Who needs a diesel?

Recently I’ve been overhauling my own QL, which is what I called a BMW boxer engined special that I made around 1983/84 (if you’re interested to see more have a look at motochassis.com in the articles and gallery sections ).  In fact I made about half a dozen others for paying customers as well.  On the chassis front, even with the passing of nearly two decades there is little that I want to change except for modern tyres and suspension units.

Back then BMW only made the Airhead boxer and I thought that it might be nice to upgrade to the more modern Oilhead version.  However, the whole concept and chassis design of the QL was based entirely around the older model and it would have meant starting from scratch instead of an overhaul.  I had a pair of Krauser four valve heads with re-angled inlet ports which went quite well so I decided to stick with the old engine and just blue-print it properly with minimal changes.  A pair of titanium conrods would be nice but that’s just dreaming.  So I’m left with fitting some big Dellorto carbs (maybe EFI at a later date), a nice cam and a better ignition system.  Which begs the question “just what constitutes a better ignition system?”

There are a number of electronic ignition systems on the market even for the old airheads but I chose to build my own for a couple of reasons. 

1. It was an opportunity to play with an OOPic, which is an easy to programme micro-controller based on a PIC (Programmable Integrated Circuit), using OOPs (Object Orientated Programming) techniques.

2.  I wanted to incorporate some features that don’t all usually come in the same box.

For example; rev. limiter, traction control, alternator cutoff for low speed acceleration, drivers for a LED bar graph tacho, drivers for a numeric LEDS speedo, some data acquisition so that acceleration data can be downloaded into a PC for use in a software or virtual dyno, and close to constant energy sparks throughout the rev range.

Unless you ride a diesel powered bike, and there aren’t too many of those outside of third world countries, you’ll have an electric ignition system of some description.  Now, whilst ignitions have improved greatly from the early points and coil systems, to the computer controlled systems of today, with fully mapped timing and anti-knock sensors etc., the actual method of spark production has not changed at all.  (CDI or Capacitor Discharge Ignition is an exception to this.)  All that the modern electronics do is control the switching and timing of the current flow through a traditional coil (a high loss transformer really).

Spark production

To fire a spark plug we need a high voltage, upwards of about 20,000 volts, to jump the gap and sufficient energy behind it to start the combustion process.  Once the spark has been started it needs a much lower voltage to maintain itself for a short period. 

There are various ways to produce a short duration high voltage pulse, the piezo-electric effect is one, as used in gas lighters etc. this turns mechanical force into electricity, but the standard for engines has long been electro-magnetic.  This relies on the fact that if you place a conductor (a piece of wire) in a CHANGING magnetic field then a voltage will be produced in that wire, this is additive, and so the longer the piece of wire in the field, the higher will be the induced voltage.  The way to put a long length of wire into a magnetic field is to wind it into a coil and then concentrate the magnetic field through the coil.  So an ignition coil has a long length of thin wire wound around a core of magnetic material.  This is called the secondary coil.  Now we need to create the changing magnetic field.  The greater the time rate of change of the field the greater will be the induced voltage.  So we obviously need to change this field quickly.  In practice it is easier to get a quick change when the magnetic field collapses rather than when it’s building up.  This fortunately fits in quite well with the requirements of the engine, we want a short duration spark but we have a relatively long period of time (one or two engine revolutions) in which to recharge the magnetic field ready for the next spark.  For those of you not at home with electrical concepts imagine slowly filling a bucket with water from a tap and then quickly emptying it by up-ending it.

In essence there is another coil wound around the same core, but of thicker wire and probably with about 100 times less turns than the high voltage coil.  This coil is called the primary.  Through a set of points (contact breaker) or electronic switch, battery current is passed through this coil during most of the engine cycle and this builds up a magnetic field in the core.  When the engine needs a spark the points are opened, the battery current ceases and the magnetic field collapses quickly thus producing the high voltage in the secondary winding to fire the spark plug.  A voltage of maybe a few hundred is also induced into the primary winding and that is why we use a condenser (also known as capacitor) to bypass the points in a non-electronic system.  Without the condenser we get arcing at the points and poor ignition performance as well.

The strength of the magnetic field in the core is proportional to the instantaneous current in the primary winding.  The maximum current that can flow is determined by battery voltage and what’s known as the resistance of the primary coil.  This maximum current puts a limit on the strength of the magnetic field that can be created in a given coil.  However, when the points or the electronic switch close it takes some time for this current and magnetic field to build up to this limit, it is not instantaneous.  This time depends on the winding resistance and what’s called the coil inductance and for engine applications may typically be between 5 to 20 milli-seconds.  At 12,000 rpm. the time for one revolution is 5 milli-secs. so it doesn’t take Einstein to figure out that at high revs. ignition performance is likely to fall off, simply because there is not enough time to recharge the magnetic field fully.  However, when idling at say 1000 rpm. we have 60 millisecs. available, but if we apply the coil current for most of this time then we will be passing peak current for a longer time than necessary.  This draws more average current from the battery which gets replenished by the charging system by taking a little more power from the engine, but more importantly it heats the coils and electronic switches more.  This increases the resistance in the circuit which reduces the coil current and thus ignition performance as well.

As a compromise between coil heating at low revs., spark performance at high revs and the mechanical accelerations involved in opening and closing points it became usual not to close the points again immediately after the last spark, there was an idle period with no current through the coils followed by what’s known as the dwell period when the points close and current flows.  The dwell is usually expressed in terms of degrees of crankshaft angle.  So for a given dwell angle the points will be closed for longer at low revs. and shorter at high revs.

A brief history of the primary coil current.

 ‘A’ indicates when the points or electronic switch open and halt the current in the coil, leading to the collapse of the magnetic field and induction of the spark voltage in the secondary coil.  The points close at ‘B’ and the current builds up with time until it peaks out at ‘C’ charging the magnetic field in the core.  At low revs. as shown here, a steady current flows from ‘C’ until the next ‘A’ when the next spark occurs.  The period from ‘C’ to ‘A’ is not necessary and is wasteful of  battery energy and leads to unnecessary coil heating.  Efficient production of sparks occurs when ‘C’ and ‘A’ coincide.

Let’s put things into practice

The airhead BMs had various ignition systems depending on their year of manufacture.  Initially a simple points system with mechanical rpm. advance was used, after that they went to what’s been called the canister system, firstly this contained points but these only controlled an electronic switching module and so the current through the points was greatly reduced and switching efficiency improved.  Later the points were discarded in favour of a hall effect electronic sensor to determine the firing point and the same or similar switching module to look after the current flow in the coil.  My engine had this latter system.

You’ll probably have realized by now that I’m a person that doesn’t like guessing much, if I can I like to calculate the outcome but if I can’t then I want to test and measure to remove the guesswork.  So if I was to build a better ignition system I needed to look carefully at what I had to start with. 

The BMW fires both cylinders together, one spark is a wasted one each rev. this is accomplished by joining two 6 volt coils in series.  So the system needs to make one spark per rev.  With cold coils the maximum current measured was 4.5 amps, and it took approximately 15 milli-seconds to reach this value.  The idle period was 150 degrees and so the dwell was 210 degrees.  At 1000 rpm. the dwell time was then 60 * 210/360 = 35 milli-secs.  and at 8000 rpm. (the maximum that any non-super-tuned-racer airhead need spin) the dwell is only 4.375 milli-secs.  obviously this is way short of the time, 15 milli-secs, needed to fully charge the coil and so ignition performance at high revs was obviously suffering, along with excessive dwell time at low speeds.

So what could I do about it?  Well I aimed for what I referred to above as a constant energy system.  What I mean by that is; that instead of the dwell being a function of crank angle I wanted to make it a fixed time period throughout the whole rpm range.  That way, at low revs. the average current would be reduced along with coil heating and if the period was long enough I’d have improved performance at high revs.  This is easy enough to do when the ignition is controlled by micro-computer.  Basically you delay the start of passing current through the coil for the difference between the time for one engine rev. and your desired dwell period.  Actually it’s a little more complicated because in low gears a motorcycle can accelerate the engine so quickly that there is significant rpm change from one rev. to the next, but that just means that the computer has to calculate engine acceleration as well.

Even though it took approx. 15 milli-secs. to nearly reach coil saturation current (98%) I decided to go for a dwell or charge time of 10 millisecs. because with that charging period the peak coil current reached was close to the maximum anyway (4.2 amps against the maximum of 4.5 amps or 93%), and was equal to the standard BMW system at 3,500 rpm.  However, at 8,000 rpm a complete revolution only takes 7.5 milli-secs. and so it’s obvious that I had to allow for some ignition performance drop off at high revs.  After firing a sparkplug, things take a little while to settle down again before it’s wise to restart the flow of primary current.  If you restart too soon you’ll cut the spark duration for example.  To be on the safe side I decided to wait at least 1 milli-sec. before reapplying current, therefore at 8,000 rpm I was left with 6.5 milli-secs. in which to rebuild the magnetic field.  However, that compares with the 4.375 milli-secs. of the standard system, that’s 50% longer.  In terms of performance that 50% longer charge time resulted in a 20% higher peak current, and so it seemed a worthwhile improvement.

Basically the new system has slightly inferior performance to the BMW system up to 3,500 rpm. but it maintains constant performance from idling up to 5,455 rpm and then gradually tails off.  Measured spark duration times, not shown here, reduced markedly from about 6,500 rpm with the standard system but remained unchanged up to 9,000 rpm with the new system.

The graphs show three parameters for comparison, viz; Dwell time, average coil current and peak coil current.  The peak coil current is a good comparison to indicate actual ignition performance, both from the point of view of maximum available spark voltage and of spark energy available.  At 8,000 rpm. the new system has a value of 3.7 amps with the original system down to 3.1 amps, a worthwhile difference.  This improvement can be better put into perspective by noting that the new system produces the same strength spark at 8,000rpm that the original did at about 5,500 rpm.  In general we can see how the standard system suffers as the revs. rise due to a fixed dwell angle rather than a fixed dwell time.

The average current is a guide to coil heating and below 3,500 rpm the current drawn by the new system is lower in keeping with the need for less sparks at low rpm., whereas the original works against logic and draws much more as the need for sparks decreases.  Above 3,500 rpm the new system drinks more juice in keeping with the improved ignition performance.  There few free lunches.

Dwell times of the two systems.  This shows the unnecessarily long dwell time of the standard system below 3,500 rpm. due to the constant dwell angle.  The new system shows how the desired dwell time of 10 milli-secs is maintained up to 5,455 rpm.

Average current.  The new system in general draws more current as the demand for sparks increases, the fall off after around 6,000 rpm reflects the fall off in ignition performance.  The high current at low revs. with the standard system is a result of the excessive dwell time.

Peak current.  This is a good indicator of ignition performance.  Note how the new system maintains level performance up to 5,455 rpm before starting to drop off, although at low revs. the current is a bit less than the standard system because of the choice of 10 milli-secs. for the dwell period.  The standard system has lower performance above 3,500 rpm and drops quickly.

The photos are of oscilloscope (a device that lets you ‘see’ electricity) traces of the instantaneous current through the coils, at three different rpm., the BMW system is on the left.  The captions explain what’s happening.  Note that the time scale has been adjusted to show just a bit more than one rev. at each of the three rpm values.

1,000 rpm.  The dwell time of the BMW system is 35 msecs and the excessive saturation time is easily seen.  This results in unnecessary coil heating.  On the right the new system charges the coil for only 10 msecs and the peak current and hence the magnetic field is only slightly less (93%).

3,000 rpm.  The two systems have equal dwell times at 3,500 rpm and so these two traces are very similar.

8,000 rpm.  Here we can see that the peak current for both systems is reduced from the levels at lower revs. but the 210 deg. dwell of the standard system, 4.375 msecs, gives a significantly lower value than the new system with 6.5 msecs.  charge time.

I’ve talked about ignition performance in this article, and referenced it to how near to the maximum we can charge the coil in the time available between sparks.  This is a good indicator of the maximum available spark voltage that the system can generate and also the amount of energy available to be discharged into the spark.  We’ve seen how with the same coils this performance can be improved considerably, 20% at 8,000 rpm. and the spark strength at 8,000 rpm is as good as it was at about 5,500 rpm with the original system.  However, we mustn’t get too excited.  Ignition performance doesn’t translate directly into engine performance, I’m not expecting a 20% increase in engine power.  A spark is needed to start the process of combustion, if a given spark starts that process correctly then a “better” spark will make no difference.  So why bother?  Well, most mass produced articles are built within severe price limits and that often means in this context that some standard ignition systems are border-line, there is often room for improvement.  It is under adverse conditions that the deficiencies show up more, starting in freezing conditions, misfiring with dirty plugs or mixture mal-adjustment.  Fuel delivery systems, particularly some older carburettors, often make a right mess of fuel metering under conditions of rapid throttle movement and acceleration.  A strong spark is more tolerant of fuel mixture errors.  Higher compression ratios and other tuning features also place higher demands on the spark’s department.

By modern multi-cylinder standards the BMW boxer is a slow revving motor and we’ve seen that the standard system is struggling at maximum rpm., so how do 16,000 rpm motors manage?  There are a variety of answers to that question.  The BMW fires each cylinder together and so needs twice the rate of sparks compared to a system which only fires each plug when needed, i.e. once every two revs.  In addition the BMW uses two six volt coils, but 12 volt coils generally perform better.

The detail design and materials used in a coil can affect it’s performance and high revving engines are often fitted with higher performance coils.  To this end a growing trend will be modular coils and switching modules fitted directly to the spark plugs, eliminating the plug cable and other interconnections, this is to generally cut down on losses and inefficiency inherent in a more spread out system.

Sometimes the highest spark rate demands come from quite everyday sources, the traditional American V8 had just one coil and an 8 way distributor.  That means that it required four sparks per engine revolution, that’s four times the BMW requirement at the same rpm.


By achieving a longer dwell period at high revs. I’ve managed to improve ignition performance quite considerably, extending the rpm range for the production of strong sparks.  By limiting the dwell time to 10 milli-secs. the power requirements have been considerably reduced at low revs., reducing also the coil heating.

Could I have done better and should I try harder?  Well I wasn’t able to attain my aim of constant dwell time (with a near fully charged coil) throughout the full rpm range but this was due to the characteristics of the coils being used.  Now if I can find some higher performance coils that reach current saturation in 6.5 msecs. or less then I can make a system that will give equal performance throughout the whole rev. range.  I actually have some coils that charge quicker but the stored magnetic energy is much lower and so overall performance is less.  It’s easy to achieve the maximum when the target is lower.   Anyway, even though I’m doubtful that further ignition improvement will be reflected in better engine performance I’m going to look for some “better” coils to try, but you can be sure of one thing, there’ll be no guesswork in the final selection.  I’ll study their spec. sheets, do some calculations and then test them. 

I’ll keep you informed as to how I get on.  When the bike’s back on the road (the tests above were all done on the bench) I’ll use the virtual dyno. that I mentioned in the introduction to compare the actual road performance between the different systems.  In fact software or virtual dynos. can be quite a useful tuning tool for those without a real dyno. and in the future perhaps we can look into how they work and what’s reasonable to expect from them.

2022 Note. I now use an ESP32 micro for this job with programmable timing.