I am hoping, beyond what I can search here, someone has or has seen a
good scope trace of the Duty C signal. In a perfect world, there would
be a trace from an '00-'03 and one from an '04, when apparently the
signal polarity was changed.



Before '04, +12V to the duty C disabled the AWD, '04 and above, +12V fully activates the duty-C and the AWD.



What I am hoping is that I can get reasonably close to the proper signal by using a few transistors to 'invert' the signal.
 

The duty-C is modulated according to throttle position, inhibitor switch, gear, and VSS. And then if a slip is detected it will add more pressure.

 

So if you want to compare apples to apples, the signal should be measured under the same conditions, and not just in park.

 

I imagine the function was inverted hydraulically somehow, and simply inverting the electrical signal will not be ideal because the hydraulic response is non-linear.

IIRC it's a 50hz square wave. I had pictures and a video of when I had my 98 legacy automatic, I'm tearing my hair out trying to find them. This sis annoying. Friend of mine has an 02 legacy auto, next time I'm up there I can do some probing with the scope while he's driving.

Heh, I'm not going for 'ideal', I'll settle for 'good enough and nothing gets damaged'.

 

If full lockup with the kitbashed setup isn't sufficient, I'll also implement a 'full lockup' switch.

 

The 2004 signal is a 50HZ signal, at 50% duty cycle it's a symmetric square wave, and the higher the duty cycle the less time it spends at 0V (or just as easy, floating, not driven).

Nevermind 

Edited February 8, 201313 yr by presslab

You are correct that the '03 signal might not be the '04 signal but inverted.

 

At least when I drive home tonight in the snow I know to put in the FWD fuse, which only for my sube, will indicate 'Four Wheel Drive'.

I recall the TCU had a bunch of 5 pin devices on it, which I th9ink are the drivers:
HA13705C



If this is the case I could pull the input pin (CMOS level stuff) and wire in an inverted signal from an inverter, and the solenoid signal would be inverted.

OK, some progress.

Have my TCU out, have traced the circuits for the Duty C output.

Device is a Sanken Electric SI 5151s.

Similar to the hitachi device above, but opinout is:

1- GND
2- SIGNAL IN
3- Vo (output)
4- Diag (what the TCU uses to tell if the solenoid circuit is shorted or open)
5- Vs (presumably +12V)

SIGNAL IN can be a TTL or CMOS input, I'm thinking of snagging an inverter (will have to measure if these are +5V or 3.3V circuits) and just invert the signal to the driver.

This driver can source 1.8A, 1.5W as installed without a heatsink (18W with).

I wonder if I can find an inverting driver I can use instead?

Other driver devices on the board are 5155s (2.5A) and 5153s (2A but built-in zener).

Links I found that led to the info-

http://wikitest.pgmfi.org/twiki/bin/view.pl/Library/OBD1CivicIntegraECUs

http://wikitest.pgmfi.org/twiki/pub/Library/OBD1CivicIntegraECUs/SI-5151S.pdf

http://www.littlediode.com/datasheets/pdf/Datasheets-SI5/SI5155S.PDF

http://www.datasheetdir.com/SI-5153S+download
 

Finding pwr and gnd would be a lot easier if I could ID any of these ICs:

Mitsubishi M59550 FP, and M59350

No logo, marked 5223, I don't think it's a MAX5223.

The 59350 might be a watchdog timer/power IC.
The connections make sense.
http://pdf1.alldatasheet.com/datasheet-pdf/view/249323/RENESAS/M59350FP.html


 

A thought occurs to me- if I invert the signal going to the solenoid driver, I may also have to invert the monitoring signal from the driver to the microprocessor.

I just noticed something else- the FWD fuse does NOT connect the TCU pin to +12V, it connects it to ground.

Also, to avoid activating a code/flashing AT Temp light, a resistance of something less than 30k ohms needs to be between the solenoid output and ground, the open circuit sensor has a max trip resistance detected of 30k ohms.

 

No power resistor required, and need never be switched out of the circuit.

Well, my temp mod is done.
Wired up a DPDT center-off switch so I have:
- normal connection
- transfer solenoid disconnected to give me front-wheel-drive
- FWD switch active, which gives me 'locked' AWD.

I went by the specs on the driver chip, which said a 30k resistance would keep the open-circuit alert from tripping, wired in a 10k, apparently iot does need to be less than that.

I have GND and the sig from the TCU coming up to where I can get to them, so I can add a resistor and see where the cutoff is.

I also have a set of voltage measurements to compare against (at some point) for all gears, at idle, mid-throttle, and full-throttle, measured AC and DC.
 

Looking at the valve bodies, 2003 vs 2004, no visible differences.

There are a few external differences on the bellhousing, though.

Seems like an awful lot of trouble to go through if you're not even sure the Duty c is working properly. Have you checked line pressure at the solenoid? I'd check everything out mechanically before going after the electronic end.

Based on how well the AWD works when supposed to assuming a reversed signal, as subaru has confirmed is used, the Duty C is working perfectly.

 

It has also been confirmed by someone else the 2004 signal is reversed. There is no question.

I'm going to attempt this on the output side first, I found an inverting driver that will take the output of the TCU and drive the solenoid directly.

http://ww1.microchip.com/downloads/en/DeviceDoc/25006a.pdf

And ohcrap, just realized while I was ordering stuff from Digikey I forgot a power resistor to terminate the TCU output...

There's always something you forget. Good thing about digikey is they're usually pretty nice about adding stuff to the order if you give them a call. They do package and ship REALLY fast though, so sometimes they can't do it.

Yeah, I got a SHIPPED notice before I even posted that...

And a diode, forgot that too, but there is a radio shack and even an electronic supply house not far away.

Heh, I think I found a diode that will work- ultra-fast 200V high current 1N3891, made for high frequency rectification.

 

I also noticed when the TCU senses the current to the solenoid is too low, it latches the error condition and does not try to drive the solenoid until after the power is reset (key off).

What is your end goal with all of this?  Is something broke or do you just want more control over you AWD?

 

Reason I ask is because it seems like alot of work to implement the "diff lock" mod...

This is not a 'diff-lock' mod, the 2004 trans I put in has the AWD respond to the TCU transfer-duty solenoid signal the opposite way as my '03 trans (car is an '03). My choice is invert the signal, or do ANOTHER trans swap. Needless to say, I'm inverting the signal.

 

If it works properly with a simple inversion of the signal, I'll never use the switch again. I never had a case where 'more lock' under manual control would have helped the tiniest bit.

I'm picking up what you are putting down, makes total sense now.

 

Wouldn't it be easier to replace to TCU? Or is there something I am missing? I don't know the new gen stuff that well...

 

But don't get me wrong, modding stuff is fun and I do it as much as possible, I am just asking because I am curious :-)

2004 TCU would require also changing the ECU, and another (at least one) piece of electronics.

 

So, easiest is invert the sig or build a manual control.