The Top DSM Community on the Web

For 1990-1999 Mitsubishi Eclipse, Eagle Talon, Plymouth Laser, and Galant VR-4 Owners. Log in to remove most ads.

Please Support RTM Racing
Please Support ExtremePSI

[RESOLVED] Can't do pulls without downshifting.

This site may earn a commission from merchant
affiliate links, including eBay, Amazon, and others.

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

clean96GST

10+ Year Contributor
462
0
Jul 14, 2008
Arlington, Washington
So I bought the car about 6 months ago, mechanics checked it out and said it was in good shape but I can never go full throttle without the dam thing downshifting. Is there anything I can do? Also when I get up past like 5000 RPM it sounds way too loud.... not like exhaust it just doesn't sound good on my engine like it's going to blow up LOL.
 
With your automatic, it will always kick down a gear when acclerating, your best bet is to put it in second and go from there.

To your engine noise, mine engine gets a bit loud when up in the rpms, yes louder than the exhaust, you will just have to know what it sounds like when it good and what doesnt sound good.
 
If you go 30mph or a little more, it shouldn't downshift hard if you're trying to get a clean run. The hardest kickdown is around 25mph.

If you're not trying to get a clean run in, why dont you want it to downshift?
 
If you go 30mph or a little more, it shouldn't downshift hard if you're trying to get a clean run. The hardest kickdown is around 25mph.

If you're not trying to get a clean run in, why dont you want it to downshift?
Because when I'm one the freeway doing 70 and put it on the floor it drops into second...
 
Understandable, buy a shift box or see if you have a tcu that's eprom. Mabye you could (or have someone) burn a chip for you that changes that.
 
2nd gear pulls on the freeway ftw!
 
This isn't a problem, downshifting is a good thing since you'll move faster.

Pay attention to your WOT shift point from 2nd to 3rd, then in the future you'll know why it's downshifting and it won't be so annoying since it's helping you accelerate faster. :boring:
 
The DSM Auto uses fuzzy logic to try to figure out what the driver wants to do. If you have been driving hard the transmission will shift down quicker and hold the lower gear longer after lifting the throttle.

When you nail the throttle to WOT the TCU thinks that you want to accelerate and accelerate fast. If the next lower gear will be less than 7000 RPM the tranny will shift down. This is a good thing. If you want to stay in the same gear and a lower acceleration rate is ok then go down on the throttle gently and the car will accelerate gently in the same gear.

After 300,000 miles in two AWD Talons the TCU can now read my mind and is always in the right gear.
 
Ok thanks for the help guys, next time I have a chance I will test out the WOT shift point, although it scares me because it sounds like it will be my last pull.
 
Your 96 GST should have about the same shift points as my 98 TSI.

Up shifts are always at 7000 RPM at WOT.
First to second is 40 MPH.
Second to third is 70 MPH.
Third to fourth is 110 MPH.
 
The DSM Auto uses fuzzy logic to try to figure out what the driver wants to do. If you have been driving hard the transmission will shift down quicker and hold the lower gear longer after lifting the throttle.

When you nail the throttle to WOT the TCU thinks that you want to accelerate and accelerate fast. If the next lower gear will be less than 7000 RPM the tranny will shift down. This is a good thing. If you want to stay in the same gear and a lower acceleration rate is ok then go down on the throttle gently and the car will accelerate gently in the same gear.

After 300,000 miles in two AWD Talons the TCU can now read my mind and is always in the right gear.

That is absolutely NOT how the TCU works.

There is a table that dictates where to shift up or down based on current gear, throttle, and shaft speed (which can be loosely correlated to RPM).

There is zero fuzzy logic or learning. There are trims for the line pressure but they hardly make a difference.
 
Last edited by a moderator:
From 1998 Mitsubishi Eclipse Press Release VIS www.autopricing.com :

Or search this site for "fuzzy logic", remembering the OP has a 2G.

Because autopricing.com knows everything and experienced users know nothing, right? You really trust some random internet article written by a desk jockey with no automotive knowledge over an experienced DSM brain?

That is just a fancy way of saying it has a computer controlled trans and EFI. It is "adaptive" but not in the way you described, at all. The adaptives are just there to adjust line pressure for different temps basicly.

But what do I know, I only program them :rolleyes:.
 
Last edited by a moderator:
You did notice that the reference was from a Mitsubishi press release right? If you were programming TCU's for Mitsubishi in the early 1990s you should set your old employer straight.

After 180,000 miles in the 2G Talon I can report that it responds just like the "desk Jockeys" say is should. When I drive like a blue haired lady saving gas the car expects to save gas. When I drive like a DSM owner should it holds the lower gears on throttle lift and down shifts at any hint of my wanting to go faster.

Tomorrow when I have a little more ambition I will search the USPTO data base for Mitsubishi patents for adaptive transmission control. For now here are a few of the thousands of references to the Mitsubishi Fuzzy Shift control.

Other references from this site:
http://www.dsmtuners.com/forums/automatic-tranny/149611-tranny-rebuild-update-john.html
http://www.dsmtuners.com/forums/automatic-tranny/148944-big-turbo-t-bad-streets.html
http://www.dsmtuners.com/forums/newbie-forum/55116-just-me-do-i-have-problem.html

For the Mitsubishi Galant:
Full Test: 1999 Mitsubishi Galant ES
1997 Mitsubishi Mirage Review - 1997 Mitsubishi Mirage Road Test & Research
"Mitsubishi's Adaptive Transmission Control Management (ACTM) system. The ACTM's computer control actually catalogues and learns driver habits and adjusts the transmission's shift points to obtain optimal performance."

A history of adaptive transmissions where Mitsubishi is referred to as an early adapter of the technology.
The Truth About Adaptive Transmissions | The Truth About Cars

The 1995 Mitsubishi version of fuzzy logic transmission control:
1995 Mitsubishi Galant Road Test Performance Review, Engine, Transmission & Handling at InternetAutoGuide.com

Describes the Mitsubishi adaptive transmission control first introduced in 1992. My Talon drives as described here.
NationMaster - Encyclopedia: INVECS

Mitsubishi specs for 1997 Eclipse with "Fuzzy Shift"
1997 Mitsubishi Eclipse, Based grade Modification - all pics, specs, parts and prices
 
Last edited by a moderator:
In the "is it just me or do I have a problem, he's saying that he doesn't know when the "fuzzy logic" kicks in. Read post #3, the car shifts differently everytime. If he goes over that hill every day, wouldn't the "fuzzy logic" have that down by now? I read some of the other ones, and it looks like everyone who has posted here in the forums about it was someone who doesn't know exactly what it does.

Doesn't it make more sense that it has pre-progammed shift points based on things. I can tell when my car is going to shift because I know it shifts at a certian RPM or if I'm going up a hill and have to push on the gas more to keep the same speed, it will shift to help keep the car moving foreward because my foot is on the gas.

Granted, I don't know a lot about the auto TCU, but I would trust someone who works on them and progams them. ("fuzzy logic" doesn't make sense to me either. why/who the hell calls it "fuzzy logic" anyways?)
 
I do not program them for Mitsubishi, I program them for customers. I guess there must be something wrong with the Dozens of TCUs I've reprogrammed, that do as they are programmed, every time? :rolleyes:

I don't think so. I don't think you understand the context they are using "adaptive" in. I've done Galants, DSMs, and Mirages, and they all work the same. I have the TCU code dissassembly and edit the code and tables to do what I want. The only "adaptive" variables in the trans are the line pressure trims, which it adjusts based on conditions like temp, etc.

When you are driving, it has a lookup table for the shifts, it checks your throttle point and looks at that point to see what it's upshift and downshift thresholds are for that throttle point. Depending where the output shaft speed lands, it upshifts or downshifts as needed. There is also a "shift speed" trim that is sort of a hysterisis. The line pressure is looked up in similar fashion, and adjusted for environmental variables as well.

Believe what you want, short of teaching a programming lesson I'm not sure what else I can provide evidence wise, but I invite you to pull the code off a TCU chip, dissassemble the code yourself, and prove me wrong.

I program them with the shift points I want and shift pressures I want and they do it every time on the dot with no "learning" at all.
 
Thanks for stirring me up and forcing me to dig for more. I have always suspected that my 2G Talon adapted its shift points to cornering forces and braking forces as well as throttle control and gear selection, but have never found that documented. Until now.

I appreciate your offer to view disassembled code but of all the many ways to document the original intent of the software designer, that particular technique is at the bitter bottom of my list. Top on my list is the Patent process. To get protection for the intellectual property of any invention the intent must be disclosed in a patent application is enough detail to allow the invention to be reproduced.

One of the Mitsubishi patents for their “Fuzzy Shift” invention may be found at System and method for the control of shifting of vehicle automatic transmission - US Patent 5544053 Description. That the invention “learns” and adapts can be verified in the “Summary Of The Invention”, and all of the nitty gritty details that only programmers could enjoy is disclosed later in the “shift control programs” section. This is the first place I have seen disclosing that the algorithms have input from cornering forces.

The mapping you described is disclosed here and much more.

I don't think you understand the context they are using "adaptive" in.

...short of teaching a programming lesson I'm not sure what else I can provide evidence wise, ...

Cough, cough, Kidzuku Patents, snort. But That’s ok, personally I stopped believing the “experts” back in the mid seventies.

I have no doubt that you can change values in existing TCU’s, that cause changes in the way the transmission responds. I also have no doubt in Mitsubishi’s ability to program a TCU that senses and responds to many variables as disclosed in their patents. I’m not so sure that there is any real conflict here. Your description of the TCU is an accurate sub set of the patent description.


("fuzzy logic" doesn't make sense to me either. why/who the hell calls it "fuzzy logic" anyways?)

I never had any use for that term. It was first used in the early sixties to describe a programming method where some key parameters were determined at run time instead of being fixed by the programmer such as look up tables.

In the context of this thread the decision to down shift could be a function of many variables including rate of change of the throttle, recent cornering forces, and recent deceleration due to braking.
 
That document is just an extremely complicated version of what I said above. Their "fuzzy" term seems to just be their name for the gearshift hysterisis. It has nothing to do with "learning". It just explains, simply as I did, that there are conditions programmed in the TCU to cover most every instance and thus for every action there is an appropriate action for the trans to take place that corresponds to those conditions specifically. There is no learning in that sense of the word, however all available relevant variables are taken into account when deciding how to shift or not to, so all driving styles are covered by their appropriate behavior. It is all pre-programmed reactions for various circumstances.

and FYI, patents are design ideas, just because someone patents something doesn't mean they implement it in everything they produce.
 
Ah finally we agree on everything except the word "learning". I'll decline that debate. I used to think that people who didn't write software could not call themselves integrators. Then I had the sudden realization that I only use the English language, not control it.

Agreement feels so good that I feel an adaptive tranny story coming on:

In 1997 when the TSI was brand new I was driving my new baby in the right lane of a four lane divided when a pickup came through a red, across his merging lane, and tried suicide by Talon. ABS was not enough but an emergency lane change onto the empty merging lane worked. When I released the ABS the Talon had made a two gear down shift and was in second gear with RPMs up, boost up and ready to go. While passing the pickup at WOT and 7000 RPM the transmission took third with a bang. By the end of the merging lane I was clear of the pickup, back on the highway and lifted the throttle. The Talon held third gear for a long time seeming to ask if I really wanted to use overdrive, or wouldn't I prefer to go back down on the throttle.

If Mitsubishi wants to call that behavior adaptive or fuzzy or learning or just multi-variant control of hysteresis I'm still alive and not offended.

You're probably right about patents and implementation in general. But that Mitsubishi patent accurately describes the behavior of my Talon. Personally I have tended to implement more in the product than disclosed in the Patent. That leaves the copy cat with an inferior product.
 
Last edited by a moderator:
Support Vendors who Support the DSM Community
Boosted Fabrication ECM Tuning ExtremePSI Fuel Injector Clinic Innovation Products Jacks Transmissions JNZ Tuning Kiggly Racing Morrison Fabrications MyMitsubishiStore.com RixRacing RockAuto RTM Racing STM Tuned

Latest posts

Build Thread Updates

Vendor Updates

Latest Classifieds

Back
Top