Table resolution and size

[Token]

I have been doing pretty good with the current resolution by giving needy areas more resolution at the cost of less needy areas. IMO couldn't hurt to have a little more though

[AbatelliCristian]

Is possible increment the size vvt table? now is 8x8. I think is good increment the rpm row

[masterx81]

I know that this is an 'hot topic', but i know some tuners that don't want to use the ms due to the 'limited' ve and spark table sizes.
Maybe would be nice to use ve and spark 2 and 4 tables (for who don't use dual table and blending) to extend the ve/spark 1 and 3.
With this mod, the switch-map feature would still be possible (as now the only way that i know to extend the maps is to switch 2 tables with load or rpm).

[tpsretard2]

my gems that is used in a customer prodrive rally car has one more load site than the MS and about 3 or 4 more in the rpm and the figures do not change much.

I started to use the rpm triggered map switch, after still fighting i was then found that it was a problem with the map sensor!!
Now i want to go back to the 16x16. Unless you are going to crazy boost and or crazy rpm it really is not needed. another omex based rally car i tune not much more resolution, actually has less on the load sites, only 11.. however i have a 500bhp rally car that runs up to 9k tuned on it..

Them saying it dos not have enough resolution or limited ve. they are idiots and i would not trust my tractor to them to map it...
if you know what you are doing it is enough. only thing that would be cool is for the people that to use the rpm or load site switch for wat ever reason is if tuner studio could know this and somehow show both to make tuning easier..

[masterx81]

There are also bike guys that can take advantage of the ms3 system, and some more breakpoints can be helpful. In any case, the tables are already here (table 2 and 4 when not used for other things)...

[subwoofer]

Reviving an old thread here with a wish 180 degrees in the opposite direction from most here. The flat-4 I am tuning now is ending up with a VE map as smooth as a baby's bottom, so I am only using 12 of the 16 bins either axis. This is still a lot more than I think is really necessary, could 8x8 be an option for a later release? As long as nothing wild happens, the interpolation will take care of everything and tuning will take a lot less time.

[knightrous]

I did a quick search but couldn't find any discussion on it.
Could we have an option for those of us who don't use the switched tables to use that memory/allocations to get a 16X16 AFR / Spark table for use on MS2 (I know this is in the MS3 forums, but there isn't a thread like this in the MS2 section ). As a new person to tuning, I have some difficulties getting my tune nice* when all my table scalling are different. I believe I would find it easier if there was a 1:1 alignment between AFR/Fuel/Spark

At the moment, I'm looking at back scaling my fuel table to 12X12 to achieve this, bit of a back step IMO, but I think it will help me out.



* This could all be down to my specific setup, motor, gearing and limited skills.

[frank_ster]

the 6 or so engines i have played with have had very flat ve tables. 12x12 or even 8x8 would work fine.
especially with pid ego control, the emissions i don't buy that. the acceleration enrichment cant be perfect enough to always have the perfect airfuel ratio.

the little inconsistency is what the cats are for.

for tuning a engine for the first time 8x8 is great. fine tuning 12x12 is good. the 16x16 tables i find i'm just using the interpolate button all the time.

for the people who say ms is crappy because the tables are not big enough are tards. i say give them 50x50 tables and they wouldn't be happy with that either.

[Foolee]

Well I'm new to MS. And I plan to invest in MS3 Pro for my turbo street truck shortly. I can understand why you guys want to keep it 16x16. The best argument I have seen against going 32x32, 24x24 or even 20x20 is how much memory it will use up. That being said Megasquirt is designed to accommodate a large crown of enthuses. I don't see why there cannot be a compromise. I don't see why it could not be had both ways. Enable 32x32, if future updates/supported features comes out that require the memory space then force the user back to 16x16 or something similar. Put the choice in the hands of the user on how the memory of their MS3 is being used. Not everyone uses all the features on MS. Not everyone uses, NOS, VVT, NLS, Boost Control, Knock Sensors, ect, on their build. The one thing I love about MS is the fact that you can customize it to meet your needs. This is something that other ECU manufactures don't offer. That's just my $0.02

-Darren Lee C.Tech

[muythaibxr]

Unfortunately with our current memory layout, it's not possible to do 32x32.

16x16 was already our "compromise" as most of us thought that 12x12 was big enough.

Ken

[dontz125]

I was helping a fellow out last night who was trying to get his head wrapped around table switching. I think I got him clear on swapping between VE1 and VE2 using rpm or tps or kPa, and how this could make a 32x16 or 16x32 table (30x16 or 16x30, allowing for overlap), depending on how you set it up and which trigger data you used.

How hard would it be to allow switching based on two triggers? Swap between 1 and 2 based on rpm, swap between 1 and 3 based on tps / kPa; table 4 is selected when rpm AND tps / kPa setpoints are achieved.

Thoughts?

[frank_ster]

uh is the ve table like a moguls ? can you provide screen shots as to why that many tuning points is nessasary ?

[dontz125]

Heh - no, this was a curiosity question, not something that I particularly need ...

[frank_ster]

the only time i would of needed something like this is when the sequential injection squirt would hop over the intake valve closing threshold it was causing some weird issues until i figured out why.

any one requiring more than 16x16 is probably needs it as a band aid solution to another problem.

[maik21]

new to megasquirt ecu. i purchase a holley commander 950 some years ago and holley ecu dont interpolate. it want hard to make engine work fine with a 16x16 fuel an ignition table. 1 year ago, i put a ms2 v3 in a nissan 2.0 engine. put ms extra code and it works veeeery fine and it was more easily than the holley system.

past week i finish assemble 2 megasquirt for my car(get rid of holley system) and a car`s friend .

[mUnky]

I want perfection for my tune. If I can demonstrate that sections of the table are subject to significant non-linearity, would that not be evidence that linear interpolation is not enough for that particular situation?

To what extent do neighboring cells affect the interpolation? I saw somewhere that the interpolation is weighted... could you guys expand on that a bit? I've been jumping pretty hard into engine management theory (and practice!) and it would appear that for really perfect operation, you really do just need a large number of tuning points.

I guess it goes without saying that if larger tables were implemented, it need not be mandatory? Nor do they necessarily have to double to 32x32...

[Peter Florance]

I want perfection for my tune. If I can demonstrate that sections of the table are subject to significant non-linearity, would that not be evidence that linear interpolation is not enough for that particular situation?

To what extent do neighboring cells affect the interpolation? I saw somewhere that the interpolation is weighted... could you guys expand on that a bit? I've been jumping pretty hard into engine management theory (and practice!) and it would appear that for really perfect operation, you really do just need a large number of tuning points.

I guess it goes without saying that if larger tables were implemented, it need not be mandatory? Nor do they necessarily have to double to 32x32...

Actually 32 x 32 is four times as large, not double.

[wes kiser]

I want perfection for my tune. If I can demonstrate that sections of the table are subject to significant non-linearity, would that not be evidence that linear interpolation is not enough for that particular situation?

To what extent do neighboring cells affect the interpolation? I saw somewhere that the interpolation is weighted... could you guys expand on that a bit? I've been jumping pretty hard into engine management theory (and practice!) and it would appear that for really perfect operation, you really do just need a large number of tuning points.

I guess it goes without saying that if larger tables were implemented, it need not be mandatory? Nor do they necessarily have to double to 32x32...

How do you intend to measure perfection?

[mUnky]

32x32 has four times the tuning points, yes, but I was referring to the length of the axis' and they're doubled

I'll measure perfection the only way I can; to the point at which I'm happy

I've just spent some time looking at old dyno graphs for my engine (with the old ECU a terrible microtech) and I'm having trouble designating tuning points, taking into account interpolation and still seeing optimal control. I guess I can say that a close-to-perfect tune will operate the engine at it's maximum efficiency, adapting to the physical parameters of the engine to such accuracy that the fitted sensors can not detect deviation. An unrealistic outcome for the home enthusiast yes, but still worth striving for, I feel.

It looks like the argument against larger maps is that the difficulty of implementation will outweigh the small benefits of finer tuning resolution.

That said, the "don't fix what isn't broken" philosophy is not one that I think is valid

[Peter Florance]

32x32 has four times the tuning points, yes, but I was referring to the length of the axis' and they're doubled

I'll measure perfection the only way I can; to the point at which I'm happy

I've just spent some time looking at old dyno graphs for my engine (with the old ECU a terrible microtech) and I'm having trouble designating tuning points, taking into account interpolation and still seeing optimal control. I guess I can say that a close-to-perfect tune will operate the engine at it's maximum efficiency, adapting to the physical parameters of the engine to such accuracy that the fitted sensors can not detect deviation. An unrealistic outcome for the home enthusiast yes, but still worth striving for, I feel.

It looks like the argument against larger maps is that the difficulty of implementation will outweigh the small benefits of finer tuning resolution.

That said, the "don't fix what isn't broken" philosophy is not one that I think is valid

You should try tuning with current table size. I bet you will find it will work. I tuned an LT10S setup on a Diasio; I'd could have gotten it much closer with MS3 in about 25% of the dyno time.
I can't speak for developers, but if I were them, I would make you show them, with MSQ and datalog, how the current take setup doesn't work. And be willing to try suggested changes to your tune.
Also, I think if we really needed bigger tables, I think several active tuners (like Wes above) would have demonstrated it a long time ago.