Members Page/Kevin Black/VWGTI Another

MembersPage/KevinBlack/VWGTIAnother (2008-07-03 21:21:55)

Project page for (another) 1994 VW GTI Turbo 16v

[ Engine]

[ ECU Order]

[ Wiring]

[ Dynos]

[ News]

Dumps:

Mcd:

http://www.vems.hu/files/KevinBlack/mcd-saved.txt

Mct:

http://www.vems.hu/files/KevinBlack/mct-saved.txt

Custom EC36/EC18 Wiring harness

12v IGN: 14 AWG to pin 25
GND: 14 AWG to Pins 5,21,22,32,26 (ridiculous overkill) , sensor GND wires on branch (per UserGuide) to pin 26, all GND wires then go to 10 AWG wire to Engine Block, at OEM location for 1993-1998 VW 2.0 engine.
Injectors: Pins 7,19,8,20
Ignition: Originally used Pin 35 Ch0 to dummy ignitor, eventually used Pins 34 and 36 (Coil 2 and Coil 3)
IAC wire ran to Pin 17, never used or plugged in to anything(tied up, no chance it was shorted to ground)

Analog 5 Pin EC18-16, Launch Control input using 2700 ohm pullup to pin EC36-29
Analog 7 Pin EC18-12, tested for WOT shift but eventually used for input from Autronic Air/Fuel Analyzer.
Autronic Air/Fuel Analyzer analog ground is connected to chassis ground (afterthought)

Eventually switched pins between 16(now Wideband) and 12(now LC), to allow the wideband to be logged in Megatune "Misc Voltage 1"

Ch4 Output Pin 3 - Tach, 1000 ohm pull-up to 12v.
Changed to Ch 0 Pin 4 (to test output, after that found bad instrument cluster and changed)
Ch5 Output Pin 15 - wire triggers OEM fuel pump relay and additional relay powering 1) Injectors 2) Wideband Pin 3 and Flyback to VEMS and FJO Driver.
Ch6 Output Pin 31 Fan Relay
Ch7 Output Pin 30 Shift Light

Stepper outputs - wired as spares in harness, never connected to anything.

also of possible note, I was using a DC power inverter to power the laptop
- hopefully not. Assumedly EC36pin26 GND was connected to notebook RS232 pin5 (GND).
  - The COMM wiring is as recommended so yes DB5 connects to GND.
  - How can power inverter damage the ECU? And why is there no warning for it? Better yet, how can the durability of the ecu be improved and the damage be avoided?

custom FJO fuel injector driver

Connection = (4) in, (4) out, (1) GND (12 AWG) and (1) INJ Power (I assume for Flyback)
Flyback wire to VEMS is still present
- do you know that injector outputs can only pull to ground ? (that is pullup might be needed, perhaps internal in the FJO injector driver, did you think about this ? Measure voltage when inj output inactive)
  - Yes, and No. All I know is that the FJO Injector Driver is a commercial unit being used by thousands of users, many by (modified) OEM Honda ECU applications (If this helps answer your question?) and are installed the same way as this one.

dig up some FJO docs
Measure the inj output
- Ignition ON (FJO supply present)
- ECU injector output disconnected from FJO input
- measure DC voltage on FJO input (if it's +5V or +12V).
- optional: pull FJO input to ground with a 1k resistor,
  - again: measure DC voltage on FJO input (if it's +5V or +12V).
    - TODO on Wednesday

VEMS v3 webshop assembled

1.0.73 Firmware

MSQ File Here:

http://www.vems.hu/files/KevinBlack/Track5.msq

Custom .ini file for Megatune (could not upload, emailed to Webshop)(I changed/organized the tuning function titles, no major changes)

Symptoms that might have predicted a problem

1) TPS signal very very dirty For the first day of running/driving, the TPS signal was erratic and caused very poor driveability. I moved the spark plug wires away from the TPS but it didnt do much. The problem "went away" the next day or so with no changes to the TPS enrichments or the VE table.

2) Wideband/lambda readings very spikey The lambda readings from the beginning were ok at and near idle rpm, but get very dirty when driving. The sensor is new and has been calibrated to 20.8-20.9% O2 and the Pumpzero_PW and Nernst values were supplied by the Webshop.

Problems

Many problems! (Likely related to the mcp3208 damage)

Problem #1 Twice, shortly after making a change (what change?) in Megatune, the ECU "locks up" and causes the injectors to stay open and fuel pump to run, hydralocking the motor.

- Second time change was the Fuel pump power-up time, from 5 seconds to 60 seconds.

- First time, I do not remember, likely Start Fuel change though as I was tuning the startup.

In my case it cost me a USD $280 Bosch LSM11 O2 sensor!! Also new oil for the engine (twice), and several hours lost (work and sleep).

How can it be prevented?
- properly planning an install before doing the wiring is the way to go, it often saves a lot of bucks.

- I don't know the difference between "planning" and "effective wiring". I am not a hobbyist, I do this professionally for 2 years and overall for about 8 years. I did plenty of planning, please see the wiring information.

That means to plan

wiring (inputs and outputs)

choose firmware and trigger setups to start from (reference to similar setup)

- firmware and trigger same as VEMS #857, which has been running for about a month.

matching outputs h[0] and h[2]. If you instruct the ECU to activate an output and the outputs control injectors and fuelpump (it is possible to activate max 8 injector outputs at the same time, see GenBoard/Manual/DigitalOut/Table )

- please see wiring details and MSQ file in link for output setup.

in case of any output issue, all output channels are interesting (mcd dumpDone, variables ending with _channel except shiftcut_channel that is not output but input selection)

It is best to publish for review (split to subsystems at least trigger, inj and ign) before doing wiring and powerup.

- Felt it not necessary due to other successful project (that I made myself as a 100% plug in ECU)

HW connection problem is also possible

bad grounding ?

- Overkill, please see Wiring details above

eg. "stealing input switch ground" (not from EC36pin26 ground but) from chassis while some or all grounds are missing (or chasses to engine ground missing ?)

- Engine to Chassis Ground present

- Only thing I can think of is I used chassis ground for my Autronic air/fuel analyzer

flyback ?

- Connected to EC36-23 and to FJO injector driver.

overloaded output ?
Shift light to EC36-30 is close to max (~300 mA)
- this should not be a problem. Max current for one p259 output is actually 1A (if others are not loaded). Shiftlight is not inductive load, which makes it even less likely to cause problem, even with misconfiguration ("prelling")

extra analog input out of range (pullup to +12V or direct +12V applied ?)

- Autronic meter supplies 0-5 volt only - confirmed.

- Launch Control Pull-up is to Pin 29 5v

perhaps using battery break switch to disconnect battery while engine (and alternator) is running ?

- No switch used

This is only for accident. (especially if there is no external max 1A fuse and 18V transient diode in the harness)

other possible (though unlikely) causes:

using different megatune (1.1.x for 1.0.x firmware ?)

- I am using a custom .ini file (by me). I will e-mail this to the Webshop as the File Manager will not allow it to be uploaded.

- At the top of the .ini file this appears:

"vems.ini for Megatune2.25 + vems firmware 1.0.x (stable1_0 branch)"

almost flat battery ?

- Yes, the car has a small battery that went low 20+ times in a week. Battery charger was used, jumper cables were used only AFTER the second ECU lockup.

MCP3208 damaged

After turning the power OFF (how exactly?) for a few minutes, everything was fine the first time, however the second time I think the (edit) MCP3208 was damaged.

- Simply turn off Ignition Switch and wait the first time. The second time I unhooked the Battery because I was fearful of the ignition coil possibly being stuck ON (wired to battery that day, long story). (Ignition is now a VW VR6 2x3 coil pack(using only 2x2), using coil drivers 2 and 3, 3.0 to 3.5 ms dwell)

this shows that quite certainly something was connected wrong. Bad grounding ? or >5V (+12V ?) applied to extra analog input ?

- Please see wiring details and timeline below.

I now have no 1) EGT 2) Launch Control stuck ON
- no surprise, this confirms mcp3208 was damaged. It is also possible that the main processor got partially damaged (could also explain the injector outputs)

- Note that the fuel was injected Twice, the first time (days before) there was no issue with the analog inputs, so maybe this timeline will help you diagnose the chain of events?

_______________________________________________________

Problem #2 - perhaps also related to the same damage that damaged mcp3208 (although mcp3208 is not directly involved in wbo2 control).

The wideband O2 readings are OK at idle and low engine RPM, but jumps around greatly at medium RPM and higher, making the sensor unusable. The pumpzero_pw and Nernst values were used as supplied with the VEMS, and the sensor was calibrated to 20.9% O2.

What could be the cause for this problem?
- first, compare wbo2 settings to a [known good] (pumppw zero, nernstdc target and wbo2 calibration will be your own of course)
- most likely: doing the wiring before planning it => killing mcp3208 and perhaps other internal with bad connection

- Wideband problem was from the start, mcp3208 did not "die" until 7 days later.

possible note is I used the Fuel Injector relay output (terminal 87, shared with fuel injectors and the FJO fuel injector driver), to power Pin #3 of the Wideband O2 Sensor
- good note. Unlikely to be the cause of the problem though

To allow the car to be tuned, I used an analog input for my Autronic exhaust analyzer. I set the analyzer to 0-5v, 10-15 AFR so I could view the Megatune logs in voltage and add a "1" to get my AFR. However I could not do this after the (mcp3208 was damaged.
- this is one of the operations that could damage the chip.

- Any more information? Are you thinking the voltage spiked over 5v? How can I prevent this from happening?

Minor Questions

1) Map Multiplication

What does the "Map Multiplication" in Megatune do exactly?
- the pulsewidth would be multiplied "behind the scenes" with MAP. The fuel VE table often becomes "flat", but not always very flat especially at low kPa. VE value approx. 50% at 40 KPA, 200% at 250 kpa.

- Injector Settings:

InjOpen: 0

Rampup: 1008

Battfac: 4080

Battfac: (us) 304

PWM DC: 0% (should this be 100%???)

PWM Peak Time: 25.5

Divider: 2 (using 2 Injector Outputs 10(8+2)and 5(4+1))

Fire Banks: Alternate

2) Ignition Output Power

I found that the logic level Ignition output does not trigger an MSD ignition, I had to use a 12v power output.
- this is well possible, actually a common method to use IGBT power output

- TODO: find out is 255 ohm works and then specify that from now on for all orders (I have many applications using VAG COP coils)

- I would prefer to use logic level (or stepper) to trigger MSD rather than IGBT (less heat(clamping platye is tricky to get "right" and not "bow up" in middle, more reliable as less components are being used)

I ended up changing the ignition to a Webshop 2x2 coil pack using 2 12v drivers

3) TPS Enrichment

The TPS Enrichments are not working great (often adds fuel on small throttle transitions)

what are the tps bins used? (the lowest one is most interesting, obviously). Is the TPS signal otherwise clean ?

- TPS signal was originally very dirty, caused poor driveability. thought cause was possibly TPS near spark plug wires, moved the wires away from each other but this did not directly co-inside with "fixing" the problem. The problem "went away" by itself.

- Maybe that was a "red flag" to a bigger problem?

- Current TPS Enrich Settings:

- Bins: 1,5,20,40

- Amount: 0,1,2,4

Old News

Car is up and running on VEMS

Currently using 60-2 crank trigger only, semi batch fire injectors

Rochester Low-Z 1000cc injectors, FJO external Injector Driver unit

VEMS 400 kpa External MAP sensor