MIVEC disassembly notes
 

Just getting started on decoding the MIVEC algorithm. First tidbit is the MIVEC range. My recollection is that people are saying its 0 to 30 deg. The software code limits it to 0 to 28.8 deg, so anything higher than that in your MIVEC maps gets pulled down to 28.8 deg.

More later...

subscribed {thumbup}

Interesting. Its the factory service manual that said 0-30 degrees.

maybe -1.2 to 28.8?

I thought it only went to 28.8

LOL at all these people saying they use "30s" and can feel a difference.

Maybe they have guitar amps that can go to 11? :D

The limits are definitely 0 to 28.8 in the ECUFlash scaling of the hex values. I still haven't verified the scaling used by ECUFlash though. Most of the scalings that I've checked have been correct, but there have been a few errors here and there.

Hey JB, does the FSM say "0" to "30"?

subscribed {thumbup}

The factory service manual says that the intake valve opens (variable) 0-30* BTDC and closes 80-50* ABDC. I might have the actual piston event labelled incorrectly but I listed it in the original MIVEC tuning thread somewhere.

As far as the wags making reference to Spinal Tap, while I cant explain why it would matter if the hex is limited to 28.8* (I assume a value of 255) there are dyno'd torque improvements running 30*. A certain tooner that has left for the green pastures of BMW and Porsche used to set his maps to 33* and it would throw a MIVEC out of range code (like P1175 I think).

I remember that. Maybe its this...

I knew I posted it in there somewhere:

https://www.evolutionm.net/forums/sh...5&postcount=99

Re: Spinal Tap. Was just trying to be funny...not intended to be a swipe at anyone's expense. And I thought I read somewhere said tuner would input values even higher than 33*.

That certain tooner also had maps that had 20+ knock count.

That would line up with my experience with said tuner, although I only had 12-16 IIRC. :cool:

lol your signature is priceless. Rediculous, but priceless. {thumbup}

I'm pretty sure that I've found the RAM variable that contains the actual MIVEC position. For the USDM Evo IX, it looks to be 0xFFFF7650. Below is a plot of actual MIVEC vs the MIVEC trace through the my current MIVEC map. Actual MIVEC is not reaching 30. The max value is hovering around 28.2 - 28.6.

I've been thinking about the MIVEC CEL code that people have mentioned. Even with the 33 deg request in my MIVEC table, I've yet to see the MIVEC CEL in the year or so since I started tinkering with this tune. And the ROM code definitely has safeguards that clip requests greater than 28.8 deg back down to 28.8 deg. So, I'd be somewhat surprised if anyone is getting a MIVEC CEL from running values greater than 28.8 deg in their MIVEC table. If they are, I'd like to see their ROM.

On another note, the 28.8 deg limit built into the software is not "hardwired" into the code. It can be changed. If someone could measure a MIVEC gear assembly to see if more than 28.8 deg are available, then this value could be bumped up.

https://www.evolutionm.net/forums/at...hmentid=113795

Awesome find mrfred. I'd try that value out on the way to work today, but I'm hella tired. Maybe tomorrow ...

Thanks for the hard work.

And to add insult to injury...his sidekick refers to knock sum as just noise that is not relevant to knock.

Thank you very much mrfred. This is very exciting.

When will you be able to post a how to modify the rom and Evoscan to log mivec?

+1 {thumbup}

Subscribed!:D

Turns out the intake cam position variable (FFFF7650) is a 2-byte value. If you want to log MIVEC (aka intake cam advance) with a USDM Evo IX, you'll need to point to two unused (unneeded) MUT ports to: 0x7650 (MIVEC high byte) and 0x7651 (MIVEC low byte), e.g., your ECUFlash ROM xml file you'll need an entry such as:

<table name="MUT40 MIVEC high bit" category="MUT" address="3ef86" type="1D" level="1" scaling="Hex16"/>

<table name="MUT41 MIVEC low bit" category="MUT" address="3ef8a" type="1D" level="1" scaling="Hex16"/>

And then after launching ECUFlash, you'll need to insert 0x7650 into MUT40 and 0x7651 into MUT41.


For EvoScan, the formula is:

<DataListItem DataLog="N" Color="" Display="MIVEC" LogReference="MIVEC" RequestID="40" RequestID2="41" Eval="72-0.6*x/32" Unit="deg" MetricEval="" MetricUnit="" ResponseBytes="1" GaugeMin="0" GaugeMax="33" ChartMin="0" ChartMax="33" ScalingFactor="1" Notes="" />

In this example, I used MUT ports 40 and 41.

When logging MIVEC, you'll likely see spikes in the data. I haven't taken the time yet to determine if the spikes are real or correspond to a high byte change over (same as what happens with 2-byte load spikes).

Looks like there's a possibility of increasing the MIVEC advance range. In the post listed below, Shiv alludes to changing the MIVEC advance limit.

https://www.evolutionm.net/forums/sh...20#post3670320

I'll give it a try on my way home from work. I'll probably try for 33 deg just to see what happens.

Now I've pretty much convinced myself that approx 28-29 deg is the max advance. I grabbed a drawing of the MIVEC assembly from the FSM and measured the amount of rotation possible. Assuming the drawing proportions are reasonably correct, then it looks like there is room for up to 50 deg of advance. However, there is a nub on one of the MIVEC arms that limits the MIVEC advance to 28.8 deg. Seems like that nub would have to sawed off get more than 28.8 deg.

https://www.evolutionm.net/forums/at...hmentid=113813

What would the advantages be from more cam timing? Any down sides? the factory put it there for a reason, but if more power could be made, I would be willing to saw it off!!!

Is the 4G63 an interference motor? If so, how may degrees of cam advance until the piston and the valve go boom? If they use the same MIVEC assembly for other cars, the length of that stud is probably used to set the safe limit for each car.

^ It is an interference motor.

that's what I was thinking... I think that 30 degrees is a mechanical limit, not a software one.

very interesting work mfred{thumbup}

I agree that the mechanical limit on advance is likely to protect the valves from running into each other or into the piston. There must be some safety cushion though because Cosi and GSC cams must have either longer duration or more lift or both. I imagine that there must be at least a few cam degrees of safety cushion.

What if you only shave off like 5-6mm to get like 35 or 40 degrees of timing? Or get one made?

I made it sound as if hacking off that nub was a simple as grabbing a hack saw. Its definitely more complicated than that. Cutting anything off would throw the assembly out of balance, and the issue of when the valves start smashing into things would take some time to figure out. And if it could be done, there's no guarantee of any signficiant gains.

I mentioned in the thread I started on the various load variables used by the ECU, but I'll mention here again that MIVEC uses the same load variable used by the stock boost control system. This load variable can track as much as 20 load units below the load used for fuel at high IAT temperatures.

I understand, I think it would be better if we had one made that was balanced and all... Im sure our major companies in japan have looked at making aftermarket cam gears for the IX for even more adjustability.{thumbup}

very interesting stuff mfred. I personally think what we have available is good enough. There are so many factors into manipulating something like that as fred said that it is probably not worth it in the end.

I have yet to see a company come out with something for the STi which has be using this same technology for years so i doubt we will see anything in the near future.

Subscribed!

At lunch I raised the software upper limit on the MIVEC advance from 28.8 deg to 30 deg. The software limits are two 1D tables found at 0x293A and 0x293C on a USDM Evo IX. They are two-byte values. These can be viewed as uint16 with a scaling of 72-0.6x/32.

The result is that even with values of 30+ deg in the MIVEC table and 30 deg in the software upper limit, the actual MIVEC advance does not exceed 28.8 degrees. See table below. So, I'm very certain now that the true MIVEC upper limit on the Evo is 28.8 degrees. All you people with values of greater than 28.8 deg in your MIVEC table can turn those values down to 28.8. :-(

https://www.evolutionm.net/forums/at...hmentid=113839

From what I read it sounds like the valve controlling oil pressure to the cam gears is on an independent feedback controller. If you set the setpoint at 28.8 it will control the valve so the cam advance is right around the bump stop. Setting it to 30 will most likely force it to try to control to 30, so the control valve will go wide open and use as much oil pressure as it can let through to force the cams against the bump stop. This may reduce flutter on the cam position....? I doubt it'll make any real difference.

This is all speculation, but it would be interesting to compare a 28.8* log vs a 30* log to see if the amount of position variation changes.

I really don't get why the Evo uses so many different load variables. I wonder if it makes sense to have the ECU use one load variable for everything. It would certainly make tuning easier.

The MIVEC max advance setting can be thought of as the feedback on the hardware limit. Your assessment of what happens when the upper limit is set to 30 is mine as well. Not sure about the flutter. Could be some, but hard to say as some of the variability in the MIVEC logs could just be resolution related. Without some knowledge of the capability of the MIVEC hardware, it worries me to let the valve be held open continously at full advance (normally the valve would close upon reaching 28.8 deg), so I am not willing to leave the upper limit set to 30 deg.

Not sure why you say its speculation. I've logged MIVEC with the upper limit set to 28.8 deg (factory value) and 30 deg. Both plots are in this thread. In both cases, MIVEC did not exceed 28.8 deg.

I'm just waiting for mrfred to find the mivec link to the timing 1,2,3 decision tree :P then I wont have to do some pretty graphs to prove it :P

I meant the part about there actually being a difference in how the feedback controller sees an input of 28.8 vs 30. If the output is a 1-5V signal and the scaling is set so that 28.8 in the ECU is 5V then it won't make any difference either way. It was my way of saying that I'm making a WAG.

Found out this evening that while 28.8 deg looks for certain to be the upper limit on MIVEC, 0 deg is not the lower limit. I was able to push the MIVEC as far as -1.6 deg by altering the lower limit and adding negative values to the MIVEC table. So, it looks like the total range is around 30-31 deg, with it spanning from around -1.6 deg (logged) to around 28.4 deg (also logged).

Not sure how much value there is to be able to go negative by -1.6 deg. I suppose the place to use it would be at higher rpms.

If you want to see the MIVEC upper lower limits, add the following two entries to your 88590015 xml file:

---
<table name="MIVEC Bounding Minimum" category="Timing" address="293A" type="1D" level="1" scaling="ValveTiming16"/>

<table name="MIVEC Bounding Maximum" category="Timing" address="293C" type="1D" level="1" scaling="ValveTiming16"/>
---

Then add the following entry to your evo9base.xml file (if you're using post 1.29 version of ECUFlash, use "big" instead of "little" endian):

---
<scaling name="ValveTiming16" units="degrees" toexpr="72-x*0.6/32" frexpr="(72-x)*32/0.6" format="%.1f" min="-1.8" max="28.8" inc="0.1" storagetype="uint16" endian="little"/>
---

Interesting, so you had to alter something for the car to actually do -1.6 when you had -1.6 in the table?

Yep. See my edited post above. Besides the MIVEC table, there is a bounding upper and lower limit setting in the ROM. No matter what you put in the MIVEC table, the MIVEC will not exceed the bounding limits. And of course there is the physical limitation on the MIVEC range which trumps everything else. :-)

Cool stuff. mrfred for pres! :)

Could be possible, but I think the MIVEC signal is measured from the cam position sensor which is a pulsed signal. This would mean that there is no voltage limitation say like for a MAP or MAF sensor. It would be more like RPM. And MIVEC is represented in the ROM by two bytes with the 0 - 28.8 deg range using up only a small fraction of the available range of two bytes.

Yesh, suggested it and Tephra is looking into finding the perfect load variable to use. Also making sure that using one variable wont mess something up.

TB's post FTW.

Indeed.

What values should they read in Ecuflash?

Any ideas for rom 88570008?

SK

0.0 and 28.8

So I tried out -1.6 (which actually you can't input into ecuflash, it just snaps to -1.8) on the dyno today. Was interested in seeing what it would do as I usually run 0 up top on the 9's. Didn't really do anything power wise but this Evo was a un-standard case 9, using a exhaust adjustable cam gear.

I'll try it out on some more 9's and report back. :)

Lol. That's why I get the big bucks ...

My MIVEC map is actually running just a little advance up top, so I'm looking forward to your results Razorlab.

What position was the exh cam gear set to?

Well theoretically the idea still holds true that alot of early advance will help spool but I am not sure what the range would be where there are diminishing gains. I remember reading that when Subaru AVCS came out it was supposed to feature 40* of adjustability. Looking at Subaru maps there definitely isnt 40* of adj being used however.

It was pointed out earlier that the 4G63 is an interference motor, but the piston is moving away from the intake valve so there isnt really a chance of interference unless it was extremely over advanced BTDC. As far this is concerned you could look at the specs for the biggest MIVEC cam now available and see how much advance BTDC it has taking into account it has the safety margin for 28.8*. Add the 2 numbers together and you should come up with a reasonable idea of what the max limit is. I will look at the cam engineering for the Cossies again and then see what we have for a rough idea. I would assume its only in the 45-50* range though (total max advance).

mrfred- One other question about the "nub" on the MIVEC gear. I am thinking this is behind the plastic cover and IN the spool valve assembly yes? Once my car is back from getting the transmission rebuilt or replaced I am gonna do this just because I have to know whether it helps or not.

Bryan- Without having to disclose the actual cam gear setting has this helped spool at all? Optimum MIVEC changes with the lobecenter so altering the effective engineering requires a little time with a dynosim and then you can see what will help or not. I can run it on my lappy later if you want some ideas (unless you already got it in the bag), just PM if you want it done.

Based on the diagram, it looks to me like one of the MIVEC arms has the nub molded onto it. Those arms are the arms that see the oil, so the MIVEC assembly on the cam gear would need to come apart. Is that your take as well?

I had it all apart when I did cams but I cant remember how it actually goes together. From that diagram there yes I agree that it needs to be disassembled in order to be modified.

oops, yes I should have been more specific, all I meant was 'was the exh cam advanced or retarded relative to stock'? I could see the exh cam going either way, depending on what issue you're trying to lick. No need to disclose the actual setting.