a few issues with shifting

[Bernard Fife]

the timing of it is wrong it retards the timeing before the shift and returns it right at the shift point

ashford,

The shift sequence is:
- starts at the time a shift is requested and 'approved' (i.e. it won't over or under-rev the engine, etc.),
- the 'clutch outputs' are set,
- the pressure adjustment delay period is waited,
- all the solenoids are set, then
- the shift completion delay is waited out, and
- the spark adjustments are sent and finally
- the 'clutch outputs' are reset,
- the new gear is declared in the datalog.
I believe this is how is should be.

You can't really go by the datalog on the timing of this, as the outpc fields are not synchronized, and the timing of values in the same record can be out by significant fractions of a second.

long datalog where it lost connection a few seconds in and a few time through the log.

strang enough the vss worked perfectly when this happened

I sounds like you have CAN and/or serial issues to deal with. You need to sort those out, as these can easily corrupt the controller's memory and prevent it from communicating (at best) or operating as intended (at worst). I would check all my connections very, very carefully, and if I didn't find a solution I would look into shielding the serial and CAN cables.

Lance.

[ashford]

i built a dedicated vr input for my car with a ncv1124, mph worked well with previous settings, i dropped the smothing from 10-4 and the input masking from 25 to 70. the mph was perfect till about 35 at highway speeds it drifted about 2-3 mph. i then changed masking to 0 and i am about to look over the logs. after i fixed the mph problems the shifing was way off and i discovered the shift speed hysteris works backward from what i thought. if i have a shift point at 10 and a hys at 5 it shifts at 15 i thought it would wok it shifts at 10 and not downshift till 5. my 1-2 and 4-3 shifts are still harsh but i think it may be something in the valve body. i changed my pc pwm to 1khz and it smoothened things out abit.

cam across an annoying problem my pressure transducer likes to stick below 40 psi. to get a more accurate reading i hooked it to shop air with a regulator 9 adc at 0 psi about 290 adc at 100. if it sticks i just tap it and it goes back to 9 .

the psi guage now works with 4122 but either the adc no longer works or it needs a reset to work.

so far i am happy with the way it works.

[ashford]

got a look at the logs. the msq is the same as the logs except the input masking is set to 0. the logs also have the ms3 info. a little advice on vss settings would be apreciated.

as for the spark retard occuring too early it is evident in the ms3 parts of the log, just look at spark advance and engine rpm, timing drops, returns then engine rpm drops from the shift.

[Bernard Fife]

the psi guage now works with 4122 but either the adc no longer works or it needs a reset to work.

Ashford,

The ADC works, but there is a scaling issue I will have to think about for the next code release (the issue is that the conversion to psi/bar is done in the INI now, and there's not a third option, requiring another setting for something that very few people probably use).

For now, you can use the 2 point setting to report the ADC, by changing the settings to:

2point.GIF (5.58 KiB) Viewed 9350 times

To do this you will need the latest INI (and 4.123 code to match the INI): It is here: http://www.msgpio.com/manuals/mshift/V4code.html

Also, there is no dedicated gauge for the ADC output, you would have to create one. It would look something like this (add it to the [Gauges] section of the INI):

Code: Select all

   line_ADC       = linepressure,  "Pressure ADC",          "",     0,    1023,     1,      5,   1020,   1024,  0, 0

a little advice on vss settings would be apreciated.

There is this: http://www.msgpio.com/manuals/mshift/vss.html and this http://www.msgpio.com/manuals/mshift/V41tune.html#gv However, since you have built your own input circuit, you'll have to sort the best values yourself, since no-one else is likely to have the same set-up you have.

the shift speed hysteris works backward from what i thought

For the hysteresis, it works both ways, with your settings if you have just dropped blow 10 mph, then it will take 5 mph more (i.e. almost 15) (or a change in load that exceeds the load hysteresis value) to shift back up. See: http://www.msgpio.com/manuals/mshift/V41tune.html#gs If it didn't work that way, you could still get the trans hunting between gears.

Lance.

[ashford]

i have 2 problems im still trying to work out it almost seems like the hysteresis is not working for 3rd gear. if im at a near steady state and i get near a 3rd gear shift point it will go in and out of 3rd a few times, it happens in both 4th slowing down slightly and 2nd slowly speeding up.

The attachment 43434343.png is no longer available

second i have been changing shift completion and shift pressure delay and cannot get the shift retard and pressure adjustment to overlap the shift. looking at spark advance , line pressure and rpm show this

The attachment retard before shift.png is no longer available

graphretard again.png (43.51 KiB) Viewed 9327 times

[Bernard Fife]

cannot get the shift retard and pressure adjustment to overlap the shift

Ashford,

As I said above, you cannot check the timing advance versus the shifts using a datalog, as the data records in datalogs are not synchronized. The only proper way to check the timing is with an oscilloscope, and the last time I checked it worked as designed. I will check again when I get a chance, though. You should also be sure to be aware of each of these settings: http://www.msgpio.com/manuals/mshift/V41tune.html#ga as they can affect the timing.

hysteresis is not working for 3rd gear

It works for me here, I just tried it. Be sure to read this: http://www.msgpio.com/manuals/mshift/V41tune.html#gh as the throttle position can disable the hysteresis in some user-settable circumstances. What MShift code version are you using? If you aren't using the latest beta code (4.123), then please try that (it will become the release code soon). It has a number of fixes and improvements in all areas over v2.204 (but not so much over 4.122), and is the starting point for any testing.

Lance.

[ashford]

the reason i looked into the shift retard was because i could physically feel the timing retard and come back before the transmission actually shifts, changing settings seems to move how far ahead retard starts but always come bach right before the shift. if needed i can take an sd log on the ms3 at 10ms sampling rate to show spark advance and rpm.

as for the 3rd gear problem i am using 4123 tps shift load tables and kpa pressure. i am using table 2 because i didn not build a 4wd circuit and that is what table it is on, im not sure if that makes any difference or not.

i am temped to go back to kpa based shift tables to see if it takes care of anything, the only downside is if the turbo starts to spool ill have unwanted downshifts.

[Bernard Fife]

i could physically feel the timing retard and come back before the transmission actually shifts

Ashford,

Feelings aren't especially useful, we need accurate measurements (the datalog isn't an accurate measure of this timing, as I said previously) to see what is happening.

However, I suppose what you are seeing could happen if the shift is taking longer to complete than the controller thinks it should. The amount the controller waits for the shift to complete (and thus for the CAN shift timing retard to be reset) is called the 'Shift Completion Delay', and you might need to make that longer. The maximum right now is 500 milliseconds, but that can be made longer by editing the INI. Use notepad or any other text editors to find a line in the INI (near line #241) like this:

Code: Select all

  shift_delay      = scalar,  U16,   427,                   "msec",     8.00000,    0.00000,     10,    500,        0    ; Shift solenoid - actual shift delay

and change the 500 value, say to 1000 like this:

Code: Select all

  shift_delay      = scalar,  U16,   427,                   "msec",     8.00000,    0.00000,     10,   1000,        0    ; Shift solenoid - actual shift delay

That change would give you up to a one second delay, but you could go even higher if necessary. You can read more about the 'Shift Completion Delay' here: http://www.msgpio.com/manuals/mshift/V41tune.html#gs.

Alternatively, you can increase the line pressure so that the shift is completed more quickly (i.e. within the delay period).

Lance.

[Bernard Fife]

Ashford,

It occurs to me that we could make the shift completion delay a function of the line pressure. This could be a 6 or 9 element table, with longer delays at lower pressures (to allow the hydraulic elements to work and then the friction elements to 'grab'). The 4.1xx code is now 'feature frozen' (bug fixes only), but I have begun work on the 5.xxx series of code, and this could be added to that shortly. Let me think about how to do this the best way.

Lance.

[ashford]

if you are open for suggestions i have a few