Ever got frustrated by trying to tune your boost control or idle control?

This is a method published by 2 dudes back in the 1940s, producing parameters resulting in quarter wave decay for closed loop control. While it may not be optimal for all instances, it serves as a good basis to start.

Step 1:
Make sure the process is lined out.

Step 2:
Increase the gain in 1/2 steps of the previous gain. After each increase, if there's no oscillation change the setpoint slightly to trigger an oscillation.

Step 3:
Adjust the gain so oscillation is sustained. i.e. same amplitude. If the oscillation is increasing, decrease gain. Conversely, increase gain if the oscillation is decreasing.

Step 4:
Note the gain causing the sustained oscillation, and the period.

Use one of the following tuning equations as appropriate.

Proportional control only: Gain=0.5
PI: Gain=0.45, reset=PU/1.2
PID: Gain=0.6, reset=PU/2, derivative=PU/8

 

manual boost control (ie. a forge unit), and idle control (ie. an adjustment screw) are not closed loop, feedback systems similar to the 02 sensor function, therefore what you just posted does not apply.

Do you even understand what you just posted?

 

looks like a standalone set up, so yes it does look like he knows what he is talking about.

 

delete that, too early for 02 sensors.... actually it looks like early versions of simple proportional control. It really doesn't even have to do with cars now that I can place the name from a process control class I took in college.

and to add, who runs a standalone for boost control and idle control that you can actually go in and adjust the porportional, integral and derivative settings?

 

AEM EMS uses P&I control that is completely adjustable for closed loop A/F and boost control, should the user choose this method of boost control. This is the only method for closed loop fuel control.

 

I'm not following you to clearly, I have no experience with the AEM, that may be why. But are you saying that the user can go in, and change the gain and reset for A/F ratio, and boost control? If so, how many users actually use this function? Also, the only method for closed loop fuel control is for the user to go into the programming, and set his own gain and reset?

I would certainly hope that the AEM was already using PID, and not just PI control schemes. If not, it's dated by today's control standards.

(added) most standalones have a self tune function for various signals, whereby they do all this "learning" internally without user input. Somewho I think the algorithms that the design engineers came up with would be better than most user inputs. But from what you said, the user does this instead of the microprocessors in the standalone AEM for closed loop fuel control?

 

Well your boost pressure is controlled by a solenoid , and idle speed by a stepper motor, also in closed loop. So what I posted does indeed apply.

Did you?

 

First of all, it would be greatly appreciated if you took the time to at least learn about the topic before commenting. You mentioned taking a process control class in college. Great. Open up your eyes. Process control is everywhere. How do you think your fridge works? Air-conditioner? Heater? Microwave? The list goes on and on. It's definitely applicable to automotive technology.

Second, re: PID vs PI, this is not even a debate. One more parameter to tune is great if you know what you're doing. On the other hand if PI or PD control works fine, then why bother adding complexity?

As for the last bit, it's possible for the engineer to specify a sensor(s) for a specific application and program the control loops for it. However if you make changes to any part of the system under control then things may not be what you expect. Thus leaving some parameters to the tuner/end user is still somewhat desirable (assuming he knows what he's doing).

 

Crufty,
The Z-N method is a nice quick and dirty method to get your PID gains in the ballpark, as you mentioned. Thanks for posting this. I think many people would be surprised at how often this is used in industry where masters and Phd controlls engineers are hard to come by.

Are you proposing people tune these parameters with a fucntioning system (ie actually measuing boost resposne while tuning bc solenoid) or using a sensor or scope of some sort to measure solenoid response? The only issue I can forsee with tuning a live system is accidentally having the system go unstable...

I assume your reffering to standalone units here, as I have not seen anyone tuing control parameters in the OE computer.

 

Am I the only one who feels that all this is WAY over my head?

 

The Ziegler Nichols tuning method is way outdated and tends the force the system to the point of instability (not so great for automotive systems in which a disturbance like predetonation can easily lead to a blown engine). A similar method is the Cohen-Coon tuning method, but that tends to be not very robust (i.e. cannot account very well for disturbances in the system, which can be many in an automotive system).

There are better methods out there, like IMC tuning which give numerous different parameters for time constants, gains, etc. for the various different types of systems out there (open or closed loop, stable or nonstable processes, and processes with and without time delays) and the type of control they use (P/PI/PID).

 

Hi there...

The motivation behind the post was to help others understand how these things work, that there is a method to the "madness".

You are very correct in pointing out that bad things could happen should a live system go unstable; however having said that it's much safer by using the steps outlined earlier as a base to start off from than it is to tune blindly, as is often the case.

Happy holidays.

 

I do understand the topic quite well thank you. I agree that process control is everywhere, even on our cars, the 02 sensor closed loop is a prime example. I also understand that what you posted originally has nothing to do with boost control or idle control on our cars as stated. Name a boost controller where you can actually go in, and set your gain, reset and derivative values, such that your formulas posted work.

Also, I don't know about your stuff, but my fridge, a/c, and heater work off simple temperature switches, and it's on/off. If you call on/off operation process control (and there is defintely no PID controller mixed in there), maybe you need to learn a bit more about the topic before posting formulas like you did.

 

I'm with ya buddy. I leave my tunning to a good flash and SAFC. That's as far as i go

 

although I don't know if I would use his posted method, it is helpfull to know various methods to set PID's. My standalone uses a PID on the idle control. might even have one on the boost(though I havn't had to go there)