MR2 Owners Club Forum banner

61 - 80 of 84 Posts

·
Premium Member
Joined
·
12,780 Posts
I made a mistake when i said 86. Frank's range in the data given is 32.15 to 71.76 normalize it to start at zero and it's 0-39.61. Your range is 15.69 to 76.08 which means your range normalizes to 0-60.39 which is way wider than frank's. but interestingly Frank's throttle blade never actually goes to 86 so i suspect there's something actually limiting physical travel of the accelerator pedal in his car.
I don't think you were mistaken, Frank's Pedal E value ranges from ~32-86, so I think that's where the 86 came from. but his D value (which is what I have been using) ranges ~16-70. So both have a range of 54 on his car.

My Pedal D range is ~16-76, or a range of 60. Not a ton more, but it is some.

We both have a blade range of ~16-81, range of 65.
 

·
Registered
Joined
·
1,341 Posts
Here's some info from looking more closely at the pedal.

Previously I had made sure that the floormat was not interfering with the pedal. It was not.

However there is a layer of carpeting on the floor. I moved it out of the way. Underneath it is a rubber stop mounted onto the floor.

Removing the carpet layer increases the Pedal E travel to 88.6 (from 86 with carpeting).

However the Throttle Manifold is pegged at 81.2 when Pedal E reaches 77.3 and does not go beyond that.

Then I removed the rubber stop from the floor. Pedal E reached 91.4 with the pedal on the metal floorpan surface.

I am willing to bet any sum of money that the rubber stop is not original and it was added later possibly as part of the pedal recall work. It does not look original - we all know what original Toyota stuff looks like. The rubber piece is a slightly irregular square apparently cut from a rubber sheet. It is 1.75 inch on a side, and 1/4 inch thick. It was held on to the metal floor with rubber adhesive. There was a piece of Gorilla tape securing its bottom edge. Ha ha.

So next, if there is still a desire for this, I will take a look in the throttle body and report.
 

·
Premium Member
Joined
·
12,780 Posts
Very interesting. So between removing the carpet and the rubber pad, you gained a raw travel of ~5.4, which is almost exactly the difference in raw travel seen between your "pedal D" range and mine. So there is something to Marc's theory that the ECU is calibrated for a specific range of pedal travel that is seen in the stock application and in the swap we are removing that limit.

This could be compounded by the fact that I'm using a Camry pedal on a Rav4 ECU... could be that the Rav4 has even more limit on pedal travel in the factory application, and the ECU is tuned accordingly.
 

·
Premium Member
Joined
·
3,544 Posts
As far as I know it looks like all the pedals have the same electrical signal characteristic but yeah i had not considered that the full mechanical travel would not be used. but honestly it makes sense when i think about it. You don't want the pedal to run into its mechanical stop under any condition, you would have to build it stronger. Instead you can just let the pedal run into the firewall and the carpet and the pedal is easier/cheaper to build.


I would not be surprised if the effective travel of the highlander tune i use as a base to the 5-plug is different than the rav4 base tune i also use.
 

·
Registered
Joined
·
1,341 Posts
I have a hypothesis, in three parts:

Lemma 1. The pedal/throttle relationship is a simple transfer function, no matrix transformations.

Lemma 2. The modulo 17 tables in the binary that we attribute to the throttle behavior are in fact related to the torque convertor line pressure/lockup.

Conjecture: If this is correct, the three modulo 17 tables for torque request will bot not be found in the Camry ECU binary, because they are handled by the TCU binary.

I haven't looked yet. Just thinking out loud.
 

·
Premium Member
Joined
·
3,544 Posts
I'm wondering if some of the throttle tables are actually just blade angle minimums of some kind. I know the 2GR loves to "hang" the RPMs at certain spots when you hook up a VSS signal.

I do agree that it looks like the pedal to throttle is just a simple function.
 

·
Registered
Joined
·
1,341 Posts
I don't think you were mistaken, Frank's Pedal E value ranges from ~32-86, so I think that's where the 86 came from. but his D value (which is what I have been using) ranges ~16-70. So both have a range of 54 on his car.

My Pedal D range is ~16-76, or a range of 60. Not a ton more, but it is some.

We both have a blade range of ~16-81, range of 65.
So after thinking about this some more and looking some more at the binary there are definitely some sensor calibration 1-dimensional maps (I believe they are 22x1) that convert the pedal voltage to some "value" which then goes into the 2-dimensional throttle control/torque control maps that we don't want to mess with. I believe that editing these maps would have a similar effect as fiddling with the pedal signal electrically and give better possibility for fine-tuning the pedal response. I did not have an opportunity to test this but it seems like something worth looking into for anyone who's interested in the pedal response.
 

·
Premium Member
Joined
·
3,544 Posts
Frank, if you're talking about the maps that bitedit marks that look like this:


I'm not sure what those maps are, but they aren't the "accelerator pedal sensor" maps that bitedit claims they are.
 

·
Registered
Joined
·
1,341 Posts
Frank, if you're talking about the maps that bitedit marks that look like this:


I'm not sure what those maps are, but they aren't the "accelerator pedal sensor" maps that bitedit claims they are.
I did reference a BitEdit screengrab that showed two 22x1 pedal maps listed among other sensor maps.

I've made a near-complete inventory of high-resolution 1-d maps/sensor calibrations and I'll write something about that soon.

Meanwhile I got in my possession a 2008 RAV4 with U151F and 89661-42G00 - if there is any interest in the throttle behavior of this vehicle let me know and I'll put it through its paces.

Now all I need is a Highlander 2GR to complete the trifecta of Rav4, Camry, Highlander. I'm gonna pass on the minivan.
 

·
Premium Member
Joined
·
12,780 Posts
I picked up a Denso pedal off of ebay to do some testing with. Output seems to be the same (within a couple percent) of the CTS pedal, but it does have a stiffer spring in it (even compared to my CTS pedal with the shim removed, although the shim didn't do much on my particular pedal). Measuring pedal force by holding the base of the pedal in my hand and pressing the pedal down on a scale, the CTS pedal took 138oz to fully compress and the Denso took about 160oz. So, about 15% more. It might make the Denso pedal feel just a touch less sensitive. But electronic correction would still be required to get the full range of pedal travel to equal the full range of throttle travel.

Additionally, the Denso pedal has about 1/4 to 1/2" more physical travel than the CTS pedal, so assuming you can make it use the entire output range, that will also make it easier to modulate, and even without electronic adjustment it might be less sensitive to motion.
 

·
Premium Member
Joined
·
12,780 Posts
I feel like the Denso pedal is a lot nicer to drive than the CTS. It's a subtle, hard to explain difference. It feels more responsive at the low end, without feeling twitchy. The CTS pedal always had a bit of a dead spot at the very top of the pedal travel (right around 0 throttle) that was noticeable when pulling away slowly. And to be honest, my throttle controller made this slightly worse, but the overall effect of making use of the full pedal travel was worth it. But the Denso pedal seems to eliminate that dead zone at zero pedal, and just overall feels a little smoother.
 

·
Premium Member
Joined
·
3,544 Posts
Interesting, both pedals should be built to the same set of requirements. there must be some pretty wide tolerances in the requirements for both pedals to have a different feel.
 

·
Registered
Joined
·
1,341 Posts
It seems that all the throttle stuff is always 16bit and wide open is represented by the value 11010 (0x022B) and surprisingly that value only shows up in 26 different "clumps" so now that i have that new information it may be easier to find the related maps. but the search also pulls up stuff like this:


i have no doubt that that is DBW related but that looks like two tables that are on top of each other. it would be easy to say that's just a table with 3 axis but i know what those look like in this ECU and that's not what that looks like at all. but if we plot it out it looks like a "resonable" throttle transfer curve. little lumpy but not ridiculous:


Also, the lumpiness may disappear with the axis values. since this isn't a normal table i have no idea what the column header values are. this graph assumes they are equidistant.
I've been looking at sensor calibration curves and this grouping above is a set that I've identified as a collection of eighteen 16x1 1-d calibration curves. This is confirmed because each of the 18 curves is preceded by the dimension declaration 0x0010 which is usually the declaration for a 16x1 map. The first image is how it all looks in WinOLS.

Next by looking at the curves it's evident that there are only three independent calibrations.

Curves 1-5 and 12-18 are one calibration.
Curve 6 is another calibration.
Curves 7-11 are the third calibration.

Images 2, 3, and 4 show the numbers laid out in WinOLS for the curves corresponding to each calibration.

Finally I've plotted the three calibrations and this is shown in image 5. The three calibrations follow the same trend but with some differences in the shape.

What does this mean? Maybe this is giving some aspect of the throttle response under three different sets of driving conditions, but I don't know what.

PS> Um yah it occurs to me this could be some kind of hysteresis behavior.
 

Attachments

·
Registered
Joined
·
1,341 Posts
Sure. Sharing the whole excel file is probably the easiest.
I've overlayed the uncorrected throttle data on the characteristic curves. Normalized so 100% throttle corresponds to 11010 (the maximum value in the throttle axis), and 76% pedal corresponds to 7411 (the maximum value in the pedal axis). Innaresting. The throttle-to-pedal curve is not one of the three characteristic curves but it does suggest a hybrid of two of the curves, retaining characteristic features of both.

What I would be tempted to do next is re-scale the pedal axis in the eighteen curves in the binary by a factor of let's say 1.15 so the maximum value in the pedal axis is 1.15 x 7411 = 8523 and see what happens to the throttle-to-pedal relation.
 

Attachments

·
Premium Member
Joined
·
3,544 Posts
Frank, you should give it a go and test it then report back.

The ECU has a bunch of modes it's possible one is an econ curve and the other is a power curve and it transitions based on some other table factor.
 

·
Registered
Joined
·
1,341 Posts
I've been distracted and occupied with other things so I haven't put much time into this however there is another interesting piece to this puzzle.

First of all I think that all of the testing has established that whatever the torque request (17x17) and throttle request (21x17) maps do, it is not directly related to the pedal and throttle action. From what we've seen and measured in driving tests the pedal to throttle position is a simple 1:1 mapping.

We have candidate maps for the pedal-to-throttle mapping in the collection of 18 16x1 mappings discussed above. These resemble the actual pedal-to-throttle measurements but do not match them exactly. Mysteriously, they fall into only three distinct sets of identical mappings.

The new piece of the puzzle is this picture from a failed tuning suite that has been withdrawn from release. It's a listing of maps available in an unspecified calibration possibly 2gr-fse. This picture was presented previously about one year ago and I didn't appreciate its significance until I revisited its thread again today:
https://www.mr2oc.com/6785490-post729.html

So this picture presents a set of 13 maps labeled as "throttle mappings" of which only three are labeled: "normal," "normal 2 eco," and "snow," and the others are unlabeled. How these guys came up with these labels, I don't know. But regardless of the labels I can say this presents some parallels to what we see in our 2gr-fe calibration.

Without any more details about these mappings we can't draw any definitive conclusions other than a independent (if somewhat weak) confirmation that we are on the right track with identifying the throttle mappings. What this means is that we are one small step closer to being able to simply change the measurable pedal-to-throttle response without relying on any external signal processing simply by re-flashing the ECU. One small step for a map, one giant leap for the 2GR, in the spirit of July 20, 1969. Ha ha. Ha ha ha ha.
 

Attachments

61 - 80 of 84 Posts
Top