Hi Bob and All -
I have some long-winded input for you. That pesky rotating magnetic field flying around the engine as the magneto turns affected me the same way as it did you and a few other writers who want to place electronic sensors on their motorized bikes. I had a SenDEC Tach that worked fine on my HT engine, but became useless on my Honda 4-stroke in-frame engine.
The answer to my problem was a six-foot length of RG-58 coax cable from Radio Shack for $8. 12-foot cables are available for less than $13.
Without getting too boring with details, I built for myself a light-pulse-sensing circuit and fed the output pulses of that into an analog designed-for-automobiles tachometer (a relatively inexpensive one). My first build had the needle swinging wildly all over the tachometer face. (The circuit and tachometer run off a 12 volt sealed lead acid battery. Both have a positive lead, a negative ground lead, and a third wire, the pulse signal wire, out of the sensor circuit going into the tachometer.)
I think the rotating magnetic field from the rotating magneto was playing havoc with creating unwanted signals going into the tach; the magnetos are not shielded. To minimize the unwanted signals on the sensor wire, I bought a six-foot length of RG-58 cable. I soldered ground wires to both ends of the braided sheath and soldered the sensor wires to the center lead. The sensor signal now cannot be affected by the rotating magnetic field. I did not bother shielding the positive power wire; I thought it was unnecessary for reasons I can explain in detail if someone is interested.
The radio frequency characteristics of the coax cable seems to have cleaned up digital pulses out of my sensor as well as keep interference off of the sensor wire. I kept unshielded sensor wire as short as I could. Six feet of coax may be unnecessarily long. When I soldered the components together and saw a huge initial improvement, I just kept it and tie-wrapped the excess cable to the bike frame.
I wonder if you used a length of coax cable center for your tach signal around the spark plug wire and grounded both ends of the braided sheath to your engine, that would be useful. You would probably have to solder a short stub of tach cable on to the other end of the coax center. If you try that, let us know what the results are. (An afterthought: grounding only at the engine may be the only one needed; try both options and let us know.)
Today, I took the bike and analog tach circuit to the road. It works almost perfectly. The needle still swings back and forth across a 200 rpm range, but considering the tach mechanics were NOT made for the severely vibrating handlebars of an engined bicycle, I can overlook that. I limit my engine speed to 4000 rpm indicated.
That mechanical tach and the sensing circuit both take a severe beating when I ride. The handlebars are anything but rock solid steady. On the way back today, almost home, the tach needle stopped working 60 miles into the ride (I rode 65 miles today). Analysis showed that a solder joint wire broke loose from the solder. Physical vibration broke the copper ground wire from the sensor circuit. I disassembled and re-soldered the ground back in place. Vibration will be the death of the sensor circuit; I expect this to happen again.
The tachometer needle now also reads 300 rpm higher than when brand-new. I used florescent lights to calibrate the tach at 3600 rpm. The needle reads 3900 rpm. I can mentally compensate for the higher reading. This tach circuit is an inexpensive answer for my "I want one" mentality. A dead-on accurate analog tachometer circuit is out of my financial realm. If I needed one, somebody else would have to pay for it.
Attached should be some photos of the sensor and the tach mounting. As the crankshaft turns, pulses are read by the photocell (LED and photocell and all other parts are also from Radio Shack.) It was a bit tricky getting circuit resistance values just right. Physically mounting the circuit took me a few days to work that out. The circuit container is a camera film canister. (When was the last time you saw one of those?) A slot in the lid keeps out some unwanted light; too much solid-intensity light affects the circuit output. The tachometer mount is a piece of wall paneling that I covered with electrical tape. This was quick and easy. An expensive mount would not have been any better.
The first photo shows the circuit canister mounted next to the MaxTorque clutch.
Te second photo show the broken ground wire I had to repair.
The third photo shows the circuit I had to assemble on the very small circuit board.
The fourth photo shows the size of the circuit next to a common house key.
The last photo shows the analog tachometer mounted in place. Also seen are the solid-state speedometer computer, two headlights and the marine horn.