The Hall Effect pickups are located on the front of the speed reduction unit. Shown is the forthcoming mounting for the Kitfox.

Camputer Dual Electronic Development For The CAM 100 Auto Engine Conversions and By PETER BROOKE have about ignitions prompted Canadian Ignition Systems Canadian Airmotive, Inc. Airmotive to develop a dual electronic 7400 Wilson Ave. ignition system for the CAM 100. The It is easy to understand why there is Delta, B.C., V4G 1E5 CAM 100 is the leading 100 hp auto en- a lot of interest in auto engine conver- _____ Canada ______gine conversion and an excellent sions as replacements for traditional replacement for the Continental 0-200. aircraft engines. like something is missing. You don't Many pilots are very happy with the have that key on the instrument panel Designing the Specifications significant advantages offered by con- to check on run up or the security of a versions. The relatively low cost of dual system. It is true that modern au- We found that the development of parts, the benefits of water cooling and tomotive electronic ignition systems, if the CAM 100 ignition system was not the reduced fuel consumption are ap- carefully selected, are very reliable. It as easy as anticipated. An old fash- pealing improvements. In most areas is also true that plug fouling in auto en- ioned points ignition with one coil and of North America an easy to install gines is virtually a thing of the past. a is simple to comprehend. cabin heat system will greatly extend However, it is hard to argue that a dual A modern dual ignition system with the flying season. The absence of shock system would not be better. one coil per , electronic ad- cooling is a definite advantage. Signifi- Some auto engines are factory vance, engine rpm limiting and cantly improved fuel consumption and equipped with dual ignition. These ig- optimized dwell time proved to be a being able to use unleaded gasoline are nitions are generally engineered for very significant project. big pluses. The list of advantages goes economy or pollution control, not pri- The first job was to separate myth on and on. marily for reliability. They often use a from reality and establish the system A feature of auto engines that con- distributor which eliminates a lot of the specifications. Starting with the spark cerns some pilots is the ignition system. dual effect. Also, they are not generally plugs, did they have to be dual? This If you are used to having traditional found on engines that would make the could be retrofitted but would be dual magneto ignitions then change to best conversions. costly, partly negating one advantage an engine with one system, it seems Hearing the concerns some pilots of the conversion. Our survey of auto SPORT AVIATION 55 RANGE OF ACCEPTABLE TIMING FROM DYNO TESTING

40 '" ^JL CAM 100 'ELECTRONIC IGNITION / >\ ADVANCE CURVE s^~ -\_ 30 // d / 1 III y/ O | 20 // Q < // g 10 /y/ 0 0 1000 2000 3000 4000 5000 6000 ENGINE RPM (WITH PROP LOAD)

FIG. 1 CAM 100 ELECTRONIC IGNITION ADVANCE CURVE repair records and mechanics indicated showed the engine would still produce tage we would share with magnetos. In that plug failures were almost nonexis- 65% power. all other respects it appeared the CAM tent. This is apparently due to cleaner Further investigation revealed we 100 system would be an improvement. fuel, better plugs and the strong spark could improve reliability by providing Because the system utilizes an indepen- provided by modern electronic igni- one coil per and making all dent coil for each spark plug, all tions. We became convinced that dual pickups and electronics dual and sepa- switching for the coils is low voltage plugs would not add significantly to re- rate. This would even allow running (12V) and is done electronically. liability. Proper maintenance was all one system on a small backup battery, There is no distributor to switch the that was required. In the unlikely preventing an aircraft electrical failure high voltage to the plugs, which means event of a plug failure, dyno tests from stopping the engine. This advan- no potential failures from cracked dis- tributor caps or eroded contacts. Compared to magnetos, the CAM 100 system would be lighter and less ex- pensive, have a proper ignition advance and a hotter spark. Theoreti- cally it should also be more reliable. The next specification to establish was the operating temperature range. We decided that the ignition system would have to operate flawlessly from -40 degrees C. (-40 degrees F.) to 100 degrees C. (212 degrees F.). We have a customer in Alaska who uses his Full Lotus floats equipped plane from the snow in wintertime to tend his trap line, which goes to prove that some people do actually fly at -40 degrees. 100 degrees could be attained in the engine compartment where the sen- sors and coils are located. The ignition control computer would be installed inside the cockpit to reduce the tem- perature extremes it is subjected to. The heart of the CAMputer 100 ignition system is the state-of-the-art dual digital The system should work properly at computer controller featuring fully programmed advance, anti-kickback control and low voltage. We established that 8 rev limiting. volts might be all that was available

56 DECEMBER 1993 3.5 mS DWELL (COIL "ON" TIME)

COIL TURNS SPARK OCCURS AT T.D.C. SPARK OCCURS AT 34° B.T.D.C. ON AT 21 "B.T.D.C.

3.5 mS DWELL (COIL "ON" TIME)

COIL TURNS ON AT 139° B.T.D.C. B.D.C. B.D.C.

SPARK TIMING @ 1000 R.P.M. SPARK TIMING @ 5000 R.P.M.

FIG. 2. THE EFFECT OF DWELL AS ENGINE SPEED INCREASES with a very low battery that was just monitored power output and fuel con- were many potential solutions so we turning the engine over. At the upper sumption at various rpm while started with the simplest and most limit 20 volts could occur as a result of changing the . We also cost effective possibility. a regulator failure. Outboard motors did comparisons with different quali- without regulators can raise the sys- ties of fuel. Acceptable timing was Experimenting To Find tem voltage to 18 volts with a fully defined as the range within which rpm The Best System charged battery and no accessories was maintained within 50 rpm of the turned on. If the battery was acciden- maximum for that particular The first approach was to design a tally disconnected, the system voltage opening. We also established that the system utilizing existing automotive could go even higher. We would in- preferred fuel was premium unleaded, reluctors, pickup coils, and electronic corporate protection to prevent although lower quality fuel was ac- modules as found on some other con- destruction of the ignition in case ex- ceptable in an emergency. versions. We discovered that pickup treme over voltage ever occurred. A These tests generated a graph indi- coils from Dodge or Ford V-8s worked proper advance curve was mandatory cating a band of acceptable timing. well when combined with the compact to optimize power and efficiency. No The band we used was based on regu- electronic modules from GM distribu- advance at very low speeds and a lar unleaded fuel (see Figure 1). tors. All these components are well strong spark would aid hand propping. Premium unleaded fuel produced a proven, inexpensive and readily avail- Rev. limiting would prevent engine wider band of acceptable timing. The able. We fabricated our own test damage. The ignition system would stock timing was very advanced at reluctors as the stock units were too have to function properly at very low 5000 rpm and then started to roll off small in diameter to accommodate the rpm, as might occur with hand prop- and retard at higher rpm. We did not four pickup coils we needed. (The re- ping. In the event the engine stopped find this desirable as the timing ap- luctor is a metal rotor on the and the operator forgot to turn off the proached dangerously close to the distributor shaft that spins by the ignition switch, the system would limit of detonation. We selected a tim- pickup coils. As a ridge on the reluc- have to turn itself off. This feature ing curve that ramped up from O tor passes a pickup coil, it induces a would prevent unnecessary battery degrees at 1000 rpm to 34 degrees current in the coil.) drain and possible coil damage due to BTDC at 4000 rpm. Above 4000 rpm The spark advance system would be overheating. advance remained at 34 degrees BTDC. a mechanically rotated plate. The The advance curve was established This curve was within the limits estab- pickup coils would be mounted on the using a variable timing system. A lished by testing. plate. The plate rotates as the throttle mechanism was built that allowed the After establishing the performance is opened. This type of advance is ignition timing to be adjusted while a specifications we searched for an off- used extensively on outboard motors. CAM 100 test engine was running on a the-shelf product. There was nothing The whole combination did work but test stand. To approximate the load available that was ideal. All options had limitations. while flying, the engine was fitted were very expensive. A key objective The reluctor type of trigger puts out with a propeller selected to allow max- with the CAM 100 is affordability a signal somewhat similar to a sine imum permissible rpm at full throttle. which depends on all components be- wave. The modules turn the ignition The test stand was configured as a dy- ing very competitively priced. coils on as the wave crosses up through namometer to allow horsepower We decided to design and manufac- zero. They turn off, and the coil measurements. During test runs we ture our own ignition system. There sparks, as the signal next crosses zero,

SPORT AVIATION 57 speed reduction housing. The two magnets are mounted in the flywheel. One sensor assembly triggers cylinders 1 and 4, the other cylinders 2 and 3. Hall Effects are very reliable and once set up at the correct timing will need no further adjustment. We had already established the op- timum dwell and ignition advance curve so the next problem was discov- ering the best way of translating it into a system that could be pro- grammed. A common approach is to use a lookup table to set the degree of advance based on the engine rpm. We came up with a more elegant solution that may be unique, certainly we could find nothing published like it. We use solid state counters to translate the time it takes for the flywheel to do part of a revolution into the amount of advance and the relative number of degrees of dwell without using a chart or lookup table. The CAMputer ignition system. Because the system utilizes an independent coil for We also incorporated other desir- each spark plug, all switching for the coils is low voltage (12 V.) and is done elec- able features, all part of our system of tronically. using counters. We designed the tim- ing so the spark will remain at TDC (top dead center) up until idle speed on the way down. The second zero low speeds the slope of the signal from when it starts to advance. This en- crossing occurs when the reluctor the reluctor/coil combination was so sures that someone hand propping the ridge and the pickup coil are lined up. flat between the peaks that the module engine will not experience a kickback This automotive system produces had trouble deciding when it was cross- (a welcome feature for anyone experi- undesirable dwell times; dwell being ing zero. This sometimes resulted in an enced in hand propping). We did not the time the coil is turned "on." The extra spark at the wrong time. want any possibility of the spark over spark occurs as the coil is turned "off." After much experimenting with advancing and causing detonation. To As the engine turns faster the time be- various setups, we abandoned this ap- prevent this we located the magnet tween turn "on" and turn "off" (the proach as being inadequate for our that causes the Hall Effect to go low at dwell) decreases, ultimately resulting application. 34 degrees BTDC (before top dead in a weaker spark at high speeds. This center), our maximum advance. When is not a linear relationship with speed, The Computer Solution the spark advances to 34 degrees, at but at very low speeds the dwell is about 4000 rpm, it locks onto this sig- very long - long enough to seriously As this low tech approach wasn't nal and cannot advance any more. overheat our lightweight coils during a successful our next approach was a Detonation caused by over advanced prolonged idle. high tech one using a programmable ignition can cause serious engine dam- We also needed a good spark at hand onboard computer controller. age and the telltale pinging might not propping speeds. A reluctor setup that First, we had to determine the best be audible in an aircraft. We prevent would produce enough signal at very method of triggering the ignition. Our this from occurring by providing a me- low speeds would overload the module research indicated that latching hall chanical cue to the ignition indicating at high speeds and cause double spark- effect switches triggered by flywheel that it has reached full advance. ing. The high signal output from the magnets would be the preferred sys- Another advantage of the counters pickup coils at high rpm could be elec- tem. Hall Effect switches have been occurs if the engine were to stop with tronically controlled but that did not used for years to trigger automotive ig- the ignition still switched on. The solve the coil heating problem. nitions. Latching Hall Effects go low counters run out of capacity after In the stock (four cylinder) automo- as one magnetic pole passes them and about 2 seconds and shut off any coil tive ignition system the reluctor stays low until the opposite pole that may be energized. This prevents produces a signal which causes the coil passes them, when they again go high. coil overheating and damage. With a to produce a high voltage. This process (Low means the input terminal is ef- resistance of only 1.5 ohms the coils repeats four times per revolution or fectively connected to ground. High will absorb about 100 watts and soon every 90 degrees. The high voltage is means it is isolated from ground.) As get very hot. then routed to the correct spark plug we wanted a dual system, each pickup We also set up the dwell time so it by the distributor rotor. In our modi- assembly had to have two completely was longer than normal at cranking fied system each plug would fire once independent Hall Effect switches. speeds to ensure a strong spark even per revolution, an arrange- These are not readily available so we at low battery voltage. At higher ment called a waste spark system. This had to design and manufacture our speeds dwell is optimized around 3.5 results in 180 degrees between signals own. The two Hall Effects switches mS. to give a good spark, reduce coil when the reluctor is driven by the cam, are mounted on a circuit board and heating and save electrical power. meaning that the signal from the potted inside an anodized aluminum At about 6500 rpm the sparks start pickup coil is stretched out twice as far holder. The holders mount in adjust- to cut out to limit the engine to safe as in a normal four cylinder engine. At ment slots in the front of the prop maximum rpm. This works very well.

58 DECEMBER 1993 Running the engine up to 6500 rpm on will start but will not accelerate it may induces a high voltage in the secondary the dyno and then backing off on the be a spark problem. At idle the com- windings. However, this high voltage load (to allow over revving) results in pression pressure is low and a weak is also hard on the coil insulation. only a slightly different exhaust note. spark will jump the plug gap. As soon Limiting the spike to 250 volts allows The engine refuses to rev any higher. as you open the throttle the engine a high enough spark voltage to jump We initially built the circuit on a takes a gulp of air, the compression the plug gap while limiting the magni- breadboard using discreet CMOS logic pressure rises and the weak spark no tude of the voltage the coil insulation chips to prove it would work. When longer jumps the plug gap. must withstand. we were satisfied with the circuit we Using a big heavy duty coil can pro- We experimented with current lim- then translated most of the hardware duce a spark that will jump 1" or more iting to protect the output transistors in the logic system into a program run in free air. Dyno testing the engine from shorted coils but found it pro- by the micro controller. did not show any performance advan- duced too much voltage drop. The tage to a spark this hot. We did find fuses (breakers) mentioned above pro- Some Interesting Discoveries that excessively high spark voltages tect the transistors without any caused problems with cross firing and voltage drop. During the development we discov- radio interference. As the electronics in our computer ered some things we hadn't realized Performance was also not effected controller are completely dual we had before. The time that the coil is on by reducing the plug gap from .040" to have a method of switching the out- (the dwell time) actually works against to .030". The reduced plug gap will put to the ignition coils from one side the advancing timing (see Figure 2). If allow an engine to run with a lower of the board to the other. Each coil the coil is turned on at 21 degrees secondary ignition voltage. Increas- could be driven by both transistors BTDC at 1000 rpm with a dwell time ing the plug gap eventually causes (main and backup) at the same time. of 3.5 mS. the spark will occur at TDC. misfiring. However, if one transistor failed it In other words, it takes the crankshaft could pull down the output from the 3.5 mS. to turn 21 degrees at 1000 Finding the Best Components other and prevent the coil from firing. rpm. At 5000 rpm the engine will Our first solution used a large 4 pole turn five times as far during the dwell Investigating coils, we looked at switch. This worked well but the cable time and the spark will occur 105 de- many shapes and sizes. Most were from the controller to the instrument grees after the coil is turned on. The eliminated because of cost or weight. panel was too bulky. We decided to spark will then occur at 105 - 21 = 84 We did not want to use dual coil units move the switching onto the board. degrees after TDC. The dwell time will (the ones with two spark plug wires), Some research uncovered a small relay effectively retard the spark 84 degrees to avoid the possibility of a failure in rated for the electrical loads we needed between 1000 and 5000 rpm. In order one side affecting the other. We set- that would function at a vibration to advance the spark to 34 degrees tled on a proven lightweight coil that level of 10G. The relays are switched BTDC at 5000 rpm the coil on time is used on jet skis, snowmobiles and by a simple single pole toggle switch will have to advance to 34 degrees. light industrial engines and is avail- on the instrument panel. The Main The coil on time will also have to ad- able from several manufacturers. side of the switch keeps the relay coils vance the 84 degrees that the dwell This coil was bench tested with a energized. If there is a failure in the retards the timing for a total advance simple transistor setup driven by a fre- relay coils or selector switch, the re- of 84 + 34 = 139 degrees BTDC. quency generator. Watching the lays will unlatch and select the Backup The automotive reluctor and pickup pattern on an oscilloscope and moni- side of the board. coil system gets around the above ef- toring current draw and spark length, Both the Main and the Backup sides fect by always having the spark occur we established the optimum dwell, or have their own 5 volt power supplies as the reluctor ridge and pickup coil coil on time. A dwell under 2 mS. did and filters. A surge suppressor is used line up. The advance is achieved by not allow enough time for the coil to control any large spikes that could moving the pickup coil relative to the windings to saturate and the spark get onto the supply lines. The sup- reluctor position. The coil on time is was weak. Over 4 mS. had little effect pressor also serves as reverse set far enough ahead so that even at on spark length but did increase coil connection protection. If the power high rpm enough dwell remains to heating and power consumption. We wires are hooked up backwards the produce a spark. This, combined with also verified that the coils would oper- suppressor will conduct and immedi- the reluctors non-linear feature with ate well at a frequency equivalent to ately blow the fuses. speed (the dwell time does not reduce over 6000 rpm. Both the Main and the Backup sides as fast as the rpm increases), produces To protect the rest of the system of the board have their own micro con- an acceptable dwell time for automo- from being effected by one coil failing, troller. The unit we use, an 18 pin I.C. tive systems. each coil has its own fuse (or breaker, made by Microchip, has extensive uses Another interesting discovery was if you wish). In the unlikely event of in industrial controls and also has the the length of spark (in free air) re- one coil shorting out, it would imme- virtue of low cost. Each micro con- quired to run the engine. A common diately blow the fuse and have no troller has its own quartz crystal that test of an ignition system is to haul effect on the supply to the other coils. regulates the "clock" frequency. It out a spark plug, place it on the block, The output transistors that control runs fast, at 16 MHz, a speed only crank the engine and watch for a current to the coils are a critical fea- fairly recently exceeded by desktop spark. I have done this and assumed ture. We are using MOSFET computers. It uses a simplified set of that seeing what looked like a healthy transistors that switch quickly and 33 instructions; even so a printout of spark meant that the system was O.K. have a built in zener diode that limits the program for this ignition fills six In fact, a spark that will only jump over voltage. When the coils are pages (see Figure 3). During produc- .040" in free air will probably not even turned off and the magnetic field col- tion, programming a new chip is get an engine started. We discovered lapses, a high voltage spike appears at simple. A master chip is placed in a that a minimum spark length for this the transistors. The built in diode lim- programmer socket and the read but- engine is 3/8" (in free air) and 1/2" is its this spike to 250 volts. This spike ton is pressed. The programmer reads desirable. If you have an engine that helps to produce a strong spark as it the program into its memory and veri-

SPORT AVIATION 59 of difficulty in the design process was SYSTEM TICK HAS OCCURRED (30.5 MICROSECONDS) essentially determined by the high 0018 004 CLRWDT ;CLEAR WATCHDOG TIMER performance standards set by Cana- 0019 201 MOVF RTCC.W ;GET NEW RTC VALUE dian Airmotive, that we had to meet. 001A 028 MOVWF CAPTURE ;UPDATE MEMORY It would have been an easier job if we accepted a lower standard, but that DO EDGE DETECTION ON TDC AND BDC INPUTS would have meant compromises in the 001B COO MOVLW MASKINPUT ;GET INPUT DATA POLARITY MASK performance of the ignition. For ex- 001C 185 XORWF PORTA, W ;GET CURRENT PORTA DATE ample, if we had used heavier coils we 001D 029 MOVWF A_SAVE ;SAVE IN MEMORY would not have had to be so specific about the dwell times. A timing curve 001E 18A XORWF A_OLD,W ;TEST FOR CHANGES that ramped up as soon as the engine 001F 02B MOVWF A_CHANGE ;KEEP CHANGES started turning would have been sim- 0020 149 ANDWF A_SAVE,W ;KEEP "HIGH" CHANGES pler, but wouldn't have offered 0021 02C MOVWF A_HIGH ;SAVE IN MEMORY protection against kickback while 0022 249 COMF A_SAVE,W ;GET INVERTED DATA hand propping. If we left off the rev 0023 14B ANDWF A_CHANGE,W ;KEEP "LOW" CHANGES limiting someone might have damaged 0024 02D MOVWF A_LOW ;SAVE IN MEMORY their engine while test running with 0025 209 MOVF A SAVE.W ;GET PORTA DATA AGAIN no propeller. 0026 02A MOVWF A_OLD ;UPDATE MEMORY The entire development project took us twelve months, which was PERFORM ENGINE RUNNING TIMEOUT DETECTION longer than expected. The effort has 0027 20B MOVF A_CHANGE,W ;GET ANY CHANGES proved to be worthwhile. The CAM- puter has proved to be a winner, 0028 743 BTFSS STATUS.Z JEST CHANGES, SKIP IF ZERO especially in terms of reliability. In all 0029 A33 GOTO RUNNING ;ELSE PROCEED AS RUNNING our testing, right from the time the 002A 20E MOVF COUNTO.W ;GET LOW COUNT BYTE first production unit was produced, we 002B 10F IORWF COUNT1.W ;INCLUDE HIGH COUNT BYTE have yet to record a single malfunction 002C 643 BTFSC STATUS.Z ;SKIP IF RESULT IS NON ZERO while flying. This is a prime example 002D AOO GOTO START ;ELSE ENGINE STOPPED, GOTO START of the benefit of state of the art tech- nology when applied to aviation. 4 FIG. 3. PART OF THE MICROCONTROLLER PROGRAM fies it. The master is then replaced in tion box was mounted on the shaker the socket by the new chip and the table and operated for 250 hours while program button pressed. The pro- being shaken drastically. During the grammer writes the program to the endurance test the ignition module op- new chip and verifies that it is correct. erated at an elevated voltage. The It then blows a built in "fuse" in the system passed the test with a faultless chip, making it impossible to copy the performance. program. This eliminates the neces- sity to pot the circuit, making it Production serviceable at the factory if necessary. The CAMputer circuit boards are The Testing Program produced professionally using wave soldering to achieve the most reliable We established a rigorous 50 hour solder joints. The assembled board is testing program which served as the sprayed with a conformal coating to initial verification for the Popular Fly- protect it from moisture and corrosion ing Association in the United damage. The completed board is then Kingdom. The PFA regulates home- mounted in a conductive alodined alu- builts in England and their engineers minum housing. helped with the formulation of the test An electric powered flywheel is program. used to test each ignition throughout As well as 50 hours of flight testing the entire CAM 100 rpm range. Cur- About the Author there were other comprehensive tests rent draw, spark advance, spark required. We did extensive testing on voltage and rpm limiting are all tested Peter Brooke, A.S.T., is a found- the dyno at -40 degrees C. (-40 de- for both sides of the dual system. ing member of the Applied Science grees F.) to 100 degrees C. (212 Each completed CAM 100 engine is Technologists and Technicians of degrees F.) from 8 volts up to 18 volts. test run on the dyno before it is British Columbia (ASTTBC), a mem- Bench testing included extended run- shipped. The complete CAMputer 100 ber of the Society of Automotive ning checked with an oscilloscope and ignition is retested at the same time Engineers (SAE) and manager of the most demanding of all, the "shake with the engine it will be installed on. R&D for Canadian Airmotive. He and bake" endurance test. A shaker was the development team manager table driven by an electric motor was In Hind Sight for the CAM 100 and built. A heat lamp mounted above the the CAMputer dual digital elec- table did the baking part, subjecting Developing the CAMputer 100 igni- tronic ignition. He has twenty years the ignition module to a scorching 100 tion system, in hind sight, proved to experience in the design of mechan- degrees C. (212 degrees F.). The igni- be an extensive undertaking. The level ical and electronic products.

60 DECEMBER 1993