V12 with ITB's turbocharged

[turbojoy]

The goal is to control a V12 with sequentional ignition and injection.
This is a brief overview of the process with the problems I ran into. Also the solutions are given.

Engine setup:
V12 with itb's and turbo

Sensors/setup flywheel:
6 teeth on crank
1 teeth on cam
sensors both VR
coils: replaced by VAG COP coils (brand: NGK)
injectors: high-Z (13ohm)

Ordered a Vems with no options. (actualy did not look good and forgot to pick them)
This was not good as the info to build it up for my setup was not easy to find and some were confusing.

Ordered a 2nd Vems with options:
VR for crank and cam
EGT
Knock
8 ign + 8 inj

Wiring:
As Vems got only 8 outputs I wired the injectors in six pairs (6x2). Controlled sequentionally.
Ignition in wasted spark; two coils firing at the same time. Real 'wasted' spark thus ...
The goal was to control the V12 with one Vems.
This was a no go due to the 2stroke mode be obsolete now. 
Solution: use of a 2nd Vems. (I had already bought them anyway, so easy to decide)
In order to do so I needed to place a 2nd cam sensor.
By doing this we actualy split the V12 into two 6 cylinder engines.

Build of Vems Left and Right bank:
Soldered the LM1815 on the 2nd Vems (right bank) for the 2nd cam sensor.
Ran a wire from the output of the LM1815 (for the crank signal) of the first Vems (left bank) to the crank input of the 2nd Vems.
The crank signal is for both Vems boards the same. Logical. 
In order to mount the 2nd cam sensor we fabricated a new cover to hold the original cam sensor and the new one.

Theory:
The V12 fires each 60 degrees. So the new cam sensor should come 30 degrees after the original one.
This was unfortunately not possible due to a mounting stud being exactly on that spot. 
So mounted the 2nd cam sensor 90degrees after the original one.
60 degrees after is not good as then the sequence comes together again then with the left bank.
If you devide the degrees you shift the 2nd cam sensor by 30 you will need an uneven outcome to sync with the right bank, if it is even you sync with the left bank here on this V12.
i.e. shift 30 degrees = 30/30= 1 (uneven) ok for sync with right bank
i.e. shift 60 degrees = 60/30= 2 (even) ok for sync with left bank
Igntion output stages were also rebuild for use with the VAG cop coils.
These need a positive 5V signal to charge/ignite.
I used a small PNP transistor (BC640) to do the job and soldered these on the underside of the Vems pcb.

Testing:
When testing the ignition outputs of the first Vems I got this problem:
output 6 (05) retriggers itself after 25msec.
All other ouputs were good.
The same happened with the 2nd Vems: also output 6 (05) retriggered itself.
Tried to solve it with a diode accros the output (to kill the small flyback voltage, seen on my scope)=>no go.
Flyback was gone, that worked!
Tried to solve it with a extra pull up resistor =>no go.
In the end:
The solution was a 10nF cap on the +5V line towards the BC640's. 
I also placed a 1000µF cap on the +12V feed for Vems board (directly on the pins of the ecu connector)
This helped to eleminate a lot of noise from the on/of switching of the injectors. (seen on my scope)
Athough this noise did not produce problems, but I didn't like it to see. 
To control the VAG COP coils correctly I needed to check the box 'invert' ignition outputs in Vemstune.
When testing the outputs in the car I ran into this problem:
When powering up Vems board the igntion coils are energised for about 4 seconds.
This is way to long to charge a coil so the fuse blew.

Why does it take 4 seconds to invert the output when powering up Vems board? 

My solution for now (didn't have more time to investigate deeper) was building a 'power on delay relay' into the +12V power feed to the ignition coils.
Maybe I can solve this also by adding another NPN transistor in my output stage and choose back 'normal' for igntion output.
In meantime I found a picture of someone who used a TC4427A chip to control ignition.
By looking to the datasheet this seems a far more better solution, I'll will test it out.
Vems should be able to control these VAG COP coils directly but I didn't want to risk that as the logic output does not have any protection against overvoltage/reversed polarity/spikes/whatever...
Will be tested/tried later.

Software setup of engine in VemsTune (very briefly) to be able to startup:
12 teeth for full engine cycle
use of cam-sync
TDC reference: 50 degrees (which was not correct, as found out later)

Used for startup a mapping of a Ford Sierra Cosworth that I found somewhere on the web (forgot where exactly).
Of course changed the settings for my ign/injection crank/cam setup.

Live test of Vems boards on the engine:
This was not as hoped although the engine fired up and ran!!

Problems:
- Lots of misfires
- No response on the throttle at all
- Right bank stayed rather cold on exhaust compaired to the left bank (after the 2 first problems were solved)

Solutions:
The first had 2 causes and solved the 2nd problem also.
The 3rd was caused by me (made a mistake). I think I need to learn again how to count...

The first: this was mainly due to bad spark plugs.
After replacing these there were less misfires.
After resetting TDC reference to 30 degrees in VT: no more misfires.
Also the engine responded now on the throttle.
So 2nd problem also solved.

The 3rd was somewhat more hidden.
The exhaust gasses of the right bank where not as hot as the left bank. Hmmm.
Theory.
We put the cam sensor 90 degrees after the original cam sensor. This implies that the right bank starts at the 2nd cylinder.
I wired it so that it started  (the ignition sequence) at the 3rd cylinder...
The ignition occured in the correct order only 60 degrees to late. This gave a lower temperature of the exhaust gasses of the right bank.
After shifting the igntion sequence it was good.
2 hours later of some (fine)tuning the engine ran good, started good (warm and next day cold) and revved good.
Also idle was stable, but the engine died as the throttle was released fully after letting it rev high, or after a 'blip' on the throttle.
This was due to not be able to control the idle control valve.
With itb's you need a shot of air just before the idle rpm is reached...
Played with the igntion advance (idle control with ign adv) but it was not able to catch this up.
Now time to go to the rollers for the rest of the mapping.
As the car is a 4 by 4 it's not that easy to find a rolling road here.
But will have access to one within a few weeks/months. (I hope)

Questions still remain:

How to setup the Y axis of the VE table for my setup?
Does anyone have an example?
From 0 to 100%(tps) or from 0 to 250(kpa)?   

I have itb's so for the NA part it's Alpha-N and speed density when boost comes up.
I know that there is a boost compensation table for the boost.

The VE table is 16x14; the igntion table is 12x12.
How to change the ignition table to 16x14? 

Boost table is 8x8.
Can this be enlarged to i.e. 12x12? 

Another big problem that remains:
The idle motor is something special.
It is controlled by reversing polarity.
Original it is controlled in this way:
rpm to low: on
rpm to high: on with reversed polarity
rpm at setpoint: off

Can this be controlled by Vems?
If yes, how? (settings in VT)

Any help is welcome, would love to finish this nice project with all problems solved.

Newbie question: how to post images here on the forum?? when clicking on the icon only 2 times 'img' (between brackets) is displayed.
I should have expect that there would be a window pop up to upload an image...   

[mattias]

Personally I would have gone with a much simpler solution.
- One ECU
- Hall sensor for the trigger and only primary trigger (no cam sync) and used 12-1 or 24-2 or some other pattern (and physical wheel and tooth size) that the sensor and software can handle.
- Wasted spark using either six of the IGBT outputs converted to logic signals with stronger pull-ups to drive two coils. You could also make do with only converting two IGBT outputs and used the four stepper outputs for the remaining 8 cylinders.

I have done  two V12s like that, only they had twin spark plugs per cylinder, which required four (4) coilpacks (3x2) to fire 24 plugs, I used external igniters and it worked out well.

I can understand some of your decisions were forced by the engine and hardware. I had free hands on the engines I described.

I don't see how missing a "2-stroke" mode has got anything to do with your decision to go with two ECUs.. ?

What firmware are you using?

[turbojoy]

Hi Matthias,

Yes, my decisions were forced by the hardware and not allowed to change them.
Unless it was realy necesary. Also time to build/modify was short. Very short. 4 days...
The Vems boards are also build inside the original ecu box; meaning no modifications on the original wiring.
It was allowed to put an extra wiringloom for the coils as that was the main reason to go with Vems.

Don't see the point for choosing hall over VR. Both have their (dis)avantages and I was forced to use the original sensors.
Even for the 2nd cam sensor I used the same brand/type as the original to keep it uniform.
But agree, if I had a free hand, I would done it otherwise.

The logic level outputs from the Vems board are not that ridgid to current/voltage spikes.
That's the reason I used a simple transistor a drive for the coils.
Hope to rebuild the output stages with a more proper solution: TCA4427A.
If I have them before next Monday.

As far as I understand if 2 stroke mode should still work I would get at every tooth of my crank wheel a spark event.
This is enough to run a V12 in wasted spark with 12 tooth wheel on the crank. Not?

Anyway, it worked the way I did it now.
Have you any answers on my other questions?
I'll be working further on it next monday, would be great to know by then.

Cheers,
Tommy

[Riff]

To answer the question on the images, you need to upload them somewhere, and use the link to them here.
Photobucket is one of the places you can use, and has img codes that you can insert rigth into the text box.

[VEMS]

Hi Tommy,

Perhaps you missed this (sticky) thread in General Discussion "How To :- Display photo's, Video's & Screen Shots in this Forum": http://www.vemssupport.com/forum/index.php/topic,112.0.html ?

Best regards, Dave

[turbojoy]

Hi all,

This is an update of the project.

This is changed in meantime:

The Vems boards are rebuild with the tca4427 as ignition output stages.
This works very well and let me get rid of that timer I had to use previously.
Made an idle driver for the idle motor with the use of an L298 and a basic stamp who reads the pwm signal of the vems board.
The engine is fully mapped for cold starting, warm starting and idling + revving.

Problems now:
1) the wideband O2 sensors does not work (never did, actually)
2) left bank: exhaust gasses cold (no, other problem than shifted igntion, it was good for a while)
3) pwm signal jitters...
First the 2nd problem.
When mapping for idle we encountered -after a while- a problem on the left bank.
The lambda has dropped to 0.86 and we had to adjust a lot to try to get it back to 1.
In fact it was not possible to get back to 1. (we measured lambda with the use af a professional lambda meter ecm1000).
Not seeing directly what's going on I measured the exhaust temp directly after the cylinders on the exhaust headers.
We got +/- 80°C on the left bank and 200°C on the right bank.
Also when revving there were sparks flying out the left exhaust and it smoked black.
The turbo's that were free spinning into the air gave not as much flow as the right bank.
In the end our conclusion was that the exhaust of the left bank was blocked.
As there were still 6 cyl that were running good it was not that obvious to find out.
Now question: is it possible that the cat's are (partly) melted at idle???

2nd problem WBO. (maybe move this topic to an other section on this forum?)

The WBO on both my Vems boards don't work.
I checked every connection and performed the calibrations with the 2 100ohm resistors.
Also checked the heater output with a lamp of 12V/10W. (later with a led as in the vems manual=>same result)
All was good, I got the same as the values as written on the paper that came with the boards.
There was one thing however that I find very odd.
After about 35 to 45 sec the heating of the sensor shuts off.
It does so when powering on; when the engine starts and when calibrating the WBO.
WHY?
When calibrating, after the heaters shuts off, I click 'stop calibrating' and then 'start calibrating'.
I have to do this several times until the sensor is heated up and gives an O2 reading.
Then the heating stays on (regulates).
Also several times restarting the engine gives the same result, but WBO does not work inside exhaust on running engine.
Maybe because the cold exhaust gasses cool the sensor and after +/- 45sec the heater shuts off.
I'm using firmware 1.2.10
Is there something wrong with this firmware about the WBO heater???
Both my boards have exactly the same problem.
As I'm a little affraid to upload other firmware (a lot can go wrong) I haven't done it.
Shortly, we will have a new left exhaust mounted and we will continue with mapping.
I hope to solve the WBO problem by then with some ideas from this forum.

3rd problem: Jitter on pwm signal for idle valve.
I put the I and D value of the idle regulation to 0 to get only a pwm proportional with the (mis) value of the rpm.
This signal is read out with a basic stamp.
I made my program like this way: pwm < then 45 =>activate motor to lower rpm
pwm > 55 =>activate motor to raise the rpm.
and  between 45 and 55 => do nothing.
This works surprisingly very well; only I get sometimes erratic behaviour.
This is because the pwm signal of the vems is jittering.
I had to lower the deadband from 45-55 to 48-52 to get a better working and the jitter disturbs the correct working.
Is there a way to get rid of that jitter??
The I and D didn't do anything.
The jitter can clearly be seen when measuring the signal with a scope.

Best regards,
Tommy

[fphil]

I made my program like this way: pwm < then 45 =>activate motor to lower rpm
pwm > 55 =>activate motor to raise the rpm.
and  between 45 and 55 => do nothing.
This works surprisingly very well; only I get sometimes erratic behaviour.
This is because the pwm signal of the vems is jittering.

In my case  (3 crank triggers only effective) I have also noticed that the rpm is very noisy. I only use the I action. In your case you cannot have a big I term, P and D 0, because, the I term would be still integrating in the dead band
Luckily your setting has a dead band for P, I have advocated for this as fw parameter. 
Indeed the bang-bang control with deadband that you have programmed is known to be very effective (optimal in fact).

[Kamuto]

it seems you are overengineering it, I have started one v12, 2 wasted spark coilpacks, 0 problems, both widebands working and so on...

[turbojoy]

Hi,
Overengineering it? Maybe, maybe.
But I'm glad to read how to control this V12 with one Vems board with this setup:
6 tooth on crank; 1 on cam. => not allowed to change this!
As far as I have read and discovered the Vems can only have 8 events for one engine cycle (=720degrees, 2turns)
I need 12 and due to the setup of the crank/cam signals wasted spark gives me only 3 events at every 120degrees instead of 6 every 60degrees.
Due to this setup it's controlled like a 2stroke 6cylinder.
Vems can only control a 4cyl 2stroke with such a setup.
If there were 12tooth (or more) on the crank then yes, control with one Vems possible.
Correct me if I'm wrong, me too would love to use one Vems if possible.
The fact that the wbo2's are not working have nothing to do with (maybe) overengineering it.
I'm hoping I did not setup something properly; but what?
I posted my wbo2 problem in another category on this forum.
Any help is much appreciated. All ideas are welcome even if they look stupid.
Best regards,
Tommy

[Kamuto]

it's like 2 m20 engines, one wasted spark coil for one engine. 3 ignition outputs used for one side,wasted spark coil wiring 1&6 7&12 5&2 11&8 3&4 9&10 configure it like 6 cylinder engine, coil type, number off teeth 12, then 0 2, reference 0 2 4 6 8 10 and use 6 injectors and 6 ignition outputs to control all that

[turbojoy]

Hi Kamuto,

Well, you see, it didn't work for some reason.
I use cop coils but can fire two at the same time.
I did use one vems in the beginning and wired it up like you've said.
Maybe I did do something wrong because it didn't run good on 12cylinders.
Like said before, I had only 4days to complete the install.
In meantime I have made use of some extra outputs on the other vems for my idle control.
I did so because I discovered that injection output 07 is dead on the first vems I bought.
The pin on the cpu itself does not go low/high when sending a command.
As I had plenty of others I did not bother.
Anyway, if I get the chance and time to review/rewire it, I will definately try it out.
Because indeed, one vems simplifies the tuning (now need two laptops).
Thanks for your thoughts/advice.

Regards,
Tommy

[turbojoy]

Hi all,
Invested some time to upload photos.
Here are some of them.

Ignition coils on right bank in test configuration
[/URL]

Home made idle control.
Motor shield for arduino and basicstamp.
[/URL]

This is the idle control on the engine.
Its just pushing open the throttle.
[/URL]

View of the Vems inside the car.
[/URL]

One of the Vems boards inside the ecu box.
[/URL]

Best regards,
Tommy

[Kamuto]

you are crazy..

[PeepPaadam]

Very exotic, a Bugatti V12! Indeed understandable why altering the trigger is no option.

Any more info/pics of the whole car?

[fphil]

Very nice and clean setting. I should have put the genboard in the old ecu box as you did, that's the best. Luckily you had double high box.