Shift Target sets to -99 sometimes

[mill3833]

Ok, so I got 4141d loaded and running on the car.

Same problem occurs. When shifter is in M3 and it is slapped into M1, Error code 33 is set as well as mGear goes to -99. cGear is stuck at 0. tGear follows the shift map as speed increases and decreases, but cGear is stuck at 0. A power cycle of module is required to reset it.

I've also got the fault to set slapping it up from M1, so it doesn't appear to be direction sensitive. I also tried max lever position change fro 1 to 3, but that didn't appear to fix it.

Then I tried the "Ignore Lever Errors" option, and that did it. Don't know if that is the correct way to fix the problem or not.

MSL is attached.

Dave

[Bernard Fife]

Dave,

If the 'ignore lever errors' settings helps, by all means use it (that was the way all the code versions were until fairly recently). However, there is something else going on in the code, because an error should be cleared as soon as a valid gear pattern is input. So I will keep looking for the cause!

Lance.

[Bernard Fife]

Dave,

Here is some 4.142 test code that might be better (it performs better on my bench, but that's hardly a fool-proof test environment!).

Monitor_4142.abs.s19

(72.87 KiB) Downloaded 452 times

GPIO_MShift_4142.ini

(218.87 KiB) Downloaded 498 times

Lance.

[mill3833]

Just tried 4142 in the garage, up in jack stands. No joy.

A quick slap from M3 to M1, and error 33 sets as does mGear= -99. This is with Report Lever Errors = T. If I set it to Ignore Lever Errors, then it resets and works. Found that out a couple of versions ago.

Log attached.

Dave

[Bernard Fife]

Dave,

Why are you 'quick slapping' the gear lever? Do you get the same behavior if you shift the lever more slowly? The code doesn't expect the manual lever to shift two positions and also have errors in between, and that might be the root cause.

The code should be able to handle this, of course, and I will work towards that, but I need to understand the conditions under which it the undesirable issues happen.

The code has been designed from the beginning so that all manual shifting is done with the manual shift buttons, and the shift lever is only used to set the highest possible gear that can be attained. If we are using the code in a different manner, then the code needs to be changed, and we need to know exactly what is expected of it.

In the meantime, I would keep the 'Ignore lever errors' set.

Lance.

[mill3833]

Well, slapping is just a way to get the fault to set. This is a floor shifter car, sometimes when shifting from R to D, you over shoot D3 and land in D2 or D1. I do not get the fault if the lever is moved more deliberately. This car never had "paddle shifter" buttons and I'm personally not a big fan of them, so didn't incorporate them. So far, I haven't had the problem with the error reporting turned off. So for now, that is how I'll use it. Just took it out for a 5 mile lap and it all seemed to work nicely.

Dave

[Bernard Fife]

Dave,

Okay, that's good info. I had been looking for single lever errors (without much success), and I suspect a couple of faults in a row is causing the issue. The code should be able to handle that, but it isn't, obviously. I'll have a look at that issue (I have some ideas), and in the meantime you can leave the error reporting off.

Lance.

[mill3833]

No rush. This "bug", if you want to call it that, really doesn't appear to have any functional effect as long as error reporting is off. But you know the code, so I'll let you be the judge of that. I feel safe driving it. Am slowly working up to WOT shifts, am tuning the VE table etc. So we'll see what happens when we get to WOT 1-2 and 2-3 shifts. This will be done in Auto mode. The computer never misses a shift .

[Bernard Fife]

Dave,

The attached 4.143 code *should* be a bit better with respect to the issues above. If you feel like trying it, give it a go.

Monitor_4143.abs.s19

(72.94 KiB) Downloaded 511 times

GPIO_MShift_4143.ini

(218.87 KiB) Downloaded 471 times

Lance.

[mill3833]

Lance,

I tried 4143 this evening. With Lever Error reporting "on", it still sets error when shifter is moved quickly from D3 to D1. It set on the first try, same as previous versions.

To help troubleshoot, I took TS data, connected straight to MSShift via serial, in an attempt to get highest cut rate data. Ended up getting about 25 hz. File is attached. With engine running, on the ground, foot on brake, first I slowly moved it through the gears from D4 to D1 and back up to D4 to establish a baseline. Then I moved it briskly from D3 to D1 and error 33 sets 0.625 sec later, as well as mGear = -99. It then goes through what appears to be a random "reset" period that varies between 1.6 secs and 3 secs. The fault clears for one cycle and then sets again. The only way to "fix it" is to turn off Lever Error reporting.

I then turned off the Lever Error reporting, and drove it around the block and it seemed to shift ok, as before.

Again, no rush. I have lots more to do on this car before it is ready for the track, including working up to WOT shifts. I did a 40% throttle 2-3 shift, using the lever to hold it in D2, punching it, then moving to M3 at which time trans shifted about 0.8 secs later. All that felt the same as a stock PCM tune, so that's good.

Dave