Deciphering the igniter...

Having successfully decoded the PGM-FI Main Relay's DNA, I'm now on the
warpath for the Igniter, in preparation for another update to the FAQ.

Some basics:
1) The igniter simply replaces the old make/break points in a Kettering
system.
2) The role of the vacuum/centrifugal advance, and the distributor cam is
now performed by the ECU. The ECU performs this function by adding or
cutting ground as necessary, just like it does with the Main Relay.
3) The igniter has an activation power feed from the ignition switch. The
coil's power goes through the igniter as well, but can't go anywhere until
the igniter is powered by the ignition switch and grounded by the ECU.
4) When the ECU grounds the white wire, the flip-flop in the igniter
closes, allowing the coil's current to travel through the same white wire
to ground in the ECU, thus charging the coil.

Now: When the igniter gets broken, what exactly goes wrong with it? Does
the transistor fuse in the open position? Does some other component go bad
and prevent current from flowing? Is it cracked solder again? Anybody know?

 

Extrapolating from my experience with power transistors...

Junction type semiconductor devices invariably short when they fail. The
junction develops a hot spot - a tiny area that is hotter than the
surrounding junction and therefore has more current flow. That produces a
nearly instantaneous runaway condition, melting the junction in that area.
If the junction is back-biased, like the collector junction of a transistor
is, the voltage ensures the PN junction is homogenized.

(Still assuming the power switcher is a bipolar transistor) The collector in
power transistors is usually soldered to a heat sink. The base and emitter
leads are fine wires. When the collector shorts, the fault current flows
through emitter lead and often burns it out, leaving the device open
circuited. Some power transistors have heavy emitter leads and will just
stay shorted, even with fairly high currents, but it would not be good to
have the ignitor fail in a shorted condition.

If the switcher is actually a power CMOS device, I don't know enough about
them to say.

Mike

 

Extrapolating from my experience with power transistors...

Junction type semiconductor devices invariably short when they fail. The
junction develops a hot spot - a tiny area that is hotter than the
surrounding junction and therefore has more current flow. That produces a
nearly instantaneous runaway condition, melting the junction in that area.
If the junction is back-biased, like the collector junction of a transistor
is, the voltage ensures the PN junction is homogenized.

(Still assuming the power switcher is a bipolar transistor) The collector in
power transistors is usually soldered to a heat sink. The base and emitter
leads are fine wires. When the collector shorts, the fault current flows
through emitter lead and often burns it out, leaving the device open
circuited. Some power transistors have heavy emitter leads and will just
stay shorted, even with fairly high currents, but it would not be good to
have the ignitor fail in a shorted condition.

If the switcher is actually a power CMOS device, I don't know enough about
them to say.

Mike

 

1. One could view the old contact points as a combined crank angle
sensor and coil actuator. Some early electronic ignition systems
actually still had contact points (and vacuum and centrifugal advance),
but the points were only used as an input to the electronics. The actual
switching of the coil was done by the electronics. The electronic box
would simply ground the coil whenever the contact points were closed,
and open-circuit the coil whenever the contact points were open. The
only benefit of this vs. the non-electronic system was that the points
saw very little current and virtually no inductive loading, thus there
was no arching and the points would not wear.

The ignitor (ICM) does exactly what the electronic box did in the
example above, it simply grounds and open-circuits the coil at the
falling and rising edge of the signal from the ECU. Theoretically the
ignitor could be mounted inside the ECU, but then you would have a very
nasty signal (high voltage spikes etc.) running all over the place
rather than being nicely tucked inside the distributor.

2. Yes. There are 3 sensors in the distributor, the CYP, or Cylinder
Position Sensor, gives one pulse for every complete rotation of the
distributor rotor (presumably around TDC prior to the power stroke on
cylinder #1, but that is a guess on my part), the TDC, or Top Dead
Center, sensor gives 4 pulses for each complete rotation of the
distributor rotor, one pulse each time a piston is at TDC at the
beginning of its power strike. Finally there is the CKP, or crank
position sensor. This sensor gives a large number of pulses for each
rotation of the distributor rotor.

The CKP gives the crank angle with very fine resolution, but it is a
relative measurement only. I.e. by looking at the CKP output you can
tell that the engine now has rotated e.g. 76.3° since you started
counting, but you can not tell what specific position it is at. Then
what you (well, the E-C-you, that is) do is you reset your counter every
time you see the TDC signal. Now you know the exact number of degrees
you are past the latest power-stroke TDC, and since you know the rotor
goes 90° between each power-stroke TDC, you can easily do the
subtraction and find out how many degrees /before/ TDC you are. The ECU
calculates the desired ignition timing based on throttle position, RPM,
manifold pressure etc., and then simply turn on or off the signal going
to the ICM at the exact right point in the rotation as read by the CKP
and the TDC signals.

The ECU really does not care about /what/ cylinder gets the spark, the
finger in the distributor does that in the conventional way. The ECU
/does/ need to know what injector to fire, however, and that is what the
CYP sensor is for.

The above is relevant for my '94 Civic. Newer systems have gotten rid of
the distributor finger as well, and then you need the CYP sensor to get
the right ignition sequence.

I don't know about Honda, but some other newer cars have only one
sensor, essentially the CKP. The absolute position information is
indicated by missing pulses on the CKP signal. Imagine an ABS wheel
sensor where you file away a couple of teeth in the right locations.
E.g. you remove 4 teeth, each 90° from the next removed tooth. Then you
remove a 5th tooth at some other location. now you can deduce the CKP,
CYP and TDC information all from one sensor.

Incidentally, the CKP signal can also be used to detect misfires.
During the power stroke, the crank will accelerate a bit, then retard a
bit around TDC. This can be measured as a slight variation in the
frequency of the CKP signal. If you see that there is no acceleration at
the time when you know there is a power stroke, you conclude there was a
misfire. This type of detection is required in OBDII equipped cars.

3. It is my understanding that the output transistor fails in the
ignitors. I can not find any reliable reference on this, however. I
don't think the main relays fail because of poor soldering from the
factory. I believe the solder fatigues over the years, since solder is
the only /mechanical/ fastening for a fairly heavy relay. Solder, if
subjected to stresses and strains (like shaking a relay for mile after
mile), will fatigue. If you have a stranded copper wire, tin the leads
and clamp it down really good in a screw terminal, you can come back a
few years later and see that the solder has been reduced to dust and
that the wire now is loose. I do not believe the ignitor has any heavy
components in it.

 

1. One could view the old contact points as a combined crank angle
sensor and coil actuator. Some early electronic ignition systems
actually still had contact points (and vacuum and centrifugal advance),
but the points were only used as an input to the electronics. The actual
switching of the coil was done by the electronics. The electronic box
would simply ground the coil whenever the contact points were closed,
and open-circuit the coil whenever the contact points were open. The
only benefit of this vs. the non-electronic system was that the points
saw very little current and virtually no inductive loading, thus there
was no arching and the points would not wear.

The ignitor (ICM) does exactly what the electronic box did in the
example above, it simply grounds and open-circuits the coil at the
falling and rising edge of the signal from the ECU. Theoretically the
ignitor could be mounted inside the ECU, but then you would have a very
nasty signal (high voltage spikes etc.) running all over the place
rather than being nicely tucked inside the distributor.

2. Yes. There are 3 sensors in the distributor, the CYP, or Cylinder
Position Sensor, gives one pulse for every complete rotation of the
distributor rotor (presumably around TDC prior to the power stroke on
cylinder #1, but that is a guess on my part), the TDC, or Top Dead
Center, sensor gives 4 pulses for each complete rotation of the
distributor rotor, one pulse each time a piston is at TDC at the
beginning of its power strike. Finally there is the CKP, or crank
position sensor. This sensor gives a large number of pulses for each
rotation of the distributor rotor.

The CKP gives the crank angle with very fine resolution, but it is a
relative measurement only. I.e. by looking at the CKP output you can
tell that the engine now has rotated e.g. 76.3° since you started
counting, but you can not tell what specific position it is at. Then
what you (well, the E-C-you, that is) do is you reset your counter every
time you see the TDC signal. Now you know the exact number of degrees
you are past the latest power-stroke TDC, and since you know the rotor
goes 90° between each power-stroke TDC, you can easily do the
subtraction and find out how many degrees /before/ TDC you are. The ECU
calculates the desired ignition timing based on throttle position, RPM,
manifold pressure etc., and then simply turn on or off the signal going
to the ICM at the exact right point in the rotation as read by the CKP
and the TDC signals.

The ECU really does not care about /what/ cylinder gets the spark, the
finger in the distributor does that in the conventional way. The ECU
/does/ need to know what injector to fire, however, and that is what the
CYP sensor is for.

The above is relevant for my '94 Civic. Newer systems have gotten rid of
the distributor finger as well, and then you need the CYP sensor to get
the right ignition sequence.

I don't know about Honda, but some other newer cars have only one
sensor, essentially the CKP. The absolute position information is
indicated by missing pulses on the CKP signal. Imagine an ABS wheel
sensor where you file away a couple of teeth in the right locations.
E.g. you remove 4 teeth, each 90° from the next removed tooth. Then you
remove a 5th tooth at some other location. now you can deduce the CKP,
CYP and TDC information all from one sensor.

Incidentally, the CKP signal can also be used to detect misfires.
During the power stroke, the crank will accelerate a bit, then retard a
bit around TDC. This can be measured as a slight variation in the
frequency of the CKP signal. If you see that there is no acceleration at
the time when you know there is a power stroke, you conclude there was a
misfire. This type of detection is required in OBDII equipped cars.

3. It is my understanding that the output transistor fails in the
ignitors. I can not find any reliable reference on this, however. I
don't think the main relays fail because of poor soldering from the
factory. I believe the solder fatigues over the years, since solder is
the only /mechanical/ fastening for a fairly heavy relay. Solder, if
subjected to stresses and strains (like shaking a relay for mile after
mile), will fatigue. If you have a stranded copper wire, tin the leads
and clamp it down really good in a screw terminal, you can come back a
few years later and see that the solder has been reduced to dust and
that the wire now is loose. I do not believe the ignitor has any heavy
components in it.

 

Michael Pardee wrote:
| Extrapolating from my experience with power transistors...

Thanks for the insights.

| If the switcher is actually a power CMOS device, I don't know enough
| about them to say.

Extrapolating from my experience with Power MOSFETs (not much), I would tend
to agree. However, I have seen both happening, the MOSFET developing an
open-circuit and "wire replacement" I wouldn't be surprised if short
circuit between drain and source would eventually overheat something inside
and cause an open circuit.

Don't know whether current igniters are MOSFET or bipolar. MOSFET would make
sense due to the ease of driving them, the ridiculously low losses (perhaps
not that much of a problem? I tinkered around with a MOSFET that takes 180A
(amps, not milliamps!) and has something like 15 mOhm (milliohm) "on"
resistance). On the other hand, car manufacturers tend to be rather
conservative with electronics, so they might hold on to bipolar for another
while?

From a page on Toyota igniters,
http://alflash.narod.ru/New_Folder/58846406.gif (from
http://alflash.narod.ru/P1300.htm) it seems like bipolar is still in
fashion.

Now, if they use IGBTs, then someone else needs to give their input

 

Michael Pardee wrote:
| Extrapolating from my experience with power transistors...

Thanks for the insights.

| If the switcher is actually a power CMOS device, I don't know enough
| about them to say.

Extrapolating from my experience with Power MOSFETs (not much), I would tend
to agree. However, I have seen both happening, the MOSFET developing an
open-circuit and "wire replacement" I wouldn't be surprised if short
circuit between drain and source would eventually overheat something inside
and cause an open circuit.

Don't know whether current igniters are MOSFET or bipolar. MOSFET would make
sense due to the ease of driving them, the ridiculously low losses (perhaps
not that much of a problem? I tinkered around with a MOSFET that takes 180A
(amps, not milliamps!) and has something like 15 mOhm (milliohm) "on"
resistance). On the other hand, car manufacturers tend to be rather
conservative with electronics, so they might hold on to bipolar for another
while?

From a page on Toyota igniters,
http://alflash.narod.ru/New_Folder/58846406.gif (from
http://alflash.narod.ru/P1300.htm) it seems like bipolar is still in
fashion.

Now, if they use IGBTs, then someone else needs to give their input

 

Lots of stuff can go wrong. It's a brutal environment.

Ozone can carbonize dirty surfaces near the HV and create a conductive
path into sensitive circuits. The coil's winding insulation can fail
and overheat the coil. Excessive voltages caused by cracked wires, worn
points, or a failed capacitor results avalanching in the power
transistor, the semiconductor equivalent of an arc-over, and accelerated
aging of the coil insulation. Avalanching slowly degrades the silicon
crystal into a more passive form, like a resistor, that will overheat.

Transistors usually fail as a short circuit or a low resistance. If the
fried transistor heats enough to melt a lead wire, the flyback power
released from the coil can lead to a small explosion at the break.

Typical ignition systems are flyback types. Inductors act as constant
current devices. Applying 12V causes current flow to gradually
increase. Break the current rapidly and voltage shoots in the opposite
direction in an attempt to maintain the current flow. That's a few
hundred volts on the primary and tens of thousands on the secondary.
Flyback transformers have a problem with imperfect magnetic coupling
between the two coils. Some of the power spikes back into the primary
even if the secondary discharges into a spark plug. It's the
capacitor's job to dampen the primary's voltage so it doesn't arc over
the switch/transistor or rise faster than the switch/transistor can turn
off. If the secondary doesn't discharge into a spark plug, the flyback
voltage will rise until something gives way. Hopefully its a protective
avalanche diode or spark gap.

 

Lots of stuff can go wrong. It's a brutal environment.

Ozone can carbonize dirty surfaces near the HV and create a conductive
path into sensitive circuits. The coil's winding insulation can fail
and overheat the coil. Excessive voltages caused by cracked wires, worn
points, or a failed capacitor results avalanching in the power
transistor, the semiconductor equivalent of an arc-over, and accelerated
aging of the coil insulation. Avalanching slowly degrades the silicon
crystal into a more passive form, like a resistor, that will overheat.

Transistors usually fail as a short circuit or a low resistance. If the
fried transistor heats enough to melt a lead wire, the flyback power
released from the coil can lead to a small explosion at the break.

Typical ignition systems are flyback types. Inductors act as constant
current devices. Applying 12V causes current flow to gradually
increase. Break the current rapidly and voltage shoots in the opposite
direction in an attempt to maintain the current flow. That's a few
hundred volts on the primary and tens of thousands on the secondary.
Flyback transformers have a problem with imperfect magnetic coupling
between the two coils. Some of the power spikes back into the primary
even if the secondary discharges into a spark plug. It's the
capacitor's job to dampen the primary's voltage so it doesn't arc over
the switch/transistor or rise faster than the switch/transistor can turn
off. If the secondary doesn't discharge into a spark plug, the flyback
voltage will rise until something gives way. Hopefully its a protective
avalanche diode or spark gap.

 

Honda shows MOSFETs in their ignition schematics of the Civic but
doesn't show what is used in the Accord.

MOSFETs are kind of tricky to use for flyback circuits. There's a
strong capacitive coupling between the gate and the source. Several
amps coming off the gate from that capacitive coupling isn't unusual.
It can lead to nasty RF ringing or destruction of the driver. The
solution is of course to use pair of power MOSFETs to drive the HV
MOSFET. The driver is complex but it fits on a chip.

Bipolar transistors have less capacitive coupling but they need lots of
holding current. The driver is trivial but it needs power resistors
that run too hot to be on a chip.

MOSFETs can't handle much current when they're designed for high
voltages. A small 30V MOSFET can handle 50A to 150A but a small 500V
MOSFET handles 4 to 8 amps. Automotive ignition coils is pretty much
their limit. Larger HV power supplies use bipolar transistors or IGBTs.

An IGBT has impressive current and voltage abilities but one might be
too slow for an ignition coil. The IMA, A/C, and other 144V 3-phase
motors are driven by IGBTs the Hybrid Accord.

 

Honda shows MOSFETs in their ignition schematics of the Civic but
doesn't show what is used in the Accord.

MOSFETs are kind of tricky to use for flyback circuits. There's a
strong capacitive coupling between the gate and the source. Several
amps coming off the gate from that capacitive coupling isn't unusual.
It can lead to nasty RF ringing or destruction of the driver. The
solution is of course to use pair of power MOSFETs to drive the HV
MOSFET. The driver is complex but it fits on a chip.

Bipolar transistors have less capacitive coupling but they need lots of
holding current. The driver is trivial but it needs power resistors
that run too hot to be on a chip.

MOSFETs can't handle much current when they're designed for high
voltages. A small 30V MOSFET can handle 50A to 150A but a small 500V
MOSFET handles 4 to 8 amps. Automotive ignition coils is pretty much
their limit. Larger HV power supplies use bipolar transistors or IGBTs.

An IGBT has impressive current and voltage abilities but one might be
too slow for an ignition coil. The IMA, A/C, and other 144V 3-phase
motors are driven by IGBTs the Hybrid Accord.

 

I agree. Mounting of heavy components by solder is a dubious practice, and
the vibration makes fatigue an "if, not when" proposition.

When I worked in avionics, we would often see a popular transponder with
fuse failure. We replaced the fuse every time and only once saw a unit come
back. The fuse was a slo-blo glass fuse, and we could see the solder
attachment at the end had crumbled, leaving rough edges, instead of the
melted ends of blown fuses. We never saw it in twin engine planes, only
singles.

Mike

 

I agree. Mounting of heavy components by solder is a dubious practice, and
the vibration makes fatigue an "if, not when" proposition.

When I worked in avionics, we would often see a popular transponder with
fuse failure. We replaced the fuse every time and only once saw a unit come
back. The fuse was a slo-blo glass fuse, and we could see the solder
attachment at the end had crumbled, leaving rough edges, instead of the
melted ends of blown fuses. We never saw it in twin engine planes, only
singles.

Mike

 

This brings up one mystery - how do electronic ignitions get away with no
capacitor or equivalent energy shunt? Points needed a capacitor to give the
contacts time to get a little air between them before the voltage peaked, so
a capacitor isn't needed (strictly speaking) with an igniter. But if the HT
wires open up, the energy has to go *somewhere*. I know in Integras it often
goes into the coil with bad results, but many systems aren't damaged when
the HT lead is open.

Mike

 

This brings up one mystery - how do electronic ignitions get away with no
capacitor or equivalent energy shunt? Points needed a capacitor to give the
contacts time to get a little air between them before the voltage peaked, so
a capacitor isn't needed (strictly speaking) with an igniter. But if the HT
wires open up, the energy has to go *somewhere*. I know in Integras it often
goes into the coil with bad results, but many systems aren't damaged when
the HT lead is open.

Mike

 

Power MOSFETs also get around the thermal runaway problem. My hunch is they
are some sort of MOS devices.

Ditto - I barely know they exist!

Mike

 

Power MOSFETs also get around the thermal runaway problem. My hunch is they
are some sort of MOS devices.

Ditto - I barely know they exist!

Mike

 

When my ignitor failed, I didn't take it apart but perhaps should have
so see what's inside.

It could be that they wirebonded the dies and put thermally conductive
gel around it all -- the same way they make hockey puck Solid State
Relays, for instance.
They make enough of these ignitors to justify the one time charges and
would make them pretty cheap in quantity.
Not sure if they actually did this, but mechanically and electrically
it is a superior method of making a very reliable product.

 

When my ignitor failed, I didn't take it apart but perhaps should have
so see what's inside.

It could be that they wirebonded the dies and put thermally conductive
gel around it all -- the same way they make hockey puck Solid State
Relays, for instance.
They make enough of these ignitors to justify the one time charges and
would make them pretty cheap in quantity.
Not sure if they actually did this, but mechanically and electrically
it is a superior method of making a very reliable product.

 

Wow, what a response!

With the knowledge amply herein exhibited, maybe this group ought to be
called rec.autos.makers.honda.electronics! I'm impressed. This is the sort
of thing that makes a good FAQ possible. Thank you.

I'm responding in this particular message only because my original post has
been pushed off the stack by my news provider.

With the wealth or information in everybody's responses, I've saved them
all and will have to review them. I will post a draft once I figure I
understand all this.

For a picture of the insides of an igniter, see here:
http://www.gcw.org.uk/rover/igniter.htm#userreport
(you'll have to scroll up a bit)

For more pics:
http://www.markl.f9.co.uk/howto/electrical/igniter/igniter.htm

More questions:
1) Can anyone tell from the foregoing pics what kind of transistor is in
the igniter pictured?
2) What's this thing about the attachment of a heat sink by "wires"? I must
be dense or something, because I thought heat sinks were firmly fastened by
screws or solder blobs.
3) Can somebody explain a "flyback circuit" in terms a layman can
understand?