93 Impreza downshift in power mode

[remcow]

Hi all

When I step on the accelerator, one would expect an automatic
transmission to downshift and impreza has this 'power mode', which
makes kinda more fun to drive.

My question is: can the sensitivity to how much the pedal is jammed
down be adjusted? It takes very little at lower speeds for power mode
to take over and it seems that power mode is often not really
required. Sometimes the pedal is really not pressed further at all,
yet 'power mode' is active.
At higher speeds (>40m/h) the car behaves normally.

If it can be caused by a defective part, what direction should I look
into?

Thanks for the help.
Remco

 

[hippo]

<<When I step on the accelerator, one would expect an automatic
transmission to downshift and impreza has this 'power mode', which makes
kinda more fun to drive.>>

<<My question is: can the sensitivity to how much the pedal is jammed down
be adjusted? It takes very little at lower speeds for power mode to take
over and it seems that power mode is often not really required. Sometimes
the pedal is really not pressed further at all, yet 'power mode' is
active.
At higher speeds (>40m/h) the car behaves normally.>>

<<If it can be caused by a defective part, what direction should I look
into?>>

<<Thanks for the help.
Remco>>

Am I missing something here? If the trans is in "Power" mode, isn't it
*supposed* to hold gears longer and change down more readily?. What
happens if you set the switch to "Normal"? Does it still do that, or show
more restraint? If that changes things, then maybe you should only put it
in "Power" when you want to press on or play and leave it in "Normal" the
rest of the time. Cheers

 

[Danny Russell]

Hi all

When I step on the accelerator, one would expect an automatic
transmission to downshift and impreza has this 'power mode', which
makes kinda more fun to drive.

My question is: can the sensitivity to how much the pedal is jammed
down be adjusted? It takes very little at lower speeds for power mode
to take over and it seems that power mode is often not really
required. Sometimes the pedal is really not pressed further at all,
yet 'power mode' is active.
At higher speeds (>40m/h) the car behaves normally.

If it can be caused by a defective part, what direction should I look
into?

I used a pair of resistors in a voltqage-divider configuration to reduce the analog throttle
position sensor (TPS) voltage signal being sent to the TCU. This reduced the "downshift-happiness"
of my '02 OBS.

It works by making the TCU think that the pedal has not moved as far or as fast as it actually has.

This will however lower the full-power upshift points, and if you reduce it too far, will also lower
the line pressure which can cause the lockup clutch to slip under power (NOT good). I experimented
for a week to arrive at a result I was happy with. -Danny

 

[remcow]

Hey Danny
That sounds like a good idea worth trying. Mine just seems ultra sensitive.
Do you remember what value resister one can use? If the current isn't high,
I wonder if a pot can be used..
Thanks again for that suggestion.
Remco

 

[Danny Russell]

Hey Danny
That sounds like a good idea worth trying. Mine just seems ultra sensitive.
Do you remember what value resister one can use? If the current isn't high,
I wonder if a pot can be used..
Thanks again for that suggestion.

It's been a couple years since I did it, so I had to go out and look at the values. Basically I
ended up 90% going to the TCU from what it originally was. So if the original full throttle signal
was 4.5VDC, now it is: 4.5 x .9 = 4.05. The resistor values I used were 114k for the series (top)
leg of the divider, and 1M for the shunt (bottom) leg. I created the top leg value by paralleling a
470k & 150k.

I used a 1M linear taper pot and a digitag voltmeter during the original experimentation with
values. Once I got what I wanted, I swapped-in fixed resistors for reliability.

Small changes will give you large results, so the pot will be very sensitive to small adjustments.
I made sure to tap into the wires under the steering column (not sure on your model year), not at
the TPS under the hood. The reason is that I wanted to affect the signal going to ONLY the TCU, not
both the TCU & ECU. You'll need to consult a factory wiring diagram to locate the physical
connector in the harness which feeds input to the TCU, as well as the specific pins on that
connector that concern you.

Be very careful not to reduce the voltage too much, so as not to reduce line pressure at high
throttle openings (firm shift) to that of low throttle openings (soft shift), or the lock up clutch
will slip as a result and you will quickly burn it up (they're not very robust). Ironically, I
performed this mod to help save the lockup clutch by preventing it from constantly dissengaging and
reengaging. -Danny

 

[remcow]

Great Danny - thanks! I really appreciate you looking into it.

I have the schematic, so will see what needs to be done where.

Since these are huge resistances (i.e little current), I could use two
resistors and a multiturn pot, scaled according to the numbers you just gave
me.. The pot's swing will just be very small and will allow me with some
tweaking if that is what I need.

Thanks again for that idea!
Remco

It's been a couple years since I did it, so I had to go out and look at the values. Basically I
ended up 90% going to the TCU from what it originally was. So if the original full throttle signal
was 4.5VDC, now it is: 4.5 x .9 = 4.05. The resistor values I used were 114k for the series (top)
leg of the divider, and 1M for the shunt (bottom) leg. I created the top leg value by paralleling a
470k & 150k.

I used a 1M linear taper pot and a digitag voltmeter during the original experimentation with
values. Once I got what I wanted, I swapped-in fixed resistors for reliability.

Small changes will give you large results, so the pot will be very

sensitive to small adjustments.

 

[Danny Russell]

Great Danny - thanks! I really appreciate you looking into it.

I have the schematic, so will see what needs to be done where.

Since these are huge resistances (i.e little current), I could use two
resistors and a multiturn pot, scaled according to the numbers you just gave
me.. The pot's swing will just be very small and will allow me with some
tweaking if that is what I need.

I wanted to keep the values high enough to avoid causing any current loading of the TPS signal which
is derived passively off of the 5k wirewound TPS pot. Conversely, putting too much more than 100k
in the series leg would risk adding noise, stray capacitance, etc. I chose comfy pocket in the
middle.

I considered getting really nutty and fitting a crude, RC time-contant feature that would allow me
to tune the rise-time of the voltage as a result of suddenly depressing the throttle, but then I
came to my senses (well almost).

I probably don't need to remind you to make sure that you bring the ground side of your divider back
to the SENSOR GROUND on the same connector. It might be called something else like TCU SIGNAL
GROUND or whatever.

The multiturn pot inside of range-limiting resistors is a great idea.