Another injected Elan

[peg_pilot]

Here are some photos of my injected Elan. It is running quite well, but I spend all my time adjusting it anyways.

The throttle bodies are from the Kawasaki GPz-1100 Motorcycle. I threaded them with my mini lathe and made up adapters to connect them to the Weber flanges on the cylinder head. The cold air box is a piece of PVC drainage pipe (very elegant, no ?) and the intake side adapters for the throttle bodies connect to it via standard plumbing slip joints.

The fuel system is closed-end, I pulse-width-modulate the drive to the fuel pump to control the pressure sensed by this transducer:

I think this is much more elegant than having to run a return line back to the tank. It also reduces current draw, which is a consideration since I am still using a 20 amp Lucas generator. The system pressure is set by software, which is much more flexible than a mechanical regulator.

I made up a surge tank by machining an extension for an SU carburetor float bowl that fits between the body and the lid. I replaced the needle valve with a larger one I found that is intended for use with a Chevy truck carburetor since I didn't think the flow rate of a single SU carburetor would be enough for the twink. You can see part of the surge tank in this shot, it is in the nose and bolted to the air cleaner housing.

The surge tank is fed by the usual mechanical pump on the engine, since it has a float valve, there is no need for a tank return line. The high pressure pump is bolted to the floor of the nose so it gets some positive head from the tank.

Instead of an idle air controller valve, I decided life would be simpler (because of the independent throttle bodies) if I just made the mechanical throttle stop adjustable, to this end I have a small stepper motor linear actuator as seen here:

Having closed loop idle speed control really makes driving the car more enjoyable.

The control computer is my own design, I based it on the Motorola 68332 32 bit microcontroller, which is a fossil, but has a nice coprocessor for generating timing signals. This latter makes sequential injection very easy. I also is NOT a BGA package, it was hard enough hand soldering a 132 pin flat package to the circuit card. I adapted some open source software to run on my machine. Despite the independent throttle body setup, I am getting good results with the speed-density control algorithm. I generate a manifold air pressure signal by searching for the minimum value over an engine cycle.

[Europatc]

Looks like you have been busy Peg Pilot
all the best
Stuart

[john.p.clegg]

Peg
Nice....same throttle bodies as mine...for the moment...am aiming to convert to single throttle and plenum.

John
Megasquirt Extra

[elanman999]

Peg,
A nice installation with some clever ideas. I like the fuel rail pressure control and the cold idle speed adjustment.
One question, why did you go for manifold pressure over throttle position for load measurement?
Cheers
John

[peg_pilot]

It was my impression from perusing the Megasquirt website that manifold pressure is a more linear indicator of engine load than throttle position. In particular they claim that at lower engine speeds there is often a large range of throttle positions which all correspond to essentially the same load, e.g. "wide open throttle" may be reached well before 100% physical throttle position.

[elanman999]

Peg,
Whilst I would agree that when using throttle position as the load signal it is non-linear to air flow this is not a problem as the system will be mapped after installation and so this can be accounted for. Particularly as most modern ECU's have setable load (and/or RPM) points allowing one to add more points where needed if necessary. Above all the signal is repeatable.
Against this when using ITB's, as you are, the manifold pressure signal at low loads is far from steady. I would be interested to know exactly how you are overcoming this issue.

When I installed my system all of the (UK) after market ECU manufacturers recommended using throttle position when using ITB's, it's not perfect but it is reliable.

Cheers
John

[peg_pilot]

John,

I sample the manifold pressure at quite a high rate (~3 kHz) and then I take the minimum value over the period of time corresponding to one engine cycle (which I can determine from my crank angle sensor, which is a variable-reluctance type that I have mounted in the bell housing to detect the teeth on the flywheel ring gear). The minimum value turns out to be quite repeatable at idle (and other conditions) and I further low-pass filter it to get my final signal for lookup in the volumetric efficiency table. I find the minimum pressure varies from about 32 kPa at idle to about 80 kPa with the throttle all the way open.

Because I have a lot of extra memory on my controller, I sometimes save all the 3 kHz pressure data in this spare RAM (I have written some software that lets me upload the contents of the controller memory to my laptop). The shape of the pressure vs. time curve at idle is quite interesting, there is a substantial peak when the exhaust valve opens, which seems to indicate that there is a lot of backflow at idle. I am surprised I don't get backfiring. Again since I use only the minimum value of the pressure this peak doesn't affect the fuelling calculation.

I do not know if this approach is "better" than using throttle position - I have not tried it to see. In any case I am still fiddling with the VE table to get as close as I can to the ideal air fuel ratio. I find it very challenging to establish steady state engine operating conditions and drive the car at the same time.

Dan

[john.p.clegg]

Yes,it's not easy keeping one eye on the laptop and one eye on the road...

John

[elanman999]

Dan,
Thanks for the very full reply - right, I understand now, a very novel approach to filtering the signal.

John- Yes so true. I found it best to set the laptop to log the data, drive, then stop to view the data and make any changes.

Have fun
Cheers
John

[crypto]

Hi Dan

Very interesting project (and eager to see/learn more). What was the main reason for the conversion ?
I believe many engine ECUs did use the 68332 for this purpose because of the TPU.
Was it pure joy of tinkering or why didn't you use a commercial ECU ?

Any figures about power-output or fuel consumption ?

I once had a car with a mechanical bosch injection (OPUS) which I intended to convert to a electronic/sequential system. When the engine was very hot, and I turned it off, I often had problems starting again because of vaporization of the fuel. In this case I was glad to have the pump running a moment to get "cold" fuel from the main tank.

Peter

[peg_pilot]

Peter,

The reason for the conversion was just the joy of tinkering, I am an engineer and can't look at a piece of machinery without thinking about how else it might have been designed. I especially wanted to do some electronics design, since I had not done any in the past. I am most proud of the tachometer drive circuit I came up with (the nice thing about the custom ECU is that it can take on more duties in the car, it also flashes the turn signals and handles turning the radiator fan on and off) which is based on something I found in a fabulous book: "The Art of Electronics" by Horowitz and Hill. It is a temperature-compensated current source, so the deflection of the needle is hardly affected by the temperature of the meter coil.

I was quite concerned about fuel vaporization - that is why I located my surge tank and high pressure fuel pump in the very front of the car, where there should be a good supply of cooling air. Also, the surge tank is above the pump, so there is at least a little positive pressure head to reduce the chance of boiling. Time will tell....

I have no idea about power of fuel consumption, I was looking for a place to take the car where I could put it on a chassis dyno, but I didn't find anything near me.

Do you have a J type coupe ? I have 29J, which from what I gather from the LotusElanSprint web site started life as a series 4 and then was upgraded to a Sprint when it presumably failed to sell. I think the FHC is very nicely done, very good looking.

Dan

[crypto]

Dan,

Don't understand what you exactly mean with "J type coupe". I own a 1972 Sprint FHC, which is almost totally apart at the moment. It's a UK version and I converted it to LHD. Currently I'm stripping the car down to fiberglass to repair all the blemishes and making it ready for new paint. Hope to have it ready this summer.

I like doing mechanical (and electrical) work. Repairing fiberglass is ok too, but taking the paint off was/is a nightmare and I'm glad it's not an old Cadillac or something even bigger ...

I'm an engineer too, developed hardware and software for 68k series, powerpc 8xxx and now for the QorIQ family. And as you sad, the engineers way of thinking is to improve things with more modern technology. I wouldn't go as far as you by developing an ecu from scratch, simply because with the time available, I never could finish it. But using a megasquirt is a good basis, guess there is plenty of hardware and software around.

Btw. can you post a photo of the crank / cam sensors.

Peter

[peg_pilot]

Peter,

Elan serial numbers (after some point in time) that end in J designate a FHC for the American market. If I had been thinking, I would have looked at your profile and seen that you are not in the US and therefore would not have asked the question. How do you get parts in Switzerland ?

Here is a picture of my cam sensor, it is an infrared photo transistor (made by Crane Cams) mounted in the distributor. You can see the shutter wheel, it has four slots, I widened one of them with a file to distinguish cylinder #1. I put a bolt through the upper and lower parts of the distributor shaft to disable the centrifugal advance.

Here is a shot of the crank sensor, it is not very illuminating, I'm afraid. You are looking at the flywheel cover plate below the oil pan, which latter is just visible in the top of the frame. The sensor looks like a short piece of threaded rod with two wires coming out the end. The small cap screw you see on the bottom of the bell housing holds a small cube of aluminum that is unseen behind the flywheel cover. This forms a mounting bracket, I tapped a hole through it to hold the sensor and the nut on the flywheel cover keeps it from unscrewing itself. The sensor is looking at the side of the ring gear teeth. The sensor is industrial rather than automotive, it is made by a company called AI Tek.

What sort of engineering do you do ? Given that you worked on 68xxx systems, would you by any chance know where I could get the source code for a TPU debugger ? I have an old DOS executable, but it doesn't do me much good because I use Linux and while I have a DOS emulator I don't think it gives access to the ports.

Dan

[crypto]

Dan,

Thanks for the pics, looking good !

Regarding the parts, typically I order them from the well know suppliers in England or via ebay (as the elan shares so many parts with other British car brands). They are easy to get but import to Switzerland is sometimes expensive because of fees, tax and import duty.

Regarding source for a tpu debugger I can't really help. I hardly can remember using tpubug (and I believe that was written in Borland C), but I think you should be able to get it running in a VM under Linux. Years ago I used some DOS software with COM ports in a VM and that was working perfect.

Peter