subaru "5eat" also used in nissan g35 and titan

[Bernard Fife]

ClimberD,

5 shift solenoids (but perhaps could settle for 4) - the standard code has up to 4 outputs, but the forth one eats up a spare output. The fifth solenoid (for engine braking) could be done using the other spare output.

TCC solenoid (preferably w/ LU Feel) - this is in place. In the 4L60E, the LUF is a separate solenoid (and uses a spare output). Does your trans adjust the feel by PWM'ing the TCC solenoid? If so, then nothing new is needed, the standard code will handle that.

PC solenoid - no problem, this is standard in the code - PWM frequency and duty cycle are user-settable.

Spare On/Off Output to activate a relay only during a shift - a spare output can be used as a clutch output, but this depends on having some outputs left to use, of course.

With a bit of custom internal wiring, some of the LED outputs could be used as solenoid outputs (the code is ready for this), potentially releasing some of the spare outputs. This depends on how many LEDs you want to activate. The other possibility is to use a voltage based shift lever indication scheme (the 'Ford' scheme in the manual) and one or both of the other two inputs could be switched to use as outputs (again requiring some custom wiring). I am thinking this through now, and we'll decide what to do shortly. You can also check this: viewtopic.php?f=4&t=143&p=1491#p1491

For more info on the GPIO board circuits, see: http://www.msgpio.com/manuals/index.htm

Lance.

[ClimberD]

Lance,

There is a 7-segment display already in the middle of the console, used only to display the gear when in "Sport/Manual Shift" mode. I would love to take control of that for MegaShift, but will need to do a lot more reading to see if that is even reasonable for me to tackle. If it is not reasonable, then using 1-less LED would be perfectly acceptable.

I will do more reading to see what type of shift lever indication scheme Subaru uses. My gut tells me it is just one signal wire with various resistors (I have seen this in other controls), but I will need to confirm for the actual shifter position outputs.

Regarding the Toyota 8-speed link you posted, I'm afraid I will need the MAP/MAF/TPS analog input for Load reference. Frank will be able to use his MS controller with CAN to read load, so perhaps there is his spare output for solenoid-actuated pressure dump during shifts.

I clearly have more digging to do, and I will do so hopefully without too long of a delay

Edit: 4th Option, is to buy a MegaSquirt controller and just use CAN. But that would double the cost and add complexity. I'll try to think up some trick benefit to justify it, but for now I can think of nothing since the stock ECU is sufficient for my needs, and is thoroughly hacked and tunable. Is there a "super diet" version of the MegaSquirt controller, which would just receive Tach, MAP, MAF and TPS (or at least a couple of these 3 load values), and transfer it to MegaShift via CAN?

[Bernard Fife]

ClimberD,

It sounds like a voltage based shift lever would be best for you, then we could turn the other two digital inputs into outputs for shift solenoids.

There is no 'baby GPIO'. It wouldn't be much cheaper. By far the most expensive parts of a GPIO are the ampseal connector and the case. You might mount a second GPIO board without the connector in the same case, and run wires directly to it through a grommet. If DIYautotune will sell you a board with the case and connector, that's an approach that might work for you. But lets try the shift lever scheme first - it's cheaper, and makes the code more flexible.

Let me know when you find out how the OEM shift lever works.

Lance.

[ClimberD]

Thank you Lance!

Separately, is there a preferred sensor or sensorXsensor table for referencing Load for MegaShift with turbo cars? I would love your feedback.

Some quick thoughts:
TPS would give the earliest shift response, but is the least accurate indicator of engine torque.
MAP would be the slowest to respond (since relatively slow to build boost) but would be the most representative of power.
MAF would fall somewhere in the middle.

Am I right about any of this, and what is your experience with this? (since it looks like we will be using just 1 analog load input).

[Bernard Fife]

ClimberD,

If this was mine, I would use a MAP sensor as the load sensor. They are inexpensive, common, and easy to package (and no worries about leaks on install or over time).

MAP sensors more responsive than you need in a trans application - the load has to be averaged over time anyhow or the trans will shift far too often. However, the code is written so that if the MAP jumps up, the average immediately goes to that value, so there is very little delay for the appropriate shift.

Lance.

[ClimberD]

ClimberD,

If this was mine, I would use a MAP sensor as the load sensor. They are inexpensive, common, and easy to package (and no worries about leaks on install or over time).

MAP sensors more responsive than you need in a trans application - the load has to be averaged over time anyhow or the trans will shift far too often. However, the code is written so that if the MAP jumps up, the average immediately goes to that value, so there is very little delay for the appropriate shift.

Lance.

I can tap into analog TPS, MAP or MAF 0-5V signals from existing OEM sensors, so no need to buy an extra sensor. Won't even need to run a single new wire through the firewall

Here is a 4-page PDF of every Stock TCU Input/Output.

Trans Control Module TCM I-O Signal 5EAT Subaru Legacy GT.pdf

5EAT Inputs-Outputs

(43.91 KiB) Downloaded 1730 times

I cannot find an actual schematic of 5EAT shifter position inputs. But look on the 3rd and 4th page of the attached sheet.
4 Separate Inhibitor Switch Inputs are listed.

EDIT: I found this page re the Inhibitor Switch, though they do not correspond to the I/O Inputs attached above:

5EAT Inhibitor Switch.pdf

Inhibitor Switch Page

(42.5 KiB) Downloaded 915 times

Just my quick $0.02, we might only need the 1st and 3rd Inhibitor Switch inputs.
Inhibitor Switch 1 = Park is high, Neutral is low
Inhibitor Switch 3 = Reverse is high, Drive is low

Edit: 5th Inhibitor switch wire, called "Inhibitor switch 3 open circuit monitor input." No clue what that is.

So I have no idea why there would be 5 wires to tell the TCU where the damn shifter is positioned, unless Subaru just carried forward the 4AT TCU 4-wire format each on/off for P, R, N, D, except one more odd wire, and no ON/OFF on the 5AT.

Does this free up that last on/off 'during-shift' output?

Edit: I see there is a VSS output, which I may not need if my stock TCU (which will also be connected) will report it to the BIU. Not sure how all that works yet...

Lastly, every solenoid is 800Hz, which means we probably have everything we need to set it up. What does P-VIGN stand for?

[ClimberD]

Alright, after lots of edits, the post above should have all the information we talked about before.

Except: I think I figured out a point of confusion. I have read the Subaru manual and all the MSGPIO.com pages many times now. As far as I can tell, the Subaru 5EAT does not have any sort of outputs that indicate gear position in D. According to the wire signals from the inhibitor switch, you are either in P, R, N, or D. No 1,2,3,4,5th gear signal. I think the TCU calculates this based on the two VSS signals and engine RPM, or something like that. I see for the 4L60E and other transmissions, it has a separate inhibitor switch signal set for each P, R, N, 1, 2, 3, 4, etc. The Subaru 5eat is tiptronic, so you only get to choose D, and then you can press the +/- switch to gear up or down from your present gear.

Can we make this work on the Subaru 5EAT without these gear position inputs?

[ClimberD]

More on the "physical swap" side of things, it appears after more research that anyone doing this install might choose to completely pull out the stock TCU. The LegacyGT thread on swapping to a manual transmission is very revealing. http://legacygt.com/forums/showthread.p ... 32374.html

This may take some experimenting when the MegaShift controller is installed.

[Bernard Fife]

ClimberD,

Thanks, I will have a look when I get a chance.

The code (which will be numbered 2.009) is complete to add two shift solenoids to the code (replacing Input2 and Input3 when used with a voltage based shift lever).

External components will be required (likely a minimum of a base resistor, and driver, and something to handle the flyback).

The code allows PWM on these solenoids (same as input 1, 2, and 4) If PWM is used, the processor pin output is reduced by 2/3 to avoid interfering with the ADC measurements. I need to test the code a bit before putting it up on the site (http://www.msgpio.com/manuals/mshift/V2code.html) and I will also post notes here: viewtopic.php?f=4&t=20&p=1439#p1439.

The user must select a voltage based shift lever ('Shift Input Patterns/Configure Lever Type') and must also select to use one or both of the shift lever inputs as shift solenoids ('Shift Input Patterns/Configure Inputs/Input2 and Input3 Usage') for the added outputs (5 & 6) to work. Changing these settings requires a reboot for them to take effect.

I will work out the details of the driver circuit shortly.

So the code now handles up to 6 shift solenoids (with one spare output left if using all 6), and up to 8 forward speeds.

To answer your earlier question, we don't need to know the position of the gear lever except for PNRD (you can downshift manually using the shift buttons, or let the gear table do this automatically), and we don't need to know the actual gear we are in (MShift can figure this out from the state of the solenoids).

Lance.

[ClimberD]

Lance,

I am indebted to you.

I'm going to further research some of the things you have written to better understand the language.

Separately, it appears that no matter how I might connect MegaShift, the AWD lockup solenoid is always going to be without stock TCU control. If the stock TCU will think it is in D, then it will try to pull timing and cut throttle when it shouldn't. If it is told to be in Neutral, then no power will go to the center diff solenoid. Disconnecting the TCU will cut control of the center "AWD" clutch completely.
I am happy to get the rest of the controller worked out before we tackle AWD control. But basically, I am thinking there must be some sort of GPIO format that can process the following inputs: MPH, lateral G sensor, MAP, and possibly steering angle sensor, and produce a PWM AWD clutch output, or even a simpler ON/OFF, not sure what on/off would do to it yet. There is also the front VSS and rear VSS, which get compared to determine the difference in front vs rear spinning.

Anyways, with E85, roughly 530 chp will enter the trans, and roughly 400 will result at the wheels, at peak. I would hate to have a properly controlled trans, only to have all the power not being able to hit the ground in proper proportion. Do you have any thoughts as to a simple way to use the GPIO board to control a center diff? I have thought about a second MegaShift controller dedicated to this purpose. It appears the 1st MegaShift controller may already be tapped out of inputs and outputs. At any rate, I pledge to do another round of research into AWD clutch controls, including the aftermarket solution for the STi 6-speed transmission. Then I will bring those findings to the table.

Thank you again for so overwhelmingly supporting my interest in such a feature-heavy transmission. I looked at the Subaru 4AT diagram, and laughed out loud at the small number of solenoids, and complete ease through which it is controlled electronically. If only life could be so convenient...