Tests of CDI Ignition Modules

[DKGrimm]

I've been using S/S CDI ignition modules for years. With Roy Sholl retiring a lot of people are wondering where to go for substitute CDI modules. At the same time there is quite a bit of enthusiasm in HMEM and other places for inductive coil systems and for the Sage-Gedde driver module for those. Following discussions with several people at the Black Hills engine show in September I decided to launch a project to study and compare both kinds of systems to see just what the advantages and disadvantages of each system may be. There's a long way to go before I get enough data to answer questions, but I started by looking at CDI modules. I begged, borrowed, bought -- but did not steal -- as many of those as I could lay my hands on to run the tests. I ended up with sixteen modules, four of which were broken, which gave me data on twelve.

The tests involved some highfalutin test equipment including a sampling oscilloscope with a 50 MHz bandwidth and 1 gigasamples/second sampling rate. I used data from that to compute the total energy that ended up heating the plasma of the spark itself. That energy is measured in millijoules, which I think everyone has at least heard of. (A joule is a watt-second.) One of the most interesting things I learned was that only a quarter or so of the total energy spit out by the module ended up in the spark itself. I have read suggestions on HMEM that a CDI pulse is too short to really give a hot spark. Right now, I think there may be some truth in that.

This is all preliminary stuff, but I summarized the work on CDI modules in a short report, attached below. I'm looking for critique on all this stuff. Let me know what you think.

Don

 

[oldengineguy]

WOW ! Just Wow! Of the 7 I.C. engines I have (5 built by me ,2 inherited) the 2 that have CDI and miniature plugs will not start or run reliably to the point that they are never displayed. Maybe I'm just a dinosaur but give me a blue, fat, hot- points, coil and condenser spark any day. Great work on that report.
Colin

 

[Zeb]

Love this. I take it you are running a Rigol? The text is very readable.
It would be interesting to see the effects using different gaps and compression. A bigger gap for the same setup would increase the current, correct?

I used to gap and test plugs with the old Champion tester. Some plugs would work fine until compressed air was applied.

I could see using an off-the-shelf arduino and a hall sensor for crank position. The capacitive discharge circuit would be the real tricky bit for me.

 

[DKGrimm]

Love this. I take it you are running a Rigol? The text is very readable.
It would be interesting to see the effects using different gaps and compression. A bigger gap for the same setup would increase the current, correct?

I used to gap and test plugs with the old Champion tester. Some plugs would work fine until compressed air was applied.

I could see using an off-the-shelf arduino and a hall sensor for crank position. The capacitive discharge circuit would be the real tricky bit for me.

Zeb: Techtronix TBS 1052B with a homemade 1000:1 compensated voltage divider. Still a problem with scope amplifier saturation and overload on the peaks if I use enough gain to really show detail on the rest of the waveforms, especially with the 0.025 gap plug. I've had to put in some low pass filtering at about 20 MHz to be able to see anything at all. Risetimes on the spark initiation are unbelievable.

Don

 

[DKGrimm]

Love this. I take it you are running a Rigol? The text is very readable.
It would be interesting to see the effects using different gaps and compression. A bigger gap for the same setup would increase the current, correct?

I used to gap and test plugs with the old Champion tester. Some plugs would work fine until compressed air was applied.

I could see using an off-the-shelf arduino and a hall sensor for crank position. The capacitive discharge circuit would be the real tricky bit for me.

Zeb: forgot to reply to your first question. I tested a whole bunch of spark plugs that way when I was working for my dad at his garage as a teenager. To tell you the truth I really have no data about the effects of pressure and gap size. I plan to do some testing to find out, but that's a ways down the road. Right now, I'm guessing that the current is mostly driven by the CDI module, while the voltage is determined by the arc as the current and gap vary.

 

[dsage]

Don:
With all the fantastic work you have done in the past on Magneto's and your very scientific approach to that work. I am sure we can trust your results. It's something I always wanted to do and have quite a setup here that could accomplish it. But I just never took it past testing the Sage/Gedde modules.
I'm glad you are taking interest in it. We will all benefit from your findings.
Thank You.

 

[DKGrimm]

Don:
With all the fantastic work you have done in the past on Magneto's and your very scientific approach to that work. I am sure we can trust your results. It's something I always wanted to do and have quite a setup here that could accomplish it. But I just never took it past testing the Sage/Gedde modules.
I'm glad you are taking interest in it. We will all benefit from your findings.
Thank You.

Thanks for the kind words. The design of the Sage/Gedde driver module has impressed me greatly. The hallmark of a good design lies in its simplicity. I'm just now trying out some of that yellow transfer paper to bond the resist pattern to some blank boards to build two or three for myself. We'll see how that works.

 

[dsage]

In the end it is just a driver. Although it will drive any coil. So the results will be totally dependent on the coil you choose. Many are being used to drive COP (coil on plug) coils because they are relatively small (compared to the old school can variety). The two-wire Ford coils are about the cheapest. Even if you get knockoff's that are sub-standard, the output is plenty for a model.
One fellow in California is using 8 COP coils and 8 drivers (no distributor) for his 8 cylinder engine. The ultimate I guess.
Personally my test setup is with the old school can variety because they are more demanding on the module and I can compare the output to a typical car performance and waveforms.
To be honest I've used the S/S CDI modules on almost all of my models because I like to hide everything under the base. But I find them a bit limited under demanding conditions. One of my engines is two cylinder with about 8:1 compression and the S/S module struggles. Also with the bizzare waveforms it's difficult to tell where problems lie.
I have wound a few coils of my own in order to reduce the size but it's too much trouble.
If the S/S modules are going to disappear then we will all be looking for a replacement. Or just a usable small coil that can be used with a driver (my preference).

 

[dsage]

Thanks for the kind words. The design of the Sage/Gedde driver module has impressed me greatly. The hallmark of a good design lies in its simplicity. I'm just now trying out some of that yellow transfer paper to bond the resist pattern to some blank boards to build two or three for myself. We'll see how that works.

Have fun. It should be fine.

 

[petertha]

Following along. I'm going to have to cross this bridge eventually with multi-cylinder ignition (gasoline) but have almost mentally resigned myself to buying my way out of the problem with a CH ignition box, mostly due to lack of electronics intelligence. But I wanted to throw these links out FWIW to discussion. I'm not sure if Ken followed some of the Bob Shores methods or modified on his own. I have Bobs book (Ignition Coils & Magnetos in Miniature) but most of it is over my head and/or ambition level.

http://modelicengine.la.coocan.jp/index english.htm
http://modelicengine.la.coocan.jp/c060701.htmhttp://modelicengine.la.coocan.jp/k080401.htmhttp://modelicengine.la.coocan.jp/k080402.htmhttp://modelicengine.la.coocan.jp/j5171201.html

 

[peterl95124]

This is all preliminary stuff, but I summarized the work on CDI modules in a short report, attached below. I'm looking for critique on all this stuff. Let me know what you think.

Don

Don,
after looking at the spark plug waveform in your paper, its so noisy that I'm surprised you have any confidence in any numerical integrations based on it. I started thinking perhaps an alternative would be to measure the post spark ring down energy and subtract that from the original resistive load energy, I don't know if you can do that without knowing the internal resistance of the circuit that's ringing down, you can measure the coils secondary resistance easily enough, but that's not everything if its coupling to a circuit in the primary)

my own take on CDI is that its an off-the-shelf solution for hobby machinists that aren't also hobby electronics tinkerers, but that there are no other advantages (except maybe for RC you can get away with a smaller lighter coil). In theory CDI could fire faster than induction discharge ignition, and that might be helpful for folks building multi-cylinder engines, except that these units probably weren't designed for it, so it would be interesting for you to try to measure their maximum sustainable(!) firing rate.

lastly I think the most important bit of information in the article is that a smaller spark gap results in less energy delivered to the spark (at least if we believe your measurement technique), because I've been making really small spark plugs with .015 gap for my RR Merlin V12 and maybe that's not enough for reliable ignition, although it seems to run AOK anyway (for which I'm using two Sage-Geddy electronic ignition circuits one for each bank of six cylinders).

anyway, I love it when people do try to use science and technology to understand what's going on in our hobby, keep up the good work !!!

Pete.

 

[dsage]

I'm from the old school of being able to see the waveform have a defined HV peak followed by a horizontal line burn time where you can see (and analyze) the actual energy conducting in the fuel mixture. Ignition waveform analysis for coil ignitions are readily available on the internet. Such as:
https://www.motor.com/magazinepdfs/052005_04.pdf
With the ignition waveform on a scope you can analyze an awful lot about the operation of the engine.

The wacky CDI waveforms don't give me any confidence that much happens other than a short spark.
This is why CDI systems have fallen out of favor. If they were any good car manufacturers would be using them. Instead they have gone to one coil per cylinder (with some exceptions). The key to a good ignition is a long spark full of energy (sort of an arc) not a short hot spark.
The other pit fall of a CDI is the repetition rate for high revving multi-cylinder engines was hard to obtain because it takes too long to recharge the capacitor. Reliability was not there.
IMHO the energy output from a CDI is too short lived. Hot maybe. But if it doesn't hit first try then it's game over.
I think effort is better spent finding a suitable small coil that can be driven by a coil driver.
The only thing CDI has going for it is possibly the size of the unit. But again if time was spent on trying and analyzing (as Don has done) a few readily available, small and cheap coils I think the results would be better.

 

[DKGrimm]

Don,
after looking at the spark plug waveform in your paper, its so noisy that I'm surprised you have any confidence in any numerical integrations based on it. I started thinking perhaps an alternative would be to measure the post spark ring down energy and subtract that from the original resistive load energy, I don't know if you can do that without knowing the internal resistance of the circuit that's ringing down, you can measure the coils secondary resistance easily enough, but that's not everything if its coupling to a circuit in the primary)

my own take on CDI is that its an off-the-shelf solution for hobby machinists that aren't also hobby electronics tinkerers, but that there are no other advantages (except maybe for RC you can get away with a smaller lighter coil). In theory CDI could fire faster than induction discharge ignition, and that might be helpful for folks building multi-cylinder engines, except that these units probably weren't designed for it, so it would be interesting for you to try to measure their maximum sustainable(!) firing rate.

lastly I think the most important bit of information in the article is that a smaller spark gap results in less energy delivered to the spark (at least if we believe your measurement technique), because I've been making really small spark plugs with .015 gap for my RR Merlin V12 and maybe that's not enough for reliable ignition, although it seems to run AOK anyway (for which I'm using two Sage-Geddy electronic ignition circuits one for each bank of six cylinders).

anyway, I love it when people do try to use science and technology to understand what's going on in our hobby, keep up the good work !!!

Pete.

I did try to do something like what you described in you first paragraph, and you are right about not knowing enough about the internal circuits to understand all the losses. The unknowns are stacked too deep. I got some ridiculous answers trying to do that.

Sampling the waveform every 20 nsec gives a pretty good base for tracking the energy even through the high frequency oscillations. It's really important that the relative phase of the voltage and current in these ringing fluctuations be sampled together. I was encouraged to find a pretty consistent range of results for most all the modules tested, suggesting that the method was at least in the ballpark. The worst problems came when the initial input spike drove the scope amplifiers into saturation and made major biases in the data during the recovery times. I had to keep the gain down to avoid that, so sometimes the voltage and current data was down in the mud of quantization errors in the sampling.

I had a conversation with Paul Knapp about plug gap. His experience was that sometimes increasing the gap actually led to less reliable engine firing. I think it's true that large gap can absorb somewhat more of the CDI energy, but under compression it may be that the pulse voltage from a CDI is not high enough to fire a large gap consistently. Maybe others might have experience with comparing different gaps under different conditions.

 

[DKGrimm]

Following along. I'm going to have to cross this bridge eventually with multi-cylinder ignition (gasoline) but have almost mentally resigned myself to buying my way out of the problem with a CH ignition box, mostly due to lack of electronics intelligence. But I wanted to throw these links out FWIW to discussion. I'm not sure if Ken followed some of the Bob Shores methods or modified on his own. I have Bobs book (Ignition Coils & Magnetos in Miniature) but most of it is over my head and/or ambition level.

http://modelicengine.la.coocan.jp/index english.htm
http://modelicengine.la.coocan.jp/c060701.htmhttp://modelicengine.la.coocan.jp/k080401.htmhttp://modelicengine.la.coocan.jp/k080402.htmhttp://modelicengine.la.coocan.jp/j5171201.html

I spent some time looking at all the wonderful machine work in your attachments. As for the problems you are having with oscillations in the two ignition modules, it is not obvious what is going on there. One thing occurs to me: The two modules may be triggering each other by coupling through the common power supply (shared battery). It would be relatively east to try a separate battery for each module to see if the oscillations go away.

 

[Zeb]

I was looking at this moped website where the IC would read a discrete input for the crank position (hall+magnet) and fire the trigger based on a mapped RPM. Not sure if any of you analog gurus had thoughts on their SCR circuit.

The nice thing about using the IC as the trigger, is that you can map and advance the timing on start or run injectors. I don't think their arduino could handle the refresh rate without borrowing from a separate clock.

Dumb cart before the horse question...
Other than going to dual ignition, is there a practical way to stack capacitive outputs through a rectifier, so the spark(s) is more evenly distributed?

 

[william_b_noble]

Don,
after looking at the spark plug waveform in your paper, its so noisy that I'm surprised you have any confidence in any numerical integrations based on it. I started thinking perhaps an alternative would be to measure the post spark ring down energy and subtract that from the original resistive load energy, I don't know if you can do that without knowing the internal resistance of the circuit that's ringing down, you can measure the coils secondary resistance easily enough, but that's not everything if its coupling to a circuit in the primary)

my own take on CDI is that its an off-the-shelf solution for hobby machinists that aren't also hobby electronics tinkerers, but that there are no other advantages (except maybe for RC you can get away with a smaller lighter coil). In theory CDI could fire faster than induction discharge ignition, and that might be helpful for folks building multi-cylinder engines, except that these units probably weren't designed for it, so it would be interesting for you to try to measure their maximum sustainable(!) firing rate.

lastly I think the most important bit of information in the article is that a smaller spark gap results in less energy delivered to the spark (at least if we believe your measurement technique), because I've been making really small spark plugs with .015 gap for my RR Merlin V12 and maybe that's not enough for reliable ignition, although it seems to run AOK anyway (for which I'm using two Sage-Geddy electronic ignition circuits one for each bank of six cylinders).

anyway, I love it when people do try to use science and technology to understand what's going on in our hobby, keep up the good work !!!

Pete.

there are a couple of much cheaper solutions, you may go to my web site (wbnoble.com) and look for my discussion of a CD ignition for my older Porsche - probably cost me $5 in parts to build. Or if you ask I will pass on info to contact a guy who has designed a new CD ignition for older sports cars that is more modern and probably easier for those who are less electonically inclined. the article on my web site supposes you have at least a clue. I used a scrap 400 Hz transformer, an oil cap, and a cheap SCR. I had a version with more parts, but it didn't work any better.

 

[DKGrimm]

I was looking at this moped website where the IC would read a discrete input for the crank position (hall+magnet) and fire the trigger based on a mapped RPM. Not sure if any of you analog gurus had thoughts on their SCR circuit.

The nice thing about using the IC as the trigger, is that you can map and advance the timing on start or run injectors. I don't think their arduino could handle the refresh rate without borrowing from a separate clock.

Dumb cart before the horse question...
Other than going to dual ignition, is there a practical way to stack capacitive outputs through a rectifier, so the spark(s) is more evenly distributed?

Not sure how to approach that one. Do you have an idea of how you would do it?

 

[DKGrimm]

there are a couple of much cheaper solutions, you may go to my web site (wbnoble.com) and look for my discussion of a CD ignition for my older Porsche - probably cost me $5 in parts to build. Or if you ask I will pass on info to contact a guy who has designed a new CD ignition for older sports cars that is more modern and probably easier for those who are less electonically inclined. the article on my web site supposes you have at least a clue. I used a scrap 400 Hz transformer, an oil cap, and a cheap SCR. I had a version with more parts, but it didn't work any better.

I looked at your site. Looks like you have a CDI design that works. Hard to say if it gets more energy into the spark without building and testing it.

 

[dsage]

I guess it comes down to how much energy is enough. If you look at commercial units such as MSD. They are not small units and - don't forget they still drive a standard ignition coil or one of their super coils.
Inside you will find a very healthy high voltage supply of about 450 volts with a fairly large transformer and a lot of electronics. They specify the output joules of their various units so some reverse math can be applied.
Certainly one of those is going to kill a model engine but maybe an appropriate scale down can be achieved.
The (older un-potted) S/S modules are easily reverse engineered. The power supply can be measured, the capacitor measured and the coil measured. It's all very low energy stuff.
I'm sure components could be chosen to make a happy median between those two products. If that's what is required.
I've started on a few in the past few years but it quickly becomes a cost vs size vs benefit exercise so I've put them aside. And just use the S/S units. If the S/S modules disappear then hopefully someone will come up with one a bit more powerful at a reasonable cost.
Size is my main driving force. And, as I mentioned, it's why I don't even use my coil drivers much. The coil is too large.

 

[DKGrimm]

I guess it comes down to how much energy is enough. If you look at commercial units such as MSD. They are not small units and - don't forget they still drive a standard ignition coil or one of their super coils.
Inside you will find a very healthy high voltage supply of about 450 volts with a fairly large transformer and a lot of electronics. They specify the output joules of their various units so some reverse math can be applied.
Certainly one of those is going to kill a model engine but maybe an appropriate scale down can be achieved.
The (older un-potted) S/S modules are easily reverse engineered. The power supply can be measured, the capacitor measured and the coil measured. It's all very low energy stuff.
I'm sure components could be chosen to make a happy median between those two products. If that's what is required.
I've started on a few in the past few years but it quickly becomes a cost vs size vs benefit exercise so I've put them aside. And just use the S/S units. If the S/S modules disappear then hopefully someone will come up with one a bit more powerful at a reasonable cost.
Size is my main driving force. And, as I mentioned, it's why I don't even use my coil drivers much. The coil is too large.

Dave, you've hit the nail on the head with your comments.

I've been grappling with this question about what could be done to get more of the available energy from the CDI capacitor into the spark gap. I can't help but think of trying to match a radio transmitter to a very low impedance and very reactive antenna. It takes a matching network of some kind. That matching network will be a combination of capacitors and inductors exchanging energy back and forth between themselves to transform the output power into high current and low voltage while neutralizing the reactive load component. OK . . . I get that this is a little far afield, but the analogy is thought provoking.

The voltage I measure across a 0.025 spark gap is just in the order of 100 to 200 volts when it's carrying a half an amp or so. The only way to get more energy into the plug is to stretch out the pulse over a period of time longer than the 40 microseconds or so that these CDI modules produce. I'm starting to think that the real difference between coil ignition and CDI ignition may well be that CDIs use smaller coils that can't store all of the energy from the capacitor and meter it out over a longer time.

Like you, I have used S/S CDI modules for most of my models. Not overly expensive, and mostly trouble free on an engine that doesn't need a really hot spark. But my Atkinson Differential engine would not fire at all until I hooked up a full Kettering automobile system with a 0.045 gap plug. And every engine where I have tried a higher energy ignition system has run better, leaner, cooler, and more reliably. I don't know the answers to any of this yet, but I'm having fun looking to learn something.

Don