Tech: The 944 / 951 Ignition System.

[Rogue_Ant]

The automotive ignition system is commonly one of the least understood parts of the powertrain. Unlike mechanical operation, it can be difficult to visualize exactly what is required to successfully fire the spark-plugs.

To make things more difficult, doing any quantitative testing of the ignition system requires specialized tools that very few people posses.

That said, I have been doing a lot of ignition system, and ignition coil testing lately, and (as always) I would like to share what has been learned.



To get started, let's go over the ignition system.

The ignition coil:



This is the Bosch "black coil", which is the OEM unit for both the 944 and 951. The purpose of the ignition coil is two-fold. Firstly, the coil has a primary and secondary windings. By having many more turns on the secondary winding than the primary, the coil acts as a "step-up" transformer. This steps-up the voltage to many thousands of volts - enough to fire the spark-plug. The secondary purpose of the coil is to store the energy needed to actually fire the spark-plug, and to keep the spark existing for a short time (typically less than 1/1000th of a second).

In the 944 & 951 (and nearly any electronically-controlled ignition), the coil is supplied with a constant +12 volts whenever the ignition is turned on. The DME controls the "ground-path", by a large transistor (the silver colored can-style transistor in the DME). To operate the ignition coil, it first must be charged, which the DME does by turning on the transistor, providing a ground for the coil - completing the circuit. Then, once the coil is charged, the DME turns-off the transistor, which opens the circuit causing the coil to fire. (yes there are more details, but they are not necessary for this thread - feel free to search/learn more about coils/inductors)

This digital (the transistor) control gives the DME very precise control of when the actual ignition event (spark) happens.



When the DME turns on the transistor controlling the ignition coil, current (amps) starts moving through the coil. Due to the property of coils, they resist changes in current flow (amps). So, initially when the transistor is first turned on, there is very little current flow, but as time increases, the current flow rises. Generally as long as the coil is not saturated, this current flow rise is a linear relationship with time (and voltage). Measuring the current rise is not an easy thing to do (especially without altering the charging circuit). However, it can be done, and here is a graph of the stock ignition coil being charged:



The cyan/blue trace (CH2) is the digital logic which turns on the transistor for the ignition coil. The important data is how long the logic pulse is "low", which is the length of time that the DME is charging the ignition coil for. Here each horizontal division (block) is 1mS (milli-seconds), the DME is charging the coil for 9.78mS. The yellow trace (CH1) is the coil current. Each vertical division is roughly 1.5amps worth of current.

So, as you can see, when the DME first turns on the transistor to charge the coil, there is very little current flow. But, as time increases, the current through the ignition coil starts to rise. After a few milli-seconds, the current reaches a peak and will not rise anymore - about 9 amps in this case. This is actually due to the internal current-limiting circuitry inside the DME. (more on this in a bit)

As one might imagine, all things remaining equal, more current means more ignition energy. So, it is ideal that the coil receives as much current as possible, to result in as powerful of a spark as possible. However, lots of current through a transistor, wires and coil can heat-up and generally be hard on the components. Look again at the graph, peak current is reached well in advance of the total charging-time. In-fact, peak current is reached nearly 3mS before the end of the charge-cycle. During these final 3mS, the coil is not gaining any energy, rather it is needlessly conducting current, which can heat-up the coil, wires, and DME transistor... So, what should we do? This:



Now the coil is still receiving the same peak current flow (9amps), but it spends no time holding at this peak current. If you inspect the blue trace data we see that the charge time is now 6.46mS (down 3.1mS from the previous graph). This is an ideal charging setup - we hit the maximum energy possible, and minimize the impact to the coil, wires, and DME transistor.



Now that we understand basic coil charging, and understand that it takes roughly 6.5mS to fully charge the factory ignition coil, the next thing to think about is how much time is actually available to charge the ignition coil. In the 944, one ignition coil must fire all 4-cylinders, which means there needs to be two ignition events per revolution. Therefore, as RPMs go up, the available time to actually charge the coils becomes shorter.

Here is a graph of the possible coil charge time, with actual ignition event time account for:



What should be peculiar is the fact that the possible charge time after roughly 4000rpm is less than the time needed in the earlier graph to reach peak current. Well, because there is less time available to charge the coil, the DME must reduce the amount of time it charges the coil for. By 6500rpm there is only ~3.6mS of time available to charge the coil:



A quick count of the divisions shows that at this shorter charge time, the coil is only reaching ~5.25 amps of current, which is nearly 4 amps less than the possible peak! Furthermore, coils store energy according to the calculation:

1/2 * Inductance * Current^2

Yes, that is current-squared. So, the amount of current is very important for the final coil energy value. Given our current data, we can see that at 6500rpm, our coil is down to ~34% of its peak possible value! Is it any wonder why we must run tight spark-plug gaps in order to prevent mis-fires.



So, what is the solution? Lets try a different ignition coil. The next coil I tested is the MSD Blaster coil:



This coil has nearly identical physical dimensions to the factory Bosch coil, which makes it an easy swap. Testing the blaster coil, we see this graph:



For this coil, we see that it takes 4.8mS to reach a peak current of ~9amps. This is a significantly shorter amount of time than the stock Bosch coils time of 6.5mS. So by using this coil it has time to fully charge until ~5200rpm - which is a significant improvement over the Bosch coils 4000RPM. Even though this coil charges faster than the Bosch unit, by 6500rpm it is also down on potential energy:



Since this coil charges faster, even at the short charge-time of 6500rpm, it has more current flow than the Bosch coil. We see it has ~7.5amps of current, which means this coil is only down to ~70% of its peak possible energy - a HUGE improvement over the factory Bosch coil (which was down to ~34% energy at the same RPM).



There must be a catch, right? Well, yes there is. Because the MSD coil charges faster, that means it needs less charge time to hit peak current. Remember the first graph, where the coil hit peak charge, and then maintained it for a significant amount of time? Well, that is exactly what will happen if we install the MSD coil without changing the charge time (known as dwell). The DME transistor, wires, and coil are fairly robust. So one could install the MSD coil without any other changes, and benefit from improved ignition energy, but this is needlessly hard on these components. What needs to be done is to change the charge-time data in the DME software (easy for us to do, and is user-available in the DME Tuner software).

(continued...)

[Rogue_Ant]

Even though the MSD Blaster coil is a huge improvement, I was not satisfied with the loss of ignition energy as RPMs increase. So, after a bit of research, I picked-up another MSD coil to test. This is the most powerful, inductive coil that MDS offers; the HVC-II:



This coil is obviously not the same physical dimensions of the factory coil. However, it utilizes newer coil design, and is not difficult to retrofit for use in the 944 / 951. Testing this coil:



Immediately, we see that this coil charges extremely fast! Even with a little bit of hold-time at the end of the charge cycle, this coil hits peak current in a short 3.3mS! This is an improvement of 1.5mS over the MSD Blaster coil, and nearly half the time of the factory Bosch unit! Furthermore, this coil will not run short on charge time until 7000RPM. This means that there is no ignition energy loss for the entire rev-range of the 944 / 951!

This is an awesome coil!

But, like the problem mentioned earlier with the MSD Blaster coil, this coil charges very fast, and if the ignition dwell map is not adjusted to account for the fast charging, the DME transistor, wires, and coil will suffer. Further, this coil is much more likely to damage the DME transistor or wires than the MSD Blaster coil - the dwell map MUST be changed to account for this coil.

Let me say this again:

DO NOT RUN THIS COIL WITHOUT CHANGING THE IGNITION DWELL MAP TO ACCOUNT FOR THE FAST CHARGE TIME.

Seriously - don't do it, you will kill the DME transistor (just a matter of time).


Now let us finish-up by revisiting the energy equation earlier:

1/2 * Inductance * Current^2

We now know current, but what about the inductance? Like current, measuring coil inductance requires a special tool - lucky that I have such a tool.

Inductance is, essentially, the energy storage potential of the coil. More inductance does mean more potential energy. And though it would be tempting to make a coil with as much inductance as possible, inductance also adds to the overall coil impedance; higher impedance means slower charging. So, it is a bit of a juggling act of coil inductance vs charging rate.

That said, the stock Bosch coil has an inductance of 5.85 mH (milli-Henries). Now we can complete the calculation for peak energy (at peak current):

1/2 * 0.00585 * 9^2 = 237mJ (milli Joules)

This is actually quite a bit of energy! But remember that by 6500RPM, the Bosch coil's current has been significantly reduced:

1/2 * 0.00585 * 5.25^2 = 81mJ

Wow! That is a huge reduction in coil energy from the peak potential...



The MSD Blaster coil has an inductance of 4.36 mH, which is less than the Bosch unit, but since the MSD Blaster charges quite a bit faster, will there be an overall improvement?

First the peak energy (at peak current):

1/2 * 0.00436 * 9^2 = 177mJ

Interesting that the MSD coil actually has less energy at peak current, but lets see it as RPMs increase; here at 6500rpm:

1/2 * 0.00436 * 7.5^2 = 123mJ

Even though this coil doesn't have quite as much peak current energy as stock, it maintains a much higher level of energy as RPMs increase. By 6500rpm, the MSD Blaster coil is 50% more powerful than the stock Bosch unit.



Finally, the MSD HVC-II coil. This coil has an inductance of 6.23, which for how fast it charges, is a lot of potential! So, the coil at peak energy (peak current):

1/2 * 0.00623 * 9^2 = 252mJ*

WOW!!! This is by far the most powerful coil tested. AND remember that this coil does not lose any energy for the entire rev-range. So by comparison, this coil at 6500rpm has twice as much energy as the MSD Blaster coil, and three times as much energy as the factory coil !!!


*The coil was starting to saturate, so it will most likely have a little less energy than the linear calculation predicts*

[Eric_Oz_S2]

Joshua

If the coil is charged with a 'constant 12v', why does the dme mapping adjust dwell for rpm AND battery voltage? Does the dme include a voltage regulator to cap the voltage to 12v?

[Rogue_Ant]

Originally Posted by Eric_Oz_S2

Joshua

If the coil is charged with a 'constant 12v', why does the dme mapping adjust dwell for rpm AND battery voltage? Does the dme include a voltage regulator to cap the voltage to 12v?

Perhaps that was a poor choice of words on my part.
The coil is supplied system voltage with the key on. So ideally this is 13.8 while the engine is running, but could be much lower (such as when cranking), or even higher if your voltage regulator isn't quite right.
This is why the DME adjusts dwell depending on both RPM and system voltage - it must cover the entire dynamic range of possible operation.

[John_AZ]

Thanks Rogue_Ant-again.

Bookmarked

EDIT--Some great discussion on this thread:
http://forums.rennlist.com/rennforums/944-turbo-and-turbo-s-forum/740107-tech-the-944-951-ignition-system.html

J_AZ

[Gawernator]

Very informative!

[Ederd]

Wish I had this info a month ago. Thanks

[Aussie944S]

I read this understood like a few words such as coil, spark and the letter A..... but I'm keen on improving my ignition system and was planning on upgrading my coil to the msd blaster but from what I can understand I need to fix the "dwell" in the DME.... this I have no idea how to do... can it be done at home.
Car details
1988 944S, Max HP chip in DME.

[Rogue_Ant]

Originally Posted by Aussie944S

I read this understood like a few words such as coil, spark and the letter A..... but I'm keen on improving my ignition system and was planning on upgrading my coil to the msd blaster but from what I can understand I need to fix the "dwell" in the DME.... this I have no idea how to do... can it be done at home.
Car details
1988 944S, Max HP chip in DME.

The only way to adjust the dwell is by changing the appropriate information inside the DME (on the chip / EPROM).

The Blaster coil isn't that significant of a switch, so you will probably be fine without changing the dwell - it just won't be ideal.

[VirginiaF1]

Originally Posted by Rogue_Ant

Given our current data, we can see that at 6500rpm, our coil is down to ~34% of its peak possible value!

Amazing post.. So many thanks from an eager novice.

So does the 66% reduction account for my engine fighting to rev beyond 4,500 rpm..?
If we assume for a theoretical moment that vacuum issues are not part of the problem.

(Specifics: Shell 93 AKI/Octane, New 2013 Bosch OEM Coil, Beru wire set and distr. & WRDC07+ sparks; 2-year old timing belt; ..)

TIA!

[Rogue_Ant]

Originally Posted by VirginiaF1

Amazing post.. So many thanks from an eager novice.

So does the 66% reduction account for my engine fighting to rev beyond 4,500 rpm..?
If we assume for a theoretical moment that vacuum issues are not part of the problem.

(Specifics: Shell 93 AKI/Octane, New 2013 Bosch OEM Coil, Beru wire set and distr. & WRDC07+ sparks; 2-year old timing belt; ..)

TIA!

Well, the stock system is definitely enough to do the job on a stock car. And we know it will work even with modified cars - however, that doesn't mean it is ideal, and at some point it is not acceptable at all (in-consistent ignition, misfiring, ect).

The 944NA has lower ignition energy requirements than the 944 Turbo, so I doubt the coil charge is the reason for your car's issue...

[VirginiaF1]

Fair enough.. Thanks.
Just trying to understand if/how or how not your post applies in my old girl's case.

Plenty of other things to be corrected then in order to get another 257k out of her..
Cheers!

[guru944]

Rouge Ant:
Thanks for the technical feedback.
Great graphics and discussion.
I'm using the Nology PFC-M75 for many years with excellent performance with my MSD6BTM.
Keep us posted.
Later...

[skyhawk5421]

Are the changes that need to be made possible with an ostrich by someone less experienced than yourself? The 944 at the local junkyard has one of those coils, probably what landed it there.