Megasquirt V3 ?, Just a few questions for the MS crowd Options

[alpha434]

Does it support direct port injection yet?

That's the first one on my list.

[DNHunt]

It will do port injection or TBI. With the right injectors it might do direct injection. On more than 2 cylinders it is batch fired. The channels can be fired either simultaneously or may be alternate. It will not do sequential injection of more than 2 cylinders

Dave

[Mueller]

no idea, but here is some neat info on the subject from one of our customers.....

direct injection

[alpha434]

That was my question. I need sequential for eight cylinders.

Oh well.

[Mueller]

QUOTE (alpha434 @ Mar 14 2006, 11:59 AM)

That was my question. I need sequential for eight cylinders. Oh well.

yep...gotta spend the big bucks for that one....I think Haltech (TimT is a rep) is the most cost effective units for that.....

[lapuwali]

No, it doesn't do direct injection, because it doesn't do sequential injection.

Just to clarify, the versions don't go 1, 2, 3. The BOARD is currently at version 3. There are two processor options, however, the 8-bit HC08 (aka MS1), and the 16-bit HC12 (aka MS2). Either processor will fit in any of the boards.

MS1 will very likely never support sequential injection in any form. MS2 MAY, in the future, but it doesn't do so now. I don't know if anyone is actively working on it. I doubt they are. It would require add on hardware, as the boards only have two on-board injector drivers, to support up to two banks of N injectors each. Most people only use one bank and fire all the injectors together, much like L-Jet does. Some use alternate fire on each bank, but there's little advantage to doing so. As it stands, the hardware can't do sequential for more than two cylinders.

To get full sequential for 8 cylinders, you're going to need to look at the high-end ECUs, like Motec.

[DNHunt]

James

They are actively working on it. Sequential injection and coil on plug for 8 cylinders will be available with router board. But, don't hold your breath. These are being developed by Bowling and Grippo and they tend to test a lot more than some of the other people who bring stuff along.

Alpha

For now you are going to have to open your wallet. Another thing to consider, if this is going on a costomers car, this is experimental and there is no warranty. So if it goes lean or there is a catastrophic failure you know who might have to eat the costs. It's great for a hobbyist but, I sure wouldn't sell it.

Dave

Dave

[alpha434]

What about two MSII boards used in tandem?

That would hit 4 cylinders.

[alpha434]

My own project.

No need to involve something like this on a (paying) customer car. They can get the high dollar crap like everyone else.

[DNHunt]

Alpha

True enough, that would give you 4 injector channels and you could tune 2 separate maps. But then you need to sequence them relative to the cam and inputs would have to be identical. It's probably doable but, more than I want to take on.

[Tom Perso]

Why are you so interested in sequential? Only real benefit is idle emissions. Or, are there more benefits than that?

Tom

[Mueller]

QUOTE (Tom Perso @ Mar 14 2006, 01:02 PM)

Why are you so interested in sequential? Only real benefit is idle emissions. Or, are there more benefits than that? Tom

if you have a motor with "real" direct injection, you have to have seq.

direct injection places the nozzle of the injector right into the combustion chamber..........

[jsteele22]

Hey Alpha,

Did you in fact mean Direct Injection, which as I just learned means squirting fuel straight into the chamber, or merely sequential multiport injection (MPI) where an injector is located near (but upstream of) each intake valve ?

The DI scheme sounds like it has a slew of design/research challenges, as mentioned in the link by Mueller. But sequential MPI is, in principle, a piece of cake. The only reason that MS does not have it right now is that in order to control each injector, two PWM/timer channels are necesary : one to determine when the injector is open vs closed, and the other PWM value to determine if the injector should be driven hard (during opening) or less hard (to keep it open). The injector driver gets the AND (NAND, actually) of these two digital signals. So there are a number of strategies in the MS world. One is to use the mysterious "router board" which is talked about (but not described or offered for sale) on the MS site. I think the idea is too keep the board electrically identical to the main board, and just add more injector driver outputs. As mentioned, the development & testing that B & G put into their products is, well, time-consuming. Another approach is to pick a different chip from the same 16-bit family and lay out a new board for it. I've been toying with this, since its kind of what I do (home-brew electronics for scientific applications). The actual MS code should just port over with a few modifications (read : "it'll buff right out..."). The MC9S12DT128 has 8 timer/counter channels plus 8 PWM outputs, so with the same NAND logic and double the injector drivers, it could do sequential injection on 4 cylinders, and even 8 cylinders if you're willing to drop ignition and PWM idle control.

The main problem I see with this approach is that MegaSquirt kind of tries to be "Open Source" (Like Linux, FireFox, etc.), but they're not really. They say its a real no-no to use any circuit boards that aren't manufactured by them and sold by them or an authorized distributor, and if anyone uses non-sanctified hardware, then they better not ask any questions. So technically, you are prohibited from innovating and sharing, which is the driving ideology of the Open Source movement. But it's a tricky issue to manage even a software-only project, and adding hardware to the mix raises a ton of new issues. I'm not aware of any hardware + software projects that have really kept to the Open Source model and managed to be widely successful; MegaSquirt is actually the best effort I come across so far.

Anyway, Alpha, I don't think I've run across you before. I look forward to (figuratively) bumping into you sometime if I ever get my teener on the road...

[lapuwali]

I think you're reading rather too much into the "not invented here" board situation wrt to B&G. The schematics are freely available, and plenty of people have made other boards. What's discouraged on msefi is support for those boards, for obvious reasons. B&G didn't have anything to do with making those boards, so they have absolutely no idea to what level of quality they were made. These guys are hardly making a killing on these boards, and are basically offering support gratis, as are all of the other regular volunteers there. They do encourage hacking new features onto the boards, and many of the best features were added by people other than Al & Bruce. The "extra" code, for example, came primarily from a couple of guys in the UK that have pretty much completely rewritten the software, and packed in WAY more features than were there in the first place.

The chief problem I see with B&G is they tend to be too much the perfectionists, and generally don't release stuff until it's nearly done. This isn't always true, and Bruce openly discussed the design of the v3 board long before it became available, esp. with things like the tach input circuit, which was the most troublesome aspect of the earlier boards.

Strictly speaking, the PWM is only for low impendance injectors. It's not required (or used) for high impendence injectors.

It HAS been frustrating that real details of the "router board" haven't been released. I haven't been able to figure out what the thing is actually. It looks like it might be a set of programmable timers that can run free of the main processor and handle things like spark control independently of the processor (with only advance being controlled by the processor). This is, in essence, very similar to the TPU used by Motorola in the 68332, which was the processor chosen for the DIY EFI controller than preceded MS, and which ultimately failed as a project because too much was bitten off in one chew, and the whole thing got out of control.

If you weren't around in the DIY EFI days, you missed a lot of this, and most of it explains why B&G operate the way they do. They (and anothers) initially started with a far grander and more sophisticated system that would have easily been able to do full sequential with coil-on-plug for at least 8 cylinders. After the collapse of that project, B&G went off by themselves and did MS as a simple 8-bit thing that did fuel only. Once they had it working and useful, it was slowly advanced to add ignition (distributor only), then DIS with EDIS, then other ignition systems. MS2 is still way far away from what the DIY EFI board would have been had it been successful. The "Ultra MS" that's still only talked about (using a PowerPC based microcontroller with a TPU) would finally reach the same goal that was originally laid out in the late 1980s!

And if you really think "Open Source" isn't generally about closing ranks around one system and one way of doing things, you haven't paid much attention to the actual politics of most Open Source projects.

To the ever-ambitious alpha: since mechanical things are more your line, I'd concentrate on the mechnical aspects of direct injection, and just buy an ECU that can already do full sequential now. Trying to do the whole thing yourself pretty much ensures the project will never get finished. As far as I know, no one is making the mechanical bits required to do aftermarket direct injection on an engine that wasn't designed for it. Theoretically, any ECU that can do full sequential all the way to redline on your engine (and it's possible none of the aftermarket units will do this, not even Motec) can be programmed to handle real direct injection. The main difference is that in DI, the injection all has to take place between the time the intake valve closes and the time the plug fires, which is far shorter than the full load pulsewidths usually employed by port injection. At, say, 8000rpm, you'll have about 2ms to inject 100% of the fuel required for that cylinder. For proper idle mixture, you'll need less than 10% of that fuel, so the injector has to be able to operate with an open/close time under 0.2ms, which doesn't describe any off the shelf port injectors.

[jsteele22]

Hi lapuwali,

I'm glad to read your response. I didn't know any of the history or personalities of MS, and I'm glad to hear that other people have indeed worked on different hardware and not been attacked/ostracized for it. I totally understand the desire on the part of B&G not to want to have to debug a bunch of hardware variants that they've never set eyes on. Without any personal knowlege of that project, it's hard to differentiate between that (very sensible) concern, and a much less noble, and very common, NIH attitude. Like I said, MS is the best example of a hardware + software system I can think of that approaches being both Open Source and successful. And its not like they have some moral duty to be Open Source ™, anyway - it's still a very cool project.

Also, I hadn't noticed that the PWM business was only used for low impedance injectors -- so for high-Z, there's that much less to worry about. Also, going w/ EDIS looks like it cuts down on a lot.

End of hijack, back to, uh, work.

[Tom Perso]

QUOTE (Mueller @ Mar 14 2006, 03:10 PM)

if you have a motor with "real" direct injection, you have to have seq. direct injection places the nozzle of the injector right into the combustion chamber..........

Ah yes, grasshopper. We're talking injection much like the new 2.0T VW motor, correct?

Sounds trick, although it seems that those injectors are running uber high pressure, and getting the injector pointed at the right angle would be a treat.

But, I won't rain on anyone's projects. I can't talk, the mess I've made in the garage makes me a fool just like the rest of us.

Oh well, back to (IMG:http://www.914world.com/bbs2/html/emoticons/beer3.gif)

Tom

[DNHunt]

Tom

Don't worry, now that you got that root canal done you'll start thinking straight.

Ultra high pressures like a die.sel

Dave

[lapuwali]

There's some information on the megasquirt.info site on the Router board. It uses the very processor you mentioned, jsteele, the MC9S12DT128. It gets injector and ignition timing hints from the main processor (which has to be MS2) via CAN, and handles the crank and cam signals entirely on its own, providing 8 injector and 8 ignition signals, so you can do full sequential on a V8 with COP.

The one downside to this is there's no way to have per-cylinder control of injection or ignition, so no fancy stuff like ION-sensing ignition. All cylinders will get the same injection and ignition timing. Given suitable injectors and the rest of the mechanical stuff to back it up, however, you could do direct injection on a V8 with this. Think of it basically as a user-programmable EDIS controller that can also do injector control.

The boards can be slaved (this is a feature of CAN), so you could have two of them to handle up to a V16, if you just happened to have one of those lying around. And if you're planning on injecting some old aircraft radial, you can put in a third to get 24-cylinder control. MS at the Reno Air Races...

I suspect the really advanced stuff like ION-sensing they're either thinking of hacking onto the router board at a later time, or waiting until they get to UltraMS.

[jsteele22]

Wow, I hadn't heard of ion-sensing before. my first google hit was a pretty good read :
Ion Sensing Description

One thing I've thought about from time to time is using a fine-toothed crank sensor (360 teeth, say) and using the angular acceleration of the crankshaft after each spark to measure output power on a cyclinder-by-cylinder, spark-by-spark basis. This would be way to much work for a 16-bit MCU, but very accesible to an FPGA, which could do lots of calculation in parallel. At the very least, this could auto-fine-tune the relative amount of fuel to each cylinder, but could even do something like the "peak power position" determination mentioned in the reference above. And I do think I read that Ultra-MS is gonna have an FPGA on board.

If only replacement engines were cheaper !

[lapuwali]

Ion-sensing has been around awhile, too. SAAB actually had controllers in production as long ago as the early 1990s. In the DIY world, much work was done by an Australian, and there's some mention of him and his work in the MS stuff. IGOR was the code name for the project, as I recall.

Reading a fine-toothed wheel gives you all kinds of new stuff to play with, like being able to detect misfires and isolate them to a particular cylinder, so you can trim ignition/fuelling per cylinder to operate just rich of the lean misfire limit for maximum economy.