Shrinking Dwell

[TheMonkey]

ignition output pulses are sequencing. i set dwell for LS2 coils as per the info detailed in the LS2 coil write up. 5.5 ms max dwell, 2.0 max spark duration, with voltage adjustments.

i wanted to witness the overlapping dwell from 2 ignition events in a row. they do in fact do that, but dwell is shrinking with high revs, and i can't figure out why. i zero'd out the voltage adjustments, and accelerator adjustments, and fooled around with 'max spark duration' but it still does it.

the dwell sits right at the 5.5ms, revs start to go up and there is some overlap in dwell periods in consecutive spark outputs, then at about 4650 revs the dwell starts shrinking. my max revs will be low 6,000's. i should get full 5.5ms dwell up there, but it shrinks down to 4ms ish. why is it shrinking?

thx, Scott.

[Bruce Bowling]

The shrinking of the dwell is there from the dual spark MSII mode. There are two spark outputs that are routed to one of 8 ignition outputs. And each can be dwelling or spark burn (or off). In the MSII code there are two variables called (I think, check the MSII docs) MAX_DWELL and MAX_Spark (or something to that effect). The dwell is the desired maximum dwell time. The spark time is the burn time phase when the spark happens and its in plasma glow stage.

When the RPMs increase there may be not enough time to dwell and to burn before its time to dwell again. So what the code does is shrink both the dwell and the spark duration values in proportion. As the RPM goes up this will keep on shrinking.

Now, for the sequencer mode things are better because when the sequencer is pointing to one coil it can dwell and then spark, and immediately point to the next coil and start dwelling it. Its not the same coil anymore, it now points to the next one, so the issue of burn time goes away.

So what you can do is shrink down the burn time number, this should help with the dwell shrinking somewhat.

Now, there are only two ignition outputs that get routed, so there can be some inherent overlap. In your case of 6,000 RPM and an 8 cylinder, there is a spark every 90 crank degrees. So, assume 0 degrees MSII processor ignition 1 channel fires, then at 90 degrees channel 2 fires. At 180 degrees channel 1 has to fire again. Even though the router is pointing at different coils the multiplexing of 2 channels causes some dwell reduction - there is no way around it.

So, for your case of 6000 RPM and 180 degrees, this gives 5 milliseconds (millisec = (166.7 * deg)/rpm)). In reality this means that you can get a 4.9 ms dwell and a 0.1ms burn, with the burn time really now meaning the time it takes for the router to point to the next coil. The resolution is such because we are dealing with a 16-bit processor and this is what units made sense.

For the LS2 coils, they are monster beast that give a spark at 5 ms that will kill 100 horses hooked up in series (or any other animal). If you run the coil at 4 milliseconds the coil energy is reduced a tad, but its still enough to kill off, say 99 horses. In other words the LS2 coil, at 4ms dwell, is still much more than many other coil setups I have seen. So even at the reduced dwell at 6K RPM that you have right you will still have a wicked spark

- Bruce

[TheMonkey]

Bruce-

thanks for the great explanation. so, with 2 source IGN signals that get routed, at most we can overlap 2 consecutive dwells. they can overlap until end of first dwell in que is .1ms from beginning of T+2 dwell. for V8 requiring 5ms dwell, this gets up to 5900 revs at fell dwell. my max of 6200ish would get 4.7ms dwell. for SBC or other guys, not sure where you redline, but 7,000 revs would get 4.3ms dwell. compared to max dwell of about 1ms with 1 coil at that speed.

your horse killing analogy is a good one. none of my comments are complaints, i'm just trying to completely understand everything (about this project anyway ;).

[PJM38]

Hello,

I ended few days ago the megasquirtII conversion of a 84' 944na Porsche with the help of the information found in this excellent forum (thanks Jlturpin, M. Grippo, etc...).
A quick summary of my configuration: To re-use the 2 stock trigger sensors (130 tooth crank wheel without missing teeth and 1 teeth crank wheel), i transform my V3 board for the dualspark mode. So i add a second VRinput module (LM1815 based) and a second output module to drive wasted spark type coil (renault twingo). For the dual spark mode i choose "M-0 wheel with falling crank sync." and for trigger offset -301.5.

...And it works not so bad. I can start the engine, i have not tach loss until red line but I DONT UNDERSTAND why dwell is start shrinking around 2000 rpm and continue to shrink until redline (my settings Max dwell duration : 4ms, Max spark duration : 2ms).
On my 4 cyl. there is a spark every 180° crank. At 2000 rpm it's every 15 ms.If we consider two coils it's one spark every 30 ms on each coils (?). Well, during this time it's easy to dwell 4 ms + 2 ms spark no?
Hope someone can explain me this mystery.

Thanks in advance for your help (and sorry for my poor English)

Patrick, Grenoble-France

[grippo]

The shrinking dwell is most likely due to the way the timing wheel is configured, that is, the trigger offset and delay tooth settings. If you look at the below timing diagram you will see tach pulses for IGN1 at 2 places. Trigger offset of -301 deg means that IGN1 tach comes in -301 deg after TDC for the cylinder connected to IGN1. Since you want to spark before TDC you can see that you don't have a lot of room for dwell + advance - a fixed, approx 60 deg, and the time for this dwell + advance becomes less and less as rpm goes high. The way microsquirt is set up you must start dwelling at the first IGN1 and no sooner. That is why the dwell is shrinking.

..................................TDC............-301deg.............................|
.........................|____________________|________________________|
.......................IGN1........................IGN2..............................IGN1

The solution for this is to change delay teeth by adding say 80 deg (however many teeth that is) and then change trigger offset to say -20 deg (it must remain negative). This will push the first IGN1 to 20 deg past TDC and you can redraw the timing diagram as:

...........................................................................TDC..-20d..|
.........................|____________________|________________________|
.......................IGN1........................IGN2..............................IGN1

Now you should have plenty of time for dwell + advance.

Hopefully you don't have an odd-fire or that would have to also be taken into account.

I would first try this with the injectors unplugged and just cranking, and 'calculated' not 'trigger rise' starting, and check with a timing light that the spark is where you want it. This could avoid a backfire if things aren't right.

[PJM38]

Merci beaucoup!
Thanks!!
Explanation was clear and now...
It works fine with Trig. Offset = -30° (before -301.5) and Delay Teeth = 30 (before 0) ...No more dwell shrink. See attachements.
My next problem is the tach loss you can see on data logging. I will try to fix it by myself with the forum database.

Thanks one more time for your quick and precise answer.

Patrick

[grippo]

I see 1 tach loss when you get over 6000 rpm - is the loss consistent at high rpm or can it happen any time. We normally see this with a missing tooth wheel, but you don't have that. I also see a very sudden dip in the battery voltage just before the tach loss. You might need an audio cap across your battery if that happens whenever you lose tach.

[PJM38]

Thanks for you analysis, i plan to buy a 2F audio capa in the next day to finalize my installation.

But, i think i fixed my issue. I had tach loss at always fixed rpm mostly at 1700, 3400rpm. Signal measurement with oscilloscope shows me that the crank sensor has the bad polarity. The logic output time of my LM1815 (around 0.3 ms) was added in the calculation. When the time between 2 teeth was in the same order or a multiplier of the logic output time, 131 teeth instead of 130 was seen between two reference inputs. Since i change from falling to rising, i have no experience tach loss anymore. Hope my analysis is right.

Patrick

[grippo]

It probably is - we have seen many times that a simple switch in polarity makes it work.

[PJM38]

Hello again,

I have a new amazing issue with my Porsche 944 Megasquirted in dual spark mode. Sometimes an offset of around -30° is added to the advance read on megatune. Example, at idle, the advance read on Megatune gauge, is 8° but the advance read with the strobo is 39°. When i do a 'burn to ecu' with no specific change, the advance read with the strobo come back to 8°. This offset appears or not appears after cranking.
I assume that there's a problem with the cranking, but the cranking next pulse tolerance changes (from 20 to 80%) have no effects.

Hope you understand the description of my problem and that some one have a idea to solve this.

Thanks

Patrick-Grenoble