Richard Paul

iTrader: (5)

RotaryGod, If it is nessasary we can always make the timing wheel ourselves.
If they are not available or need to much work to modify let me know.
Also if it is nessasary to keep the stock one and run another at the same time.
If you can figure out how it fits and draw it I can make it.
Doesn't have to be a CAD drawing just show me what it looks like, I'll do the cad work.
This is an offer only if you can't get one. It is always cheaper to buy it then make a one off part.

nijelSC

Bump to keep it alive.

Richard Paul

iTrader: (5)

Now the forums back up, after I sent RotaryGod an E mail. Just so it is documented and we get everyones input I'll repeat myself here.
I question the idea of moving the intake temp sensor into the intake manifold. If you read the hot wire function but use the temp of the compressed air you will get a false reading. I think if left alone as long as there is enough room in the program everything will work right. In that case there is no reason to tell the engine it is supercharged. How would it know? It just thinks it got bigger or you drove below sea level.

brillo

don't worry its still alive, I've got the eccentric wheel sourced and will order it on Monday. I'm thinking about other potential issues we may have and trying to find ways around them this weekend. After speaking with the nice folks at Speedsource Racing, I learned that the power steering is PCM controled, so I'm not sure what may happen when we take over some of the controls, hopefully, like the throttle, it won't care.

Speedsource uses Motec computers that can actually control the DBW, but they race with no power steering. I'm guessing thats for weight and reliability reasons rather than computer issues, so we will see.

I'm pretty confident that i could run the car on a standalone now, but I would like to do it using as many of the stock systems and keeping the creature comforts as well.

GiN

From what I've heard, the power steering system is run from another dedicated PCM and communicates via some sort of ultra-top-secret CAN scheme that Mazda doesn't want anyone to know about - not even the folks at Motec. Either that, or Motec's CAN spec isn't compatible with what is used on the RX-8. Hence, you might also want to consider the loss of stock dash / instrument panel functionality when upgrading to a full standalone. The Speedsource cars use Motec ADL displays in place of the stock dash.

Longhornxtreme

Gin... You need to read the ENTIRE thread....

rotarygod

We aren't going to have any issues with interior gauges and I am pretty sure power steering won't be an issue. Remember the stock ecu will still be reading all of the engine's signals. It just won't control ignition or fuel. The rest it should do just fine.

brillo

RG and I were talking earlier and RG mentioned that the Megasquirt ignition usually works in wasted spark mode, which could be an issue for the stock coil packs. We might be able to find away to get it to fire in a different format, but I'll have to look into this carefully as the Canzoomer folks were cooking coil packs pretty easily.

Not a major hurtle, we could aways adapt some 2nd gen coils, but I'd like to use the stock ones if possible.

Jaguar

Ignition Solutions makes some upgraded coils for the 8 that they have been selling the crap out of in Japan. They have been featured in alot of car magazines with positive comments and gains. Here is a link to thier site....www.ignitionsolutions.com. They don't list the part # on thier site but it is PDC-5313-01. The only down side is that they cost $580. But they do come with warranty. Check the media tab at the top, it has links to articles done up about their products.

With regards the the MS. What are the dimensions of the housing box? They look about the same size as desktop computer from the MS site. But I can't imagine that it would be that big.

Thanks.

brillo

the box is about 6in by 4in by 2in? its small, bout the size of the emanage. interesting site, to bad they misspelled Efficiency right on the front page

rotarygod

Well I guess after a little more research we are going to encounter a problem with the power steering, VDI, auxiliary ports, etc. after all. The stock crank trigger wheel can at this time only be read by a Motec ecu. Megasquirt can not read it. The solution to this was to install a 36-2 trigger wheel that it could read. If it were only that simple. The problem with this is that the stock ecu can't read the new wheel. Why is this an issue? Well it must stay hooked up in order to control all of the rpm related features of the car. This includes all that I just listed. It gets this signal straight from the trigger wheel sensor itself. If we send the new wheel pattern to the stock ecu, it will be speaking a different language and we will completely lose all of these features. I don't know about you but I'm not to keen (is that word still used anymore?) on this idea. We could probably get away with installing the new wheel somehow in addition to the old one and then have another sensor added so the MS can read it but that is really getting away from the plug and play aspect that we are shooting for. The goal is to not have to modify anything. Even changing the crank trigger wheel wasn't along the lines of the original intentions as that too requires more than jstu plug and play but it would have been acceptable if it had been the only thing additional that needed to be done. Apparently we will face more issues with that. Alan had these suspicions through e-mail conversations with Speedsource Racing. They are running a Motec ecu on their race cars. They originally tried to pseudo piggyback/standalone the Motec onto the stock ecu much like we are attempting with the MS and they ran into this same crank trigger issue. Their solution was to run a 36-2 trigger wheel but then the other issues came up. At least we are finding this out through research rather than trial and error! Luckily for them Motec very quickly updated their ecu so that it could read the stock trigger wheel pattern and all is well.

It has been suggested that the new Haltech E8 can read the RX-8 wheel but I'll wait to actually hear that someone has done it before I totally believe it. Nothing against Haltech. The info comes from a Haltech person who claims the trigger wheel of the late model WRX is the same as the RX-8. I have pictures of both and can tell you that they are not the same so I wait confirmation to see if I'm just missing something somewhere. If it does control it Haltech runs about $1400 or so right now. MS will always be under $200 and has the best online support group out there.

So where does that leave us? The simple solution is to find a way to get Megasquirt to read the stock trigger wheel. This is the same scenario that many people have encountered with different cars into the past that is subsequently be added over and over again to new software revisions of Megasquirt. We need to find a way to write the code so that it can read the stock trigger wheel. Since neither myslef nor Alan can do this with any degree of competence and since we are the only ones trying to integrate a Megasquirt onto an RX-8, that leaves us with a slight problem. It looks like it's time to get a crash course education in computer programming. OK we'll find someone that can do it and tell them what we want. That's where we are at right now. If this gets worked out, we should not have to add anytning else to this system other than to just wire up a harness and literally plug it into the car and go. What a great day that will be. Worried about emissions testing? It should pass but if it is a concern, just open the hood, unplug it and drive off. We'll get there some day.

BaronVonBigmeat

What kind of programming language would you have to know?

Richard Paul

iTrader: (5)

Fred, what if you made a new pulley with the notches cut into the front rim?
I don't know what it looks like down there so can't see where the sensor would fit without getting into trouble with the belt. If there is room you could point it from the front to the rear. Remember you can design the pulley anyway you want it. The pulley could extend or put the timing wheel between the two belts. Use your imagination.

You never answered my q about the temp sensor.

Longhornxtreme

I'd be willing to contribute a small amount of funds to contract the programming out to someone with the necessary education. If you can find out how large the project will be I'll try and contact my friend with a CS major, I bet he could use a little cash...

rotarygod

Richard I was out of town over the weekend. Sorry I didn't get back to you on that. You could try to keep the temp sensor in front of the blower but I really like the idea of having one after it. My personal opinion is that I only need to know what the temperature of the air is that is actually going into the engine. After all that's what tuning is based on and what correction is all about. I don't need to know what it is before a blower modifies it.

I don't know a whole lot and am still getting questions answered on the Megasquirt site. I do know it is in assembly code but beyond that I have no clue. We're still researching that. Even after we figure that out, we still need to figure out how to build a test bench using a stock trigger wheel and pickup sensor. I'd rather test it on a bench using a drill or electric motor than sit there and flood Alan's engine with fuel while we crank it over with no lubrication. I'll let you know when we get this figured out. The hope is that we can take some existing code and modify it rather than write entirely new code.

brillo

I just spent the last 2 hours reading up on the MS programing. In a nutshell, the processor runs on assembly language, which is good and bad. Its good, because its real efficient, but its also the least intuitive. I have some base programing skills, enough to read some code and get a feel for what its doing, I'm attempting that now. The good news is I'm not needing to write a whole program, just alter a variable or two (I hope). I'm going to run through the code as best i can to at least id the problem area so i can point someone in the direction of where to add / fix the problem.

we have some other issues to work out in the mean time so we'll work around this for right now. we are still working out the remaining wiring issues and signal testing.

Richard Paul

iTrader: (5)

Rg, you are wrong, you are going to mesure the mass flow then tell it that the fluid was hotter? you can't do that. You can use a map sensor and disregard the MF using speed density maps. But you can't take the flow and use the exit temps it will get the wrong idea.

If left alone the MF will operate correctly no mater how much goes through it up to the limit of the programing and other hardware. Change the temp of it's readings and it will think the air is thinner and lean out. Just what you were trying to prevent.

rotarygod

I understand what you are saying now. Wasn't thinking that through as well as I thought apparently. That might work just fine then. It sure does make life easier to use something that is already there. Strangely enough I'm only focusing on the temp sensor rather than the map which is why I'm thinking that way.

Longhornxtreme

Wouldn't it just be better to add the GM IAT sensor that works with MS right around the same area we add the MAP sensor? We need to know the temp of the charge air... not the temp of the drawn through air at the stock MAF

rotarygod

Actually using the air temp sensor in the maf is perfectly fine.

rkostolni

Its been a while, but in college I did some Intel assembly coding and some PIC assembly coding. Send me any documentation you guys have found and describe exactly what the processor needs to input, from where, and what it needs to output, and I'll try to see if I can help.

My email address incase it is more convenient is rkostolni@yahoo.com

rotarygod

I'm going to get into some newly learned tech now for the guys that can follow along and are interested in my findings. To those that can't, just trust me! Alan: I just figured, what the hell. Just post it all.

After studying that chart for a while and the older rotary crank triggers, I have figured out how the RX-8 wheel works and can honestly say that there is no way that the late model WRX wheel is the same. Haltech will not be able to read this wheel at this time. If you look at the picture you can see 3 separate "flat" sides. On the right and left we can see 2 of the flat sides are exactly 180 degrees apart from each other. Above the flat side on the right is another one spaced exactly 30 degrees off of it and firing later. Let's think about this for a second. How does the rotary ignition system work? Leading then trailing plugs fire one after another. That's what the 2 flat sides represent on the right. The bottom one is leading and the upper one is trailing. If anyone has read anywhere about the timing split on the RX-8, it isn't 30 degrees as these 2 sides would indicate it to be. The RX-8 timing split is actually between 5-15 degrees depending on rpm and load. Mazda decided to take a reverse approach to controlling this. Instead of starting at the same point as the leading and then programming in negative numbers to increase the split, they started at a fixed split number far out and are adding numbers to decrease split. They add anywhere from 15-25 back into the trailing side within the code. Never thought of doing it that way but apparently it works.

The strange thing about this system that is different from the RX-7 crank trigger systems is that the RX-8 only uses 1 crank pickup sensor whereas the RX-7's used 2. They had 1 sensor for the leading and 1 for the trailing. The RX-8 ecu somehow reads both of these flat sides and then determines which side is actually passing it and fires the appropriate coil. This will be the biggest area of programming in the Megasquirt. We will need to find a way to have it read what it already does using 2 sensors but instead learn how to do it on only 1 sensor. The ignition circuit that I will have to build for the RX-8 will be the same though and will be exactly half of what the RX-7 ignition circuit is since they use 2 of the same circuit on 1 board. That's the easy part.

Now you're probably wondering why I didn't mention the opposing flat side on the left side of the trigger wheel. What can that be for? Remember that it is exactly 180 degrees out from the leading flat side opposite it. What does this mean? The RX-8 has a wasted spark leading ignition system! Yes it does. However there have been people that have tested it and not found this to be true. Why? Simple. Mazda has the code written which allows them to disengage wasted spark whenever they want to. It is only programmed to run wasted spark where it needs it. Since wasted spark does nothing for power and is only emissions, it is used at lower rpm's and loads. That's it. At higher load levels and rpm's wasted spark isn't needed. This is because rotaries have their worst emissions at slower speeds and don't need as much emissions control at higher rpms.

The single most important reason why they cut out wasted spark at higher rpm's though is not because of fantastic engineering. It is because of the budget department at Mazda. Anyone ever seen how pathetic and small those ignition coils are on the car? Some people have already killed them. Wasted spark fires the leading coils twice per revolution which obviously works them twice as hard and gets them much hotter than if they weren't running wasted spark. I have no idea at what rpm or load they disengage wasted spark but let's just say it is at 3000 rpm. I made this number up so don't memorize it. Let's say it fires twice as often up to this rpm. This means that the leading coils are actually firing 6000 times per revolution. That's fine as running in normal mode which is non wasted spark, they are still firing 50% more often than this at 9000 rpm. Got it? At 3000 rpm they would fire at 6000 times per minute but at 3100 rpm they would only be firing 3100 times per minute. Remember I only made those numbers up for the sake of example. I can't be that far off though. The trailing coils never fire in wasted spark mode so they always fire once per revolution.

Going back to the trailing side of things. Remember I said that the ecu adds numbers back in to decrease the split. This is undoubtedly based on map points within the stock ecu. At idle the RX-8 fires the trailing coils BEFORE the leading. Actually 5 degrees before the leading. Since the leading and trailing sides of the trigger wheel are 30 degrees apart, this means that the stock ecu is adding 35 degrees back to the front side of it's base number which results in it firing trailing first. I now actually have my doubts that the ecu fires the trailing first at startup as well. I now suspect it doesn't. I suspect that leading and trailing either fire together or in a very close split for startup and then immediately after the engine warms up and settles down to it's normal idle rpm, trailing then fires first. If trailing fired before leading at startup, we may have some very serious issues in terms of how easy it would be to start the car. It would probably be next to impossible without getting a large hill involved.

What this basically boils down to in terms of what is needed to get the Megasquirt working with this setup is really nothing more than code that can decipher which face that is passing the sensor at any particular time belongs to which coil and when to fire it. It needs to know that the 2 faces that are 180 degrees apart always fire the leading coils. This really only makes 1 leading coild output which means you can't wire the leading plugs backwards with this new setup. You can right now though so don't try it! You never know we might just be able to keep this 2 separate circuits for each leading and just be able to progam it to shut off wasted spark if we wanted to. I don't know yet but it is probably also only a coding issue. The ecu just needs to know that the side that comes 30 degrees after only 1 of the other sides is the trailing coil. Then it needs to know which trailing coil to fire as every revolution around it will be alternating this one signal between 2 trailing coils. This is already being done in the code now but with 2 sensors and circuits. We need to figure out how to integrate that all into one sensor circuit.

The only piece of concern is that once we do figure this all out, it is unlikely that we will be able to shut off wasted spark mode. As I stated already, we may find a way around this or we may not. This would basically mean wasted coils mode very shortly afterward. They are pretty small and can't dissipate heat fast enough but more importantly they can't build up a charge and release it fast enough to fire consistently at higher rpms. We are basically in essence trying to fire them at a rate of 19000 rpm at peak engine speed. I don't see that happening. The other option would be to offer an ignition coil pack upgrade as part of the ecu kit. This is what some others offer as an add on such as Microtech. I am not opposed to this idea at all since good coils aren't that expensive, especially compared to the stock coils, and a nice stronger coil upgrade would be a welcome addition to the car.

This is it for now. The next step is to work on the code but at least this is another thing figured out.

swoope

iTrader: (5)

great read rg,
you make my brain hurt in a good way.

beers

Longhornxtreme

RG... Would we be better off looking at a completely new ignition system such as something from MSD? I know alot of your research is helping RP find the best way to manage his AFSC, but what about those of us thinking BIG turbo?

My knowledge in the ignition control area is limited as the last two engines I had used individual coil packs and my modifications were pretty small as it didn't take much to add a couple hundred horses to a late model chevy small block... I pretty much let the stock computers suss out all those issues... Just used a MSD retard box when on the juice...

rotarygod

A new set of coils may be necessary and this doesn't bother me. I really don't want to bother trying to tell people how to integrate 3-4 ignition boxes onto their cars though.

Alot of this is aimed at potentially helping Richard but he still wants to do it differently than I do so it may not end up helping him at all. What I want to do is to finally give RX-8 owners full control over their engine so they can not only tune it to max performance whether that be stock or ported but also as a good option for those going forced induction. Megasquirt can read up to 20 psi but I don't recommend high boost with it. The reason is that it only has 12X12 maps. You divide these up into vacuum/pressure load in one column. This is only 12 points. For a nonturbo engine, you encounter only vacuum in the system. You don't have nearly the range to look over and control so you don't need to have as many columns of control. The more boost you run though the more pressure range you have in the system. You will now have to divide these same number of columns into a far wider area of coverage. This means that the control resolution dwindles and alot of your tuning is based on averages between 2 points. It's not to say that it can't work good but I personally am under the opinion that boost when used with MS should be kept within reason and not approach the 20 psi mark. I'd be comfortable at 10 psi. That's personal preference though as you could probably carefully tune to higher levels with a great deal of success. As the MS gets upgrades in the future and processor speed goes up, these maps will also increase in size and resolution. You've got to start somewhere.