I thought I would start a different thread for this.
After running for about 15 - 20 minutes the ECU cuts out sometimes both cylinders sometimes one. I thought it might be a bad temperature sensor so today I substituted fixed resistors in place of the sensors.
Heres what I discovered
First of all I loose 1 cylinder. On feeling the transistors I find the two spark FETs are quite warm, not enough to burn but getting up there. The transistor in the center of the board is warmest. I put this down to the distance from the ground plane at the edge of board. I assume this is the one that cut out.
My FETS have the two little heatsinks that Cliff provided along with some heatsink compound.
The injector FETs stay nice and cool.
After sitting for a couple of minutes the bike will start but soon cuts out again.
I have no cover on the ECU at the moment.
It was a very hot day ~40 C.
I have 5.3 nF 1000 V (ncd472m1kvz5) ceramic disc capacitors wrapped around the pins going to the two large tracks.
Could it be these capacitors?
Do I need a better heat sink i.e. a thermal connection to the case?
Hope someone can help. :(
Roy

Something's not right. They should not be getting warm at all. At most it is possible to feel a slight warmth.

Have you modified the coil charge times in the map?
Coils should be 0.7ohm

This is the circuit for each coil driver
http://www.cajinnovations.com/MyECU/CoilDrivers.htm

Are both the gates at ground ?

Both gates read 0.027 V to ground (Fluke 87), there's zero on the other two terminals (R5 is not energized)
With the bike idling I read ~0.3 on the gates
Yes coils are 0.7 Ohms
I will hang my scope on the gates tomorrow and report back.
No, I made no changes to the coil charge time but I will look at that also.
Thanks
Roy

Cliff,
       OK, I checked out the pulses with my scope, they look pretty much as expected, nice square pulses about 7mS wide 13 V high starting at zero.
The trace from the transistor tab though is a bit puzzling, pulls down from 13 to almost zero then over the gate on time rises back to about 10. At first I thought this must be normal but on further thought shouldn't it stay low the whole time gate is high.
I will send you a pdf of what the traces look like, I haven't figured out how to post them here.
I should have looked at the injector traces, they must be somewhat similar.
Roy

7ms sounds a bit excessive. It should be around 6ms from the map entry coiicharge0=6000.
This is for starting only.
Once running it will revert to the coilcharge1=2700.
Check this is the case otherwise they will overheat.

I'm surprised the voltage is ramping up so high. That does seem too high. Those FETs don't have a low Rds because they are high voltage FETs. I'd expect about 6V across them.

The 6000 looks like it may be too high. I think everyone might benefit from a reduced value  to say 4000. Or it could be your test scenario because normally the voltage is less than 10V due to cranking.

Yes, I don't know what i was smoking, tonight when I tried again I got 3mS for the Gate pulse. I updated the pdf and took some photos of the trace, I will send this to ecu000@cajinnovations.com :)
Is it possible for the ECU to stick in cranking mode, that would explain why I saw a longer pulse, otherwise I was out to lunch. The trace at coil connection looks very similar to the stock ECU, I couldn't figure out which was the Gate connection, so I don't have that.
I reset the crank duration as advised and will try it tonight, hopefully it's come right.
Roy

Ok, I gave the map a 20% increase in injector time 2-5000 revs.
The bike seems much happier now I went for a long ride and it only cut out once after a run at 3/4 throttle. I measured the transistor temperature, only 35C so that's not a problem.
Here's my theory the previous run it was a very hot day, the map was way too lean, it was a hot day, the air sensor was hot, the oil sensor was hot so the bike cut out on lean mixture.
Today was quite a bit cooler and the mixture was richer and the bike ran OK until I ran at wide throttle where the map has not been richened.
Does that make any sense at all? :-/
Can someone explain the effect air and oil temperature have on the map, does hotter sensor make the mixture richer or leaner?
My mixture is still too lean at low revs and low throttle setting misses quite badly but as soon as I soon as I wrap the throttle or the revs are over 3000 it settles down and pulls away.
Needless to say I am much happier  :D

Kiwi Roy wrote on 07/14/10 at 15:48:47:

Can someone explain the effect air and oil temperature have on the map, does hotter sensor make the mixture richer or leaner?

This is the raw map lines affecting it:

# OilT is the permanent choking boost
#Temp C   -30   -20   -10     0    10    20    25    30    40    50    60    70    80    90   100   110   125
OilT%   +64.8 +53.9 +44.5 +35.9 +29.7 +19.5 +10.2  +9.4  +5.5  +0.0

# AirT is the barometric adjustment for air temp(based on ideal gas law)
#Temp C   -30   -20   -10     0    10    20    25    30    40    50    60    70    80    90   100   110   125
AirT%   +24.7 +19.8 +15.2 +11.0  +7.1  +3.4  +1.7  +0.0  -3.2  -6.2


At 50 degrees oil temp, and 30 degrees air temp, these sensors will not affect the injection times. At lower temps, you will have enrichment. At higher air temp, there will be a little leaning. Values are in percent.

AFAIK, the not-filled-in values will be a repetition of the last stated. I'm not sure. Anyway, you will hopefully not have an ambient air temp of above 50 degrees  :o

So much for that theory then, the temperature sensors don't have a lot of effect, I assume the +# is microseconds injector time so for an injector time of say 3050 +5.5 is nothing, or is it %? Never mind, I see it's percent boost.
So going from ambient 20°C to 30°C I loose 10.1 + 3.4 = 13.5% fuel
I am still puzzled over the map table Flags. If I look at some maps it's mainly populated with o0 & o1 (derived from Sport11Stock) wheras other maps have a wider range o0,o1,o2,o3,o4,05 (Derived from code as flashed by Cliff - I think). My understanding of flags is bits set by the program but the character o "oh" isn't binary or hex. What sets these flags? Is this perhaps refering to the O2 targets e.g. o4 calling for O2 target of 0.80. There are 8 O2 targets, which is o0 and which is o8? :-/
Running open loop shouldn't all the targets be o0 or does it just ignore that line in the code?
I think my ECU Controller always shows closed loop even though I have selected None for sensors under MyECU Config.
Bear with me, I'm learning here, I'm sure others are as well  :-[
Leave me alone long enough and I will probably figure it out.
I must ask Cliff for spellcheque  ;D
Roy

The targets are specified earlier in the map

MyECU Cfg O2=0.00V,-2.50V,-2.34V,-2.17V,-2.00V,-1.84V,-1.67V,-1.50V,

Setting sensors to none will run open loop.

Thanks Cliff, I found the good "Maps Explained" topic on the Technical Site. I will study this before I ask any further :-[
Perhaps a button linking to that site would save a lot of questions.
Still making progress, the bike runs better with each change I make.
Update July 15th
I found a major faux-par tonight. When I started out I changed the pressure module setting and it read correctly but I must have accidentally overwritten the file because tonight I noticed the barometer pressure showing 580 mb, obviously this is having a major effect on the map and accounts for my having to en-richen it across the board.
Actually I think it was caused by software problem with the 10 year old laptop I am using, if I open the file in notepad when I save it it changes to a text document that I can't see with the MyECU software. No matter what I tried it always does that.
I will fix the file with a better laptop and start again.
Cheers
Roy

Here
http://www.jefferies-au.org/MyECU/technical_site_map.html
I'm sure you have been there before.

Found it - thanks

Kiwi Roy wrote on 07/15/10 at 21:25:13:

if I open the file in notepad when I save it it changes to a text document that I can't see with the MyECU software. No matter what I tried it always does that.

Try using Wordpad instead of Notepad. I don't usually use Windows (ECUController works just fine under Linux too, using Wine) but I recall having edited maps with whatever was available.

http://www.jedit.org/
is worth a look, I also had problems initially with the text editor on my MacBook. Took a while to work out why my modified ecu files were being saved but not running - all sorts of bollocks were being added. Anyway this editor doesn't add any superfluous codes or formats and can be used in Windows or Mac OS environments. AFAIR it's free.

On the other hand, a forgiving ECUControl program should really not care if lines end with CR or LF or any combination of them, or even a mix of those line endings in different parts of the file.

I thought I addressed that issue. If not send me a map file that has problems and I will look into it.

Thanks MF, Raz,
                      I bought home a better laptop, the one I was struggling with is about 10 and has a dud battery so I was using it attached to the bike battery.
I will check out the jedit software, using a Mac would be perfect for me.
Has anyone tried using one of those small handheld PCs, blackberry size. I have an old one with serial port.
Check out my other thread, "Possible Software Glitch" re the baro sensor, I feel such a fool  :-?

Mad Farquhar wrote on 07/18/10 at 17:17:38:

http://www.jedit.org/ is worth a look, I also had problems initially with the text editor on my MacBook. Took a while to work out why my modified ecu files were being saved but not running - all sorts of bollocks were being added. Anyway this editor doesn't add any superfluous codes or formats and can be used in Windows or Mac OS environments. AFAIR it's free.

You can also run Terminal out of the Utilities directory under OSX and use the "vi" editor. It's probably the most basic editor in existence. Changing the EOL's to DOS compatible might be necessary though.