www.neons.org

Steve Peterson and I performed testing and compiled the following information a month ago, and were going to wait until we completed some life testing on the Fairchild IGBT we found before we posted this info. However, this subject has come up, and we felt it beneficial to post this info now. We're still going to do some life testing, and will edit this post with those results.

As almost everyone is aware of, the VB921 ignition module has not been a stellar performer overall. Although it works fine for engines that do not demand a very hot spark, some applications that are more demanding (turbocharged engines mainly) experience problems with spark blow-out. People have then just cranked up their spark dwell to get more spark, and the result is a burnt VB921. It's very clear that a more robust solution is needed for demanding applications.

Overview

For those who don’t want to read this whole document: We have a new IGBT solution, it works well at 5 ms of dwell (which duplicates the stock Neon PCM output), and we can upgrade your RS Autosport ECU for you. The cost is $20 plus shipping. Please see the “Repairs and Upgrades” section of our catalog for detailed instructions on how to order.

Technical Details

A few months ago, we decided to start looking into other ignition drivers out there that we could use instead of the VB921. We found an ignition IGBT that has good specs, and we have done some fairly involved testing of this module.

The first objective was to compare the output of this module with that of the VB921 and the stock Dodge Neon PCM. The Neon’s stock ignition has been shown to work well with turbo systems, at least up to the point where the stock ignition map gets too advanced for the boost levels being run. We were able to exactly match the performance of the stock PCM by using our new IGBT module with a dwell setting of 5 ms. On an oscilloscope, both the PCM and the new IGBT exhibited the same dwell (5 ms) and the same peak coil charging current of 6 amps. The VB921 could achieve the same current, but at about 5.5 ms of dwell. We do know is that if a VB921 is operated over about 4ms of dwell for any duration, it will fail. This has been proven many times in peoples’ cars.

Following are three images of VB921 testing. The first shows the voltage across the emitter and collector of the transistor (Vce) at 5 ms of dwell, which shows a) the transistor turning on (voltage goes down), and b) the transistor turning off and the spark firing for about 2 ms. The second image shows the same information at 7 ms of dwell. Notice the pulses at the end of the transistor’s “on” period… this is the VB921’s current limiting at work. The third image shows the tail end of the coil charging current curve (10A per division) at 7 ms dwell. This image shows that as of about 6ms, the VB921 met its maximum current of 7A (it’s rated at 6.5-7.5A).



As we all know, the VB921's get pretty hot. How hot? Running for an hour at 5000 RPM and 5 ms of dwell in a v3 system and mounted to the heat sink, the bodies of the VB921’s were 190 F. Increasing the dwell to 5.5 ms was necessary in order to match the new IGBT's performance at 5 ms of dwell (6A peak current), and resulted in a module temperature of 240 F. This is also the performance level of the stock Neon PCM. These temperatures were measured at an ambient temperature of 80 F. Increase that ambient temperature to 140 F in a very hot car, and the modules would be at about 300 F. This could explain how some people have melted carpet with their VB921-equipped MegaSquirt ECUs.

Next, we wanted to see how much heat is created by the new IGBT during operation at 5ms of dwell. Running for an hour at 5000 RPM and 5 ms of dwell in a v3 system and mounted to the heat sink, the bodies of the IGBT’s were 144 F. The tabs were 148 F, and the heat sink was 137 F (using thermal interface pads between the tabs and the heat sink, not grease). There were no injectors running. Based on these measurements and the fact that this IGBT is rated to operate at 150 C (300 F), we feel that the amount of heat generated by these modules (about 1 watt each under these conditions) is very manageable and would not be an issue. Given the very small amount of heat dissipated (little energy wasted by heat) and the comparison of current and dwell measurements between this IGBT and the stock PCM, we conclude that this IGBT at least duplicates the performance of the stock PCM.

Shown below is a capture of one coil charging cycle. Disregard the different shape of the spark firing portion of the curve… this test was done on the bench, not on a running engine. The spark gap’s resistance changes drastically on a real engine.



Finally, we wanted to see what it would take to get the same spark energy as the MSD DIS unit. The advertised spark energy of this unit is 190 millijoules (mJ). Based on the measurements taken in this testing, the Neon coil appears to have an inductance of 7-8 mH (someone with an inductance meter should verify this). At 6 amps of current, the energy delivered is between 120 and 140 mJ, i.e. the stock Neon level. Assuming a 7.5 mH inductor, increasing the dwell to 6 ms (which increases coil current to 8A) delivers a little over 200 mJ of energy. We set the dwell to 6ms and the IGBT’s only heated up to 180 F, which is well within their range. Bump the ambient temperature up from the 80 F in the garage to 140 in a hot car, and the new temperature is 240 F. Specific testing would need to be done with low-z injectors in order to determine if the heat sinking in the v3 is sufficient when the injector flyback damping heat is added to the mix. Beyond the concerns for heat sinking the ignition modules, the coil gets pretty hot at this operating point. The temperature was at 200 F and still climbing. Our conclusion is that the stock coil can run at this setting just fine, but it may need to be done on an occasional basis, such as in drag racing. Another thing that might work well is boost-dependent spark dwell, so the dwell only increases to 6 ms if it’s needed based on a MAP sensor signal. That said, we hypothesize that since an MSD box drives the coil at 400V or so (thereby making its dwell time TINY in order to achieve 190 mJ of energy transfer), it will actually heat the coil more than the IGBT will when driving the coil at 12V. This is because the peak current is higher with the MSD, and heat due to resistive losses is proportional to the square of the current. For this reason, more heat is lost when transferring a given amount of energy into the coil since it’s done at a higher current. In other words, if a coil pack can run at 190 mJ spark energy on an MSD box without overheating, it should be able to do it more easily with a standard 12V coil drive. However, as this puts more heat into the coil pack than the stock PCM does, RS Autosport does not guarantee long-term reliability at greater than 5 ms of dwell.

Following is an image of a spark cycle at 6 ms of dwell at 7200 RPM. Section 1 is the firing of a spark (2 ms in duration), section 2 is the waiting time before the next coil charging even happens, section 3 is the coil charging period (6ms dwell time), and section 4 is the next spark firing event. As can be seen here, at 6 ms of dwell and 7200 RPM, there’s almost no extra time left (0.3-0.4 ms or so). If the motor spins much faster, either the charging dwell or the spark duration will be impacted and spark performance will suffer as a result, depending on how the firmware deals with this situation (we didn’t test it). Aside from this RPM limitation, this operating mode *should* match the MSD DIS unit’s performance, provided it’s not doing multiple spark discharges at high RPM.



At 5 ms of dwell and the resulting 2 ms of spark duration, a 4-cylinder wasted spark engine can rev to 8600 RPM without the “spark event window” closing in on the 7 ms total spark event. If stock Neon ignition energy output is sufficient for a given vehicle, this would be the maximum RPM at which this level of performance can be achieved. However, at 6ms of dwell, the maximum RPM where performance is not impacted is 7500 RPM. Herein lies the benefit of a coil-on-plug system… the allowable RPM is doubled, as the spark event window is doubled with respect to a wasted spark system.

All of this having been said, we do not consider ourselves to be experts on ignition systems. If anyone out there sees anything that we've said wrong or incompletely or that we're incorrectly drawing conclusions, by all means let us know! It would be much appreciated.

Retrofit Service

RS Autosport will offer a retrofit service if anyone would like to have these IGBT’s installed. The cost is $20, and we only work on units built by us. For now, these IGBT's are not offered as a kit. Shipping charges will be calculated at the time of ordering. The unit will be shipped back to you on the Monday following the day on which we received it.

steve, i'll be sending you an e-mail about throwing these in my box while you're handling that pesky injector problem

boost, we need a sticky!!

Wow, great writeup Rodney. I'm sad to say that I have these units installed and I've only been able to idle them in the garage. I promise to get some real results soon. I've been extremely busy with other BS lately.

So a few of these lil guys?

Arg and I just finally got my MS from you guys...

Just got done replacing some on one of your units for a local guy.
Wish we woulda had this info then...

Now its $225

Would you retrofit one of your units even though its been modified (only for idle control)?

This retrofit service is for anybody who has an RSA built Megasquirt. Warranty status doesn't come into play here.

As far as that goes we allow user modifications to the unit with the understanding that if the problem is directly related to those modifications then it's not covered.

Steve

Once I get my DD running so I don't have to drive the fast car, I'll be sending you mine.

??????

you get a new car `

Ok.. So I can't buy the IGBT's from you, I can't send my MS in (not that I'd want to, it's my only car right now), and you won't tell us what they are so that we could order them ourselves.. Hmm.. Ahh well, I'm not running boost. Great writeup though.

This isn't something we've gone out to MSEFI or wherever and ran with someone elses idea. We took the time to find a part, order test units and spent many hours testing. Should we then be expected to give the part number out?

Really, whether to give the part number out is something we've been struggling with. I think from the work we've put into it we should be able to keep the part number to ourselves. There are plenty of other IGBTs out there and they may work just fine. But this one has some testing behind it at least.

Steve

????

Steve

Increased price of new DIY kits just a joke

Hehe, totally lost on me since we don't sell DIY kits

Steve

Isn't the idea of sharing your work and experiences with others the whole bases of Megasquirt and why is has become what it is?

Either way, that is some very nice work. Do you have any opinions as to if the new parts will provide sufficient energy to fire an NA engine up to 9K? I do understand that the dwell setting will need to be lowered farther still.

Is there a benefit over the VBs if you are running an MSD or other ignition box in conjunction with MS for spark?

Right now I have v3.0 controlling fuel and spark with along with an MSD. I wasn't sure if the MSD took the load off of the VB's enough that this change would be benficial.

Thoughts?

RS. You guys have to at the very least make them available for purchase from you guys. You aren't the ones that did all the work to get the MSextra code writen that made your product even possible. How many hours did boostjunkie spend writing the highres code to run large injectors? How big of risk did Una take by running the early MS2 extra code? RS as a business has bennifitted greatly from the work of others that you got for free.

I do greatly appreciate the effort you guys put into find this much needed product. In the same way I appreciate the first guy to show a 3.0 intake or show how to move the cam megnet to make 2.0 cams work in a 2.4.

In my post above I tried to convey that we're still on the fence about this. The reasons you guys have posted for releasing the part number were part of our previous discussions on this matter. Aside from the reasons I posted before, a big reason we were leaning towards not releasing it at this time is that availability is somewhat limited and we are concerned with keeping them available for our customers.

After talking to Rodney we've decided to release the part number. It is a Fairchild ISL9V5036P3.

Please keep in mind that our test results are preliminary and we have not yet performed life testing on these.

Steve

You won't see any benefit with these when running an external ignition box.

Steve

Hey, in comparison, it's fast.

Thanks!! It was the right thing to do!!
And at $20 ea retrofit you wouldn't exactly be rolling in dough!

It would cost more in shipping and time down. Id rather buy the component then solder them in, in a few minutes rather than ship my box from MD to CA for a 20 dollar service thats "not sure" if it is better.

Ill wait because im going to be running N/A for a little so the dwell issue shouldnt effect me. When I get the engine in the car with the turbo it might be another story though, but by then there should be some real world testing

I understand. But the same thing can really be said about the whole Megasquirt project. It's all experimental.



Steve

Yup. I cant wait to see if I should switch out my "old" ones or just leave them be.

Um, er... ahh... you still using the 330 ohm or, ah,
a different value?

This rocks

He's referring to my thread (in sig).
Wouldn't increase the price but by $4. Still under $200.

Was just an effort to troll and should be deleted.

Been using the IGBT's for about 3 months now. They are awesome. I accidentally setup up my MS the wrong way and they survived a 5 second blast, even though the little circuits on the MS board fried.

And this is why a fuse is absolutely essential when using these.

Steve

Steve and Rodney,

Let me be another on the long list to give a resounding THANKS for your work in finding a solid alternative to this problem. Excellent work, and much appreciated!


-Eric

I agree! Thanks to you guys for continuing to support the Neon community and the MegaSquirt concept! ! ! !

It's just my opinion, but I really think that by sharing the information like you did, not only do you enlarge the testing base for the development of your products, but you also continue to build good will for your business.

Hopefully you guys are actually making some money out of some of this work you do! I know for a fact it doesn't always go that way!

Thanks again,

Thad

Isn't the dwell voltage referenced in MS1? So if I set it to 5ms, that's 5ms at 12v, it'd be quite a bit less when the engine is running, at 14.4v or so. That 5ms you measured the stock ECM charging the coil at, I'd imagine you measured that at full operating voltage? So what would you have to put in the MS settings to get the 5ms running dwell?

Good point. Here's what we did:

1) Observed about 5ms of dwell in the stock system, but more importantly, observed 6A of peak coil charging current.
2) On a 400W variable power supply set at 13.8V, duplicated this 6A peak coil charging current (as measured with the same current sense resistor as the test on the stock system) by setting the dwell to 5ms in MS1 Extra. If you look at the picture of the IGBT running at a setting of 5ms of dwell (shown below), you'll notice that the actual dwell was about 4.5ms, due to the fact that we were running at 13.8V. 12/13.8*5=4.34. That's fairly close...



If the dwell is correctly adjusted by the firmware to compensate for battery voltage, the same 5ms setting should result in the same 6A peak charging current regardless of whether the battery is at 14.4V or 12V... the actual dwell time will just vary accordingly. Does this sound accurate to you? It seems to me that it'd all come out in the wash.

As an aside, also evident in the picture above is that the "minimum discharge period" should be set to 2ms.

Yeah, the way it scales makes sense. Ohms law..
Part of the reason I asked, is the dwell control in MS2 references 14v, rather than 12.. So the number you put in will need to be a bit lower. Sounds like 4.3 or so would be a good number for someone running MS2.

Good call. We'll do some measurements on our test setup and establish a dwell number based on peak coil charge current when we get around to doing more benchtop development testing.

-Rodney

http://www.fairchildsemi.com/ds/IS/ISL9V5036P3.pdf

There is a link for the specs on these new guys.

MT doesn't allow any higher than 1.5ms as the minimum discharge period.

AND Thank you very much Steve and Rodney for the support and great products you offer. The maturity and level heads you keep really show. These are the reasons I shop with you and recommend your services and products to others.

MT doesn't allow any higher than 1.5ms as the minimum discharge period.

I guess 1.5 will have to do?