Trigger hysteresis using avload instead of load?

[Bernard Fife]

Martin,

Did you have the max sLoad in Decel Mode set to 60 for the log (as in the accompanying msq)? I ask because it looks like it was set to 20 in the log, which is quite low and makes the load look very jumpy (almost digital). This setting appears to work as designed on my bench with your settings. If it was set at 60 when you took your log, then there is a problem somewhere. However, if it was set at 20, could you try setting it to 60 or 80?

Lance.

[Addicted]

Lance, it was 20kpa without doubt.

I've just been out for another run, the only thing changed is that 20kpa up to 60kpa, but more importantly the load/tps slider was set to 50%

The load traces look far better, but the driving experience was worse..

Martin

[Bernard Fife]

the driving experience was worse..

How so?

[Addicted]

Maybe 50% is too much at the moment for a cruise/daily car. Probably more suited to a more sporty drive.

For me the addition of throttle makes it more likely to downshift with an increase in throttle, it all comes down to tuning for each application. I need to spend more time with it..

I can't say the latest changes have improved the 3-4-3 shift issue though I'm afraid, it still does it..

Martin

[Bernard Fife]

Martin,

Thanks, that is really useful feedback! I am going to put an option in the code to let users skip the jumps in load when it rises (so that it will rises the same smooth way it falls).

In the meantime, there are a few things you could try:
- you could adjust your TPS table to be more 'flat', perhaps something like this:

TPStable.JPG (25.99 KiB) Viewed 17092 times

this will greatly reduce the jumpiness of the load due to TPS movement especially at lower rpms.

- you could adjust your shift table to be less dependent on load, maybe a little like this:

Shifttable.jpg (63.79 KiB) Viewed 17095 times

so that you only have 3 load 'zones' and try to put the breaks between the zones (50 kpa and 80 kpa above) where you drive the least.

I think I now understand the 3-4-3 issue. I believe it isn't a 'simple bug', but rather that code is working as intended, however the settings needs to be tuned and the code should be tweaked to handle certain situations like this better. In particular, I think the changing load (especially MAP) that occurs when shifting from one gear to another is pushing the target gear to another part of the table. We will fix it, though!

Lance.

[Addicted]

Honestly, after hours of studying datalogs etc.. I do think your code is working correctly, I think its just down to tuning and getting more hours in the seat.

Getting the relationship between load, TPS, TPS loading and speed is more critical than I imagined.

Studying the datalogs I took of the factory Bosch ECU is no help either as its only the outputs and gives no insights into how the software is working.

I'm hoping the weather improves here so I can have a good bash at tuning things.

Martin

P.S I did post in the relevent forum and got no response, but I'd love to be able to open the megashift shift table in Megalogviewer while playing a datalog

[Bernard Fife]

Martin,

Attached is code that gives you more options on how the load is adjusted under braking, decel, WOT, and load rising conditions.

The new settings are under 'General Settings -> Throttle and Load Factors'. There are 4 new settings (in bold below):

1. WOT TPS Threshold
2. Load Smoothing w/ WOT TPS
   1. Usual rate w/ high TPS
   2. Rise quickly w/ high TPS values (default value - same behavior as before)
3. Idle TPS Threshold
4. WOT Duration
5. LOAD Smoothing Factor
6. Load Smoothing w/ Load Rising
   1. Usual rate w/ increasing load (default value - NOT the same behavior as before, change to other option if you want to restore old behavior)
   2. Rise quickly w/ increasing load
7. Load Smoothing w/ Braking
   1. Usual rate w/ braking
   2. Slow load smooth rate w/ braking (default value - same behavior as before)
8. Percentage Load Smoothing when Braking
9. Deceleration Mode
10. Decel Mode TPS Threshold
11. Load Limiting w/ Decel
    1. No decel load limiting (default value - NOT the same behavior as before, change to other option if you want to restore old behavior)
    2. Limit max. load w/ decel
12. Max. sLoad in Decel Mode
13. Downshift Delay

Note that 2 of the 4 settings have a different default behavior from previous codes. In both cases, the new default behavior slows and smooths the load change behavior. You can restore the old behavior by changing the two relevant settings.

Combined with changes to the TPS/rpm load table (suggested in a previous post, the default table has been changed to more closely match that table) and the TPS/MAP percentage (to increase the role of TPS and decrease the role of MAP in determining load), these ought to help sort out the load related issues that are leading to unexpected and unwanted shifts.

Monitor_5099zxv.abs.s19

(83.33 KiB) Downloaded 720 times

GPIO_MShift_5099.ini

(279.25 KiB) Downloaded 750 times

Lance.

[Addicted]

Hello Lance,

Its taken me a while to get the chance to drive the car with the new code..

I have the TPS table as you suggested above ^^^

I've left all the new code options as default.

Just been out for initial drive and... well... its about OEM as I could get. I mean its really really good, drove perfectly!

I've attached a log and my tune file so you can see my set up.

[Bernard Fife]

Addicted,

That's great news! I am really happy this is working for you. Thanks for posting the msq and log, I will have a look. If this code doesn't turn up any new bugs in the short-term, it will become the 5.100 release code soon.

Then I'll look at adding the selective clutch/TCC addition you mentioned in another thread.

Lance.

[Addicted]

I've put a few hours on it today, it still occasionally mishifts, but it definatly needs more tuning now. I also sometimes get slip in fourth, but I'm working on the theory that its because I sometimes find myself in fourth in low rpm/load areas so the line pressure is too low.. again tuning should help!