Pre-1.4 beta 10 boost control duty cycle

[muythaibxr]

OK, it wasn't clear that was the only diff. Maybe it is doing some blending or something. This may have to wait for me to have a chance to reproduce it unless James has time.

[nathaninwa]

I can turn blending off but most of the problems/testing are done in third gear where full target is asked for

[muythaibxr]

Right but the blending stuff is clearly ramping target rather than just setting it which is causing your problem as far as I can tell.

[nathaninwa]

back dated to pre-1.4 beta 5 Was having an issue that shut the motor down prematurely (switched back to beta 10 and all is well) but before the motor shuts down you can see that at the Delta the dc is nowhere near the bias table value, and the slider is set to 0 in basic mode

Remind me, tuning the bias table is with closed loop on, and either in setup mode, or in the older firmware the slider set to zero to numb out correction?

Heres a log and the tune I used.

[muythaibxr]

The bias table should be tuned in setup mode. I suggest you disable blending until you figure out your issue.

[nathaninwa]

I did, that last 1.4.5 tune had all blending off and single table target of 280kpa

And the blending works fine in open loop mode

It's just not using the bias table values so I can tune it proper

[muythaibxr]

I will take a look when I get to a computer again.

[muythaibxr]

OK, I see what's happening. Between 2000 and 2500 RPM, you go from a target of 0 to a target of 280 kPa. Remember that the target is interpolated, not just 0 then 280. Because of that, as soon as boost is 50kPa (your delta setting) away from the *current* target. Boost control PID kicks in.

I don't think setting the slider to 0 is doing what you think it is. Set the slider somewhere normal, set the tuning mode to advanced, and set the PID gains to all 0 (on that code rev) to get the behavior you're looking for. Also, set the boost target at 200 RPM somewhere closer to what you want from 2500 up, or set those two bins REALLY close together if you want it to go straight from 0 to 280 kPa.

The code is doing what it's told, not a bug as far as I can tell at this point.

Ken

[nathaninwa]

Gotcha. I'll fix that and make it all 280 for testing, but how do we not get the pid to work. So I can at least tune the pid table. That's what's been plueging me. I'm trying to tune the bias table manually to set it up, but it seems pid keeps working after my 280 target and uses larger than bias table numbers

I'll log tonight with the corrections and post

[muythaibxr]

Set the gains to 0 in advanced mode. Leave it that way until tuned.

[muythaibxr]

Or go to the newer firmware and use setup mode.

[nathaninwa]

That's what started all this. My .10 has been in setup mode and the .5 had the slider at zero in basic mode, but the dc values were always greater than what was in the bias table

I'll make the whole target 280 tonight and go take a quick log and post up what the changes are

[nathaninwa]

Took a couple logs today. Im not understanding how the 0-280 target messed with setup mode tuning, but all seems well now when the table is targeting 280 all the way across. Im happy with the results so far. Its still a little wet and cold here so traction is limited in some areas of road testing

Thanks for keeping pacients with me.

[muythaibxr]

It was basically slow ramping the target to 280 when you started at 0 for the low rpm target. I think the rest is just down to odd/incorrect methods of trying to keep PID from working.

[nathaninwa]

Did some more tuning today. Ill post 3 logs, MSQ is the same as yesterdays in the above post. Changes are blended boost targets, boost by gear and gear based table switching. Gear based table swtich and blended boost targets are not working properly and behaive as before using way to high DC percentages when still in test mode. Boost by gear seems to be working properly

Im going to tune using boost by gear so I can tune 1-2-3 all in one pass and tune there VE values based on its acutally loading

I would like to run blended tables so I can target near 310kpa at peak tq then drop out the top a bit to keep header drive pressures down.

[muythaibxr]

I will have to reproduce when I get time (possibly this weekend).

[nathaninwa]

No problem. I should be able to tune with boost by gear, I just wanted to report the odd behavior. I know many don't do it the way I do and when I get th bias table close I'll try it in actual pid controled boost

[nathaninwa]

Just an update. Made a 1-2-3 gear pull this morning and boost target by gear is working as it should

[jsmcortina]

Did some more tuning today. Ill post 3 logs, MSQ is the same as yesterdays in the above post. Changes are blended boost targets, boost by gear and gear based table switching. Gear based table swtich and blended boost targets are not working properly and behaive as before using way to high DC percentages when still in test mode.

Could you highlight where in the datalogs you are seeing unexpected behaviour and the precise settings in use at the time.

James

[jsmcortina]

OK, pretty sure I see the problem.

By design we wanted there to be one bias table. Blending is between the two boost target tables.
Say you have 150kPa target in one and 200kPa in the other and you are partway, you might end up with 180kPa.

The code should then lookup the duty cycle needed for 180kPa from the bias table.

However... there's a mismatch. The ini file is doing that and showing one bias table, but the code is actually using both bias tables and blending between them. (Oops.)

Can you think of any circumstances when it actually makes sense to use two bias tables? Do different fuels alter the turbo operation enough to warrant it?
Having two gives you more flexibility, but is more tuning. Our aim is to keep things simpler if possible.

Experienced users - please chime in!

James