Topic: Transistor Ignition mods (improvement?) - opinions desired!

[PeWe]

I describe stock point plate with stock condensors on plate.
HM module is connected between this plate via the blue, yellow leads to coils.

Hondaman ask if anyone would like to skip stock condensors on plate. Have same functionality built-in the HM module.
This means that there is a capacitor between blue to ground. Same for yellow.

As a stock point plate has on the plate.

[MauiK3]

I vote for the mod. Seems like a great idea.

[maxheadflow]

I describe stock point plate with stock condensors on plate.
HM module is connected between this plate via the blue, yellow leads to coils.

Hondaman ask if anyone would like to skip stock condensors on plate. Have same functionality built-in the HM module.
This means that there is a capacitor between blue to ground. Same for yellow.

As a stock point plate has on the plate.

Ok got it.. The condensers form an RC filter that helps with point bounce.

[Deltarider]

RC is abbrevation for?

[maxheadflow]

Resistor/capacitor.   It's basically a low pass filter made from those.   Commonly used to filter out switch bounce, points in this case.  The resistor is inside the box. Cap next to the points.

[HondaMan]

I'll bring up another naive question.   

On a transistorized ignition, do you even need a cap/condenser?

When using points, the cap is there to prevent arching at the points.  I've seen old automotive manual someplace that basically said the value of the cap should be based on how the points transfer metal at the contact point.  When the cap is the right size, no transfer.   This means that the cap is there to control the collapse of the coil at an exact rate to minimize arcing at the points.

With a transistor switching the coil, the switch is solid state and no arcing occurs.  The only thing to worry about is the peak voltage that the collector sees.  New ignitions don't have caps that need replacing so I expect that they don't even have caps across the coils..

I would guess that the cap does help tune a older higher higher impedance coil so that it's peak oscillation is optimum?

You're pretty close to it: the condensor's normal job is to absorb the coil's kickback (called back-EMF) energy when the points open, to lower the voltage spike to a point where it does not have enough energy to arc the points, causing their eventual demise. The value of the capacitor (in capacitance, rated as uF values) depends on the coil's inductance, time-constant of spark decay (i.e., how many times the AC waveform it creates during collapse travels back-and-forth across the sparkplug, the rest of which travels back to arc the points) and impedance (resistance) of the primary windings. However, the inductance of the condensor (they are wound inside, like a coil, as are some poly-mylar capacitors) is part of all of this, forming an LC circuit (called a "tank" circuit) to generate a little bit more voltage at the coil for 'free' in its resonant RPM (frequency) range. The inductance of the condensor is usually different (and a lot more) than is found in most plate-type capacitors, and is one of the major differences between "condensor" and "capacitor", and the reason why they are not direct substiturtes for each other in these bikes.

Whew.

The ideal capacitance for our TEC (OEM) coils works out to 0.245uF, and they can kick back 400 volts if not absorbed (then it is closer to 200 volts). If too much is stored, then the points have to dump it anyway after they re-close, making them arc then, too. If it isn't dumped, then you get the Dyna S syndrome where the coils can never fully discharge (above 3800 RPM in their design), making them run hot and shortening the spark duration. The ideal inductance is about 50 nH (nano-Henries) at 120 Hz (around 7200 RPM), important to know when selecting capacitors to replace condensors.

Electronic designs all have a 'flyback diode' clamped across the coil. This is done to both protect the transistor doing the switching and to cool off the coil by helping control its discharge rate: this extends the spark a little longer (just a few mS). It is sort of self-balancing in that as the coil heats up and lessens its inductance and impedance it also lowers it back-EMF, reducing the heating that would otherwise occur. In most cars now, this is how it is done: I haven't looked at bikes lately, but they are probably very similar.

CDI systems are a whole 'nuther thing, and too complex to talk about in a paragraph...

[HondaMan]

...But, still having the mechanical breakers for switching the control current, eventual bouncing cannot be ruled out, which would trigger multiple sparking.   

Exactly right, Delta: the points do not open all at once. At varying RPM, they open once, then their spring resistance pushes the points cam away (it does have some mechanical slop between it and the post it rides on) and often re-closes the points momentarily. On the Honda 350/360/450 twins this is an epidemic problem because those advancers relied on a set of 2-4 thin plastic shims under the points cam to push the cam outward against the washer on the bolt that holds those advancers to the camshaft (now those have melted away from engine heat and years). There must be less than 0.006" (0.002" is best) between that washer and the points cam, or else it bounces away (tilts) from the points foot (from points spring pressure) so much that it double-sparks the coil, so to speak. This loses some of the coil's energy in the first attempt, then holds it for a few milliseconds, then opens it again, making the second spark weaker. When the Transistor Ignition is then attached, it dumps the coil's whole charge at the first opening, so when it recloses too soon the coil's back-EMF is fed right back to it, killing the collapse and the spark. The result is a very weak spark. This happens once in a while on the 750 advancers (very high mileage, like mine at 120k+ miles this started showing up).

But, it can be fixed by putting the shims back under the points cam to reduce the clearance and hold the cam straight. Works like magic!

[HondaMan]

Max, I am pretty sure PeWe means, without the condensers being used the HondaMan Electronic Ignition box can have misfires above 7k rpm.

On the input or the output to the box?

Yes, above about 6500 RPM the points themselves start chattering (bouncing) on closure. Because of the high speed of the transistor, they in turn try to switch the coils just like the points are telling them to, but there isn't enough spark energy for a spark. Instead, it develops a back-EMF situation that resists the inrush of power to recharge the coils for their next round of spark.

When we built these engines to reach past 12k RPM (midget racer cars) one of the primary things that received attention was the opening/closing ramps on these spark advancers, unless they were being driven by the custom-made cam-driven points (inspired by the 350 Honda, said the owner) - which was on the same shaft that was driving his oil pump, off the cam (because the trannys were all cut off the engines). They (the crank-mounted ones) were OD ground on offset cams to make their opening ramp 3x longer, then the slots on the points plates were made VERY wide (like almost 1/4 of the circumference of the plate) so the timing could be shifted accordingly. At low speeds (starting, idling) the spark was temperamental because of the slow-opening points, so the transistor ignitions because vital to making the engines start. Many tricks were employed by various builders, all to reach 12k RPM reliably. Some used transistors driven by those points, some used SCR dwell extenders, and some used photo-electric devices (not the nice ones we have today) driving transistors, with hand-built round sheet metal drums with slots cut in them to trigger the spark. Cool stuff!

BTW: the guy who drove his points and oil pump off the cam (engine's right side) could also run 14,500 RPM...with stock rods and pistons. He wasn't a great driver and seldom placed, but boy, he was fast on the straights!

[maxheadflow]

I'll bring up another naive question.   

On a transistorized ignition, do you even need a cap/condenser?

When using points, the cap is there to prevent arching at the points.  I've seen old automotive manual someplace that basically said the value of the cap should be based on how the points transfer metal at the contact point.  When the cap is the right size, no transfer.   This means that the cap is there to control the collapse of the coil at an exact rate to minimize arcing at the points.

With a transistor switching the coil, the switch is solid state and no arcing occurs.  The only thing to worry about is the peak voltage that the collector sees.  New ignitions don't have caps that need replacing so I expect that they don't even have caps across the coils..

I would guess that the cap does help tune a older higher higher impedance coil so that it's peak oscillation is optimum?

You're pretty colse to it: the condensor's normal job is to absorb the coil's kickback (called back-EMF) energy when the points open, to lower the voltage spike to a point where it does not have enough energy to arc the points, causing their eventual demise. The value of the capacitor (in capacitance, rated as uF values) depends on the coil's inductance, time-constant of spark decay (i.e., how many times the AC waveform it creates during collapse travels back-and-forth across the sparkplug, the rest of which travels back to arc the points) and impedance (resistance) of the primary windings. However, the inductance of the condensor (they are wound inside, like a coil, as are some poly-mylar capacitors) is part of all of this, forming an LC circuit (called a "tank" circuit) to generate a little bit more voltage at the coil for 'free' in its resonant RPM (frequency) range. The inductance of the condensor is usually different (and a lot more) than is found in most plate-type capacitors, and is one of the major differences between "condensor" and "capacitor", and the reason why they are not direct substiturtes for each other in these bikes.

Whew.

The ideal capacitance for our TEC (OEM) coils works out to 0.245uF, and they can kick back 400 volts if not absorbed (then it is closer to 200 volts). If too much is stored, then the points have to dump it anyway after they re-close, making them arc then, too. If it isn't dumped, then you get the Dyna S syndrome where the coils can never fully discharge (above 3800 RPM in their design), making them run hot and shortening the spark duration. The ideal inductance is about 50 nH (nano-Henries) at 120 Hz (around 7200 RPM), important to know when selecting capacitors to replace condensors.

Electronic designs all have a 'flyback diode' clamped across the coil. This is done to both protect the transistor doing the switching and to cool off the coil by helping control its discharge rate: this extends the spark a little longer (just a few mS). It is sort of self-balancing in that as the coil heats up and lessens its inductance and impedance it also lowers it back-EMF, reducing the heating that would otherwise occur. In most cars now, this is how it is done: I haven't looked at bikes lately, but they are probably very similar.

CDI systems are a whole 'nuther thing, and too complex to talk about in a paragraph...

This is interesting stuff.  I pretty much figured the the coil / condenser to be nothing more than a tank circuit. It will have a natural oscillation once excited by the points opening.   Once the spark jumps the gap, everything changes and you have basically a transformer with a load in the circuit. In that case, the inductance changes drastically and so does the natural frequency of the circuit.   I'd imagine  it goes up. 

I'd guess that the condenser is chosen to give the highest possible Q of the circuit to create a maximum peak oscillation amplitude.  That amplitude is needed to initiate spark across the secondary (spark plug). 

I've not thought much about the energy decay in the ignition system and your discussion of it is very enlightening.  I can see where a diode across the  would help ensure that the circuit oscillation is clamped.  That way the phase of the oscillation is not too negative when the points close causing a longer saturation time. I can see where that helps with spark performance.

I'm not sure I understand the difference between a capacitor and condenser.  I'd expect in the case of a Mylar cylinder cap that there is alternate sheets of conductors wrapped into a cylinder with the insulator (Mylar) in between. The sheets are staggered so that one sheet hangs off one end and the other off the other. The conductors are fused to wire leads and the inductance would be very low inside the capacitor itself.  The big issue would be lead length.   If 40 nH is needed, leads would be the contributors. I assume the 40 nH is in series and not parallel.   Example. The lead inductance of the old DIP IC can be 10 to 16 nH and the single ground lead on a oscilloscope could be as high as 200 nH.

I've not thought of the fly-back diode as a way to dump heat external to the inductor but most of the stuff I dealt with was  relays and solenoids where switching times were slow.  It make sense tho as you have a fixed amount of energy in the inductor and you need to dump it somewhere external.

I've not dealt with CDI designs but figured that they were nothing more than a 300 volt power supply dumping directly into a coil.

[maxheadflow]

Max, I am pretty sure PeWe means, without the condensers being used the HondaMan Electronic Ignition box can have misfires above 7k rpm.

On the input or the output to the box?

Yes, above about 6500 RPM the points themselves start chattering (bouncing) on closure. Because of the high speed of the transistor, they in turn try to switch the coils just like the points are telling them to, but there isn't enough spark energy for a spark. Instead, it develops a back-EMF situation that resists the inrush of power to recharge the coils for their next round of spark.

When we built these engines to reach past 12k RPM (midget racer cars) one of the primary things that received attention was the opening/closing ramps on these spark advancers, unless they were being driven by the custom-made cam-driven points (inspired by the 350 Honda, said the owner) - which was on the same shaft that was driving his oil pump, off the cam (because the trannys were all cut off the engines). They (the crank-mounted ones) were OD ground on offset cams to make their opening ramp 3x longer, then the slots on the points plates were made VERY wide (like almost 1/4 of the circumference of the plate) so the timing could be shifted accordingly. At low speeds (starting, idling) the spark was temperamental because of the slow-opening points, so the transistor ignitions because vital to making the engines start. Many tricks were employed by various builders, all to reach 12k RPM reliably. Some used transistors driven by those points, some used SCR dwell extenders, and some used photo-electric devices (not the nice ones we have today) driving transistors, with hand-built round sheet metal drums with slots cut in them to trigger the spark. Cool stuff!

BTW: the guy who drove his points and oil pump off the cam (engine's right side) could also run 14,500 RPM...with stock rods and pistons. We wasn't a great driver and seldom placed, but boy, he was fast on the straights!

Back in the day when quality parts were easy to get, I used cb400 points and advance units to get stable points at higher RPMs since the redline on those was 10.5. Advance was a little slower tho.

[HondaMan]

I'll bring up another naive question.   

On a transistorized ignition, do you even need a cap/condenser?

When using points, the cap is there to prevent arching at the points.  I've seen old automotive manual someplace that basically said the value of the cap should be based on how the points transfer metal at the contact point.  When the cap is the right size, no transfer.   This means that the cap is there to control the collapse of the coil at an exact rate to minimize arcing at the points.

With a transistor switching the coil, the switch is solid state and no arcing occurs.  The only thing to worry about is the peak voltage that the collector sees.  New ignitions don't have caps that need replacing so I expect that they don't even have caps across the coils..

I would guess that the cap does help tune a older higher higher impedance coil so that it's peak oscillation is optimum?

You're pretty colse to it: the condensor's normal job is to absorb the coil's kickback (called back-EMF) energy when the points open, to lower the voltage spike to a point where it does not have enough energy to arc the points, causing their eventual demise. The value of the capacitor (in capacitance, rated as uF values) depends on the coil's inductance, time-constant of spark decay (i.e., how many times the AC waveform it creates during collapse travels back-and-forth across the sparkplug, the rest of which travels back to arc the points) and impedance (resistance) of the primary windings. However, the inductance of the condensor (they are wound inside, like a coil, as are some poly-mylar capacitors) is part of all of this, forming an LC circuit (called a "tank" circuit) to generate a little bit more voltage at the coil for 'free' in its resonant RPM (frequency) range. The inductance of the condensor is usually different (and a lot more) than is found in most plate-type capacitors, and is one of the major differences between "condensor" and "capacitor", and the reason why they are not direct substiturtes for each other in these bikes.

Whew.

The ideal capacitance for our TEC (OEM) coils works out to 0.245uF, and they can kick back 400 volts if not absorbed (then it is closer to 200 volts). If too much is stored, then the points have to dump it anyway after they re-close, making them arc then, too. If it isn't dumped, then you get the Dyna S syndrome where the coils can never fully discharge (above 3800 RPM in their design), making them run hot and shortening the spark duration. The ideal inductance is about 50 nH (nano-Henries) at 120 Hz (around 7200 RPM), important to know when selecting capacitors to replace condensors.

Electronic designs all have a 'flyback diode' clamped across the coil. This is done to both protect the transistor doing the switching and to cool off the coil by helping control its discharge rate: this extends the spark a little longer (just a few mS). It is sort of self-balancing in that as the coil heats up and lessens its inductance and impedance it also lowers it back-EMF, reducing the heating that would otherwise occur. In most cars now, this is how it is done: I haven't looked at bikes lately, but they are probably very similar.

CDI systems are a whole 'nuther thing, and too complex to talk about in a paragraph...

This is interesting stuff.  I pretty much figured the the coil / condenser to be nothing more than a tank circuit. It will have a natural oscillation once excited by the points opening.   Once the spark jumps the gap, everything changes and you have basically a transformer with a load in the circuit. In that case, the inductance changes drastically and so does the natural frequency of the circuit.   I'd imagine  it goes up. 

I'd guess that the condenser is chosen to give the highest possible Q of the circuit to create a maximum peak oscillation amplitude.  That amplitude is needed to initiate spark across the secondary (spark plug). 

I've not thought much about the energy decay in the ignition system and your discussion of it is very enlightening.  I can see where a diode across the  would help ensure that the circuit oscillation is clamped.  That way the phase of the oscillation is not too negative when the points close causing a longer saturation time. I can see where that helps with spark performance.

I'm not sure I understand the difference between a capacitor and condenser.  I'd expect in the case of a Mylar cylinder cap that there is alternate sheets of conductors wrapped into a cylinder with the insulator (Mylar) in between. The sheets are staggered so that one sheet hangs off one end and the other off the other. The conductors are fused to wire leads and the inductance would be very low inside the capacitor itself.  The big issue would be lead length.   If 40 nH is needed, leads would be the contributors. I assume the 40 nH is in series and not parallel.   Example. The lead inductance of the old DIP IC can be 10 to 16 nH and the single ground lead on a oscilloscope could be as high as 200 nH.

I've not thought of the fly-back diode as a way to dump heat external to the inductor but most of the stuff I dealt with was  relays and solenoids where switching times were slow.  It make sense tho as you have a fixed amount of energy in the inductor and you need to dump it somewhere external.

I've not dealt with CDI designs but figured that they were nothing more than a 300 volt power supply dumping directly into a coil.

You pretty well get it!
The condensor, though, is a sort of throwback design. It's main difference is: it does not change value or quality of capacitance with high heat, and it is self-healing because of that heat. When they are cold, their insulation is thicker and the capacitance drops (by about 10-15%), which is why engines using them seem to be "cold blooded" on cold mornings: it isn't the engines so much (provided the proper oils were used, that does have a negative effect). When I was young (and dirt was new) we used to put a propane torch on the condensors and points for a minute on those sub-zero Chicago mornings in the late 1960s if the car would not start, even with ether directly into the open carb. That worked well, provided you didn't catch the oil leaking from the valve covers on fire, too...

When the condensor heats up, the insulation becomes soft and bonds itself closely with the foil plates and the waxed paper (or now polymer paper) insulator, [re]improving the capacitance as the engine warmed it up. After a long time under the hood the wax could drain away, so until the condensors were sealed up well (they had epoxies added to their tops) they would 'age' as the wax boiled away little by little. Then the self-restoring capability would stay much longer, and until about 2002 our condensors were made that way. Now they are a miserable piece of Chinko crap, foil with a waxed paper and a rubber top seal that must be wound PERFECTLY or it will not work: it will short out thru the paper when the foil crinkles during wrap, and this appears when the foil heats up with the engine. With just the condensors and points this causes rough running, poor (or no) idle, hard starting. With the Transistor Ignition attached, it causes weak spark and fouled plugs, quickly.

So, I am embarking on modifying this design to include an additional set of condensors (400v mylar with stainless foil and molded epoxy body) in the boxes and a pair of additional connectors on the Blue and Yellow wires that will enable a simple and quick "bypass" if the box should fail. Then we can just remove the condensors, since they don't work due to Chinko negligence. All the benefits of having this dual-system, built-in backup box will make it better yet.

Now if I could just get my Explorer's freshly-rebuilt tranny to work again...

[MauiK3]

Sounds like a great mod.
Some folks call Explorers "Exploders"

[HondaMan]

Sounds like a great mod.
Some folks call Explorers "Exploders"

Mine has been a WONDERFUL car, but started showing its 190k mile age with a little tranny slip now and then on hot days in the mountains. So, I took it in to my once-favorite tranny shop (who has rebuilt my Ford and Dodge trannies expertly since 1985) for a rebuild. When I went to pick it up, I discovered that ALL the folks who used to work there these past decades are gone.

...Uh-oh...

And, yep, they did the worst tranny job I have ever seen. It worked OK, not well, for the first 4k miles, then no 3rd gear most of the time. Took it back, they replaced all the shift solenoids 3 weeks ago, now it leaks oil, was also SERIOUSLY over-filled upon return (but hasn't stopped leaking since I removed some of the oil) and worked for 3 days. Then the speedometer started jumping all over and the transmission could not settle on any gear, nor whether the convertor should be locked or not, etc., etc., so on advice of another mechanic who 'knows' these 4R55E trannys, I replaced the VSS. No change, well, not exactly: it got worse. I have now disconnected everything but the main computer (PCM) form that sensor, no change. I will next try the front ISS (Input Speed Sensor) and if that doesn't work, will have to buy another transmission.

...because the tranny shop where I have gone for decades closed last month...

[maxheadflow]

Snipped out stuff to keep from making the reply too long.

You pretty well get it!
The condensor, though, is a sort of throwback design. It's main difference is: it does not change value or quality of capacitance with high heat, and it is self-healing because of that heat. When they are cold, their insulation is thicker and the capacitance drops (by about 10-15%), which is why engines using them seem to be "cold blooded" on cold mornings: it isn't the engines so much (provided the proper oils were used, that does have a negative effect). When I was young (and dirt was new) we used to put a propane torch on the condensors and points for a minute on those sub-zero Chicago mornings in the late 1960s if the car would not start, even with ether directly into the open carb. That worked well, provided you didn't catch the oil leaking from the valve covers on fire, too...

When the condensor heats up, the insulation becomes soft and bonds itself closely with the foil plates and the waxed paper (or now polymer paper) insulator, [re]improving the capacitance as the engine warmed it up. After a long time under the hood the wax could drain away, so until the condensors were sealed up well (they had epoxies added to their tops) they would 'age' as the wax boiled away little by little. Then the self-restoring capability would stay much longer, and until about 2002 our condensors were made that way. Now they are a miserable piece of Chinko crap, foil with a waxed paper and a rubber top seal that must be wound PERFECTLY or it will not work: it will short out thru the paper when the foil crinkles during wrap, and this appears when the foil heats up with the engine. With just the condensors and points this causes rough running, poor (or no) idle, hard starting. With the Transistor Ignition attached, it causes weak spark and fouled plugs, quickly.

So, I am embarking on modifying this design to include an additional set of condensors (400v mylar with stainless foil and molded epoxy body) in the boxes and a pair of additional connectors on the Blue and Yellow wires that will enable a simple and quick "bypass" if the box should fail. Then we can just remove the condensors, since they don't work due to Chinko negligence. All the benefits of having this dual-system, built-in backup box will make it better yet.

Now if I could just get my Explorer's freshly-rebuilt tranny to work again...

Had to think about this for a while as I'm not sure I understand.  You state that the insulation is thicker when cold. I would guess that in general, as things get cooler they shrink.  This kind of threw me for a loop.  IMO it has to shrink. In that case the capacitance would go up as the distance between the plates is smaller and smaller distances create more capacitance. 

You mention that the condensers were made of thin wax paper.  The wax paper is basically paper and wax permeated through it.
 I remember in the past seeing a wax like substance coming out of some caps, some condensers, some high voltage tube stuff.  What would make more sense to me is that the paper shrinks less than the wax when the temps change and the wax pulls away from the plates in some locations. This would reduce the surface contact area with the plates and reduce capacitance.   It could also be simply due to the dielectric constant changing over temperature.

You have gone way deeper into the study of condensers then I would have for sure..   One would think tho, that a good Mylar / polypropylene cap would exhibit better temp stability than the old wax paper type.

[HondaMan]

Snipped out stuff to keep from making the reply too long.

You pretty well get it!
The condensor, though, is a sort of throwback design. It's main difference is: it does not change value or quality of capacitance with high heat, and it is self-healing because of that heat. When they are cold, their insulation is thicker and the capacitance drops (by about 10-15%), which is why engines using them seem to be "cold blooded" on cold mornings: it isn't the engines so much (provided the proper oils were used, that does have a negative effect). When I was young (and dirt was new) we used to put a propane torch on the condensors and points for a minute on those sub-zero Chicago mornings in the late 1960s if the car would not start, even with ether directly into the open carb. That worked well, provided you didn't catch the oil leaking from the valve covers on fire, too...

When the condensor heats up, the insulation becomes soft and bonds itself closely with the foil plates and the waxed paper (or now polymer paper) insulator, [re]improving the capacitance as the engine warmed it up. After a long time under the hood the wax could drain away, so until the condensors were sealed up well (they had epoxies added to their tops) they would 'age' as the wax boiled away little by little. Then the self-restoring capability would stay much longer, and until about 2002 our condensors were made that way. Now they are a miserable piece of Chinko crap, foil with a waxed paper and a rubber top seal that must be wound PERFECTLY or it will not work: it will short out thru the paper when the foil crinkles during wrap, and this appears when the foil heats up with the engine. With just the condensors and points this causes rough running, poor (or no) idle, hard starting. With the Transistor Ignition attached, it causes weak spark and fouled plugs, quickly.

So, I am embarking on modifying this design to include an additional set of condensors (400v mylar with stainless foil and molded epoxy body) in the boxes and a pair of additional connectors on the Blue and Yellow wires that will enable a simple and quick "bypass" if the box should fail. Then we can just remove the condensors, since they don't work due to Chinko negligence. All the benefits of having this dual-system, built-in backup box will make it better yet.

Now if I could just get my Explorer's freshly-rebuilt tranny to work again...

Had to think about this for a while as I'm not sure I understand.  You state that the insulation is thicker when cold. I would guess that in general, as things get cooler they shrink.  This kind of threw me for a loop.  IMO it has to shrink. In that case the capacitance would go up as the distance between the plates is smaller and smaller distances create more capacitance. 

You mention that the condensers were made of thin wax paper.  The wax paper is basically paper and wax permeated through it.
 I remember in the past seeing a wax like substance coming out of some caps, some condensers, some high voltage tube stuff.  What would make more sense to me is that the paper shrinks less than the wax when the temps change and the wax pulls away from the plates in some locations. This would reduce the surface contact area with the plates and reduce capacitance.   It could also be simply due to the dielectric constant changing over temperature.

You have gone way deeper into the study of condensers then I would have for sure..   One would think tho, that a good Mylar / polypropylene cap would exhibit better temp stability than the old wax paper type.

The poly caps do work better except for one thing: none of them can sustain 200 degrees F temps for very long. I use those inside my Transistor Ignition where they are safe: this new 'mod' I am thinking of would put another pair in there, with their own leads coming out of the box, to join to the points wire at their [normal] connection to the coils, if the box were being "bypassed". I always wish for the rider who is using my box to have a "backup plan" to let him/her ride home, should the box fail for some reason. Having spare condensors would be a big help, given that the ones available (at least now) from China do NOT work.