Upgrade to CPU MS-II. Worth the $90?
 

Hi.
I actually have a MS-I hi-res running in paralell.
I'm thinking how much can I improve if I upgrade to MS-II (I read that MS-II can do a better for control for accel (throtle vs MAPdot) and I can get a better resolution in some maps, but.

1) worth it?
2) I will have to modify too much my board or it's plug and play?
3) any one make this change before?

The CPU is definitely faster. What that means in the end, I'm not sure. The mods to the board are slight, you need to run 12v into the daughter-board and I believe the tach input circuit is different. Otherwise the regular miata spark mods and any other circuits are the same.

MS-I can do TPS based accel enrichments.

I'm fairly certain I'm going to upgrade my CPU over the winter downtown...

MS-I can do TPS or MAP. MS-II can use 40% TPS plus 60% MAP. Sound attractive to me for a smooth daily drive...

ah i gotcha, forgot about the slider...

The MS-II code seems to have a lot of bugs in it. The single biggest thing you could do to help is going to a crank trigger. There's as much as 10 degrees of slop in the timing on a CAS sensor. Ask joe Perez, but going to a crank wheel will get you error of maybe 1 degree, which will let you run more boost. Will an MS-II get you more? Some, but only a little, and only after you do a crank trigger setup on there too.
-Abe.

Really?
This is not the experience we see on the MSEXTRA Megasquirt support forum.

Please give specific examples.

regards

James
(MS2/Extra developer)

I also heard the new closed loop code is worlds better from before. I'd love to get closed loop boost control running well.

Yeah, I'd like to know what bugs you're talking about? If there are any specific bugs, I'd like to know about them.

Re: the accel enrichment, in 2.1.1 beta and 3.0.x alpha, you can use EAE with standard AE at the same time. EAE will get you much better drivability (if you're willing to spend the time tuning it), and you can use just a little normal accel enrichment to get good response on quick blips.

Some of the other changes include:
1) more accurate spark (every-tooth wheel decoder)
2) more accurate fuel (can result in smoother running engine, most of the people I've seen switch agree this is the case).
3) working closed-loop algorithms for boost, idle, EGO. I'm thinking about another change to boost to make it reach its target faster though.
4) The developers are actually working on it actively still. We're backporting the features we can from ms3 as well.

There are tons of internal improvements as well, such as the fact we do the fuel calcs in .01usec units to avoid rounding errors before dividing down to .667 usec units, and the fuel table interpolation is done in .1% units instead of 1% units. You guys don't see those but you can feel them on a running engine.

Ken
(Another ms2/extra and ms3 developer)

Wow. Very awesome to have you guys here.

Problems I'm hearing about from local cars, but not necessarily experiencing myself:

Uneven dwell to coils
Cam/Crank random loss of sync

Love the 16x16 VE table on MS2. Wish the spark table were also 16x16.

Looks like the MS big hitters are here to set you guys straight.

The only problem I've been aware of recently with odd timing and such was not a code problem.
The "Hall input" (which is really designed for coil negative triggering) was being used unmodified for the CAS input. One of the capacitors that is in the circuit to clean up the noisy coil signal was delaying the CAS signal and confusing the decoder.

Sync loss means that the input signal is incorrect.
The "sync loss logger" in TunerStudio can be used to visualise the input signal the happened right before the sync-loss and should greatly help understanding of where the problem lies.

The old motto applies - garbage in, garbage out. If Megasquirt gets a messy CAS signal... it isn't going to work well.

The Miata 4G63 CAS pattern has been used successfully with MS2/Extra for a good two years now.

James

Hopefully we are doing this constructively!

James

Last I've heard, Jerry@DIYautotune has been able to sort most of those dwell and sync loss issues as noise issues. I assume that they're spreading the word on how to fix those?

Sync loss and associated problems are almost always noise issues; especially if other people have things working properly with the same setup.

As far as 16x16 spark, we can't do that because spark tables are 16-bits per cell, and we have 3 of them... Currently at 12x12 they all fit in 1 page of flash, but at 16x16 they'd be 1.5k, bigger than our 1k table.

We did make them 16x16 in ms3 though since we don't have that limitation there.

Ken

We want to help, and we want to know about all the bugs!

Ken

Ken,
Please excuse my ignorance. Would it be possible to give up table switch in favor of a single 16x16 spark table?

I don't know how hard that would be to implement. But I'd imagine that table switch is very seldomly used. And we have a lot of turbo cars that are both high MAP and high RPM.

MS3 sounds awesome, but I'd imagine there's no upgrade path from a uS.

Technically it could be done, but the problem is that we know of people using the table switch stuff. It is in active use.

You could make a 12x22 or 22x12 spark table by using 2 spark tables both on speed density/%baro with different RPM or MAP ranges as a way to get a larger table. It's not 24 instead of 22 because you generally want to use a couple of bins to blend the two tables.

By the same method you can get 30x16 or 16x30 on fuel.

Ken

I wouldn't suggest deleting table switch permanently; I know some folks use it. I was thinking along the lines of an option.

It would be nice to be able to expand the spark table in both axis. Case in point: I'm looking at running 250 kPa and 8000 rpm, and drive the car on the street. I know it's possible to squeeze the spark map on a 12x12, 22x12, or 12x22 table that interpolates, but I'd be giving something up either way.

That being said, if we're giving up table switching, perhaps just make a single large VE table (the largest that will fit in your space) and a single large spark table (largest that will fit in your space).

I know I'm asking for the moon here, but it would be cool.

oh shit! I';m going to die.

Part of the problem here too is that we don't really think more than 16x16 is necessary with movable bins and interpolation. Engines are generally pretty linear, and normally have only a couple places where they're not. We don't really feel that you're giving up anything if you just move the bins to the places that are not linear.

Case in point, my friend (eage8 on the MS and rx7club forums) has an S5 RX7 with tables that go to 8000 rpm and 205 kPa. He gets 22 MPG on the highway (very good for a turbo rotary), and made 270 RWHP on Jerry's dyno (also good for the mods he has). He has pretty much perfect drivability and for a long time was daily driving it without a complaint (he recently got a new daily driver to replace the one that is now waiting for a 20v 4age).

Another case in point is the Huber mustang. Those guys drove that car on the street on an ms1 (12x12). I don't know how high they rev, but they were pushing 40 psi. That car had 1100 HP.

We spent a good bit of time on the dyno, but actually didn't really have to move any of his bins. Most of the time was spent just load tuning.

Also, for fuel, think about tuning 32x32 (what would fit in 1 page). That's 1024 points. For spark 1 page could fit 22x22, that's 484 points... That is a lot of wasted space and time for pretty much no benefit.

Ken

I saw this coming as soon as I read this.