Megasquirt I 12x12 tables big enough (ITBs)

Postby **NitroDann** » Mon Sep 10, 2012 12:11 am

256x256?

Why not, would be what? 8 times better than Motec?

Hahahaha.

By most people assumptions.

Dann

Si.G · Postby **Si.G** » Mon Sep 10, 2012 12:40 am

Thank you for everyone's input. I have pretty much decided to go for the MSPNP2 from diy auto tune. It seems to have everything that I need. I just need to double check that it has 16x16 VE tables, pretty sure it does
Si

gslender · Postby **gslender** » Mon Sep 10, 2012 1:23 am

Si.G wrote:Thank you for everyone's input. I have pretty much decided to go for the MSPNP2 from diy auto tune. It seems to have everything that I need. I just need to double check that it has 16x16 VE tables, pretty sure it does
Si

MS2 has 16x16 fuel/VE and 12x12 spark.

It also has has 12x12 AFR Target tables.

G

sailaholic · Postby **sailaholic** » Mon Sep 10, 2012 7:57 am

Dann, g thanks for the further advise.

I can see what you are saying, but I still haven't been able to get my point across.

Regardless of how fine the result of the interpolation equation can be calculated to (100 divisions), the equation itself is still only based on the two original map (ie table cells) points from which that equation was based from.

What your saying is that this step size dosent matter as an engines requirements are so linear that the don't change significantly from that equations over 1/12th of the rev range . Correct?

You are also saying that the code allows for the setting of map points at arbitrary points rather then fixed intervals to allow positioning of the points near "major change" locations if necessary?

Pretty sure the adaptronic does the interpolation as well, but I'll have to go back and check my notes. Haven't looked deeply enough into the others to comment.

I'll also quizz the tuner further....

Sent from my GT-I9100 using Tapatalk 2

sailaholic · Postby **sailaholic** » Mon Sep 10, 2012 8:04 am

Dann, it's not just the maps size that determines which ecu is better or not. It's just that the lower level of megasquirts have such a small map that it gets most of the attention.

But I'm sure you know this, you just wanted to make a big statement to grab people's attention.

Sent from my GT-I9100 using Tapatalk 2

Postby **NitroDann** » Mon Sep 10, 2012 8:25 am

That true.

And yes you can have any map value you want.

Also hows this for helping people see that all the cells arent needed...

Many aftermarket ecus like motec dont use all the available cells, and allow you to simply insert another row or column at a specified load or rpm. This is to help tune out issues off idle lurching and emmissions on over run.

However in default mode the table is missing half the cells it could have.

If you need all of these cells and the software has that feature built in, why dont they activate them all from factory?

Dann

sailaholic · Postby **sailaholic** » Mon Sep 10, 2012 8:59 am

Dann,

A motec has a lot more cells to start with though, and your right, I agree you could have more then needed. But it means the motec let's you put more cells in if you need them like the examples you noted.

It keeps coming back that the ms uses so fewer cells then others, not that anything less then the absolute maximum on the market is rubbish.

Another question. If the 3 is capable of such high map cell counts with the hardware involved why not allow more?

I had previously been told that ms kept the cell count down to control costs. But if the hardware is already there, then surely there can't be too much cost difference.

Sent from my GT-I9100 using Tapatalk 2

gslender · Postby **gslender** » Mon Sep 10, 2012 9:21 am

sailaholic wrote:It keeps coming back that the ms uses so fewer cells then others, not that anything less then the absolute maximum on the market is rubbish.

Another question. If the 3 is capable of such high map cell counts with the hardware involved why not allow more?

I had previously been told that ms kept the cell count down to control costs. But if the hardware is already there, then surely there can't be too much cost difference.

Most customers are stupid and think bigger = better. Most commercial businesses know this, and so make it bigger to get customers to throw down big $$$. Still doesn't change the fact that most customers are stupid.

Read this thread http://www.msextra.com/forums/viewtopic.php?f=125&t=33052

To be honest, you'd be better off buying a MS2 and with the money saved from buying a Motec, spend all that spare cash on multiple sessions of Dyno time. You'd get a better result than just a single "please tune this with my remaining $500 of ECU budget after I bought a $2000 Motec system".... LOL

G

sailaholic · Postby **sailaholic** » Mon Sep 10, 2012 9:58 am

Interesting read on the link. Only got through a small amount but will return to it.

Certainly agree that the arbitrary movable bins, would help a great deal. Will keep reading when I have time and come back with further comments.

Agree, no point spending big $ on ecu and scrimping on the tuner time. Vice versa also applies IF the ecu isn't up to the job.

Cheers

Sent from my GT-I9100 using Tapatalk 2

gslender · Postby **gslender** » Mon Sep 10, 2012 1:26 pm

If you get around to reading that link, there is a lot of good stuff and plenty of inputs from knowledgable folks in big-end-of-town racing that should know, and I think all of them are happy with 16x16 or even 12x12 tables. These are people with bikes that rev to 18,000 and high boost engines doing 7 second quater miles !!

A good point that Ken (lead developer of MS3) explains...

Again, we have seen no definitive proof that more than 16x16 is necessary, or even desirable. Even if you go to 16000 rpm, an engine is only going to have so many points where it's non-linear, and you can just arrange your bins in those areas to account for the non-linearity, and then the rest of the table will mostly be linear. If you need more, just use a second load/table, and you've got 16x30 (assuming 2 bins for blending).

Nobody has shown us proof that anything more than 16x16 is necessary. Just anecdotal evidence ("I have experience with x, y and z, and this is what is necessary). Without logs, and settings showing that 16x16 was not enough in a specific situation, we WILL NOT increase the table size. With a true VE table rather than entering raw pulse-width, and with incorporation of AFR into the math, VE tables become very flat, and you do not need many bins to cover non-linear spots.

The reason the "other guys" need so many bins is precisely because they use pulse-width in their main tables, and do not incorporate AFR into their fuel equations (since their equations are nothing more than "take this PW, multiply in corrections").

He mentions another good point, which is the Incorporate AFR, which by default creates very flat VE tables.

G

sailaholic · Postby **sailaholic** » Mon Sep 10, 2012 2:29 pm

Readers digest version of the incorporating afr ?

Also another question, you said you were working on a/c and other controls being integrated? Does it not have this out of the box or are you trying to improve this?

Sent from my GT-I9100 using Tapatalk 2

Postby **NitroDann** » Mon Sep 10, 2012 2:31 pm

Gslender writes the firmware you choose what firmware you use when you load the basemap.

Choose Gslenders for better oem behaviour.

Dann

sailaholic · Postby **sailaholic** » Mon Sep 10, 2012 2:51 pm

Yup, got that part dann, was looking more at the details of the changes, as previous posts read like certain things had poor support.

Ie I think part of gslender changes were to delay the ac compressor from kicking in straight away on button push so that the iacv had a chance to adjust for the impending load.

Other things were mentioned as well so I figured good to discuss this now while we're talking suitability / capabilities

Sent from my GT-I9100 using Tapatalk 2

gslender · Postby **gslender** » Mon Sep 10, 2012 3:08 pm

sailaholic wrote:Also another question, you said you were working on a/c and other controls being integrated? Does it not have this out of the box or are you trying to improve this?

Have a look at my latest release - I've included a comprehensive modguide book that outlines all the new features since MS/2 Extra 3.2.1
http://www.msextra.com/forums/viewtopic ... 91&t=46616

One thing worth pointing out....

There are 3 ways you can get a Megasquirt 2 ECU...

1) MS PNP2 - a pre-assembled fully plug and play Microsqurit II with electronics and wiring to suit a MX5.

2) MS DIYPNP Kit - a kit requiring small soldering and assembly that makes it a plug and play Microsqurit II with electronics and wiring you decide is needed to suit a MX5.

3) Megasquirt II Kit - a kit requiring major soldering and assembly that needs a harness connector with electronics and wiring you decide is needed to suit a MX5.

Only item 1) comes with an hardware circuit for AC that only raises the PWM IAC duty a small amount to kinda help with AC idle up. It is a poor solution with almost no control and doesn't produce the best outcomes, but without any software improvements, it would be the only way to avoid a major 200rpm idle dip.

What I've done is add software based idle-up from any input to any output, so you tell the CPU how long to wait before turning AC on, and then in the mean time, raise the idle duty and rpm target to be ready for pending AC load. Works great and like OEM. So with this available, all above solutions can have AC idle up and I would remove the wire/curcuit included 1) above.

I've added a crazy amount of other features too.... too many to list and you really should just read the modguide.

G