Project GINETTA G33


The Ginetta is equipped with a 1991 Cosworth YB engine producing standard 220 HP. The standard ECU is Magenta Marelli also know from other cars. The car was upgraded somewhere around 1992 with larger injectors (Bosch 803), larger actuator, 3 bar map sensor and an unknown performance chip. Our customer was told that the car produced 330 HP when he bought it a couple of years ago, but on the dyne the car produced only 220 HP plus we noticed that the car was running poorly at curtain rpm - other from that the car was running great, turbo is small enough to keep the car agile.

Going for the 330 HP

First of all we did calculations for the injectors, turbocharger etc. and found out that this setup was not going to give us more than 275 HP - We then did all the initial tests, like sensors, triggers, engine health, fuel pressure etc. to find out where this YB engine was missing something that could limit it to the low power output on the dyno. We ended up with a lot of approved tests, but were strange ignition and lambda maps - so we ordered a Motorsport Development performance chip for the car and according to the specification on the car - Unfortunately we also found that one of the pins on the chip was broken and welded to the socket. On the cosworth we usually use an emulator to map the car and burns the chip later on, now this was not possible anymore, but even worse we found that the installed Pectel babyboard was a non-supported version.

So instead os spending more time on trying to fix something thats not easy to fix and/or get parts for we went for the VEMS instead.

Specifications / Standard Technical Data (Cosworth YB Engine)

Specifications (For the VEMS project)

Config file:

Log file:


The Cosworth has 4 injectors that the VEMS can manage individual – The injectors is connected to a ignition plus (87) when ever ignition is on and controlled by a ground signal from the VEMS.

When reusing the original wiring harness ignition plus is already present – The following was moved from the original ecu to the VEMS

In this case we used Bosch injectors (280 150 951) witch delivers 300 cc/min.

Injector resistence is 16.5 ohm

In Base Setup > Engine Setup you'll need to define what injectors the VEMS are working with – And to do that you have to calculate the injector size.

The formula is req_fuel = 6.49 x engine cc / number of injectors / injector flowrate – And in this case 6.49 x 1998 / 4 / 300 = 10.8

Input 10.8 in Req Fuel and 300 in injector flowrate.

You also need to tell the VEMS what injectors are in use, how many etc.

Go to Base Setup > Injector Outputs Visual

In this case we set Real cylinder order to 1-3-4-2 and as a default check the boxes from the left top to the right bottom (Cyl 1 = Output 1, Cyl 2 = Output 2, Cyl 3 = Output 3 etc.) When only using four injectors you have 4 extra outputs that you can use for idle regulator, fuel pump relay etc. In this case we used Outpot 5 for the idle controller, Output 8 for the fuel relay and the rest for future use.

Finetuning this engine made us rearrange the injector order - Is set at 4-2-1-3

Note that you'll be able to se what pins are correspondent to the outputs, so these should match you wirering harness to worth the injectors.

When set you can test the outputs by choosing ON in Switch testing mode and activating the individual output. These Outputs are ground controlled so you could connect the volt meter to plus and the control pin on the injektors connector to check the signal – Same goes for idle controller and fuel pump relay etc.


The VEMS can manage the ignition in several different ways – in this case we switched out the old coil and distributor setup with a passive coil from a Ford engine (Also known as a DIS coil). This coil has 3 pins.

I could have used the wires from the TFI module, but instead i choose to go with a new set of wires - This way i'm curtain nothing goes wrong from the VEMS to the coil.

The firing order is counter clockwise 1 – 2 – 3 – 4 when the plug turn against you looking down on the coil.

Setting up the ignition default values in the VEMS for this Cosworth engine is rather easy and is just a much related to the pri and sec trigger as anything.

Go to Base Setup > Primary trigger settings

In the dropdown for Defaults choose the lancia_cosworth and press burn.

This changes the setting for the primary trigger and the timing on the Cosworth engine.

Go to Base Setup > Ignition Settings

Set Ignition Outputs In Use to 4

Go to Base Setup > Ignition Outputs Visuals

Choose 1-3-4-2 in Real cylinder order, choose Ignition Output 6 for Cylinder 1 and 4 and Ignition Output 7 for Cylinder 2 and 3. Note that Output 2 is 24 in the VEMS and Output 3 is 10 in the VEMS, witch goes to pin 1 and 3 on the coil connector. (6/8-2015 - Switched PIN 10 to 34 and 24 to 36)

Don't worry about pins 1 og 3 on the coil, if you switch then you can easily switch then in VEMS Tune.

Again, if needed you can test the outputs by selecting ON in Testing mode and activating the individual output.

Now ignition and trigger settings on the Cosworth is a nightmare - I would, anytime, recommend that you put on a new triggerwhell (This one has only 4 triggers) and new stranded wires for both the pri and sec trigger.

Setting up the triggers

Go to Base Setup > Primary Trigger Setup

Go to Base Setup > Trigger Reference Tooth Table

Before cranking you'll need to adjust the base timing - Remove the injector plugs, sparklplugs and the fuel relæ before doing this.

Go to Starting / Idle > Priming, Cranking, Afterstart Setup

Set Crank Ignition Advance to 0 (Remember to set this back at 9 degrees when base timing is done)

Note - The Cosworth engine has a mark in the engine block when this is aligned with the marker on the crankshaft ignition timing is set to 16 BDC and when the nearby tooth is aligned 0 TDC

So you'll need to set your ignition gun at 0 degrees and check if the tooth alignes with the marker on the engine block. If not, you'll need to adjust the TDC After The Trigger (Base Setup > Primary Trigger Setup). Try adjusting this value down at around 85 degrees and check again. Do this procedure until the tooth is aligned 100% and then burn the value.


In this case we are using a Bosch Wideband LSU 4.9 lambda unit.

With your VEMS you should recieve a wirering harness for the Lambda - Before connecting the VEMS, the connector on the wirering harness make sure you have the correct termination for the lambda !!! LSU 4.2 and LSU 4.9 is NOT the same !!!

Lambda Wirering Harness VEMS

And please keep in mind that the wirering harness is subject to changes in wire colors, so this documentation might not work out in all cases.

Setting LSU4.9 in the VEMS

Go to Base Setup > Wideband Settings

Controller type is Wideband if your using the LSU4.2 or LSU4.9

Is your configuring a V engine with two banks you should enable Second Chanel also, is this case we only have one Lambda so we're good.

In Channel 1 choose LSU 4.9

For calibrating the lambda go to Tools > WBO2 Calibration

Power, ground, main relay and fuel relay

As described in several documents, guides and papers you'll need to connect the VEMS power supply, fly back and ground connections secure and away from any noice like from the engine bay. In short you need separate supplies for main power and flyback and the ground is to be connected secure to the car's ground inside the car nearby the VEMS ECU.

In this case I installed a new fuse box for the project - The fuse box has one supply directly from the main relay (Controlled by the ignition switch). The VEMS power (PIN 25 EC36) is connected to the fuse box on a 3A fuse and the flyback (PIN 23 EC36) on a 10A fuse.

The VEMS has 5 ground connections (PIN 26, 5, 21, 22 and 32 in EC36). All ground connections are connected to a solid ground inside the car and with wires not longer than 50 cm. To be sure the grounding was solid I drilled a hole in the frame under the dashboard and mounted a 8 mm bolt. Also you'll, on a car like this, need to check the ground connection from the batteri to the frame and from the frame to the engine - it is crucial that the ground is steady in all corners of the car.

The Ginette has the complete wirering harness from the Sierra Cosworth ECU/Engine so i choose to reuse both the main relay and the fuel pump relay.

The main relay supplies the injectors and other components that need ignition power, so all i did was solder a new 2,5 m2 wire on to PIN 87 on the main relay - this wire is connected to the mentioned fuse box.

Main relay wirering

The fuel pump relay was not that easy - in fact i thought i could reuse the relay and the wirering and simply connect the VEMS control ground, but I thought wrong.

I ended up rewirering the relay completely

To setup the fuel pump signal you go to Outputs > Fan / Water Pump / Fuel Pump

In the buttom of the windows you need to choose INJGroup in Fuel Pump Channel and in this case output 7 (Corrosponding to PIN 17 EC36).

To check the setup go to Outputs > Output Channels

You should ne able to see this settings in the dropdown for Fuel Pump Channel - Choose ON in Switch testing mode and activate the output on the right side besides the Fuel Pump Channel - When testing unplug the relay or the fuel pump!

Air intake temperature

The sensor is supplied with a common ground and returned to the VEMS PIN 2 EC36 through the sensor.

Engine temperature

The sensor is supplied with a common ground and returned to the VEMS PIN 14 EC36 through the sensor.

Crankshaft (TDC Sensor)

The Cosworth has two triggers - crankshaft trigger on the front of the engine and crankshaft in the distributor.

The sensor is a VR type sensor where VR- is connected to a common ground and VR+ to the VEMS PIN 27 EC36.

More information under Distributor.

Distributor (Secundary Trigger)

The sensor is a VR type sensor where VR- is connected to a common ground and VR+ to the VEMS PIN 13 EC36.

In this case i converted the ignition to wasted spark with a Ford Motorsport DIS coil (Same as the Zetec DIS coil) so I mounted a alu. plate on the top of the distributor instead of the original top.

The sensor is, as described, connected to a common ground in the original wirering harness - This common ground is connected to all sensors in need of a ground and this lead to the ECU under the dashboard. The common ground is stranded so i should work out of the box. In this case I found that the sensor was generating a lot noise when cranking the engine.

Go to Tools > Record Trigger

When the window pops up wait 5 sec., then crank the engine to visual monitor the signal from the VR sensor - You'll need 15 sec. of output to be able to analyse the data.

Note that you should remove the sparkplugs, coil connector, injector connectors and the fuelrelay so that you do not damage anything.

Also i noticed that the output was not there right away - I had to crank the engine for 20 sec. before even receivingg any data from the sensor - The reason for this is that the ignition switch cuts power for the VEMS when moving from ignition to cranking … So the VEMS firstly need to reconnect to the PC before receiveng any data.

In the end i caved in and produced new wirering cables for both VR sensors - In this case i found that a XLR cable has great standing so that's what i used together with new Bosch 2 PIN's connectors.

Idle solenoid

The engine idle is controlled by a solenoid valve where the ECU rapidly switches the ground to allow the valve to adopt the required positionto regulate the idle speed.

The solenoid is connected to the common ignition+ and the VEMS PIN 9 EC36 (Also known as output 4 in the injector group.

To setup the fuel pump signal you go to Outputs > Output Channels

In the dropdown for the Idle Solenoid Channel choose Injector output 4 (7A) Grounding, InjGroup?

Choose ON in Switch testing mode and activate the output on the right side besides the Idle Solenoid Channel - When testing unplug the connector and connect a test lamp.

Throttle Position Sensor (TPS)

The TPS sensor is also not very complicated on this Cosworth - The sensor has three PIN's where it is supplied with ground, 5 Volt and outputs it's signal on the last pin.

Sensor Magnetti VEMS Other

In VEMS Tune you'll need to calibrate your TPS sensor.

Go to Tools > TPS Calibration

Follow the instructions on the screen, it's fairly easy - but very necessary.

PC Communication (RS232)

Before you can connect the VEMS to VEMS Tune on a PC you'll need to connect the RS232 cable to the VEMS.

RS232 Cable VEMS

And please keep in mind that the wirering harness is subject to changes in wire colors, so this documentation might not work out in all cases.

When connecting to a PC you'll need an RS232 port or a USB to RS232 adapter and then the RS232.


The Round display can be installed permanently in the dashboard to monitor al kinds of data, like boost, lambda etc.

The Round is to be connected to ignition+, ground and the VEMS RS232 cable - Yes, the same cable as you use to connect the PC, so both can't be connected as once.

Red wire goes to ignition+ and blue to a solid ground.



Other settings (Random order)

Go to Base Setup > Engine Setup

Set Number of cylinders to 4

Things to keep in mind

The VEMS can be configured with an internal MAP sensor or an external, if you choose an external you can reuse the Cosworth MAP sensor, but keep in mind that the std. sensor should be switched with a Motorsport 3 bar sensor.

In the Magnetti Marelli ECU you'll find the following sensors / actuators that I did not connect.


History / Debugging etc.

History (Getting this thing started)


So we got the ignition gun (Analogue) and thought we were on to the next step, even so we had problems keeping the engine running smoothly - But it turns out the ignition signals on cyl 1 and 3 is so vague that the gun can't pick up the signal, works fine on cyl 2 and 4. Also removing injector connector 2 and 4 has great impact on the engine in idle, 1 close to no impact and 3 some ... We have tried to switch out the coil and the leads + we tried inverting the connections on the primary and secundary trigger with no luck.





Hello Lars,

I have reviewed your triggerlog and vemslog.

Looking forward to your report,

Best regards, Dave

06-08-2015 Update

Hi Dave,

Thanks for responding - i apreciate all the help I can get.

  • Last time i wrote i had problems with my ignition gun - I have now (not confirmed yet) a different result so my trigger setup for trigger after tooth is now set at 100 where it was 85 before ... This means that the engine runs really great, startup is fine, idle is fine, acc. is okay, but above 3.000 rpm or some i have misfiring (As like a cyl or two does not fire properly). I'm gonna get my hands on a new gun tomorrow to make sure my base setup for the timing is spot on .. but this has nothing to do with the fact that the periods are uneven.



Hello Lars,

What was the result of the spark testing from vemstune ?

I have reviewed your last vemslog, periods seem to behave now all seems good.

Your lambda sensor indicates engine is running very lean is it connected and in the exhaust ? If the answer is yes than you might benefit from changing the req fuel like 25% to see if that improves things.

Best regards, Dave

Hi Dave,

Regards Lars

Hello Lars,

All seems well then, engine just needs some tuning; Adjust the fueling so the engine idles around 0.95 lambda and tune the rest of sites in low load rpm direction the map accordingly; That should solve your misfiring; If not please report and provide vemslog showing idling at around 0.95 lambda and misbehave when rev up and i will review.

I'm not sure why your wiki edits keep stating "HELP NEEDED" ? You are moving in the right direction.

Best regards, Dave

Hi Dave,

I would expect periods that does not differ this much - Even if i set the RPM to 3.000 i have a great difference in my period / balance view.

My problem now is (As i see it)

Hello Lars,

The period difference is not that much but seems to be caused by some unevenness in the trigger wheel on crank (difference between 3->1 and 2-> 4 is about the same), you could try inverting crank VR polarity (as a test) to see if that reduces it, take care this also alters timing calibration

Keep in mind of the units used in the relative power window (/injector calibration) it is in microseconds ->

I do not see any alarming individual power difference at idle, take care of the units: average diff is around 200-250 usec, rpm period @880 rpm is around (1/(880/60))/2 = 34msec so its only a 250usec/34msec * 100 = 0.735% difference (when taking the worst case 250usec reading) actually quite even. No need to worry

Good dwell for a wasted spark coil is usually between 2.5 and 3.5 msec.

If you have another vemslog i need to review, feel free to ask.

Best regards, Dave


Hi Dave,

Okay, i see that the balance thingy is not worth looking further in to - Also i'm pretty sure the effect of disconnecting the plug on the injectors are alike on all four cylindres.

Issue: Lambda data and VE table is way off .. Since i got this thing started I have only seen 1.83 lambda, and that continues up to around 2200 RPM, where it starts to move down.

My conclusion - When i have 1.83 in lambda the engine sound good, the idle is around 750 and the exhaust look good, but when i try to reach lambda closer to 1 my engine drowns, the rpm drops and the exhaust goes black smoke / rich mixture.

I have made to config / logs - one is with a VE table at 52 and one is with a VE tabel around 90 - When VE table is 90 the engine almost cant do idle, it smells of gasoline and black smoke is seen from the exhaust = Very rich mixture.

Please have a look

VE @ 52 - Tested at TPS 0 and TPS 7

VE @ 90 - Tested at TPS 0 and TPS 7

Hello Lars,

I see what you mean (after reviewing both vemslogs above), when enriching the engine it definitely runs worse (rpm drops, engine pulls less vacuum meaning less power), what happens when you try to lean it out by lowering the req_fuel say 25% ? When the wideband is outside the measurement range (both way to rich and way to lean) or not warmed up it will show 1.83.

We tested the out of range lean case (by enriching), so there a two options left:

Best regards, Dave


Hi Dave,

Sorry for the delay.

Yesterday i connected a new lambda to rule that out - Changing the lambda had no effect on the lambda value, it was still stuck at 1.83 below 2200 RPM.

Then i mounted a secondary lambda (To be connected to my LC-2 device), i got i connected but my serial2usb was fucked and didn't wan't to connect to my PC .. i tried different thing with my device manager, com ports etc., but no luck, by coincident i started the car (With the VEMS plugged in and VEMS tune online) and suddenly my lambda was 1 ... The rest of the weekend was spent tuning my values and in the end i really thought I was on the right track, but ... more funny stuff was going on.

Regarding the generator - The generator is defective ... it woun't start charging before engine RPM reaches 3900 RPM - I'm gonna change it, but i don't think it effects the above observations.

My concern and suspicion is that my VEMS box is corrupt - Something is not as it should be.

This keeping in mind that i have spent more than 120 hours trying to get this great machine running as i would expect it - Triggers, injectors, lambda values, temperature values, battery voltage etc. I think i have be on the phone with Peter fore more than 6 hours over the past couple of days ;)


Hello Lars,

Review of last vemslog:

Let's not jump to conclusions here, please get back to me with answers/observations/measurements on my checks above and we'll work from there.

Best regards, Dave


This vems oem vr/vr box has pre. trigger error at 4500 rpm low boost.

I changed the pull up pre trig from 27k to 10k and it looks promising for this low tooth count trigger.