Porky EFI

Rich H · Post by **Rich H** » Wed Dec 12, 2007 8:40 pm

Interesting! I would not be suprised if the table was way off, everything else is! The VE table is a percentage of the required fuel at 100 %VE

No idea what the cam duration is, its supposed to be standard. Volume is right, but the injectors are still a guess. The PWs are quite long for crank so I'm thinking I have got titchy injectors so they need much longer to get the fuel in. As a result all the tables are cock.

I'll start form scratch and see what happens. I estimated 20 lbs/hr for the injectors, I'll cut it to 10 lbs/hr see what happens.

Rich H · Post by **Rich H** » Wed Dec 12, 2007 8:52 pm

How about this, I'll up the fuel pressue aby half a bar to 3.5 bar. That will give a better flow.

What do you guys think?

steve_weegie · Post by **steve_weegie** » Wed Dec 12, 2007 8:58 pm

Have you got the datalogs from the MS? I'd be keen to look at how it was running with the MAP sensor reading atmosphere and the engine running. If it runs with the MAP sensor reading atmosphere then we can tweak the base map to suit.

Ignition timing would be affected by the MAP sensor too & the logs might tell more.

Rich H · Post by **Rich H** » Wed Dec 12, 2007 9:29 pm

No data logs (I forgot!) and the spark will be locked on 10 degrees to remove it from the equations. I'm going to up the fuel to 3.5 bar and then start again. Now I know it will run I just need to work out how...!

Got the weekend off to tinker...

Rich H · Post by **Rich H** » Wed Dec 12, 2007 9:33 pm

Even better I'll try the other injectors see if they will fit.... Might be easier in the long run and it also means less wear on the fuel pump.

robin · Post by **robin** » Thu Dec 13, 2007 1:42 am

Is the VE table 3D or 2D (i.e. does it map RPM->VE or RPM,THROTTLE->VE).

I am assuming it's the former.

Can you just post up the whole configuration file - I assume it's some plain text file ...

Cheers,
Robin

gdr · Post by **gdr** » Thu Dec 13, 2007 8:39 am

40 psi should be OK for fuel pressure. I doubt if the Duratec injectors as low as 10 lb/hr. Still sounds like the required fuel is wrong. Have you set up the correct squirts per cycle in the injection control table? Probably want 2 squirts alternating to avoid dropping pressure in the rail too much but 4 squirts batch might also work. I remember getting a bit confused with this and I think it is easy to end up with the req fuel 1/2 what it should be.

Rich H · Post by **Rich H** » Thu Dec 13, 2007 8:44 am

Its RPM/MAP -> VE and RPM/MAP -> Spark Advance, proper 3d mapping with 8x8 tables. Once its running I can map on the fly, real time adjust. None of your cheap 2d map stuff

I need to calibrate my temp senders again (it reads 120deg when the car gauge says up to temp and the IAT is 5deg high) and the timing is a tiny bit out, maybe 1/2 degree.

I'll get some logs too.
The VE map is the default at the moment

robin · Post by **robin** » Thu Dec 13, 2007 10:23 am

Ah, I see. So it uses the MAP as an input to the VE table and then multiplies the result by the MAP again.

So if your engine idled with MAP hose off then you would have been using the top left cells of the VE table to compute fuelling, whereas with the hose connected you'll be using the bottom left.

A perfect 2.0L engine would draw in it's 0.5L per cylinder on each intake stroke. That 0.5L would be at the manifold pressure, of course. So a perfect engine would get 100% in all cells.

Now for most of the VE table you will be reducing VE because your engine is not perfect.

However, for the idle region you might have to increase above 100% in order to get enough fuel in (remember a lot of the idle fuel will be lost due to fuel drop out caused by low air speed). Also, if you're not timing injection pulses off crank angle then you will generally miss the valve opening and be injecting against a closed valve; this means that the fuel will pool in the port and must then vaporise into the airflow when the valve does open - this will not be very efficient at idle due to low air flow, so pile on more fuel.

Cheers,
Robin

Rich H · Post by **Rich H** » Thu Dec 13, 2007 11:13 am

I don't think its that clever TBH, it seems to work out the required fuel at 100% VE at max RPM then scales it, so the top right of the table is 100% and the bottom left is around 30% It works purely on PW.

I have lowered the injectors to 15lbs/hr and reset everything and it coughs into life straight away (Smaller injectors so open for longer) with the MAP hose connected. It dies shortly after but no see-sawing so that is down to the default warm up enrichment table being rubbish. Looks like I might have it cracked... he says... Oh I cranked in the idle screw as well so that probably didn't help...

Got the IAT to read properly too but the CLT is still giving guff. It was reading 29 deg not 5 now its reading -29

bit of tweaking still required.

Oh and the Elise was parked behind the Porker and is covered in black soot now....

Rich H · Post by **Rich H** » Thu Dec 13, 2007 11:20 am

The MS takes a crank signal from EDIS direct then feeds EDIS with a signal to advacne or retard the spark timing. All the timing is done from the crank trigger wheel.

robin · Post by **robin** » Thu Dec 13, 2007 3:17 pm

Rich - you need to understand the fact that required fuel at 100% VE is the same no matter what the RPM (each stroke consumes the same amount of air and fuel, whatever the RPM!).

I've looked at the fuelling equation for the MS.

It looks up the VE table based on engine speed and MAP.

The VE table entry is actually two pieces of information combined - it's the relative VE (i.e. how much worse than "perfect" the engine pulls in air in these conditions) and the input AFR (i.e. how much different from 14.7 do you want it) multiplied together. Note this reference to AFR has nothing to do with closed loop running - it's just a multiplier to scale fuelling up or down at different points in the rev/load range - it's used to make a lean idle, a slightly rich mixture at peak torque and a richer still mixture at full power. The VE table is obviously made of discrete cells; the MS interpolates the four neighbouring cells to produce the correct value of VE.

It then multiplies VE by the % actual pressure in kPa (for n/a engine typically manifold pressure is less than 100kPa - at idle, much less, on wide open throttle, just a tiny bit less). So if manifold pressure is 30kPa, it multiplies the result of the VE by 0.3.

This is a crucial step - it means that the VE table should start with 100% in each cell for a perfect engine and should only deviate from this where the engine breathes less than it theoretically should, or needs to be richer or leaner than 14.7. The reduced breathing that is simply down to reduced manifold air pressure is applied after the VE table lookup and does not need to be represented in the VE table itself.

It then multiplies the result of this lookup by REQ_FUEL (which is your pulse width for full torque fuel load at 14.7:1 based on knowledge of injector flow rates, fuel pressure, engine capacity, at 100kPa pressure, ~300K air temperature).

It then applies the various corrections (warm up, throttle transient, air temperature, barometric correction, battery voltage compensation, closed loop correction when enabled).

Here's an extract from MS.info:

Code: Select all

As we saw above the fuelling equation is:

PW = REQ_FUEL * MAP/100 * VE/100 * GammaE/100 + Inj Open Time

So to see an example of how MegaSquirtÂ® calculates pulsewidths, we will look at a low rpm cruise point data from a datalog file, with MAP=40 kPa, VE=74%, GammaE=97%, and a reported pulse width of 4.0 milliseconds, with constants: Req_Fuel = 10ms and
Inj Open Time = 1.3 ms.

so: PW = 10 * 40/100 * 74/100 * 97/100 + 1.3 = 4.17 ms in this case.

In there you can see it starting with REQ_FUEL=10, scaling it back by 40/100 to correct for MAP, then by 74/100 because that's what was in the VE table for that engine speed and MAP, then the other niggly wee corrections.

So, if I were you, I would fill my VE table with 100% in all cells covering 0-2,500 RPM to start with.

Cheers,
Robin

Rich H · Post by **Rich H** » Thu Dec 13, 2007 3:47 pm

Wow. Thanks Robin! I think I follow...

I changed the injectors to 15lb/hr and reset pretty much everything to default.

It starts and idles for 5 secs then dies from stone cold, but it thinks that the coolant is at -29 degrees. I'll look at the crank fuelling at -29, fix the coolant calibration and then reduce the injector size until it matches that required fuel figure at ambient (These injectors are tiny!) That will give me a place to start from.

Then I need to start on the Warm Up Enritchment so it stays running (Assuming it dies straight away still) At this stage I'm looking for a steady start.

Would the 100% run the risk of flooding? would it not be btetter to come up from lean? Its not under load so shouldn't be getting very hot and the plugs are a pain to get out (Some tit put a coil pack in the way

)

Rich H · Post by **Rich H** » Thu Dec 13, 2007 3:53 pm

I think I might swap these injectors out before going further. If the injectors are under 15lb/hr then they are a waste of time, the cream top injectors are 20lb/hr. That assumes that they will fit...

Rich H · Post by **Rich H** » Sat Dec 15, 2007 8:52 pm

Injectors swapped, much better! However a 3 bar fuel leak caused me a fair bit of concern! It goes a flipping long way!

One of my injectors blew a seal due to a faulty clip while idling. Not good at all. Fortunaly I was reasonably on the ball and turned it off PDQ!

Dried it all out, put it all back to gether with a Zip-tie to make a mechanical lock that secures the clip that hold the injector to the rail. Much better than before.

Runs and idles alot better even on the base map. Well it did when I corrected the number of cylinders...

My coolant sender is still causing problems, so I have swapped the bias resistor hopefully to fix it. I think the original bias was 5x too high causing resolution problems and the MS was getting confused.

Starts ok but won't hot start as the coolant warms up ok but after a time it defaults to -40 which disables the warm up routines.

Question: I have a problem with run on one the ignition is switched off, I think that the Lambda control unit manages to keep the MS running, which in turn holds on the fuel pump. Any idea how I can create a hysterisis loop to kill that off? I'm probably going to fit a master ignition switch as it could be a bit scary if I need it off in a hurry adn it won't stop!

Ta
Rich