pcmhacking.net

There is adjustment for the amount of revs and time from memory

I just took him for a few km dirt rally. I've never had the back slide out from a heavy foot in this car. He has a pretty large footprint with the Porsche body kit and oversize wheels.

The TPS re-learn thing didn't work. Trying to engine brake at what feels like half throttle is exciting, there is no vac brake booster on these bugs.

The mix was getting leaner as I drove and there was a fair bit of knock at high load towards the end of the drive. I'll wire in the O2 gauge output to the ECU and see if that helps. I'll need to pull a fair bit of timing too. Even with this poor tune he's never driven so well.

Next up is migrating to 12P, Tuning the VE table, Port injection then forced induction!

Thanks everyone for the help so far!

I think we need to see a pick of this 'franken bug'...

Biggvl wrote:I think we need to see a pick of this 'franken bug'...

That's a good idea and how come it's a he?

You will need to adjust the tos voltage scalar if max volts are out of factory tollerance. Then it learns max and min pretty much from normal use, dont really need to do anything special as it gets the range from normal driving. 12P datastream will give yoy tps% and tps volts.

Sorry, not the best pic. Its still dark outside for a recent one. The tps is driven by the ecu's 5v supply so there shouldn't be any difference from one system to another. Maybe a slight variation between TPS' s.

There is no air bypass screw only a throttle plate idle screw. Maybe it is resting too far open to trigger an ilde condition. I really need 12p in there.

I was thinking about VE yesterday after restricting one of the exhusts and watching the ECU really struggle (Why? to try flood the muffler with exhaust gas and get a reading on the O2 sensor). VE on my last ECU was guestimated by using the torque curve of the engine and known idle VE's. This got it pretty close and was good enough.

I'm thinking of building a datalogger using one of my many MAF sensors I have floating about. It'll monitor RPM+MAP and log the MAF under certain conditions like no or low fluctuation in MAP etc... MAF sensors are pretty well documented and you can pull the lookup tables out of various ECU bins. Most of the hard work has already been done.

Does something like this exist in the DIY scene? Or is there a reason why they don't?

kojab wrote:

Biggvl wrote:I think we need to see a pick of this 'franken bug'...

That's a good idea and how come it's a he?

I came home one afternoon and my girlfriend told me she won a beetle on eBay. She didn't know if it ran or if it had rust. Turned out to be about 15km down the road so we went to pick it up. Max speed of around 40km/hr on 2 cylinders. The seller guaranteed us it was rust free. After getting it home the door pillars were rusted through, firewall was hanging there by the carpet that was glued to it, the rear window surrounds were rusted through. It was a mess! It did have a new floor pan welded in so it wasn't a complete write-off.

We sanded it back and begun replacing pannels and fabricating bits that weren't available. At one stage I had 'stitched' the new panels in, that combined with all the donor parts from other engines' EFI systems he became known as Franenstein, and naturally a He.

I've attached a few pics at various stages. Some of the ECU's I've built for him, The test rigs for the 36-1 trigger wheel, My original foundry setup etc...

The VE numbers themselves aren't really important, and once you have 12P talking to your wideband you can work out the VEs pretty quickly by logging with an adx that includes CalcVE.
i.e.:

6 uploads per post... here's the casting + ECU#1.

A total of 3 ECU versions were made. The first was based around the original Opel ECU where the TBI came from. It didn't have an MAT sensor which I soon realised it needed. A 20deg air temp change works out to something like 5% change in air density. The first version was a fuel only ECU.

Second version was a combined fuel and spark ECU, I corrected the mistakes I made on Ver1 and had him running and driving on it. I didn't have Injector Voltage compensation so he ran poor with headlights on etc... The little 11mhz AVR was really pushed to its limits in this application.

Third version I decided on a 32bit ARM core @72mhz. All coded in assembly I had sooo much processing power at hand. CPU load was around 4% when running and about 10% when logging. The particular core I chose had a very noisy ADC which I didn't realise until I was a month into the coding. This was still an improvment upon Ver2. A hdd crash lost all my source and I didn't feel like starting over so now I have a Delco.