Tests of CDI Ignition Modules

[willray]

I went back and looked at the video again. My module operates differently from his. Mine puts out individual sparks with spaces between them repetitively while it is activated. His produces a more or less a continuous arc while activated.

I'm not confident that the appearance of a continuous arc in his video is actually what's happening. I believe we are hearing the spark frequency in the video, and it sounds like not more than a couple hundred hertz to me.

 

[dsage]

I'm sure you are correct.
My description was meant to say that mine gives out short bursts of what he exhibits as a continuous stream.
My short bursts are at maybe 4 per second rather than a continuous stream like he has.

 

[74Sprint]

an observation about that schematic - it is very close to the Delta Mk 10 schematic from the early 70s (late 60s?). Delta used germanium transistors for low voltage drop, in my design I used silicon, today's solution would be MOSFETS. since whatever you use is driven hard into saturation or hard off, the actual dissipation is fairly small so I'm not sure it matters. Delta used a torroidial transformer, today wiht higher frequency parts, a ferrite E-core would probably be fine if not better//

I totally agree.

 

[74Sprint]

My go-to troubleshooting steps were:
1. P-Lead shield grounding. 2. Megger 3. Ohm ranger expeditions

Another shelved design is the shower of sparks vibrator circuits. The circuit had its own set of points. Might be interesting to see it brought back to lengthen the spark window, but that would be another project entirely.

Working on it.

 

[peterl95124]

Ray,
my questions were for the people doing the evaluations and comparisons between CDI and LDI,

the energy in a capacitor is 1/2 C V^2, the energy in an inductor is 1/2 L I^2, these are computable,
measurable, and knowable, but no one seems to be doing it.

the resonant frequency of an LC circuit is 1 / 2 PI sqrt(L C), again this is computable, measurable,
and knowable, and while this is sometimes being observed with a scope it isn't being compared
across designs.

so the questions still are, 1) why do CDI seem to generate short sparks, and 2) why do CDI seem
to deliver only a fraction of their energy to the actual spark (where does all the rest of the energy go)

we need to have a handle on the absolute minimum basic question like how much energy are we
starting with when comparing CDI to LDI, and how do the ring down frequencies compare, and think
about what else could be making a difference. I see circuit diagrams for CDI using SCR, and LDI
using IGBT, but they are both 4-layer devices and the difference seems to be in the gate rather than
in the power diode, so I'd like to better understand the difference in power handling rather than
what sort of trigger they require. One thing that puzzles me is that neither design seems to have
a "free wheeling diode", so how does the primary ring down without an AC circuit.

another thing on my mind is from back in an earlier life when I toyed with tesla coils, there were
lots of people working on models for the resistance of a spark based on instantaneous and
average measurements of coil power, a complicated subject due to the non-linear, and even
negative, resistance of the spark. the question being is a spark plug arc similarly complicated.

Peter.

 

[dsage]

Sorry a poor explanation but in short you really don't want a free wheeling diode per se. It will dampen the coil output.
If you use a regular transistor it actually breaks down when the reverse energy from the coil is created. Sort of like what a free wheeling diode would do.
It doesn't really hurt the transistor but it can be as low as say 100v which limits the coil energy output.
When you use an IGBT the reverse breakdown is several hundred volts because of the zener etc inside the as shown below. So you can get a higher energy output from the coil.
Maybe I've misinterpreted your question.

 

[DKGrimm]

my questions were for the people doing the evaluations and comparisons between CDI and LDI,

the energy in a capacitor is 1/2 C V^2, the energy in an inductor is 1/2 L I^2, these are computable,
measurable, and knowable, but no one seems to be doing it.

You hit the nail on the head in asking for a better understanding of where the energy goes and why does CDI seem to deliver less of it to the spark. I'm in the process of figuring that out in more detail, but I have some thoughts about just what may be going on.

First, the CDI: voltage on a capacitor depends entirely on its charge. To get energy out, you must draw current from it at whatever voltage it happens to have at the moment. Starts high, gets lower as it becomes discharged.

Spark voltage, on the other hand (after the arc has been struck and stabilized) is kind of fixed. In fact, it gets lower when you run more current through it. I'm testing that now, but my early data is showing maybe a mean value of 100 volts at 0.1 amp, down to about 50 volts at 0.5 amps for a CM6 plug with a 0.025 gap.

As you can see, the capacitor energy source and the spark plug energy sink don't match very well. The CDI step-up transformer has to make a high initial voltage to start the spark in the first place, but after it fires, the spark almost shorts out the transformer, so it makes a very short spark. There's likely more voltage drop across the winding resistance than there is across the spark itself.

For the LDI: current and voltage essentially switch roles. During the spark the inductor voltage will be the same as the spark voltage, and the rate of current decay is proportional whatever the voltage happens to be at any given instant. The voltage can be high to start the spark and can be low during the discharge. "Shorting out" the output of an inductor PRESERVES its stored energy rather than dissipating it. Thus, the spark can last as long as it takes to discharge the inductor.

I'm well aware that this explanation glosses over whole lot of important details, but I think this is the basic reason for the CDI/LDI difference.

Don

 

[dsage]

Perhaps another difference to consider. And probably another way to state what you have detailed Don::
For a given coil - say a full size ignition coil that has a heavy core:
A CDI simply dumps what voltage is on the (small) capacitor into the coil. It happens in a split second because the coil primary is on the order of only 1ohm or so. The resultant pulse never "charges up" the core.
So the coil is being used more as a transformer not an energy storage device. CDI is not taking advantage of the cores ability to store energy. In fact most model CDI's have a small HV coil with very little core and they are used simply as step up transformers.
Whereas:
On a LDI ignition system the coil is (by the ECU dwell time) purposely "charged up" until the core is (almost) saturated for up to a millisecond and then de-energized. The power to maintain a long spark (arc) comes from the core of the coil being initially energized by 12volts and 10amps (or so) and then that energy is returned when the field collapses. Then the transformer action takes place with that mount of stored energy not a quick pulse.

I can't do the math on it but dumping a 1uf capacitor at 300v into a 1 ohm load will probably not "charge up" the core of a typical coil. Whereas holding 12v and 10amps on the primary and charging up the core and then releasing it probably returns most of that energy to the secondary as high voltage and high current.

Applying 300volts to the coil primary will certainly give you a much larger output voltage than the 12v LDI system given the same coil step up ratio,
But with a CDI the amount of output time will only be as long as it takes to discharge the capacitor. Nothing is returned by the core of the coil as with an LDI.

I posted a link to understanding LDI ignition waveforms some time ago. The key is the "Burn Time" where there is an actual plasma arc across the plug for a millisecond or more. Not a short spark.
The LDI waveforms show an initial hv spike until the mixture is ignited and then a horizontal line at some voltage dictated by the conditions in the cylinder. The horizontal line is on the order of a millisecond or so long and is a plasm arc. Eventually as the coil runs out of energy the waveform drops off and oscillates as the arc is extinguished and the remaining coil energy has nowhere to go so it bounces back and forth in the coil.
A ton of information can be gleened from looking at the burn time on a scope. Rich mixture, lean mixture, turbulance in the cylinder, spark plug gap, wire condition, even if the spark ocurred outside the cylinder (like in a bad coil wire). It's a real science to analyze and understand the waveforms. It's worth looking at.
A CDI "spark" tells you nothing.

 

[DKGrimm]

Perhaps another difference to consider. And probably another way to state what you have detailed Don::
For a given coil - say a full size ignition coil that has a heavy core:
A CDI simply dumps what voltage is on the (small) capacitor into the coil. It happens in a split second because the coil primary is on the order of only 1ohm or so. The resultant pulse never "charges up" the core.
So the coil is being used more as a transformer not an energy storage device. CDI is not taking advantage of the cores ability to store energy. In fact most model CDI's have a small HV coil with very little core and they are used simply as step up transformers.
Whereas:
On a LDI ignition system the coil is (by the ECU dwell time) purposely "charged up" until the core is (almost) saturated for up to a millisecond and then de-energized. The power to maintain a long spark (arc) comes from the core of the coil being initially energized by 12volts and 10amps (or so) and then that energy is returned when the field collapses. Then the transformer action takes place with that mount of stored energy not a quick pulse.

I can't do the math on it but dumping a 1uf capacitor at 300v into a 1 ohm load will probably not "charge up" the core of a typical coil. Whereas holding 12v and 10amps on the primary and charging up the core and then releasing it probably returns most of that energy to the secondary as high voltage and high current.

Applying 300volts to the coil primary will certainly give you a much larger output voltage than the 12v LDI system given the same coil step up ratio,
But with a CDI the amount of output time will only be as long as it takes to discharge the capacitor. Nothing is returned by the core of the coil as with an LDI.

I posted a link to understanding LDI ignition waveforms some time ago. The key is the "Burn Time" where there is an actual plasma arc across the plug for a millisecond or more. Not a short spark.
The LDI waveforms show an initial hv spike until the mixture is ignited and then a horizontal line at some voltage dictated by the conditions in the cylinder. The horizontal line is on the order of a millisecond or so long and is a plasm arc. Eventually as the coil runs out of energy the waveform drops off and oscillates as the arc is extinguished and the remaining coil energy has nowhere to go so it bounces back and forth in the coil.
A ton of information can be gleened from looking at the burn time on a scope. Rich mixture, lean mixture, turbulance in the cylinder, spark plug gap, wire condition, even if the spark ocurred outside the cylinder (like in a bad coil wire). It's a real science to analyze and understand the waveforms. It's worth looking at.
A CDI "spark" tells you nothing.

Yep. Everything you say fits with testing and experiments I have done at one time or another. I've seen some engines that can run with as little at 100 microseconds of full-blown plasma arc, but that's really marginal.

 

[DKGrimm]

A CDI simply dumps what voltage is on the (small) capacitor into the coil. It happens in a split second because the coil primary is on the order of only 1ohm or so. The resultant pulse never "charges up" the core.
So the coil is being used more as a transformer not an energy storage device. CDI is not taking advantage of the cores ability to store energy. In fact most model CDI's have a small HV coil with very little core and they are used simply as step up transformers.

This dual role of the coil/transformer is the most challenging to analyze. I don't have a handle on it yet. I'm about to do a little reading on the design of pulse transformers. It's a blind spot in my experience. There may be something to learn there.

 

[peterl95124]

Wikipedia "Capacitor Discharge Ignition"
echos the claim that CDI produces a short duration spark but with no explanation

says that the spark resistance drops a lot once the arc starts, and with 4000-ohm
secondary coil resistance half of the energy goes into heating the coil, which begs
the question how is this any different from LDI (my moped coil measures 5600-ohm
which is very similar)

 

[peterl95124]

Sorry a poor explanation but in short you really don't want a free wheeling diode per se. It will dampen the coil output.
If you use a regular transistor it actually breaks down when the reverse energy from the coil is created. Sort of like what a free wheeling diode would do.
It doesn't really hurt the transistor but it can be as low as say 100v which limits the coil energy output.
When you use an IGBT the reverse breakdown is several hundred volts because of the zener etc inside the as shown below. So you can get a higher energy output from the coil.
Maybe I've misinterpreted your question.

Dave,

I was going on the fact that with points based (non-transistorized) timing there's always a capacitor across the points and the effect is the primary is an LCR circuit with a fundamental ring down frequency, so I was (incorrectly!!!) assuming the spark duration had something to do with this ring-down (mis applied Tesla Coil knowledge).

but I just looked at Bob Shores book with sketches of O-scope waveforms and realized that the arc duration was a fraction of the first half period of the ring down, and the subsequent ring down was a small amount of residual energy after the voltage gets to low to maintain the spark.


your comment about SCRs and IGBTs breaking down under reverse voltage, but at very different voltages, got me thinking about Hall Effect sensors getting destroyed. I'm a firm dis-believer in the theory that if you twist the leads to the sensor this will reduce interference to it and eliminate sensor death, instead I believe it is the severe reverse voltage spike caused by not having a points capacitor in transistorized systems that lead to sensor death. thoughts ? (I have added filter and zener elements to the hall supply voltage as recommended in an App Note for automotive use of the sensor I'm using, according to my scope they don't do much, but I haven't burned a sensor since I started adding them).

Peter.

 

[dsage]

In my experience most folks having issues with blowing sensors is because of poor wiring practice. There MUST be SEPARATE heavy and SHORT connection from the engine block DIRECTLY back to the battery negative. The biggest error is to figure it is sufficient to have a round-about ground connection from the engine back to the battery.
The hall sensor must have all three (power,ground and signal) wires un-interrupted from the driver board to the sensor as short as possible and floating. The hall sensor should never be grounded at the engine. The sensor should also be arranged so that it is well insulated from the engine. This is most difficult because the hall sensor is usually inside a distributor where it is in close proximity to the spark. It's best if the sensor can be in a separate compartment in the distributor.
You have to realize that the extremely high spark energy will be looking for a path back to the battery negative. If it can find a shortcut through the hall sensor wiring it will take that path.
Misfires because of fouled spark plugs, wires, distributor etc can cause big issues. In these cases the spark never gets to jump the sparkplug gap so the voltage rises to it's max and will be looking for other paths back to the battery other than the spark plug. Inside the distributor where the hall sensor ground connection is close by it may choose this ground path instead.
These are the reasons I try to mount my ignition systems hidden under the engine with the battery. Connections are never more than a few inches long. I've never had any issues with blowing sensors.
If you have a look at the schematic for the Sage/Gedde ignition module we recommended an Optek Hall sensor #OH090U. Digikey part number 365-1001-ND.
John Gedde found that this particular sensor is less prone to blowing. Not sure why - maybe because of how the circuitry is inside the chip. Obviously others will work but John found that without making any other changes in a situation where sensors were blowing these stood up.
Obviously making the changes listed above is a must for trouble free operation.
Yes the addition of zeners and filter caps etc can be a helpful bandaid but eliminating the source of the issues must be first.
Enough said.

 

[peterl95124]

In my experience most folks having issues with blowing sensors is because of poor wiring practice. There MUST be SEPARATE heavy and SHORT connection from the engine block DIRECTLY back to the battery negative. The biggest error is to figure it is sufficient to have a round-about ground connection from the engine back to the battery.
The hall sensor must have all three (power,ground and signal) wires un-interrupted from the driver board to the sensor as short as possible and floating. The hall sensor should never be grounded at the engine. The sensor should also be arranged so that it is well insulated from the engine. This is most difficult because the hall sensor is usually inside a distributor where it is in close proximity to the spark. It's best if the sensor can be in a separate compartment in the distributor.
You have to realize that the extremely high spark energy will be looking for a path back to the battery negative. If it can find a shortcut through the hall sensor wiring it will take that path.
Misfires because of fouled spark plugs, wires, distributor etc can cause big issues. In these cases the spark never gets to jump the sparkplug gap so the voltage rises to it's max and will be looking for other paths back to the battery other than the spark plug. Inside the distributor where the hall sensor ground connection is close by it may choose this ground path instead.
These are the reasons I try to mount my ignition systems hidden under the engine with the battery. Connections are never more than a few inches long. I've never had any issues with blowing sensors.
If you have a look at the schematic for the Sage/Gedde ignition module we recommended an Optek Hall sensor #OH090U. Digikey part number 365-1001-ND.
John Gedde found that this particular sensor is less prone to blowing. Not sure why - maybe because of how the circuitry is inside the chip. Obviously others will work but John found that without making any other changes in a situation where sensors were blowing these stood up.
Obviously making the changes listed above is a must for trouble free operation.
Yes the addition of zeners and filter caps etc can be a helpful bandaid but eliminating the source of the issues must be first.
Enough said.

Dave, so you haven't put your circuit on a scope and/or haven't observed the reverse voltage spike in the primary side ? Pete.

 

[dsage]

Yes I have. The waveforms are pretty much textbook. You can look it up anywhere. See below. They can vary a lot. But the basics are always there.
The primary side is a reflection of the what's going on in the secondary. On the primary a 12v signal level followed by a ground dwell time where the coil in energized, a release of the ground followed by a large positive spike before ionization. There is a negative spike right after the positive one but that is before ionization in the spark plug gap where the actual energy causes an arc. That is the voltage level I mentioned before where either the transistor or IGBT suppresses the coil. If you suppress that at too low a level you are taking energy away from the spark to follow. Which is why an IGBT is superior to a regular bipolar transistor (higher breakdown voltage)
Once the plasm arc is formed (burn time) the mixture acts like a resistance with a pretty steady discharge of the coil (all above ground) Following that is ringing and decay when the burn extinguishes. The ring is caused by left over energy in the coil insufficient to maintain the burn time and looking for somewhere to go.
If that's where you are seeing a large spike then I would suspect that perhaps your spark plug gap is too large the mixture is too lean and won't fire or any number of things causing the burn to extinguish early leaving a lot of un-used energy in the coil.
There should be just minimal oscillations at the end and they should decay quickly because the coil should be pretty much out of energy if it was used up properly during the burn time.
The whole burn time and ringing etc. is very much dependant on what's going on in the cylinder. For instance the steady state height of the burn time above ground is indicative of how hard it is to start and maintain the arc. A higher line might be a lean mixture or the sparkplug gap is too wide or other factors.
The rise at the end of the burn time (shown below) indicates that the mixture is becoming lean (voltage rises - harder to burn) as the mixture is burned. The roughness in the burn time is due to turbulance in the cylinder affecting the smooth burning of the mixture.
I have never seen really large oscillations after the burn time as shown below. I just grabbed that picture quickly from the internet. They may have been illustrating a fault.
I suggest you research the waveforms. There is a lot to be learned about the workings / health of your engine "system".

 

[dsage]

Here's a good video to watch. It explains a lot of things.

 

[rsholl]

Hi Don and Dave Sage,

I'd like to thank both of you for all the good advise and help you have given me over the years and also to the engine community for all of their support over the years. The main reason I'm retiring is the need to get started machining engine castings, and also my wife is having memory issues and is no longer involved with the CDI. That adds a lot of extra work trying to keep parts in stock and keeping up with orders.

Early on I did not have much luck finding anyone willing to take on the task of building my CDI as it is 100% hands on and hand assembly. Just recently I received an email from Dan Williams in New York and he is very interested in the CDI project so it looks like my CDI will not go by way of extinction. Dan is going to be at CF this year which looks like next week end. I unfortunately am not able to attend this year but am planning on next year. Dan has a degree in Electrical and Computer Engineering so should be a very good fit for continuing my CDI system. Dan is planning on bringing his Bruce Macbeth engine so he shouldn't be too hard to find, if you get a chance, stop by and chat with him.

Don, I have a lot of CDI systems floating around in my shop. Email me and we can figure out any that would work well with your testing. I am pretty sure I can find an older style CDI that I sold early on and was an Rcexl part which I modified, I also should have an original CH ignition unit and possibly an MJN.

I've been following this thread but was fighting the flu bug over the holidays and am just now getting back to what I consider my normal. Lots of good info on ignition systems and very interesting on what Don is trying to do with his testing. I have never been a big fan of the Kettering system for 2 reasons. First is the current required and second is the proper dwell time to get any kind of efficiency. Those 2 things the CDI doesn't suffer from. The most problematic part of the CDI is getting an efficient DC-DC converter to boost the 4.8 volts to at least 250 volts DC and be able to maintain at least 150 volts at normal operating speed. The next important item is the CAP and over my many years of experimenting it needs to be at least .27uf and for more all around systems about .47uf. The size of the CAP definitely has a major effect on the output. The spark coil I currently use was tested by a company asking for specific specifications which I could not supply. I sent them one to test and they placed an order. Some time later I received an email from one of the engineers with data from their testing. With 250 volts on the primary the coil was capable of an output voltage of 26 KV. The spark from a CDI is very fast and thin compared to the Kettering system. One of the biggest problems I've seen over the years is, I get a call that the CDI is not working. They do not see a spark at the plug with it out of the engine. It is very difficult to see a tiny spark at .010-.012 gap. The first thing I tell them to do is put the spark and ground wire end to end with at least 1/8" of gap and you will see and hear the spark. My final test for my CDI is to power it up with a modified extended tip 14mm spark plug that I cut the ground electrode off. It must provide a stable spark up to operating speed. I have done a lot of experimenting over the years and have fired almost every conceivable spark coil with the CDI. If you want real horse power hook one up to a standard car coil.

Don, I'm wondering how much different your readings would be if you were using a spark gap of 1/8" to 3/16" instead of the resistor or drop the resistor down to say 1K. The best test I found for power capabilities of any CDI was to side fire a standard 14mm spark plug with the ground electrode removed. If it could handle that spark plug gap it would fire any engine up to 150cc that was currently used for RC aircraft. I'll be watching with interest and know you will have more good info. I'm sure Dave will also. I do plan on staying active with the engine community. I don't know yet if Dan wants to maintain a web site or not but if the S/S (Roy's CDI ignition systems) site goes away you can contact me at ( [email protected] ).

Thanks everyone,
Roy Sholl

 

[DKGrimm]

Now that I finally have my test setups for both coils and CDI ignition going, I can certainly see why people have troubles blowing out Hall devices. I am seeing ringing transients in the wiring of almost unbelievable magnitude. I see voltage peaks above 100,000 volts and ringing frequencies over 50 MHz. It's making data taking kind of iffy because those spikes saturate the scope amplifiers and seriously distort the subsequent waveforms. The driver for this noise is the extremely rapid rise times of the current in the spark plasma when it fires. It's faster than I can measure, like tens of nanoseconds or shorter. I'm going to have to re-do my test setup to really establish a ground and shorten wires before I can trust the results. To be continued . . .

 

[DKGrimm]

I last posted about having troubles with noise saturation of the scope in making energy measurements on various CDI and coil-type ignitions I think I can report some progress on that front with a combination of better test setup, experience running the scope, and a better understanding of what I am really seeing in some of these tests. The best thing I’ve found so far is to use a resistor at the spark plug. It reduces noise pulses by about 10:1 so I can run more gain in the scope and get the signals up out of the noise.

I have some pictures attached with some sample results. The test is with an old Chevrolet coil with a Cage-Gedde driver from a 6.4 volt gel cell battery. The resistor eats about 20% of the coil energy, which is better than I expected. On page 1 you can see a pretty classic coil ignition spark discharge and you can see how the coil current decays during the spark. Page 2 shows the power that flows to the arc during the spark, and how that power accumulates to deliver the total energy. Page 3 shows the fine detail of the spark initiation with and without the resistor.

Take note of the scales on the figures on page 3. Time is in nanoseconds. There are at least two main ringing frequencies involved here, one up about 120 MHz. These are probably resonances of the inductance and capacitance of the spark plug and test lead wiring. It came as a bit of a shock to me (no pun intended) to see a peak voltage of 200 kv for the no-resistor case. The spark plug resistor knocks that down by over 10:1 by limiting the surge current when the arc first fires.

These measurements all seem to be repeatable, so I’m developing some confidence that I can go ahead with my plan to measure a bunch of CDI and coil systems and compare how they perform. The biggest problem I can see now is making sure I can keep track of what unit I am testing, where it came from, and weather it is available for others to use also. This is turning out to be a lot more work than I had planned when I started.

 

[DKGrimm]

Here's a picture of the test setup. The key item (besides the scope) is a homemade 1000:1 voltage divider sticking up in the middle that makes it possible to look at the high voltage at the spark plug itself. There's a 10-ohm current sampling resistor built into the spark plug mount.