Another Knucklehead Build

[mayhugh1]

The dual spark CDI unit from cncengines.com arrived promptly after it was ordered. At 1.75" x 2.75" x 1" this new version of their ignition comes potted inside its own enclosure but is larger than the PC board units I've purchased in the past. The second coil in this particular model may account for some of the extra size. Testing showed the outputs fire on the rising edge of the input pulse (i.e. magnet moving away from the Hall sensor) which is how the previous generation units functioned.

My plan is to place the ignition behind a control panel that I'm adding to the stand near the front of the engine. In order to accommodate the starter, a sealed 12V lead-acid battery will be used for power. Since the CDI requires something less than 6V, it will receive its power through a step-down converter installed inside the control panel's housing. A second adjustable converter will be used to regulate the flow of a recirculating fuel pump that will supply fuel to the carburetor bowl. I've used these inexpensive converters in the past with no issues so far:

https://www.amazon.com/gp/product/B07DYPMPJG/?tag=skimlinks_replacement-20

A pair of holes were drilled through the side of the enclosure above each converter board in order to access its adjustment pot and for a light pipe to make its onboard led visible from outside.

The housing will also contain a 30 amp starter relay as well as a version of the Hall indicator circuit board that I've used on all my other CDI-fired engines. This indicator uses an led to display the state of the Hall sensor without need for powering up the CDI and dealing with its high voltage. This has been very convenient for engine timing and troubleshooting in the past.

The housing for the control panel will eventually contain a mixture of small signal, high voltage, and high power circuitry all operating in close proximity. Both converters as well as the CDI contain high frequency oscillators that, along with the starter relay, will generate plenty of electrical noise. I spent time arranging models inside a virtual housing in order to come up with a placement that will hopefully minimize interferences among them.

The control panel was recessed into the rear of its housing. This will bring a number of otherwise protruding switches and indicator lights within the protective envelope of the enclosure. The real reason for doing this, though, was to make an otherwise boring box into something a little more interesting to work on. The enclosure was milled from a chunk of ebay Delrin that wound up mostly as chips. Fortunately, the block was thick enough to yield a slice for its cover.

The panel connectors for the Hall cable, starter motor, and fuel pump will be JB Welded into the rear side of the enclosure. Since Delrin's slick surface isn't conducive to good adhesion, the joints were designed with internal irregularities which will grip and stabilize the essentially potted connectors. A standard Futaba J servo connector will be used for the Hall sensor cable, but the RC aisles of a local hobby shop were shopped for the other two connectors.

In order to help test the starter system, I made a temporary clear plastic plate to support the outer ends of its driven shafts inside the cam box. This will allow me to see the chain drives in operation since they will later be hidden behind the cam box cover.

The plug wires will be brought out through the enclosure's cover in order to keep them as far away from the Hall sensor wiring as practicable. I had a bit of fun with the cover design which resulted in a lot of extra machining to make it look like a heat radiator. Even though the enclosure may seem to be larger than necessary, it looks like it will be filled almost completely with electronics. - Terry

 

[tornitore45]

This indicator uses an led to display the state of the Hall sensor without need for powering up the CDI and dealing with its high voltage. This has been very convenient for engine timing and troubleshooting in the past.

Good idea, I think will steal it for the Hoglet.
My guess is that you need an extra switch to power
the Hall-Sensor-Statuts- Indicator without powering up the HV board.
Or did you come up with some clever way to avoid it?

 

[mayhugh1]

Mauro,
Right now, as far as I can see, it requires a second switch. On my panel, the PWR switch will enable power to the control box including the indicator board. The IGN switch will turn the CDI on.

I'm working on a schematic now. The new CDIs have their own indicator led built in, but it's powered along with the CDI and closely attached to its potted enclosure. I'm currently making input measurements on the CDI to determine what its input actually looks like in order to ensure there isn't going to be any interaction between that one and mine. -Terry

 

[tornitore45]

I was thinking of placing decoupling diodes in the VCC to the sensor. Both sources can feed the Sensor but can not back feed the other box.
One never know what kind of equivalent circuit is presented by a powered down Black Box.

As of now I have remoted the original LED but the HV still needs to be active during the adjustment phase.

 

[mayhugh1]

Mauro,
I was thinking the same thing. I've determined there's a dropping diode between the Vcc line and the rest of the CDI which is what was done on the older units. - Terry

 

[ecdez]

First time seeing this thread and I just scrolled through 9 pages of awsomeness. Excellent work!

 

[mayhugh1]

Embedding the three panel connectors and their associated hookup wires into the side of the enclosure took a few days since each one required a couple steps separated by some JB Weld time. With the connectors finally located, I was able to make up the cable to the starter motor as well as shorten the sensor cables made earlier so they could be routed safely away from the flywheel.

After coming up with a wiring diagram, it was 'only' a matter of installing and soldering the various electrical components together. Along the way, I realized my enclosure had a poor form factor for how I was trying to use it. So much wiring had to be shoehorned into the spaces among the already mounted components that it quickly turned into a rat's nest.

I had planned to use one of the spare sensor indicator boards left over from one of my earlier engines. However, I discovered yet another version was needed with some additional terminals in order to avoid burying connections that might be useful later for troubleshooting. I probably should invest the effort required to learn proper PCB layout software, but with only one trivial board every couple years, I'll likely continue (mis)using SolidWorks. I tried my best to make sure that it will be possible, even if difficult, to replace failed parts. The kludge I created inside the control box isn't pretty, but it works as intended.

Unfortunately, the starter tests uncovered an unrelated problem. When finished, I could feel some roughness in the one-way bearing being used as a clutch. Under a microscope I could see some very light sprag marks beginning to show up on the countershaft. This was a nasty surprise since the starter had only a few minutes of operation on it. I had hoped the 1144 alloy used for the bearing's inner shaft would stand up to the sprags, but I was wrong and ended up machining a new shaft from O-1 drill rod. After a soak at 1475F, it was quenched and tempered at 350F. I couldn't measure any distortion resulting from the heat treatment, and so my original concerns that steered me to 1144 now seem unfounded.

The last step before our house begins filling up with family for the holidays was to machine a crow foot hold-down for the distributor. After the holidays, I'll likely start working on a fuel tank and fuel pump housing that will also be attached to the stand at the rear of the engine. I'm hoping to come up with something a bit different from the typical cylindrical tank. With the frustrating control panel wiring out of the way, it'll be nice to start making chips again.

My best wishes to all for a safe and happy holiday. - Terry

 

[stragenmitsuko]

For a milled pcb you board looks very clean .
Most of the time marks from the cutter ruin the looks of milled pcb .
That doens't affect functionality tough , its merely cosmetic .
But yours looks as good as an etched pcb .
Nice work !

Just wondering , why did you put the parts on the "wrong" side of the pcb ?
Usually for a single sided board that is , the parts are on one side , and the copper traces on the other .
That's why these pcb ar called thru hole mounted .
As opposed to smt wich are surface monted .

Pat

 

[tornitore45]

Mounting components on the circuit side has one advantage:
You can "read" the schematic just looking at the board.
Since usually a "good engineer" has the schematic memorized it help to find the right node to probe when debugging.

 

[JRD56]

Hello, New to the forum and I just started looking at your thread. I love the mix of mechanical and electrical engineering. Could you explain how the Hall Sensor triggers the CDI. What is the purpose of the 2N2222?

 

[dsage]

Hi Terry:
I assume you're using one of Roy's original CDI modules?
Don't they issue a spark on the Positive going edge of the trigger? i.e. the Hall sensor output goes low (with the magnet) and the spark is issued when the magnet disappears and the hall sensor output returns high?
I'm not saying anything is wrong with your arrangement as long as you realize that your spark will occur when your led goes off, not when the led turns on.
It's been a while since I used the modules but I thought they were made to duplicate the action of a Kettering ignition (as above).
Did he change something?
Nice job as usual.

 

[mayhugh1]

Thanks all for your comments and questions.

Pat,
I would normally put the parts on the correct side, but these simple sensor boards usually end up embedded in tight quarters and putting the solder side up make it easier to remove/replace a failed component without pulling the whole board. There's also a lot to what Mauro said.

The copper thickness on the boards I'm using is .005" and I'm removing it with a .063" 4 flute cutter running at 5000 rpm, 5 ipm, and a .010" depth of cut. I, too, am surprised the result is so clean.

JRD56,
The purpose of the little circuit board is to provide an indication of the state of the sensor without powering up the CDI. It's convenient for timing and troubleshooting that doesn't require plug firing since you don't have to remember to provide spark gaps and engine grounds for the high voltage outputs which could otherwise blow a Hall sensor or create latent internal damage to the high voltage coils. These precautions are absolutely necessary, easy to forget when your mind is focused elsewhere, and must be in place every time power is applied to the CDI. Roy provides a similar interface and indicator led for the sensor inside his unit, but the unit must be powered to use it. My PWR switch energizes my indicator, while the IGN energizes Roy's indicator and CDI. When the engine is normally running, both are energized and running in parallel. The 2N2222 is just the switch turning my indicator's led on and off.


Dave,
I'm using one of Roy's dual CDI's and it works exactly as you say. When timing the engine I use the led coming on as a warning that the magnet is 'dwelling' in the vicinity and that the spark is soon to follow. It's just a personal preference and a carry-over from the old Kettering systems where there is a dwell period and then a point opening that fires the plug. - Terry

 

[dsage]

Thanks Terry:
That's what I figured and it's the way I like to arrange it as well.
I also like your approach to the circuit board. I'll have to remember that one.
Thanks

 

[JRD56]

Thanks for explanation Terry. Back in my younger days I was a drag racer and used MSD ignition systems. With an MSD you always need a "load" if it was powered up and the engine turned over. Otherwise you'd fry it. I saw quite a few guys pull the coil wire and use the starter to turn over the engine to set the valves, etc thinking it was the right thing to do. Unfortunately they fried their MSD. Your approach definitely helps to avoid a similar failure.

 

[mayhugh1]

My goal for the fuel tank was to come up with something other than a classic cylindrical tank. A bike tank would have been novel, but it didn't seem appropriate connected to an engine installed on a test stand. YouTube videos show bike shops hanging all sorts of containers above the engines they test - some not so appropriate. While searching the web for ideas, an HDPE blow-molded tank caught my eye. After thinking about it over the holidays, I decided to create something similar in metal.

An electric fuel pump will allow me to place the tank low on the stand so it's kept subordinate to the engine. I've used fuel pumps on all the multi-cylinder engines that I've built and have been very happy with the consistent fuel delivery they provide. Although I've prominently displayed them in the past as accessories, I didn't want this one visible along side an old school motorcycle engine.

The three-piece tank design that I came up with is shown in the SolidWorks assembly photos. Although I'm trying to make the top portion of the assembly look like the fuel tank, it's actually an enclosure that I'm using to hide the pump. The tank itself is the bottom half of the assembly and separated from the pump's enclosure by an o-ring'd spacer. For safety, it's important to completely isolate the tank from the pump in order to prevent gasoline fumes from accumulating inside the enclosure around the brush motor. The tank's capacity is about 2-1/2 ounces which should provide some 10 minutes of run time. Realistically, though, the engine most likely won't be able to continuously run that long without overheating.

The fuel pump that I'm using is the same commercial product that I've used in the past. RC enthusiasts have been fueling their planes with this particular pump for many years and, remarkably, it's been available from my local hobby shop for over a decade. Although the manufacturer doesn't recommend its use with gasoline, I've been pumping gas with these pumps for years with no problems so far. The pump quickly self primes and, in recommended use, is typically located above the fuel source. I like using them in this same way because their rigid shaft seal stays dry except when the pump is running, and the risk of leaks is reduced.

The fuel pump consists of a pump coupled to a small dc motor through an Oldham coupler. The motor is designed to be operated from 6 or 12 volts. The variable voltage source inside the control panel's enclosure will be used to fine tune the flow rate of the recirculating loop once it's finally running.

I typically remove the pump and motor from their factory plastic enclosure and repackage them in a more appropriate machined metal enclosure. The machining of the recessed supports that hold the two in alignment is rather involved and essentially duplicates what was molded into the original enclosure. Several years ago, I invested effort in developing the CAD/CAM to reproduce it, and the Knucklehead's pump will be my sixth re-use of it. To protect my investment, I stashed away a couple extra pumps.

Although it complicated the design, I routed the pump's pickup tube wholly within the assembly so that all external fuel lines could emanate from the upper section and help maintain the illusion that the upper section is the tank. This required modifying the pump's input feed tube. I also plan to bring the electrical connections for the motor out to a panel connector mounted on the bottom of the assembly. This will require a pair of spring-loaded electrical contacts mounted between the tank and the pump enclosure but outside the o-ring seal. - Terry

 

[gbritnell]

Terry, Happy New Year to you and your family. I really enjoy your designs for the ancillary parts and pieces. It's almost anticlimactic when the engine runs and we don't get to follow along any more, just waiting for the next project.
gbritnell

 

[mayhugh1]

Thanks, George. I greatly appreciate your comments. May the new year bring you and yours new knowledge, satisfaction, and riches. - Terry

 

[mayhugh1]

Construction of the fuel tank assembly began with the topside surface of the pump enclosure. Even with CNC, the complex filleting that makes up much of its design also makes it a finicky part to machine. Stacking fillet upon fillet can become infectious, but wrinkly surfaces it can sometimes produce are difficult to detect except on the finally machined part.

My goal, which I only occasionally achieve with 3D parts, is to put the necessary upfront effort into making the mill do all the work to produce parts that require no additional cleanup beyond bead-blasting. The 80 grit media that I use is capable of removing machining marks up to two or three tenths deep in aluminum without altering delicate features on the part's surface. Scotch-Brite can help with another tenth or two, but I typically use 800 grit paper for defects on the order of a thousandth. Anything over that is first roughed out using fine rifling files.

My CAM software begins by sewing all the fillets, contours, and flats together into a single surface for use by certain of its machining operations. In most cases, though, it's more practical to work on individual or small groups of individual features rather than continuously machine and re-machine the entire part. This does create the potential for leaving appearance-spoiling artifacts behind in the boundaries between operations, though.

For a nicely blended result, the CAM software needs to know the precise lengths and diameters of all the cutting tools. Relative errors on the order of tenths among them will be visible in the boundaries between operations. I try to design my parts to require a minimum number of cutting tools and, if at all possible, a single ball cutter for handling all the finishing operations. The detail I designed into the top surface of the pump enclosure, however, required three different size ball cutters in order to efficiently remove the material left behind by the roughing tool. This greatly reduced any chances of me meeting my goal on this part.

In order to complete the machining in a reasonable amount of time, the CAM software also needs to be told how much deviation from what it believes to be perfection is allowed. Errors can result from a slightly different interpretations of this deviation among different contours with different shapes and slopes even when they're adjacent to one another. Spreading lengthy operations over multiple machining sessions that require re-referencing the workpiece also adds errors.

Machining of the pump enclosure began by drilling a hole through the workpiece for the filler cap. Its exact location was immediately indicated and recorded for a later sanity check. This hole will also be used as a reference when the part is flipped over for its bottom side machining. A two hour roughing pass with a 3/8" 4-flute cutter was immediately followed by three hours of finishing passes using 3/8", 1/4", and 1/8" diameter ball cutters. The lengths and diameters of these cutters were carefully measured, including the effects of runout, before they were supplied to the CAM software. The location of the reference hole was remeasured after the five hour machining marathon and found to have shifted only .0003" in the y direction and nothing in the x-direction - not bad for stepper motors.

The photo shows a well-blended final result except for some very disappointing horizontal gouges on the front surface that are several thousandths deep. These are often created by long sharp-edged roughing tools when machine and/or tool deflection and chatter combine with an unfortunate tool entry. The problem can easily and should have been avoided by leaving excess stock during the roughing pass for the finishing tool to remove. I always set up operations exactly that way, but for some inexplicable reason I did not this time. Bead-blasting removed all the other finishing defects, but filing was required to clean up the gouges.

The exterior of the actual fuel tank was machined next. Its surface is considerably simpler and was designed to be finished with a single ball cutter. The same long roughing cutter was used, but this time .010" excess stock was left behind for finishing that resulted in no gouging. The roughing tool was also used to finish the part's flat surfaces while the ball cutter finished the vertical surfaces including a huge fillet between the two. Tool measurement errors are apparent in the boundary between the two operations under the part's mounting flange. Since the error is on the underside of the part, it was left as is.

The internal surfaces of both parts were finally machined, and the pump and motor were trial-fitted. During the design of the tank assembly, I was looking forward to trying my hand at home anodizing the parts. With them finally in my hands, though, I felt they were too complex for me to properly polish for an anodized finish that I'd be happy with. I decided instead to paint the outer surfaces with Gun-Kote. Since this paint requires an oven cure, the electrical contacts must be added afterward.

A few changes for safety were added to the design shown in my previous post. The first was an array of vent holes drilled into the rear of the pump enclosure near the motor's brushes. The spacer plate will also be notched to create a weep hole in the pump enclosure in order to drain fuel in the event of a pump leak. The final steps will be to machine a filler cap and separator plate with pickup tube sub-assembly, install the electrical components, and then finally test the whole assembly. - Terry

 

[petertha]

That turned out really well.
- I've never played with those bake on sprays like Gun-Kote. Do you have a sense if parts can be touched up or entirely re-sprayed with the same system or is it kind of a 1-shot deal?
- those tabs or finger holding the motor is neat. So does that entirely support the motor kind of like a loose press in fit, or maybe will be sleeved with rubber or something?

 

[mayhugh1]

That turned out really well.
- I've never played with those bake on sprays like Gun-Kote. Do you have a sense if parts can be touched up or entirely re-sprayed with the same system or is it kind of a 1-shot deal?
- those tabs or finger holding the motor is neat. So does that entirely support the motor kind of like a loose press in fit, or maybe will be sleeved with rubber or something?

It can be touched up after it's cured and then re-baked. It's pretty durable. I made some motorcycle parts some dozen years ago and painted them with that paint, and it held up in the weather during the 80k cross country miles we put on that bike.

Those tabs were designed .010" oversize for the pump and motor that I had available when I designed them. I then file them for a snug fit as needed for the particular motor/pump combination that I end up with. I do typically use a rubber pad between the pump and the enclosure's cover plate which in this case will be the spacer plate. - Terry