Ignition related page for MembersPage/MarcellGal/EngineSwap
Ignition - wasted spark for now. Going to CDI and IonSense later.
The trigger info is on MembersPage/MarcellGal/EngineSwap/VrTrigger
Ignition advance is finally tuned.
- we did it on a "hold-speed" capable eddy dyno. With the incredible ego correction (using lambda-target table) it was fast. We used mttn than mmtf commands to cheat megatune, because MegaTunix interrogation somehow did not work. Hell, good ignition makes a HUGE difference. The car is now fun to drive, and as powerful as it should be.
I forgot to mention, I had to use mtf.. and an old megatune to set ignadv. Neither megatunix nor msns worked. After looking back, it probably worked as it should, since I had MS_COMPATIBILITY defined, and MSNS not defined in my_make.
- MegaTunix failed at interrogation. Very annoying if interrogation cannot be overriden, maybe I just don't know how.
- msns probably would have required a my_make option that I didn't think of at that time:
- MSNS defined, but MS_COMPATIBILITY not defined
TODO:
- try to decrease dwell slowly
- upload "don't do it at home" image about the ignmodule table-test
- connect and configure knock sensor
- find the camshaft sensor. Not needed for now, but it's good to know if there is none
Transformer options
Polski fiat wasted spark transformers - I have 2, but incompatible with my HV wires connection
Ford coils - looks a bit small, and I don't have connector for the primary side
Opel Astra wasted spark twin-transformer - look much better than Ford twin-transformer, I also have connector for driving it, but I didn't have specifications.
- It has 4 terminals (JPT-style connector)
- DVM suggests that it has logic level input - so I cannot drive the transformer directly without breaking the nice rugged epoxy-filled unit
- I think it doesn't have anything smart (dwell-control, or cylinder-stepping, shift-register or something), just plain IGBTs inside: when I apply 5V, it charges the coil
It's named L2 on engine drawings (H.R.Etzold book) I have.
Likely pinout
- 1 Power
- 2 GND
- 3 black-red color, (cyl1,4)
- 4 black-yellow color, (cyl2,3)
These are monoblock coils, epoxy filled. junkyard price is appr 45 Euro.
With a running engine I could go down to dwell14=0b (700usec) when it loses it's power, but the horrible high cranking and afterstart pulsewidths demand high, dwell14>0x20 (rather >0x28) even though dwell6>90.
Opel coilpack == GM 4cyl wasted spark ignition amplifier
- partnum: GM 1103872 1A23 (maybe just the cover plastic part?)
- backside number: 10487489 must be a part number, since I found a [part in a pdf list] (or [same in html] that claims to replace this
- 0D12 ???
Measurements
After some unsuccessful spec-googling on the Opel (GM) active coilpack (came from my Opel engine), I decided to give experiment a try.
- wired up supply to GenBoard/VerThree
- wired supply to the ignition module
- shorted the cyl1-cyl4 HV pins with a little gap
- shorted the cyl2-cyl3 HV pins with a little gap
- pin1=VBATT (9V in my experiment)
- pin2=GND
- pin3=short 1.5 msec dwell applied by mdn01 => fired cyl1,4
- pin4=short 1.5 msec dwell applied by mdn01 => fired cyl2,3
I first connected pin3 only and than pin4 only (therefore the same mdn01 command for both output). The other one was left open, on the open input I measured 0V.
The pulse I applied was inverted with an ES2J (fast, 600V, SMB, same as used for flyback) diode and a 2k7 resistor pullup from 5V to the IGBT output (this makes it possible to use the same output for a raw transformer with max 7A and 380V). Pulse was about 4.5V for a short 1.5msec time, and normally 0.53V. To achieve this, I used ign_out=71 inverted semantics so the IGBT is switched on normally.
The PS2 keyboard froze because of the sparks. But GenBoard continued to work without reset (even though GenBoard was closer to the spark than the PS2 keyboard).
I asked my girlfriend to issue the mdn.. commands from the terminal running on the notebook (connected to GenBoard via USB-rs232), so I could watch which transformer-output fires on a given input.
Note that I'm not absolutely sure that sparks are emitted on the falling edge, but 99% sure at least.
This module looks much better now than the Ford coilpacks (that I personally heard more about in diy-land).
- Bigger transformers
- because of the internal drivers less electromagnetic radiation in the harness and
- less heat in the ECM box.
- It is even possible to use FETs instead of the IGBT (or even a P259 output) to drive it, so the IGBT insulation issue goes away too.
- Price is about the same (appr. 40 Euro for a used but good condition 4cyl wasted spark coilpack).
Links
- Specifications for GM DIS and HEI at [1] and [2].
- units: [cheap 7-pin HEI] might be good to keep in stock, eg. for use with VemsFrontier/ArmEfi I just noticed that it's only one output channel => NOT much use
- some [4 pin HEI (is this for distributor?)] - useless for this install
- consider 4 IonSense compatible coils (small is enough for CDI)
- sparkplugs for turbo operation: I have no idea of type. I like the cylinder-symmetric ones (the type with a center electrode and 3 around it, used in Formula1, except with normal size thread), but not sure if it worths the extra cost and hunt. What heatrange should I use ?
- probably going wasted spark for now: (I don't have COP at the moment. appr. 69mm from COP coil bottom to sparkplug tip).
Config
- inverting ignition amplifier, but using signal from IGBT gates: ign_out=70
h[2]=00 01 00 01 00 01 00 01
- crank advance = 12.5 degrees: ign_crank_advance=32
See also