I think it's comming from the tranny, not the engine.
The lower gear slows the car down to the point where the transmission disengages (like when you stop at a light), then the speed increases and the tranny re-engages starting the cycle all over again.
I noticed this on our GSMNP trip while going down some moderate hills.
With the very steep hills, where it doesn't slow down, it holds the RPMs and stays in gear.

 

Hmmm, I was wondering about that. As you note, the rpm stays relatively high, seemingly tracking speed as it would if it stayed engaged, but perhaps it is disengaging. If that is the case, I have some ideas on a fix, since the trans is electronically controlled.

If it is the idle up function, the only two other possibilities I can think of are the power steering and the alternator. It certainly does feel as if the computer is applying the throttle.

 

Yes, it's very smooth, so you wouldn't think it, but I'm sure that was where it came from.
You can feel a little more drag on the system when the RPMs come up, indicating gear engagement.

 

When mine does it, the rpm is around 2000, so the trans shouldn't be disengaging like it would when coming to a stop. But it might be disengaging for some other reason.

 

Perhaps completely unrelated (and my apologies if so) but my last car also had idle-up on the power steering. I could rev the engine just by turning the steering wheel back and forth a little while stopped at traffic lights.

 

On many, if not most, vehicles with A/C, even when the A/C is turned to off, the A/C compressor periodically cycles on and off. I can't remember why, but it has something to do with keeping the A/C in good operating order. I have a Ford Ranger pickup and the A/C periodically cycles on (not for long) and off even in the dead of winter when its real cold (and I live where it gets real cold). This will cause an idle up, and then an idle down. Our Forester is my wife's daily driver. I hardly ever drive it and can't remember if the A/C periodically cycles on and off even when turned to the off position, but I suspect that it does.

 

jmacmaster, good point. However, this happens every 30 seconds or so during a hill descent. Hopefully the AC doesn't cycle that often when turned off!

The power steering isnt' a likely culprit since it happens even when descending in a straight line. So based on inputs so far, I'm thinking it is either due to the alternator, or the disengaging of the trans. Any other ideas?

 

Engine fan?!?

It had an effect on the idle speed of my old Alfa GTV (possibly the generator?).

 

Engine fan?!?

It had an effect on the idle speed of my old Alfa GTV (possibly the generator?).

MountainBiker, how subtle is the idle up? How fast are you going when it happens? (And thus how much radiator cooling would you be getting when it happens?)

 

MountainBiker, how subtle is the idle up? How fast are you going when it happens? (And thus how much radiator cooling would you be getting when it happens?)

It isn't very subtle! Imagine plonking along at 2000-2500 rpm in 1st gear, having to add a little bit of brake now and then to control the speed, maybe around 15mph or a little slower. Then when it happens, it feels like somebody else just pushed the accelerator!

The speed is slow enough that the fan could be kicking on, which would be upping the load on the alternator, which may be causing the "idle up" function to kick in. Hmmm, the interval seems very similar to how often the fan kicks on and off when just parked and idling! I know this situation doesn't always show up on every steep hill descent, and perhaps the times it does show up, the ambient temperature is higher. I think the fan is the culprit (via the alternator)!

I'll take another look at the factory service manual, but I don't recall it saying anything about the alternator driving the idle up function. The difficulty I see on the Subaru is that all the inputs go to the ECM (including a signal stating whether the AC is on or off), and then a signal gets sent from the ECM to the idle control valve. On my Toyota, the idle up device is separate from the idle control valve, so if this problem occured on the Toyota (which it hasn't), I could just disable the idle up device. On the Subie, I can't disconnect the idle control valve, because it controls the complete idle function.

 

The FSM states that two conditions trigger the "idle up" function, AC compressor on, and high electrical load. So it is very probable that the fan cycling on will correspond with my dreaded acceleration on a downhill. I'll have to do some testing, perhaps with an OBD Scantool in place so I can see what is happening.

 

I use a Taxtrix cable and ECUexplorer to log my ECU.
ECUExplorer (from Tari Racing) is not the best program around, but it speak the SSM protocol, so you get Subaru-specific variables and sensor readouts.

This what I can read:
Engine Speed (RPM),Air/Fuel Correction #1 (%),Coolant Temperature (°C),Front O2 Sensor #1 (V),Ignition Timing (°BTDC),Vehicle Speed (KPH),Atmospheric Pressure (PSI),Knock Correction (°BTDC),Fuel Injector #1 Pulse Width (ms),Throttle Sensor Voltage (V),Air Flow Sensor Voltage (V),Battery Voltage (V),Manifold Relative Pressure (PSI),Primary Wastegate Duty Cycle (%),Idle Speed Control Valve Step (steps)

In addition you can see some switches and fans (boolean variales).

Somehow I think a variable fan controller might be the right remedy for you, as it will give a more stable electrical load.

 

IMO "Idle up" should mean just that. Only when the engine is idling, it will increase the idle speed to provide enough power for the heavy drain ancilliary (A/C, fans etc). But if the engine is NOT idling but is doing 2000-2500rpm then it shouldn't idle-up (or else the term "idle up" is being mis-used).

 

IMO "Idle up" should mean just that. Only when the engine is idling, it will increase the idle speed to provide enough power for the heavy drain ancilliary (A/C, fans etc). But if the engine is NOT idling but is doing 2000-2500rpm then it shouldn't idle-up (or else the term "idle up" is being mis-used).

I'm using Subaru's term, so take it up with them if you think it's mis-used!:spinnasmiley

I agree with you in principle, and I wish Subaru engineers would've been nice enough to disable the function when the engine is running over 1000rpm.

 

Just in case anyone cares, I confirmed it. The ECM actually goes into "idle-up" mode when the alternator has a high load, or the AC compressor kicks on. This is understandable when at idle, but it even does it when at 3000rpm!

How did I confirm it? With an OBDII code reader. When you are decelerating, and the accelerator is off, the fuel injectors shut off (or nearly so), called "fuel cut mode". In this situation a code reader reports "open loop" instead of "closed loop". So I went on a long downhill, at various RPM with the accelerator off. Starting the AC or getting a large alternator load (the radiator fan was the only thing I found that triggered it), switched me to "closed loop", and I felt a lessening of backpressure. The code reader I have also reports instantaneous MPG, and it also changed, showing more fuel being used.

So now I have yet another wiring project to put on the list: cut off the "AC Sense" input to the ECM, preferably at all times, except when under 1000 rpm, but I'm not sure what that circuit would look like, and if it is worth it. I'll have to do some more research to see how to get it to ignore the alternator load above 1000 rpm.

 

hm, I been thinking about a 'eco-meter' and how to do this with the variables I can get from the ECU.
One way to do this is to use the injector Pulse With Modulation (PWM) readouts. Quick and dirty
So, is this the MPG readouts you see?

The problem is: when injector supply voltage go down, the pulse with must be increased to flow the same amount of fuel thru the injector(s). See http://www.megasquirt.info/ and http://www.megamanual.com/v22manual/minj.htm for details.

The ECU only knows on/off status for AC, radiator fan and heated windows, and calculates new PWM from this.

So if you switch off ACsense signal, and AC is on, the ECU will not adjust PWM and the engine will run leaner ..

..Its not easy to write about this in English, but here goes

 

From what I know...

It is probally from the AC compressor, or when your radiator rans kick in

 

After a reread of this thread:
Sheepish had a car that increased rpms when he used the steering wheel, and MB when radiator fans is on..

Is your battery in good shape?
Bad groundings can also cause some strange things..

 

After a reread of this thread:
Sheepish had a car that increased rpms when he used the steering wheel, ...

Yeah, but that was while stationary, and I took it to be just normal idle-up but working a bit enthusiastically. I really don't know if it still did it while I was moving.

 

Yeah, but that was while stationary, .. .. a bit enthusiastically.

I think you nail it with this ;-)
Its a workaround from the factory. The ECU have no real clue about injector power supply, so why not be enthusiastic about the whole thing?
Its not a bug, its a feature

 

The Idle-Up feature in my Subaru is fine in all conditions except one: when using the engine to control downhill speed. All other times it works as it should, and like other cars I've owned, and it is a feature.

The MPG figures are from an OBDII scangauge.

For disabling the Idle-Up when it isn't wanted, I'm not worried about making the engine lean or rich, because when this is going on, only ambient air should be running through the engine during fuel-cut mode, which hasn't blown up an engine yet, in all the years since Electronic Fuel Injection was introduced! The idle-up adds a little bit of fuel to increase idle (not to control a lean or rich condition), even though I'm not at ilde, so obviously, this is not needed.

 

MB, what you say make perfect sense

I was just skimming thru a -03 wrx ECU map definition file (xml) over at openecu-org. This behavior might be adjustable within the ECU. However, finding the right place to adjust inside the ECU is not so easy.. ..if its not hardcoded to Open/Closed loop switching...
(Very few Forester definition files and ECU dumps are availble on openecu.org.. ..mostly wrx stuff..)
..and my car is not yet supported so I never have used the flash or "tuning" tools..

I still think a variable speed radiator fan controller might be a solution.
Wire it in paralell to the orginal fan relay. That way you can keep the orginal system functional as backup. Adjust the new fan controller so the ECU never see a reason to activate fan relay.

 

Hmmm, that might work. The problem is, I don't see anything that shows a sensor connection to the ECM from the fan. However, the ECM appears to control the fan, based on temperature. So even if I unplug the fans, I may still get Idle-Up when the ECM tries to tell the fan to start. I hope that is not the case, because that will be much harder to avoid. That would make me abandon the project! So I'll pull out the fan fuse and see if that stops the rad-fan induced Idle-UP.

 

if i'm not mistaken, i remember reading that the ac (compressor) turns on whenever the selector is switched to defrost or directed to the floor even if the ac is not switched on.

 

if i'm not mistaken, i remember reading that the ac (compressor) turns on whenever the selector is switched to defrost or directed to the floor even if the ac is not switched on.

That is a very good point, because on some vehicles this is true, like my '02 Tacoma. But it is not the case on my '02 Forester.