Tuning Megasquirt

[Hybrid_From_Hell]

Cutting down logs can also cause error messages / issues

Unsure why but always use the full log

If you still have the full .XLS log and the MSQ PM me and I'll run it through my Quad core Laptop

which seems to do stuff my desktop won't I can also check to make sure nothing is not doing what it should sensor / logged wise

Nige

[Dave W]

If you are happy with the idle you might want to set the filter in MLV so it only processes entries over, say, 1200 RPM. This will also help you process larger files as the filter is applied very early on.

The fuelling map should be fairly flat and it will/can take a number of runs to iron it out. remember the fuelling map represents fuelling over/under the base figure that is calculated from the figures supplied (injector rates, capacity, cylinders, air density (air temp/MAP). You are not setting the absolute fuelling required for a specific RPM/MAP but, instead, are making small corrections to the underlying value.

When you say you used 3 logs do you mean you did a log, ran MLV, corrected the map, did a second run/log etc... ?

Remember that all MLV is doing is adjusting the fuel map to a predetermined mixture. If you aren't sure what it is doing then view the log yourself and see where the EGO level is during your run and, if applicable, see if the EGO correction when it's in closed loop is having an effect and bringing the EGO either side of the target stoich.

Because MLV relies on the EGO reading, if you are using a narrow band sensor it can only correct the fuelling so far each run if the fuelling is a long way out to start with. If your EGO correction is set to a max of 5% then the correction it can apply is approx equal to the range of the lambda sensor plus the 5% correction. For common areas of the map I find it quicker to run a short (10 min max) route that includes as varied conditions as possible (hitting as many areas of the map as you can for as long as you can). Log the route, run MLV against the log, adjust and repeat.

[Bowie69]

Hmmm...thanks Nige. The final 50 min log is actually part of a 2 hour long log - but I had to cut it down as MLV couldn't cope! Even with enable large memory switched on, it still refused to load all of the file and froze when asked to do anything.

I'll go out again today and get another hour or so.

Odd, I managed a 3 hour log on my Intel Atom 1.6GHz. CPU-d laptop... took a whle to process, but was fine

[TheRecklessEngineer]

Thanks Nige, I've emailed you the log and the matching MSQ.

Dave - Yes, 3 logs, evolving the MSQ in between. Once I've got a map that I'm happy with, I'll twiddle the idle to make it proper.

Meanwhile, avoiding the revision necessary for an exam at 9.15 tomorrow morning, I've just done another log (number 4)...

I've highlighted my areas of concern - I would expect the VE map to increase with revs, so why do mine go down?

[TheRecklessEngineer]

Odd, I managed a 3 hour log on my Intel Atom 1.6GHz. CPU-d laptop... took a whle to process, but was fine

I suspect it is more a vista related problem, rather than CPU speed. My laptop is an Intel 2 duo core machine @ 2.0Ghz and 3GB Ram. It shouldn't be short of crunching power...

[Dave W]

I've highlighted my areas of concern - I would expect the VE map to increase with revs, so why do mine go down?

You need to look at the log files yourself if you have concerns. Look for areas that correspond to the bins in question and see what the EGO is doing at that point. Running analyse in MLV isn't the only thing you need to tune the engine, you need to look at the log yourself, see what's happening when you're in those areas. If the EGO is high and Analyse is recommending a reduction in VE value then fine, if EGO shows low (weak) then you have a problem.

Remember that the VE bins are there as a correction to an underlying calculation, they are not an absolute value. There is no need, in theory, to increase fuelling as RPM increases because the fuelling is calculated/injected per revolution. Just because an engine is running at 4000 RPM doesn't mean it needs more fuel per revolution than it would at 400 RPM. The engine capacity doesn't change with RPM, the injectors don't change with RPM, all you are doing is compensating for differences caused by air intake, exhaust and ignition advance. Your peak torque and peak power positions will also have an effect on required fuelling.

The 72 figure at top left is probably there because you've never managed to get the engine into that state while logging, near full engine load/throttle at 600 RPM isn't something you'd expect to sustain so you'll never get anything meaningful in there using MLV. I'd be quite impressed if you ever managed to get the engine into that area of the map

[FridgeFreezer]

The results will also depend on your Exhaust Gas Correction settings as that determines the response of the ECU and what pops out in the data log - for example if you set the ECU to correct too quickly (short time period between corrections or large correction value) or slowly it will be throwing things off as there is a natural delay between a correction happening to the fueling, and the fuel being injected, burnt, spat out of the exhaust, hitting the lambda sensor and the lambda sensor responding. You've got a fair leeway with it but you do need to be aware of how the measurements are being made.

[Bowie69]

I suspect it is more a vista related problem, rather than CPU speed. My laptop is an Intel 2 duo core machine @ 2.0Ghz and 3GB Ram. It shouldn't be short of crunching power...

Ah, fair enough, I'm on XP by choice