legacycentral bbs

Somebody did mention to me in another thread on the BBS about switching some TPS wires, now that I recall but I didn't know what they were talking about. They didn't mention the voltage deal.

Yup, there are two models, both opposite operating.. if you can log, you can see whether the "load" goes the right way..

The load on evoscan goes up in numerical value while revving the engine as the voltage of the TPS goes down from 4.8v to 1.7v.

Do you know if I can switch the wires on the TPS to reverse the voltage so the ECU will read it properly? If this is correct, I can't believe I've driven 20,000 miles with it reading backwards. Just how much does the TPS really affect the way the car runs? Seemed to run pretty damn good this way aside from random stumbles once in a while. It's strange that it very very slowly became more frequent until it recently happens all the time and the CEL is on almost constantly.

Kimo,

First off, you can not move any wires around to change the behaivor of the TPS, i.e. chaning from a raising vs. lowering voltage. The TPS is just a variable resistor. The actual voltage comes from the ECU, goes through the TPS and the change in resistance causes a change in the reference voltage, which the ECU then reads and determines the TPS's position.

The car would be pretty much undrivable if you were using the wrong TPS. As for how important the TPS is, it is rather important and provides additional engine load info to the ECU.

Again...if you haven't checked the idle switch, do that.

You could swap the wires, and the TPI would then read the other way around, but what you need to know first is whether your ecu is expecting to see a rising or a falling voltage. The easiest way, if you aren't sure, is to drive the car and see if the load value goes up as you put your foot down on the throttle....

And, the reason that the car has still been running ok...
If the TPI has a broken wire, the ecu ignores it and uses air flow and revs to calculate injection etc. The main reason they(subarus) have the TPI is so they can calculaate the inertia of the air flow and the resulting lag before the power starts to develop.. the ecu isn't meant to inject as soon or quite as much as the air flow meter would suggest when the car is first accelerating... this is all because subarus have such long manifolds.. now a lot of manufacturers are following this practice.

So.. IF the ecu was getting the wrong voltages, it MAY have ignored the TPi and calculated using revs and air flow, which will cause stumbling...

The throtle switch can cause stumbling as Josh says, but normaly it is just as the switch opens and it's right near an idle anyway... unless it's been moved quite a bit... normal wear though, the stumble is just as you come off idle...

It would stumble only on 50mph cruising very light throttle and sometimes when taking off in 1st it would fall on its face.

Yup, wel, "very light throttle"... the switch is easy to check...

Lean mixture will give you similar symptoms... as will bad plug gap and a heap of other things..

If you can log your car, and have the manual, you should be able to deduce which parameter is not reading correctly... or, more importantly and more telling most times...is not behaving in the right manner.. which is more important than the specific figure, which do vary a little.

IIRC, the early '89-'94 Legacy and the '92-'96 Impreza EJ20G ECUs ("4-plug ECUs") did have the throttle sensor voltage reversed from later ECUs. The service manual will tell you pretty quick what the proper voltages are.

Kimo, refresh my memory...Your running a 4-plug ECU with an EJ22T throttlebody and TPS mounted to your EJ20K (late EJ20G) intake manifold with an adapter?

Actually... forget that idea about the load being an indicator of the TPI position.. it isn't.

The load is calculated from air and revs.. me bad..

All the rest is true but !!

PhyrraM wrote:IIRC, the early '89-'94 Legacy and the '92-'96 Impreza EJ20G ECUs ("4-plug ECUs") did have the throttle sensor voltage reversed from later ECUs. The service manual will tell you pretty quick what the proper voltages are.

Kimo, refresh my memory...Your running a 4-plug ECU with an EJ22T throttlebody and TPS mounted to your EJ20K (late EJ20G) intake manifold with an adapter?

I do not know which ecu it is. Rob might. Its ej20g and its converted to run wasted sark right now. My fsm for the ej22t has the specs which match this brand new ej22t(e) tps but I need to know which ecu it is and whether closed throttle is 4.8v or 0.6-1.7v for this specific ecu. Its an ej22 tps mated to an upside down ej22 throttle body bolted to a late model 97 intake manifold wired up however you wired it for me, jess. Running an unknown ej20g or k ecu. If by adapter you mean custom 1/4" TB spacer, then yes. Or are you referring to the ignition converter?

Unless there's a wiring issue I highly down its the idle switch. This is a brand new sensor and I've already had two other used ones and the behavior is the same albeit much worse with the new one and the last used one. The fact that the ecu seems to be capable of band-aiding the situation makes it much harder to diagnose.

"The fact that the ecu seems to be capable of band-aiding the situation makes it much harder to diagnose.".

It's part adaptive learning and part bandaiding... I hadn't realised you'd modified it..

So to test if the TPI is being read, and correctly, try driving along at a pretty much steady speed, and see if the IPW increases when you move the tPI....

Injector pulse width? By tpi you mean the tps, right?

Yes throttle position sensor and injector pulse width... you might log injected quantity per second..

If you drive along with the speed staying the same, even using the brakes if you need to, the air flow will stay the same and/but the fuel will change in time with the TPI..

If it's logged, to a spreadsheet, or graph I guess, you will see the small change easily... I always use spreadsheets... I think they are more accurate.

I will get a spreadsheet (csv file which can be opened with excel or evoscan) uploaded or some screenshots or both. Might be a day or two before I get the chance. Then you can take a look at it and tell me what you think. I've emailed rob to get his opinion and see if he remembers which ecu it is. I will do the idle switch diagnostic procedure too. Pretty sure I already did but ill check again.

There's a manual called "diagnostics", that I think will be what you need(josh might be able to confirm), and some other bits with diagnosis in their names you might like to read.... I'm in New Zealand and we mainly get the japanese market cars..

https://skydrive.live.com/?cid=4ca3c345 ... 7F7F%21114

Maybe you could sell me one of these mysterious jdm gc tps?

The resistance value is the same in them both, and the earth is at the opposite end to the power, with the reading in the middle on the slider. As long as you have the power going in at the end it shows diagrammatically(Pff, had to use a spellchecker on that one) in the manual, you should be right rewiring it.

Swapping two of the wires WILL reverse the reading then? Lol why does this car always have to be so challenging? I guess to force me to learn and to give me satisfaction when I overcome its many obstacles.

Yes, the readings will go the opposite way then. There might be another problem to overcome though haha... I think from memory that they are on opposite sides of the throttle body, so when they get swapped you might have to re-time them as it were... or swap the shaft over in the butterfly, or flip something... but hey, you've come so far...

And yes to the satisfaction... that is missing in so many people's lives... we need challenges.. to overcome...

Oh jesus christ. The throttle body linkage and sensor are on the opposite sides of the ej20g ones so I had to mount my ej22t TB upside down so it was oriented properly.

I don't know whether you've seen it but this is our local club...

http://www.clubsub.org.nz/forum/index.php

The tutu boys in there are always swapping and flipping and what not. Me, I'm a mechanic and try to keep right out of all that... it's a nightmare..

"tutu" by the way is a local native expression for "play around"...

Cool, never seen that site before. If I could have kept this any simpler I most definitely would have. I just wish I could run obd2 with an sti swap and be done with all this obd1 nonsense.