MS, Closed loop boost control success. It spools! FAST

[qwkswede]

I thought I might add to this thread. I have been having boost control issues on my car with an Audi wastegate. Somewhat caused by my desire to run somewhat lowish boost with a gate that is probably too small for the motor. Basically the boost just rises with rpms and boost doesn't stabilize below about 18 psi. I think it makes for a tricky beast to tame.

I wired up megasquirt's boost control feature to try and control this wasgtegate a bit better. This entailed installing an Audi VW N75 wastegate control valve, and adding the transistorized circuits to the megasuqirt unit, and wiring it all up.

I tried open loop boost control first, just setting the N75 valve to cycle at a set duty cycle, bleeding off some pressure in the signal line. But the boost creep problem remained. I could make the boost increase, but it would overboost. And if I lowered the duty cycle, the boost would just climb way too slowly the same as without the controller. And with differences in spool up in different gears, I had different boost in every gear, slamming into the overboost cut sometimes. it was ugly. I set the valve as both a bleed, and pressureizing the top port of the dual port wastegate with no success either way.

I studied up on the closed loop algorithm used in megasquirt, activated it, and a couple days later, I had a reasonably controlled boost level. At first, it was a bit bumpy, overshooting a little, then bouncing up and down 4-5 psi or so. It felt really strange, with all the surging power. But the boost wasn't running away to 20+ psi like it did sometimes.

<a href="http://www.denverspeed.com/gallery2/v/cars/irene/Electronics/boost_wiggle.png.html?g2_imageViewsIndex=2"><img width="621" src="http://www.denverspeed.com/gallery2/d/22736-1/boost_wiggle.png" alt="wiggling boost" height="645" /></a>

Below is after a little more experimenting. The boost holds relatively steady now, within a couple psi. And now I get full boost of about 15psi at about 3300 rpm. Its great! It just roasts the tires in second, and sometimes third gear when the boost hits now.

<a href="http://www.denverspeed.com/gallery2/v/cars/irene/Electronics/controlledboost.png.html" title="controlled boost"><img width="614" src="http://www.denverspeed.com/gallery2/d/22734-2/controlledboost.png" alt="steadier boost" height="640" /></a>

Here is where I ended up to get the above boost control.
<p align="center"><img align="middle" width="321" src="http://www.denverspeed.com/gallery2/d/22740-1/boost+control+settings.JPG" alt="controlled boost" height="491" /></p>

[benflynn]

will this work w/gm valve?

[qwkswede]

Hm. I don't know. I have only tried it with this valve. But in theory it should do the same thing I think. The solenoid is plumbed into the hose that pressurizes the top port of the wastegate. 100% duty cycle gives the top the same pressure that the bottom port gets. 0% bleeds that signal off to the atmosphere through the unused 3rd port on the valve.

[adrianpike]

Well damn.

I might have to dust off the Audi after all.

[qwkswede]

You gotta love $10 wastegates.

[linuxman51]

how'd you have your open loop settings defined?

[qwkswede]

how'd you have your open loop settings defined?

I set the closed loop kpa limit in the window above to 0, or near that. Here is one of the last duty tables I was playing with. I tried having equal duty cycles across the board, and that didn't work.This was an attempt to get the boost to rise a bit faster at lower rpms, then stabilize above 4000. The big problem with the open loop was that it could work in 2nd gear where the rpms rise fast, but in 3rd and 4th where you can pull in a gear for a longer time, the boost would overshoot my 21psi boost cut, and sometimes ping a little on pump gas.


http://forums.turbobricks.com/attachment.php?attachmentid=4178&stc=1&d=1197137867
I've been trying to get this thing toned down to run about 17psi max. Which is where the wastegate seems to control things at 6000rpm with the piston/valve wide open. This seems to be a good number for pump gas. The motor is happy there without detonation.. But I'd like to have that across the rpm range. Im going to keep playing with the closed loop, and see if I can't get it to smooth out a little more. I can still feel a slight power surge right now. Maybe putting a restriction on the third port of the N75 valve can keep it from reacting so fast to small changes, basically make it not dump the singal pressure to the top of the wastegate so fast.

[linuxman51]

tune it in 4th gear, I gave up on closed loop control because it would never actually lock the boost down, I wanted it to run 15psi, we could get it to go to 15, it would hang out there for a little while and then suddenly start rising and tapering until I let off the gas. open loop worked fairly well for me, but I used the entire table :shrug:

Now I have a profec-b spec2 on the car and it works great.

[qwkswede]

yeah, you can see the rising and tapering in my logs. They were pretty big at first. But I looked at the formula in the code, and realized that the proportional gain needs to be really small if you want the output adjustments to become small enough for fine control. I think I still need to go even smaller, you can see the BC Duty line swinging from 100 to 0 and back, even with the most recent "good" settings. The boost is smoothing, but not perfect. That first run, I was using something like 10% proportinal gain, and the boost control just jumps up and down rapidly around the set point. Setting it at 2% slowed it down a bit, but it still acts almost digital. Really, it needs to use some values between 0 and 100% duty cycle I think.

** The closed loop calc is
** output = output + (kpaTarget-kpa)*pGain - (kpa-kpaLast)*dGain

so imagine an output of about 100 duty cycle like when it is trying to ramp toward a boost setting of 200kpa while the current kpa is at 100kpa. put those numbers into the formula above, and you can get a feel for how big the output changes become with even a gain setting of 0.1 for pGain.

Actually, that output may be like 255 for 100% duty cycle. But I think the results are the same.