Piezo Grill Lighter Sparker experiments
Trying to Make An Ignition Device
15 March 2013:
I know people have used old grill ignitors for engine ignition but I thought I'd fiddle a bit on my own.

The principle of the piezoelectric effect is a crystalline compound that generates electricity when deformed.  Conversely, if a voltage is applied to the crystal, it will deform.  The harder you whack them, the more voltage is produced and.  Conversely, the more voltage you apply, the more they deform.  In the 1930's, Rochelle Salts crystals were used in microphones and phono pickups.  That's why they were called "crystal" microphones and pickups.  Rochelle Salts are cheap to make and produce a nice high voltage (on the order of three volts peak-to-peak for a phono unit) so a high amount of amplification was not needed.  This was a big advantage back in the tube days.  The main disadvantage of Rochelle Salts is that they are water soluble.  They need to be encapsulated and if the encapsulation develops a crack, water vapor seeps in and it's bye-bye crystal.  Originally, the encapsulant was wax.  Later, soft rubber was used.  Another disadvantage of Rochelle Salts is that the crystal elements are fragile, easily fractured.

A few decades ago, scientists developed synthetic ceramic piezoelectric crystals.  These man-made devices are not water soluble and are much, much stronger.  Today, these ceramic crystals are capable of developing thousands of volts if whacked hard enough.  The most common uses for these ceramics are ignitors for various appliances and gizmos.

As I've been told, the main problem with using the sparkers as they come off of the grill is their limited lifetime.  Apparently, the snapping mechanism which is similar to the works of an automatic center punch but much cheaper is the limiting factor.  I looked up manufacturers' web pages and as far as I can see, they're guaranteed to make 30,000 sparks.  This is fine for grills, stovetops and cigarette lighters but for ignition, especially on a throttle governed engine, you can reach 30,000 sparks relatively quickly.

First, I rooted around and found a couple of piezoelectric grill ignitors and took one of them apart.

Here are the guts to a piezoelectric grill lighter.
My idea is to reduce the ignitor to the piezo element itself and then come-up with a longer lasting "whacker" device.  After prying off the cap, the internals are removeable.  Note that the "hammer" is the piece between the two springs.  It whacks on the end of the piezo element.  I can see how the lifetime of the stock guts is limited.  The hammer is pushed by the heavier spring (on the left) until the small diameter of the hamper comes up against the bottom of the pushbutton.  This forces the hammer to cock a little and slip off of a ledge inside the housing.  It is then propelled at high speed by the heavy spring until it whacks the end of the piezo element. The spring on the right is much lighter and serves as a return spring for the hammer.  I'm sure one of the life-limiting wear surfaces is the ledge.  Although the housing is made of hard glass-filled plastic, it will still wear to the point that the hammer slips off too easily, limiting the voltage output.

Stick it in the lathe and remove all but the ceramic piezoelectric part.
Once the bulk of the housing is cut away, just enough is left to hang onto the piezo crystal unit with it's attached "anvil".

Make a snapper.
My first experiment (not shown) was to simply clamp the piezo part in a vise, hook the wire to the bottom end and hit it with a hammer, observing the spark.  From this, I learned that it takes a surprisingly hard whack to get a decent spark.  I believe the mass and velocity of the striker have something to do with how well it generates.

Mount the stuff on a board.
This last view is a first try at a "whacker" for the element.  The spring is a piece of 0.031" thick X 0.750 wide steel strapping material.  This is what I had around and makes a relatively good leaf spring.  Results were disappointing, as the element only generated about a 1/8" spark.

Next time, I think I'll try a different method of actuating the hammer.  What I think will work best is a version of the original "automatic punch" having about the same mass.  I think that a moderate weight hammer going really, really fast is the best answer for a whacker.
18 March 2013:
I got a little more done today.  First, I finished "disassembling" the sparker I used for the first tests.  Now, I have the ceramic element, the hammer and the end contact freed of their plastic prison.

The bare bones guts of the sparker.  The voltage ceramic generating element is the beige cylinder.
After getting dimensions from the sparkers, I settled down for an hour or so with the CAD.  Here's the result of the brainstorm.

First design iteration of the ceramic magneto.
I've still got to work out the screw locations and some other details but this should suffice to see if it will work.  What I'm doing here is using all of the critical parts of the original sparker - the main spring, the hammer, the anvil, the ceramic element and the end contact.

More when I feel like it.   :-)

 19March 2013:
Well, I felt like doing some more CAD work.  The design has been modified to allow the end plates to be swapped in order to reverse the direction of rotation.  I've also eliminated the bottom contact on the element.  It now bears directly on the steel base which also is one of the high tension terminals.  Other changes are chamfering the gray PVC mounting so as to clear the trip wheel.

Here's what I started with today.

Drawing and the block of PVC to make the mounting for the hammer, anvil and spring.
With a bit of work in the mill, here's what I have so far.

Here it is so far.
Again, if I feel like fiddling with it, tomorrow I'll go into CAD and dimension some more parts and print the drawings.  It's possible I'll go out to the shop and do some whittling.
 20March 2013:
I did some more CAD and made some more parts today.  If all goes well, I should find out if it works tomorrow.

It's gettin' there!
 21 March 2013:
There's some good news and there's some bad news.  The good news is that it's far enough along to test.

Here, it's assembled enough to test.

The trip approaches the hammer.

The trip cocks the hammer, starting to bind.

Trip can't release hammer due to binding.
I put it together and turned the knob.  It really takes some force to start to raise the hammer.  When the hammer is raised to near the trip point, the hammer and spring are carried toward the trip pawl which binds-up the whole thing..

I should have figured that this binding would occur and will have to work out a way to guide the hammer.  I really don't want to make a new hammer with a longer tail because the ones in the sparkers are really hard steel.  I have nothing approaching this hardness.  If all else fails, I guess I could turn an old drill bit shank into a hammer but I don't think it wouild wear well.

There are some ideas bouncing around and I'll take a couple of days to cogitate on it.  Stay tuned.

I did a bit of an internet search.  I was told that Briggs & Stratton had an engine that used a piezoelectric ignition system.  I did find that Clinton had a kind of complicated piezoelectric ignition system.  The basic idea is covered in U.S. Patent 5,291,872.  In reading the patent and looking at the drawings, what I see is a complicated system using a ferrite pot core electromagnet that is energized via electronics from a timing pulse.  One of the drawings shows the magnetic piezo device integrated into a spark plug.  On a small air-cooled engine, the heat would be a big failure factor because the ceramic piezoelectric elements would probably age quickly.

I still think simple is better with these things.  There's got to be an elegantly simple way to make it work.

24 March 2013:
Back into CAD and I may have a way to make it work that won't wear out too quickly.  Here's the simplified drawing and some description of how it works.

Here's the deal.  I now use a crankshaft. a floating "rod" and an idler crank to make the pick blade stay parallel to the tripping edge of the hammer. 

As shown in position "A", the pick blade has just made contact under the tripping edge.  As the crankshaft rotates clockwise, to position "B",  you can see that the pick blade has carried the hammer up almost to the tripping point.  At "C", the trip has occurred and the hammer has hit the anvil.

I've increased the bore for the spring and added a bushing outside the spring that guides the hammer in it's travel.  This should eliminate the binding.  My first try at the bushing will probably be made of Teflon because it's self-lubricating and should run smoothly in the rather abrasive PVC. 

The floating pick rod rod and idler crank will be made of phenolic.  The three pins on which the rods rotate are some 0.152" diameter hard steel pins that came out of something.  The lower crankshaft pin will be anchored in the crankshaft and float in the trip rod.  The upper trip rod pin will be anchored solidly and will float in the oscillating end of the idler crank.  The top pin will be anchored solidly in the front end of the case and will float in the idler crank.  I'll make a small shaft collar for the lower pin to keep the trip rod in place.  Everything else will naturally stay where it's supposed to be.  (says here in the "fine print)

The bolt going to the spring washer above the hammer serves two purposes.  One is for spring adjustment and the other is to provide an electrical path for the "hot" spark lead.  Since the all of the case except for the bottom is insulating material, there shouldn't be any arc-overs.

25 March 2013:
Last night, while carefully inspecting the insides of my eyelids, I came to the conclusion that the above design is a big............
The reason is that the upper crank can get reversed or lock-up on dead center.  On steam locomotives, this was not the case because the rod throws on opposite side drivers were slightly offset.  This kept the cranks from becoming dead-centered.

Back to the drawing board.  I've still got at least a couple of ideas that should work and may be simpler.
26 March 2013:
I just got through working up another idea in CAD.  This one uses a face cam to follow the crankshaft oscillation and keep the pick blade horizontal from when the blade picks up the hammer and when it releases it.

What I've done is to make a face cam that a 1/4" ball bearing follower runs in.  The mean radius of the cam is the same as the stroke of the crankshaft so the pick rod can remain properly oriented.  To keep from having the same lockup or dead-center at the ends of the stroke, the cam follows the crank throw through only about 90 degrees then heads-off at a 45 degree angle.  This keeps the follower from getting into a situation where it's at the point that it's perpendicular to the crank throw and can't move.

I may start making parts in the next couple of days.  Then, I'll find out if I can make the cam.

31 March 2013:
Not a whole lot of shop time for the last few days but I'm getting closer to testing the new design.

Setting up the cam on the mill and rotary table.

And here's the finished cam beside the drawing.
Once I figured out how to set the part up in the mill, the job only took a few minutes.  After setting the rotary table to 45 degrees, I squared the rotary table in the mill and found the center of the circle that describes the 90 degree curve in the cam.  Rotating the table back to zero degrees got to the starting point.  Then, using CAD, I to dimension the 0.250" end mill at the start of the cut, I dimensioned the offsets which were where to the X and Y to put the tool in the starting position. 

After that, it was simply a matter of setting a cutting depth of 0.150".  The table was run in the Z axis for 0.498 inches, the rotary table turned 90 degrees and the Z axis run until the tool was  out of the material.  Other than a couplke of tiny breakouts at the sharp edges and having to hand scrape the cam so the 0.250" bearing ran freely in it, it worked out fine.  I probably should have made this part out of brass but I had a little scrap of 0.250" phenolic sitting right there.

Here's the assembly awaiting the spring bushing and re-boring of the spring bore.
1 April 2013:
This time, it works!

The hammer, spring, spring bushing and contact.
After some thought, I decided to make the spring bushing out of some PTFE (high density Teflon).  It's bored for the spring and turned to be a relaxed fit in the bore of the PVC housing.  The cutout at the right end of the PVC housing is milled out and intersects the hammer bore.  This is to allow air to rush in behind the hammer when it is tripped.

Here's the assembled "magneto".
The "magneto" will throw a 0.200" spark, which may be enough to fire a low compression engine.  The gap you see between the output terminal and a ground lug is the distance the spark will jump.  Surprisingly, the force to turn it isn't all that great, although when I get it under test, I have the sneaking suspicion that the press fits in the phenolic could loosen up even though I used bearing set Loktite on both the "main", the "rod" and the top bearing shaft.  Time will tell.

Taking Engine Number Four apart.
I had a few minutes before going out to mow the grass so I've disassembled part of Engine Number Four so I can use as a test bed for the "magneto".  It will be mounted upside-down under the radiator and will be directly driven from the camshaft.  Because the "magneto" will be destroyed if forced in the opposite direction of it's rotation, I'm going to have to come up with an indexing one-way clutch.  It shouldn't be a big deal.  I'd really hate the mag to destroy itself if the engine backfires upon starting.

Closeup of the crankcase showing camshaft (left, small) and the crankshaft, large)
The "plan" is to drill and tap the center of the camshaft for 8-32.  The shaft extension will be screwed into the camshaft and will exit the cover plate where the one-way clutch will probably be mounted.  A simple length of 1/4" rubber hose will be used for a coupling until the mag is proved.  After that, I'd make a better one.
4 April 2013: After a couple of days, here we are again.  First of all, I finally got off my duff and ordered a cheap quick change tool post kit for my cheesy little Chinese lathe. 

My new quick change toolpost.
I was assured by the salesperson that it would work on my lathe but when I got it, the tool height was about 0.060" too high with the height setting at it's minimum.  Also the hold down bolt was altogether different than the one on my compound.

What I did was to disassemble the compound and carefully set it up in the mill where I skimmed 0.125" off of the top surface where the tool post goes.  I now have about 0.065" of fool-around height.  The bolt had to be rethreaded to 1/2-20 and the compound drilled out for threading 1/2-20.  A drop of bearing-set Loktite should keep the bolt from wandering around.

Then, I got back to the piezo mag project.  In the meantime, I figured out how to make a relatively simple one-way indexing clutch.

Parts to the indexing one-way clutch.
The driving element of the clutch shown on the right, above, has a hole drilled almost all the way through.  In this hole is a spring and a plunger that, when fully pushed into the hole, clears the O.D. by a few thousandths of an inch. 

The driven element, the big item on the left, above, has a hole drilled into one wall that is the same diameter as the plunger and in line with it.  If  you look closely at the hole in the driven part, you will see that I've ground out a ramp on one side of the hole on the inside.  The bolt is to plug the outside of the hole and adjust how far the plunger enters the outer hole when it lines-up with it.

A view of the ramp inside the collar.
The above view shows a close-up of the ramp.  When the driven part turns counterclockwise as you see it, when the pin lines up with the outer hole, the spring pushes it into the hole.  Since there's a sharp edge on the upper side of the hole (as you look at it, above) the pin locks the driving and driven parts of the clutch together in a precise location.  If the engine backfires, the driving part turns clockwise and the pin cams out of the hole in the driven element and follows the inside of the bore.  It will drop into the index hole when it passes but, if it is still being driven in the clockwise direction, the pin will again cam out of the hole and take another trip.

There is another thing that may not be evident.  In the two photos above, you see a machine screw with a diameter turned on it's end.  This screw threads into the small hole in the driven element.  When the clutch is assembled, the turned end of the screw runs in the slot that is cut in the driving element.  This keeps the pin and its bores lined up and keeps the clutch from falling apart.

Assembled shaft, clutch, fan pulley, tach drive, etc.

Another view of the assembled works so far.
When assembled on the engine, it almost looks like a really clunky shaft adapter.  The "magneto" shaft goes into the bore in the end of the shaft and there are two setscrews for locking the shaft to the "mag" when the timing is set.

One of the things I had to do was to turn down the O.D. of the fan pulley to clear the shaft.  Now, the "O" ring belt is a bit loose and I may have to get a smaller one.

I've cut a piece out of a big rusty hunk of angle iron that I'll use to mount the "mag" to the aluminum panel that the radiator mounts on.  I'll get to that in a couple of days.

6 April 2013:
This was one of those "good news - bad news" days.

Here's the ugly hunk I cut the bracket (left) out of.
After cutting out the bracket and drilling the holes, it was mounted on the engine and timed with a 10 degree lead, enough to test it but not enough to make full power.  That will take about 25-30 degrees and make it a little hard to start.

The "mag" mounted underneath the radiator on the engine.
The plug gap was set to 0.020", just to make sure it would fire with the relatively low output of the piezo element and the engine started right up and ran nicely.  I quickly grabbed the camera and made a video of it because I had the sneaking suspicion that the phenolic crankshaft wouldn't go for long.  I was right.

Here's the movie!

There were actually two failures.  Right after I'd put the camera up, I noticed the "mag" was clicking more softly than it had at the beginning.  After the engine stopped and I did a post-mortem on the parts, I found that the first failure was the pick blade coming loose from the "rod".  The little 0-80 screw threads pulled out of the phenolic,  This caused the hammer to trip off of the phenolic and with less of a "cock", which both advanced the timing about 10 degrees and reduced the output of the piezo.  The second and final failure was that the main shaft had spun in the crank cheek. 

The plan is to re-design the crank cheek and make it of steel.  I have to make it oval shaped instead of round because I need clearance between it and the hammer to eliminate the possibility of the spark jumping from the hammer to the crankshaft cheek and, thence to ground.

Examination of the pick blade revealed that it was wearing pretty fast.  Although it is made of the hardest steel I can drill (steel strapping), if you click on the picture above and click again, it should zoom-in far enough to see the wear pattern.

I'll probably re-make the rod in either brass or steel.  There will also be a spark path problem but I think geometry of the rod can be changed to eliminate it.  I think I'll use a 1/8" hardened steel dowel pin for the pick blade this time.  I think I can press the pin into the "rod" in such a way that it won't come loose.  I'll have to think on it.

 8April 2013:
Well - I guess if it was easy, everybody'd be doin' it.

The new crank cheek, "rod" and pick blade are finished, installed and working.

The new parts, made from CRS.
Making the parts wasn't a big deal and they went into the "mag" for a test.

Here it is assembled and on the engine.
Cranked-up the engine and, just as I was getting to make another video, it quit.  No spark.  All of the mechanicals were working as they were supposed to but there was no spark.  When I took it apart, here's what I found.

AND here is what happens when you have the hammer spring just a wee bit too tight.
The pile of stuff is what's left of the ceramic piezo element.  When I set it up, I set the spring pressure for maximum smack.  Maximum looks like it's a bit too much.

I took my "spare" grill lighter apart to get the piezo element as a replacement but, when I took it out, it came out of the housing in two pieces.  Drat!  Now, I'm gonna do a bit of looking around to see if I can somehow "promote" some sample piezo elements.  At worst case, I can always go out and buy grill ignitors but, since I'm cheap, I'll try to get the bare elements.

23 April 2013:
The ignitors finally arrived.  It's no wonder it took so long - they came from Dongguan, China!

And a fine bunch of ignitors they are!
24 April 2013:
Finally, what passes for success!  Instead of totally dismembering the whole ignitor to get at the piezo element, I modified one of the new ones
by carefully removing all the unnecessary stuff.

Since the last two tries with the bare element ended-up with them turning to dust, I kept the molded plastic that surrounds the anvil, base contact and the element.  The bottom of the piezo was turned down to 0.375" so it could be pushed into a stepped hole in the base plate.  Contact was made with the ignitor at the bottom by putting a little piece of brass shim stock in the hole against the shoulder before pushing the ignitor into the base.

The modified ignitor and "piezomag" base plate.
The bottom PVC piece that originally housed the anvil and the piezo element was discarded.

"Piezomag" ready for testing.
If you look carefully, you can see that I had to radius the lower left corner of the 'rod' so it would clear the plastic ignitor housing.

The Piezomag was put on the engine and the timing was set to 25 degrees BTDC.  The plug gap was left at 0.018" from the last test.

The first run was 90 minutes at 600 RPM.  The test ended when the engine leaned-out 'til it stopped due to fuel level falling in the tank.  Total sparks so far was 27,000.

The second run (after a 5 minute break to adjust fuel mixture and re-start) was 75 minutes at 600 RPM.  The test was stopped to change spark plug gap and adjust timing.  Total sparks, so far was 49,500.

The third and final run of the day was with the timing retarded to 7 degrees BTDC and the plug gap was increased to 0.025" to make the piezo element work harder.  This run was for 205 minutes at 600 RPM.  The test stopped when dinner was announced.  Total sparks at the end of the day was 111,000.

I will continue to run the engine at the present timing and plug gap until something fails then do a post-mortem.  It's still a question whether or not the Piezomag will be a winner.
25 April 2013:
Another six hours of test run was done today.  It's still clicking away!  With the same timing and plug gap, it now has a total of 219,000 sparks on it and going strong.
28 April 2013:
Another day of test running.  All parameters are the same as the last time.  The total is not 336,000 sparks and it's still going.

The engine quit from leaning-out as the gasoline level fell in the tank and I had to pull it over several times to get it to re-start.  I may have flooded the engine and, considering the piezomag doesn't put out a killer spark, it might have had trouble burning off the excess fuel.

29 April 2013:
I ran the engine another 5 hours today.  That makes the total number of sparks 426,000.  After it passes a half-million sparks, I think I'll remove the piezomag and check to see how much it has worn.

The latest engine I'm building, The Mystery Engine, may have a piezomag designed-in.
1 May 2013:
In the last two days, the engine's run for a total of 561,000 sparks.  It ran out of gas so I decided that the test run was over.  I'll tear the ignition down and check the wear points in a few days.

This science-terrific stuff is really FUN!
[email protected]
Hits since 22 April 2013