Megasquirt I 12x12 tables big enough (ITBs)

[Si.G]

I had a quick look at ITB mode, looks interesting, not sure it give me anything over hybrid alpha-n? I am getting about 66KPa at idle and as soon as you touch the throttle you are straight into the 80s. Might be worth a look though....

The big question I have for everyone is an opinion on the following.

If you rate MS1 as a zero
and a haltech sport 1000 as 10 for arguments sake...

Where is the MS2 PNP currently available from diyautotune ?? a 5?? (Or any MS2)

opinions please - just part of weigh up my options

Si.

[NitroDann]

Id rate it at least a ten because it can do everything an mx5 ever needs to do, has huge community support and is cheap as chips.

Dann

[sailaholic]

Not sure yet, will be talking to the tuner again.

Sent from my GT-I9100 using Tapatalk 2

[Si.G]

Id rate it at least a ten because it can do everything an mx5 ever needs to do, has huge community support and is cheap as chips.

Dann

Hi Dann, I totally agree with you on the above, I guess my question is really about how much better is the MS2 in comparison to the MS1?

[sailaholic]

I can't give a proper comparison as I haven't played with them enough.

However the pnp is limited from my under standing. And it might have good community support but it doesn't have good tuner support (which is important if your paying tuning time)

The maps are still limited compared to others, some say not needed, but if it's not there you can't use it.

Personally I hate how big it is.

So for me 5. It's awesome for price, but not there as a total package.

Sent from my GT-I9100 using Tapatalk 2

[NitroDann]

The big tables arent needed.
PM glslender and he will explian why.

And what tuner support?

The instant the car starts, and run on all 4, tuning works the exact same way ecu to ecu.

Its just a table full of numbers.

Dann

[sailaholic]

Last time gslender tried to explain he said it interpolates, it's still only as good as your end values and your step size.

No offence but I trust a professional opinion more then a home enthusiasts.

It's also not just about tables, but all the add on functions that comes with it and being time effective in setting it up. (time is money)

Sent from my GT-I9100 using Tapatalk 2

[NitroDann]

Gslender isnt JUST a home enthusiast, he writes the code for MS.

Patricks na6 turbo runs ms2 at 300rwhp, nice smooth afr, and not a hell of a lot more cells.

And setup isnt needed for any ecu for an mx5, just plug it in, load base map/base settings and tune.

Dann

[noddi85]

I'd be talking to ur tuner. It's what they like to work with that will cost you less.

Yes all ecus have maps and base tunes that give you a start off.
But depending what components you have can differ that base tune a lot.

Also tuning a turbo engine is a lot different from a Itb na engine.

My 4a as mentioned before didn't run rich till it hit 5000rpm
Evan then it wasn't massively running rich.
A na engine is predictable on what its going to do. A turbo car is not.
Boost spikes etc have to be considered with a turbo engine.

[Si.G]

With regards to tuner time and MS2, I intend to do the tune myself. I have found a garage that are going to hire their dyno out and are willing to support me with the tuning where necessary. He is also happy to tune it, but obviously that may take longer that a different ecu might take.

I think I am leaning towards the MS2, looked a the specs and the processor is 3 time faster than the MS1 and the fuel table is 16x16 which should be enough.

Si.

[sailaholic]

Sounds like a good option if you are self tuning. Just make sure you get the upgrades, easier to do it up front.

Sent from my GT-I9100 using Tapatalk 2

[gslender]

Last time gslender tried to explain he said it interpolates, it's still only as good as your end values and your step size.

It doesn't just interpolates, it does so with 100ths resolution.

So between 40 and 41 there are 100 different values - in just one axis! Multiply that by the 100 values in the other axis and you get the idea about how fine the tuning can be.

In my fuel table most cells are only 1 or 2 VE values apart. I really can't see how doubling or adding more rows/columns would make any difference due to the interpolation.

I think your tuner is probably used to working with ECUs that don't use this kind of interpolation and therefore really wished their ECU had more cells.

G

[sailaholic]

Yes, but the 100 different spots are still based of spot 1 and spot 100 correct?

If your map values change by very minor amounts, then I agree, more cells are unlikely to be needed.

Sent from my GT-I9100 using Tapatalk 2

[NitroDann]

What do you mean?

Hes saying that one cell right next to another, the value might be 67 and then 69 next to it, a 3% change.

If you are aiming for say, 12:1 AFR and the ecu is in the whole next cell its only going to be 12.35:1, hadly dangerous although not ideal.

If you ecu doesnt interpolate then having 24x24 tables allows this error to be 12.175 rather than 12.35, and this is assuming the ecu is actually in the wrong cell completely.

Interpolating 100 different point between these 2 cell, in this one direction alone tells us its unlikely to be out by 1/100th of 12.35.

About .1235.

Dann

[gslender]

Yes, but the 100 different spots are still based of spot 1 and spot 100 correct?

If your map values change by very minor amounts, then I agree, more cells are unlikely to be needed.

Sent from my GT-I9100 using Tapatalk 2

No. The MS2 ECU code divides the VE cell into 100ths resolution... so what you enter as 40% become 40.00% and it then creates an infinite slope (really only limited by single RPM and .01 map kPA values) to the neighbouring cell.

So it would only be limited by the number of X/Y values between the cells - I'll try and explain better with a text picture.

Code: Select all

RPM
1000  2000
40    50

In the values above, the difference between 40 and 50 is divided by the number of RPM values between the cells (in this case 1000). Using .01 resolution of a VE, the number of values evenly runs between 40 and 50 for every single RPM between 1000 and 2000. The same thing happens in the Y axis for Engine load (MAP) and equally the values are just as fine.

Most good tunes are generally flat... rarely do you have crazy spikes all over the place. The general view is that you have a smooth transition from increasing to decreasing curves - not jumps from 50% VE to 65% VE and then back to 50% VE.... that would be a bonkers tune that wouldn't be very smooth running.

I appreciate the concern the tuner is raising, but I just don't believe it even exists... but if it does, then with a 16x16 table, you can easily shift the cells to fall on the crest of change where you need greater resolution - you can have them sitting on 1923RPM and the next at 2345RPM if you wish.

Some ECUs fix the cells to 500rpm spaces, and that would also be why some tuners need 32x12 tables... some ECUs are just stupidly bad and have educated tuners to know nothing better... they just think all ECUs are just as bad and need massive tables.

Finally, when the develops of MS3 started out, they took the time to review with many tuners and engine builders the need to increase the table size beyond 16x16. They actually had enough memory and CPU grunt to do 256x256 tables if they wanted. They finally decided that 16x16 tables were all that was needed - why, because there just isn't a need for finer fuel tables then what they already had. Nuff said!

G