Hi performance boost control pwned - Miata Turbo Forum - Boost cars, acquire cats.

Welcome to Miataturbo.net   Members
 


ECUs and Tuning Discuss Engine Management, Tuning, & Programming

Reply
 
 
 
LinkBack Thread Tools Search this Thread
Old 04-09-2012, 05:40 PM   #1
Elite Member
Thread Starter
 
Join Date: Jul 2005
Posts: 6,454
Total Cats: 80
Default Hi performance boost control pwned

Being an electronics and controls nerd I wanted to see what it would take to have high performance boost control.

My wastegate can cracks open at 5 psi. Max boost is 15 psi.
Lots of people say you "can't have target boost more than 2x of your can cracking pressure". Baloney. You just have to know how to tune it.

I have an electronic (analog circuit) feedback loop monitoring (and setting) the pressure in the wastegate can (pressure sensor in line between solenoid and can). The main feedback loop (which targets manifold pressure), tells the electronic circuit how much can pressure it wants.

In feedback loop parlance, this is a "nested loop", where the inner loop is faster than the outer loop. Having a fast inner loop removes the "phase lag" aka delay due to the time it takes to fill and empty the wastegate can from the solenoid. Essentially the can pressure determines the position of the wastegate actuator. So the outer loop commands the actuator position.

The AEM ECU boost control feedback loop does PI control, sending a command (target can pressure) to the above circuit. The above circuit includes a "D" circuit in order to turn said PI into PID.

Here are two recent datalogs. Not they have different timescales.

Top half of each plot has RPM and TPS.
Bottom half shows target_MAP and actual MAP.

target_MAP is a function of TPS. So the target MAP has the same shape.


First datalog shows what happens when I put my foot in different positions.
Horizontal scale is 1 second per division:



Second datalog shows a bunch of on and off throttle to WOT.
Horizontal scale is 2 seconds per division. Worst case overshoot is 0.5 psi.

Attached Thumbnails
Hi performance boost control pwned-boost-control-1.gif   Hi performance boost control pwned-boost-control-2.gif  

Last edited by JasonC SBB; 04-09-2012 at 05:55 PM.
JasonC SBB is offline   Reply With Quote
Old 04-09-2012, 07:24 PM   #2
Elite Member
iTrader: (9)
 
TurboTim's Avatar
 
Join Date: Jun 2006
Location: Chesterfield, NJ
Posts: 6,484
Total Cats: 248
Default

build it and sell it to the masses. you rock.
TurboTim is offline   Reply With Quote
Old 04-09-2012, 08:01 PM   #3
Elite Member
iTrader: (11)
 
Join Date: Feb 2009
Posts: 2,982
Total Cats: 10
Default

No kidding, dont be selfish
miatauser884 is offline   Reply With Quote
Old 04-09-2012, 08:05 PM   #4
Tour de Franzia
iTrader: (6)
 
hustler's Avatar
 
Join Date: Jun 2006
Location: Republic of Dallas
Posts: 29,114
Total Cats: 351
Default

How does relate to us 1% TiAL guys? I tried a manual controller and saw no increase in performance. I'll have MS3 soon so I'm moderately interested.
hustler is offline   Reply With Quote
Old 04-09-2012, 09:29 PM   #5
Elite Member
iTrader: (9)
 
TurboTim's Avatar
 
Join Date: Jun 2006
Location: Chesterfield, NJ
Posts: 6,484
Total Cats: 248
Default

I plan to build the bottom end such that boost spikes/overshoot won't hurt anything. But that's not the right solution...
TurboTim is offline   Reply With Quote
Old 04-10-2012, 12:37 AM   #6
Elite Member
Thread Starter
 
Join Date: Jul 2005
Posts: 6,454
Total Cats: 80
Default

TurboTim, just make timing go lots retarded above the danger point. That'll prevent engine damage.

As for the circuit. I need to do longer term testing - check for stability, performance over time, degradation with parameter drift, etc.

I'm also thinking about some kind of overshoot arrester function in case the tables are set up wrong. And also an improved integrator algorithm that gets rid of "integrator windup" and the resulting overshoot.

The reasons I built this circuit... I noticed in bench testing with a pressure source, solenoid, and wastegate can, that the solenoid response to duty cycle is very nonlinear, and the actuator moves slowly to its final position when you put in a new duty cycle. And, the actuator "shivers" (vibrates) in time to the solenoid pulsing. Betcha you'd see the pulsing in the turbine inlet pressure as a result.

So my circuit will:

- get rid of the nonlinear effect of solenoid duty cycle vs. wastegate actuator position.

- get rid of the phase lag / delay from the time duty cycle changes, to the final position of the actuator, due to the time it takes to fill or empty the wastegate can (this is where the greatly improved response speed comes from)

- as a bonus if you change your solenoid you don't need to retune the tables, but perhaps only the 'D' term.

- it also runs the solenoid with variable frequency - at 50% duty cycle for example, it will run close to 30 Hz, because solenoids can run at higher frequency near 50% duty cycle. At more extreme duty cycles e.g. 10% or 90%, the frequency will go down to ~10 Hz. This feature reduces the "shivering" of the wastegate actuator.

Now instead of entering solenoid duty cycles in the AEM boost control table, I effectively enter target can pressure, (described by duty cycle LOL). The AEM effectively tells the circuit what can pressure it wants.

The AEM has:
- a 2D table to set up the boost control output duty cycle vs. boost target,
- a 3D one with RPM and MAP as axes. I use that to trim target can pressure vs. RPM. - Another 2D table for duty trim vs. boost error (a nonlinear 'P'),
- an 'I' gain term with limiting

- and of course a 3D table of boost target, with TPS and RPM as axes


As for building it for the masses ...
What does the MS2 have for lookup tables and PID?

If it is similar to the AEM but with 'D' built in, then the feedback loop will be in the MS, and the circuit just needs to have a duty cycle input and a pressure input from a tee to the wastegate can. If it has no 'D' then the circuit will need to implement the 'D', just like my prototype.
JasonC SBB is offline   Reply With Quote
Old 04-10-2012, 03:52 AM   #7
Supporting Vendor
iTrader: (10)
 
Reverant's Avatar
 
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,440
Total Cats: 198
Default

The MS2 has a proper PID.
Reverant is offline   Reply With Quote
Old 04-10-2012, 01:02 PM   #8
Elite Member
Thread Starter
 
Join Date: Jul 2005
Posts: 6,454
Total Cats: 80
Default

Can it do PID *and* have a TPS vs. target boost table?

If so, then my circuit can simply take a PWM "can pressure" command input, and a hose tee'd from the wastegate can.

Anyone want to play with a prototype?

Hustler:
EWG's are more linear and thus easier to tune without my circuit.

What my circuit can do for an EWG setup is perhaps make it faster and possibly make it even easier to tune for good response speed without oscillation.

What I don't know is, will it help appreciably enough?
JasonC SBB is offline   Reply With Quote
Old 04-10-2012, 01:03 PM   #9
Elite Member
Thread Starter
 
Join Date: Jul 2005
Posts: 6,454
Total Cats: 80
Default

BTW the most important improvement to boost control in my mind, is to have *adaptive* control. That is, you change something, you may start to get overshoot or under-boost, and it *learns* and adjusts. Just like the OEMs lol.

Adaptive control is a big topic which unfortunately I haven't had a lot of experience with.

Last edited by JasonC SBB; 04-10-2012 at 01:15 PM.
JasonC SBB is offline   Reply With Quote
Old 04-10-2012, 01:12 PM   #10
Boost Czar
iTrader: (61)
 
Braineack's Avatar
 
Join Date: May 2005
Location: Chantilly, VA
Posts: 72,847
Total Cats: 1,788
Default

Quote:
Originally Posted by JasonC SBB View Post
Anyone want to play with a prototype?

i'd play.

I tested an EBC from a member here, remember brian007 or something like that? maybe it was loki007? i cant remember it was in like 2005 or 2006 or something.

it was called the AXIOM EBC. had a nipple and a lcd screen and two toggle buttons. you could enter in the boost target and then some senesitivty levels (probably d term) and it would just work, it actually worked well but nothing came of it.

had it controlling a GM boost solenoid.
Braineack is offline   Reply With Quote
Old 04-10-2012, 01:16 PM   #11
Elite Member
Thread Starter
 
Join Date: Jul 2005
Posts: 6,454
Total Cats: 80
Default

But no TPS based boost targeting?
JasonC SBB is offline   Reply With Quote
Old 04-10-2012, 01:17 PM   #12
Boost Czar
iTrader: (61)
 
Braineack's Avatar
 
Join Date: May 2005
Location: Chantilly, VA
Posts: 72,847
Total Cats: 1,788
Default

not that thing no.


MS2 and MS3 does. has a tps/rpm targets table. it might have a base duty reference table now (idle does)....i havent kept up with the changes.
Braineack is offline   Reply With Quote
Old 04-10-2012, 02:25 PM   #13
Supporting Vendor
iTrader: (10)
 
Reverant's Avatar
 
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,440
Total Cats: 198
Default

Quote:
Originally Posted by JasonC SBB View Post
Can it do PID *and* have a TPS vs. target boost table?
Yes. In fact, my Enhanced MS2 has 2 different TPS vs target boost tables, which you can select from using an external switch.
Reverant is offline   Reply With Quote
Old 04-10-2012, 02:35 PM   #14
Elite Member
Thread Starter
 
Join Date: Jul 2005
Posts: 6,454
Total Cats: 80
Default

Nice.
You may want to play with implementing my circuit using code.
And my idea for a sekrit enhanced adaptive I function.
JasonC SBB is offline   Reply With Quote
Old 04-10-2012, 02:40 PM   #15
Supporting Vendor
iTrader: (10)
 
Reverant's Avatar
 
Join Date: Jun 2006
Location: Athens, Greece
Posts: 5,440
Total Cats: 198
Default

Email me, I am very open to ideas.
Reverant is offline   Reply With Quote
Old 04-10-2012, 03:53 PM   #16
y8s
2 Props,3 Dildos,& 1 Cat
iTrader: (8)
 
y8s's Avatar
 
Join Date: Jun 2005
Location: Fake Virginia
Posts: 19,038
Total Cats: 407
Default

Quote:
Originally Posted by JasonC SBB View Post
Nice.
You may want to play with implementing my circuit using code.
And my idea for a sekrit enhanced adaptive I function.
write it in pseudocode and we can see if ken/james will bite on it.
y8s is offline   Reply With Quote
Old 04-11-2012, 03:51 PM   #17
Elite Member
Thread Starter
 
Join Date: Jul 2005
Posts: 6,454
Total Cats: 80
Default

Judging from my experience pointing out the bug in the D of idle PID that GregG ran into, and watching the development of Gslender's code, they seem to be very reluctant to try anything not in their presumably very long "to do" list.
JasonC SBB is offline   Reply With Quote
 
 
Reply

Related Topics
Thread Thread Starter Forum Replies Last Post
Project Gemini - Turbo Civic on the Cheap Full_Tilt_Boogie Build Threads 57 07-19-2017 05:11 PM
LucaCarMods's Boosted Dutch Miata Build! LucaCarMods Build Threads 11 02-14-2016 07:13 AM
My solution for Oiltemp and Oilpressure input into Megasuirt (MS3) Zaphod MEGAsquirt 41 01-24-2016 01:25 PM
Going back to stock. Need some 1.6 parts. Trent WTB 2 10-01-2015 01:15 PM
Bad head gasket or ? shooterschmidty Engine Performance 8 09-30-2015 11:28 PM


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off



All times are GMT -4. The time now is 03:28 AM.