiny fan (think tail rotor on those tiny helicopters) that blows at the end of the pendulum to slow it down or behind it to speed it up.
I think I will shut up.
Excellent ! I agree !
Same remark as magnetic : no need to detect pendulum and/or reverse effect. This will auto-adjust...
Another fun way : a servo attached to the pendulum to turn a "veil"
If a servo can be attached to the pendulum, the L effect (mesured to the mass center...) is more reliable. But the aerodynamic effect is more fun...
The pulse needs to be applied in such a way that it can control the period of swing to give the desired accuracy AND it must apply a force in the correct direction to maintain the swing. Since these conditions are independent ...
IHMO, these are in fact exactly the same thing. Control period and maintain the swing are the same thing. So, there are only two possibilities. If force is not in the correct direction, reverse it.
With a coil and a magnet : reverse the coil or the magnet. The two others possibilities are :
- reverse both : you are right : this is not a good idea,
- reverse none : I have to test that to known if it work...
 
Last edited:

AllyCat

Senior Member
Hi,

IHMO, these are in fact exactly the same thing. Control period and maintain the swing are the same thing. So, there are only two possibilities.
Pulling the bob just before bdc or Pushing it just after bdc will have exactly the same effect in maintaining the swing. BUT:

If the approaching bob is slightly early then the pulse will have MORE effect (smaller distance), increasing the swing amplitude and slowing the clock timing (the required effect).

If the "early" bob is Pushed (i.e. it has moved further away at the reference time) then the pulse will have LESS efect, reducing the swing amplitude and speeding up the clock timing even more (which is NOT the desired effect).

Then there are two further possibilities (pushing before bdc or pulling after bdc) which will stop the clock (so not very useful).

Cheers, Alan.
 
Last edited:

JPB33

Senior Member
On my clock I wound the coil below the pendulum (0.5 secs duration) without a metal core. I trigger the coil from the sensor at the bottom of the pendulum swing and I've got my permanent magnet on the bottom of the pendulum arranged so that it repels the magnet when the coil is energised. I read pulsin from the sensor, the time the coil is energised is tied to the length of this pulse which diminishes as the swing increases until it reaches a preset value (22.5ms) and the energising pulse is missed, pendulum swings slightly less and the process repeats. I have'nt been able to measure accurately the amount the swing decays by witout the pulse, thats one of the jobs this winter.

Found that triggering the coil at the bottom of the swing is about right to give it a push so it builds up nicely. Tried putting delays in but only made it worse or had little effect so left them out.

Everytime I go passed I'm pleased to see its still working and displaying the pulsin--thank you Picaxe!
 

techElder

Well-known member
The OP is using a reed relay for sensing, but should remember that the same coil used for locomotion can be used for sensing.

Also, using the coil to sense can also give an analog indication of how quickly the magnet passes the coil.
 

binary1248

Senior Member
Well, the hurricane has passed with no damage to my property, just lost power for 3 days. While I had a gen set backup, apparently the internet system requires power at a remote box down the road, so no internet connection till tonight.


The OP is using a reed relay for sensing, but should remember that the same coil used for locomotion can be used for sensing.

Also, using the coil to sense can also give an analog indication of how quickly the magnet passes the coil.
Later I planned to use the coil for sensing, but for simplicity until I understood the timing I used a reed switch. The bob has a magnet in it and the coil pushes just after BDC (see my timing scope in the opening thread).
The blue pulse is the coil power kick.
The red is the reed switch. The little blip in the middle of the reed switch sense is as the magnet passes directly over te reed switch it causes a null condition and the reed momentarily releases.
 

AllyCat

Senior Member
Hi,

Glad to hear there was no damage. A reed switch is a lot easier to interface to a PICaxe pin than a small induced current or voltage. ;) And it can also indicate the bob speed using PULSIN. :cool:

Cheers, Alan.
 

techElder

Well-known member
... Later I planned to use the coil for sensing, but for simplicity until I understood the timing I used a reed switch. The bob has a magnet in it and the coil pushes ....
In actuality, it may be better to use separate coils for sense and drive due to the different impedance requirements, but it still might be possible to use the single coil. I definitely wouldn't use a reed relay for more than a prototype. Not long term for sure.

You also may be better off exposing the "pendulum" magnet to an armature that the coil is wrapped around rather than an open-air coil.

Also, don't dismiss offhand a scenario of the winding of several coils on separate armatures and switching in the armature of choice at the right time. I've seen that work in some timing applications.

PS. The coils wound on armatures could be the coils salvaged from actual relays.
 

binary1248

Senior Member
PS. The coils wound on armatures could be the coils salvaged from actual relays.
I wind my own coils using 30 gauge enamel wire on a steel cored bobbin.
Tried using an old relay (12v) but couldn't get enough power until the ps was at 35 volts or higher. Since I have a lathe it was easy to wind my own to the requirements I wanted.
I may use a IR sensor (reflective type ) as a position sensor later.
 
Last edited:
Top