Page 1 of 4
Shift to 5th from Park
Posted: Sun Mar 20, 2011 5:56 pm
by gurov
v4 failed to work for me. it would detect inputs one way (towards D) but then it would take many seconds to detect the next state. i tried multiple values for debounce with the same result.
input voltages did 4.29 v when active, same hardware that works with v2. sometimes the BURN indicator would be on for many minutes, even shifting back and forth between P and D at various intervals.
Re: Latest Code Updates
Posted: Mon Mar 21, 2011 8:13 am
by Bernard Fife
gurov,
Thanks for the observations. I believe this is related to the wait function not being updated yet, and I am working on this (it seems that some wait states can interfere with each other, so I need to have independent counters, etc.).
When I have something that works, I will check each of the items you mention to make sure they are fixed.
Lance.
Shift to 5th from Park
Posted: Tue Mar 29, 2011 7:29 am
by gurov
Lance wrote:All,
I have put up new 4.001 beta code here:
http://www.msgpio.com/manuals/mshift/V4code.html
This code backs out the changes to the shift from PNR to Drive (which, upon reflection were ill-conceived - the controller responded well when everything was right, but the changes were making things much worse if there were uncertainties in the inputs).
Instead the new 4.001 code increases the digital switch debounce factor from 80 to 400. In order to allow this, the debounce variable was changed from an unsigned char to an unsigned int (requiring the corresponding INI file). Users should update this value to at least 400 if they load an MSQ from an older code. Much larger values (up to 65535) can be used to tame uncertain values while shifting the manual lever position.
This code also implements the greatly improved wait function, which now appears to work properly with the wait values I have tried.
I have quite a bit more testing to do on this code, though so
use only with great caution.
Lance.
i will try this out later tonight, treading with great caution.
Re: Latest Code Updates
Posted: Sat Apr 02, 2011 5:33 am
by gurov
saw the same thing as last time. with debounce set to even 50, lever gets picked up fine from P to R to N to D, but once in D, gear is set to 5, counts down to 1, shifting into N/R/P makes it sit there (lever input indicators stay on the same state as when it got into D), eventually it switches back to P state, but this is after about 30-40 seconds of just sitting there.
Shift to 5th from Park
Posted: Sat Apr 02, 2011 10:17 am
by Bernard Fife
gurov,
Can you post a log and msq? I may have to send you some special diagnostic code to try so we can get to the bottom of this, as I am completely unable to duplicate it here after weeks of trying.
Lance.
Re: Latest Code Updates
Posted: Sat Apr 02, 2011 11:10 am
by Bernard Fife
gurov,
One thing to be sure of is that the output patterns in Park and Neutral are the same. If they are different, this will cause problems in the code. I plan to allow them to be different, just in case anyone needs that. The current code only partly puts that in place. It allows you to input different patterns, but expects them to be the same. I will fix this shortly.
Also, did you try increasing the digital switch debounce factor? What values have you tried? Use at least 400, but I would try up to 3000 or even more.
Lance.
x
Posted: Sat Apr 02, 2011 3:02 pm
by Bernard Fife
All,
I have put the latest 4.002 beta code here:
http://www.msgpio.com/manuals/mshift/V4code.html
This code has a large number of architectural improvements (that improve code efficiency and reduce size), but also:
- increases the default digital switch value to 2000 from 400. The minimum is now 500. Users who find that the gear position is jumping to unexpected values while moving the shift lever should use a larger value here (up to 65535).
- checks to see if the output patterns are the same for Park and Neutral. If they are not, the Neutral pattern is changed to match the Park pattern.
The code is largely untested, so
use only with great caution.
Lance.
Re: Latest Code Updates
Posted: Sun Apr 03, 2011 2:52 am
by gurov
Lance wrote:gurov,
One thing to be sure of is that the output patterns in Park and Neutral are the same. If they are different, this will cause problems in the code. I plan to allow them to be different, just in case anyone needs that. The current code only partly puts that in place. It allows you to input different patterns, but expects them to be the same. I will fix this shortly.
Also, did you try increasing the digital switch debounce factor? What values have you tried? Use at least 400, but I would try up to 3000 or even more.
Lance.
i left it at 50-80, forgetting to change it as per the notes.
park and neutral might be different, i will check in my msq.
Re: Shift to 5th from Park
Posted: Sun Apr 03, 2011 7:03 am
by Bernard Fife
gurov,
Try increasing the digital switch debounce. It should help if the problem is what I think it is.
You might also try the 4.002 beta code if you aren't already, it has a few minor fixes in this area.
Lance.
Re: Shift to 5th from Park
Posted: Sun Apr 03, 2011 9:44 pm
by gurov
Lance wrote:gurov,
Try increasing the digital switch debounce. It should help if the problem is what I think it is.
You might also try the 4.002 beta code if you aren't already, it has a few minor fixes in this area.
Lance.
i will give it a shot in a day or so. i keep reverting back to 2.0 Fx_test as my working setup.