it's scary what can be found in an ECM

[Amish Elegance]

Tacos anyone??

Don't you dare... There is some damn good controls information being spilled here, I plan on learning from it.

It sounds like you two are saying the same thing, just using different terminology. Lol.

tapatalking from droidx

At points I agree.


Ok... Never having messed with EFI Live, but growing up in a business that breathes PID controls I actually have a rudimentary understanding of what you two are fighting about. I am not an engineer (Preface) welcome corrections to understanding.

The way I see these ECM's working as they relate to the tables: The following are known's: Pumps ability to create pressure as it relates to volume. Injectors ability to deplete said pressure based upon volume with given commanded duty cycles. (This is before we muck with it) I know there are many other tables that factor in to this, but these are the only two we are talking about.

As far as the ECM is concerned, you put your foot in it. Injector duration is controlled independently of pressure commands (for purposes of this conversation). As the throttle demands dictate, the ECM reads the pressure table and sends a duty cycle to the FCA. The feedback for this function comes from the rail PRESSURE sensor. If this action is within the permitted hysteresis (I know that term is mostly used in temperature control), no derivative function is necessary. The PID controller is always varying FCA position to match the changing values in the pressure table. At steady state, the integral function assists to stabilize.

OK... PID just has to do with the Rate and Accuracy of which the ECM varies the FCA position trying to hit the pressure value on the table. Sometimes it can (commanded is equal to actual), sometimes it cant (Commanded not equal to actual)

This speaks to the comment about the set-point being aggressively above attainable pressures. The further the feedback value from the rail pressure sensor is away from the pressure on the table, the more drastic the ECM's response will be. Setting a value of 36K will snap the FCA to 100%. At this point you are not even trying to measure.

One exception... the pressure limit value. PID controls have operational bandwidths that can be set to confine them. Regardless of commanded values. Think of it as a limit function. You all say its 23k and change, seems like a good thing given the previously viewed tables.

But my understanding of PID controls, and the degree of their reaction being based upon the degree of deviation from command, commanding a set-point that much higher than the limit would produce a very unstable environment. The ECM would ramp the FCA duty cycle wildly as it approached the upper limit value. That value would kick in, drastically reduce the duty cycle, however the second it fell below the proportional value of the cutoff, you are still VERY far away from setpoint, prompting a harsh reaction by the ECM upon the FCA to correct this sending pressure readings oscillating further degrees from commanded.

As the FCA is also an actuator, is its function also encoded and PID controlled? That is to say, are its movements in response to commanded duty cycle from the ECM measured and functioning perhaps on a PID loop of their own as it relates to movement?

I marvel at some of JSP's other posts. The measurements taken from his fuel system were then CALCULATED into table values for tunes for his standalone. Maybe everybody does this, I just remember him being more open than most.

I digress. Control Engineers fascinate me and infuriate me as I work with them. But I admire them when their voodoo works. I've tried to learn from the ones around here as much as I can. How am I doing?

 

[Timbeaux]

I have no idea what you just said, but I think you just biitch-slapped somebody.

 

[Joesixpack]

By definition a valve only controls volume. Pressure is a secondary result of the amount of substance a valve passes. Plumbing 101. LOL

Its really that simple isn't it.

thing is, there's nothing in the ECM or tuning software that say..... this is the "volume" per millisecond or what ever. My understand is the FCA is told what to do based on the duty cycle it's set at. It can be called a valve, limiter, restrictor, taco, Dennis's boat.... what ever you want. It's still not "measuring" any type of volume going through it. It's just doing what it's told.

You really need to put more thought into it. The ECM is all about mm3 or volume of fuel. At any moment of time it has exactly calculated the volume of fuel it needs to inject per stroke. It uses the injector conversion table of pressure against time to understand how much time it will take to do so.

At points I agree.


Ok... Never having messed with EFI Live, but growing up in a business that breathes PID controls I actually have a rudimentary understanding of what you two are fighting about. I am not an engineer (Preface) welcome corrections to understanding.

The way I see these ECM's working as they relate to the tables: The following are known's: Pumps ability to create pressure as it relates to volume. Injectors ability to deplete said pressure based upon volume with given commanded duty cycles. (This is before we muck with it) I know there are many other tables that factor in to this, but these are the only two we are talking about about.

As far as the ECM is concerned, you put your foot in it. Injector duration is controlled independently of pressure commands (for purposes of this conversation). As the throttle demands dictate, the ECM reads the pressure table and sends a duty cycle to the FCA. The feedback for this function comes from the rail PRESSURE sensor. If this action is within the permitted hysteresis (I know that term is mostly used in temperature control), no derivative function is necessary. The PID controller is always varying FCA position to match the changing values in the pressure table. At steady state, the integral function assists to stabilize.

OK... PID just has to do with the Rate and Accuracy of which the ECM varies the FCA position trying to hit the pressure value on the table. Sometimes it can (commanded is equal to actual), sometimes it cant (Commanded not equal to actual)

This speaks to the comment about the set-point being aggressively above attainable pressures. The further the feedback value from the rail pressure sensor is away from the pressure on the table, the more drastic the ECM's response will be. Setting a value of 36K will snap the FCA to 100%. At this point you are not even trying to measure.

One exception... the pressure limit value. PID controls have operational bandwidths that can be set to confine them. Regardless of commanded values. Think of it as a limit function. You all say its 23k and change, seems like a good thing given the previously viewed tables.

But my understanding of PID controls, and the degree of their reaction being based upon the degree of deviation from command, commanding a set-point that much higher than the limit would produce a very unstable environment. The ECM would ramp the FCA duty cycle wildly as it approached the upper limit value. That value would kick in, drastically reduce the duty cycle, however the second it fell below the proportional value of the cutoff, you are still VERY far away from setpoint, prompting a harsh reaction by the ECM upon the FCA to correct this sending pressure readings oscillating further degrees from commanded.

As the FCA is also an actuator, is its function also encoded and PID controlled? That is to say, are its movements in response to commanded duty cycle from the ECM measured and functioning perhaps on a PID loop of their own as it relates to movement?

I marvel at some of JSP's other posts. The measurements taken from his fuel system were then CALCULATED into table values for tunes for his standalone. Maybe everybody does this, I just remember him being more open than most.

I digress. Control Engineers fascinate me and infuriate me as I work with them. But I admire them when their voodoo works. I've tried to learn from the ones around here as much as I can. How am I doing?

Thats bang on, The FCA has a conversion table that converts duty cycle to current to control the solenoid.

Obviously you've dealt with PID's....they can be a sweet as apple pie when they are set up.....can be mo-fo's from hell when they are out of wack.

The 36K is meaningless however as the PID is NEVER going to try a setpoint higher than the pressure limiter. No fool in his right mind would build a tune without this basic limit.

This likely being a box tune and from early limited understanding just simple max'd out the table with an arbitrary value as to ensure fuel pressure at the highest setpoint possible. So while a few folks are running around with their head cut off "36K!!!!! oh my god" all I see is a lazy tuner or with limited access to all the tables ensure max rail pressure setpoint. Hence my comment that 36k is as meaningless as 99k, its not going any higher than the set limit what ever it may be.


So I'll steal Flashscan's cruise control analogy for a min.

Cruise control is a simple PID, obviously when you go down a hill it pulls fuel to maintain the setpoint, when you go up a hill it adds fuel again to keep the setpoint. (Well shouldn't say simple as new ones do learn somewhat)....


Now we've all likely noticed sometimes the cruise takes a while to get back to the setpoint when for example we come to a fairly large hill. Every time you come to a hill it has to "learn" just how steep this hill is and how much additional fuel it needs to add.

It does this by closed loop trial and error. Explaining how each term in PID effects this is beyond the scope of this post.




HOWEVER......Now think about if at any point in time the cruise control had an exact map of our road, along with just how steep the hill was.

NOW our cruise control is not just adjusting by trial and error. It has this handy dandy map that shows it very close to just how much fuel it needs to add.

Now it works from this map and can adjust very quickly to to very rapid changes in load, and the PID loop fills in with small adjustments.


THATS what the Pressure control base duty cycle table is.

It is the Base understanding of how much fuel volume is needed from the FCA for a commanded mm3.

This table is what we call the "feed forward" term, this table is brought about from experimentation and monitoring FCA fuel volume requested and actual.

:Cheer:

 

[TheBigNasty]

I have no idea what you just said, but I think you just biitch-slapped somebody.

In a way. LOL

His explanation of pid's connected the dots that were missing in my head, and now I see what Joe is trying to explain.

More or less, Rich may be able to make a tune that works, but Joe is trying to explain that through better understanding of the purpose and implementation of the tables, his tunes would rely less on PID's corrections to produce a desired result.


Edit- Damn, tree'd by Joe... LOL

 

[zstroken]

Its really that simple isn't it.





You really need to put more thought into it. The ECM is all about mm3 or volume of fuel. At any moment of time it has exactly calculated the volume of fuel it needs to inject per stroke. It uses the injector conversion table of pressure against time to understand how much time it will take to do so.






Thats bang on, The FCA has a conversion table that converts duty cycle to current to control the solenoid.

Obviously you've dealt with PID's....they can be a sweet as apple pie when they are set up.....can be mo-fo's from hell when they are out of wack.

The 36K is meaningless however as the PID is NEVER going to try a setpoint higher than the pressure limiter. No fool in his right mind would build a tune without this basic limit.

This likely being a box tune and from early limited understanding just simple max'd out the table with an arbitrary value as to ensure fuel pressure at the highest setpoint possible. So while a few folks are running around with their head cut off "36K!!!!! oh my god" all I see is a lazy tuner or with limited access to all the tables ensure max rail pressure setpoint. Hence my comment that 36k is as meaningless as 99k, its not going any higher than the set limit what ever it may be.


So I'll steal Flashscan's cruise control analogy for a min.

Cruise control is a simple PID, obviously when you go down a hill it pulls fuel to maintain the setpoint, when you go up a hill it adds fuel again to keep the setpoint. (Well shouldn't say simple as new ones do learn somewhat)....


Now we've all likely noticed sometimes the cruise takes a while to get back to the setpoint when for example we come to a fairly large hill. Every time you come to a hill it has to "learn" just how steep this hill is and how much additional fuel it needs to add.

It does this by closed loop trial and error. Explaining how each term in PID effects this is beyond the scope of this post.




HOWEVER......Now think about if at any point in time the cruise control had an exact map of our road, along with just how steep the hill was.

NOW our cruise control is not just adjusting by trial and error. It has this handy dandy map that shows it very close to just how much fuel it needs to add.

Now it works from this map and can adjust very quickly to to very rapid changes in load, and the PID loop fills in with small adjustments.


THATS what the Pressure control base duty cycle table is.

It is the Base understanding of how much fuel volume is needed from the FCA for a commanded mm3.

This table is what we call the "feed forward" term, this table is brought about from experimentation and monitoring FCA fuel volume requested and actual.

:Cheer:

Oh feed forward...

Get even fancier and use systems that swap out the P, I and D, based on the error and you can get some systems with such tight control it will scare you.

 

[Joesixpack]

No doubt.....or PID's controlling other PID's in series can put a sane man in restraints when it's run amuck.

 

[2007 5.9]

In what instance would you specifically want to adjust the PIDs??

 

[Amish Elegance]

Not feed forward control, Dan.

The PID function is related to feedback from the rail pressure sensor. Feedback.

I believe JSP is talking about the map being a feed forward type of data input. A very very good guestimate. Duty cycle at percent= pressure. Its a conversion that happens in the background behind the rail pressure table, only in reverse. The PID loop adjusts the duty cycle commanded in this prediction and smooths transition to ever changing set points.

Did I interpret that right, JSP?

Dan, where did you experience that control? I never knew you had a background in controls.

 

[Amish Elegance]

In what instance would you specifically want to adjust the PIDs??

Let me expound and add to his questions, if I may:

First off, on a stock ECM, can you?

Secondly, if your tables have some basis in reality, would you ever need to adjust PID's?

Edit: I cannot wait to have 03-04 EFI so I can get smarter about controls using my truck. "It's not playing! Its R+D." LOL

 

[2007 5.9]

Let me expound and add to his questions, if I may:

First off, on a stock ECM, can you?

Secondly, if your tables have some basis in reality, would you ever need to adjust PID's?

Edit: I cannot wait to have 03-04 EFI so I can get smarter about controls using my truck. "It's not playing! Its R+D." LOL

I will answer real quick...

1. Stock ECM..yes. We have PID control tabs within the pressure tables.

2. I am with the understanding that if your pressure tables are smooth and concise that PID control/manipulation is un-necessary.

 

[2007 5.9]

This is what our PID controls look like...and what we can adjust

 

[BFD99]

Wow there is some good info in this thread, I'm gonna have to sit down and read this here soon, so I can educated myself some more on this topic.

 

[zstroken]

Not feed forward control, Dan.

The PID function is related to feedback from the rail pressure sensor. Feedback.

I believe JSP is talking about the map being a feed forward type of data input. A very very good guestimate. Duty cycle at percent= pressure. Its a conversion that happens in the background behind the rail pressure table, only in reverse. The PID loop adjusts the duty cycle commanded in this prediction and smooths transition to ever changing set points.

Did I interpret that right, JSP?

Dan, where did you experience that control? I never knew you had a background in controls.

I am an electrical engineer. Used to do lots of motion control systems. Would do all sorts of tuning on servo positioning systems for gantries and machining lines. Actually held the title of controls engineer for a while, but there was very little control loop tuning in that position. It was more of the sequencing of controls. Lots of PLC, CNC, servo background.

I currently use them from time to time in tuning of boiler controls and pumps and/or fans. These are a lot more forgiving than the position type controllers.

What I laugh about is I had several controls classes in college. We spent all sorts of time graphing for the models of the control system etc. I haven't seen a model yet for a control system since being out of school. Usually don't take the time to make one either, as with a general understanding of the parameters changes, you can usually tune a system pretty well by just tweaking the parameters and noting the changes until you get the desired results.

 

[chance cobb]

cool

 

[triton]

The 36K is meaningless however as the PID is NEVER going to try a setpoint higher than the pressure limiter. No fool in his right mind would build a tune without this basic limit.

This likely being a box tune and from early limited understanding just simple max'd out the table with an arbitrary value as to ensure fuel pressure at the highest setpoint possible. So while a few folks are running around with their head cut off "36K!!!!! oh my god" all I see is a lazy tuner or with limited access to all the tables ensure max rail pressure setpoint. Hence my comment that 36k is as meaningless as 99k, its not going any higher than the set limit what ever it may be.

No one is running around going crazy because of 36K. The fact is though, it's sloppy a$$ tuning. When I started the thread, I wanted to point out how nice having EFI is because now, we don't have to blindly load tunes with their RP maxed out among other things.... whether it's going to get there or not. When a box tune gets loaded.... anyone just has to hope the person that wrote it was in their right mind.

I really appreciate the info this thread turned into. With Amish here now.... it's done went 5 star

 

[madmikeismad]

Wow there is some good info in this thread, I'm gonna have to sit down and read this here soon, so I can educated myself some more on this topic.

Phuck that. This thread makes me want to stick with 12 valves forever. Might even throw the stock lift pump back on LOL

Sent from my DROID3 using Tapatalk

 

[Amish Elegance]

This is what our PID controls look like...and what we can adjust

Wow. Thanks for sharing. I'm shocked that this is the first time PID's and EFI Live has been mentioned on the boards (that I've seen.) I gotta start clicking around with this stuff.

I will answer real quick...

1. Stock ECM..yes. We have PID control tabs within the pressure tables.

2. I am with the understanding that if your pressure tables are smooth and concise that PID control/manipulation is un-necessary.

2. PID control would still come into play as the ECM transitions from one value to another on the table. (Correct?) As your inputs move about the map/table, even with its numerous data points, there are still gaps. Even smoothed, there still aren't enough values present in the table to increment progressively enough, PID loops would still be necessary to control the transition from value(setpoint) to value(setpoint). Is this correct?

The assumption I made with my statement about being necessary(or not) to tweak these values was based upon use of linear steps in a map (as opposed to sharp changes) and responses from hardware that are similar to factory.

Where I see tweaking these might be beneficial would be an application with a modified CP3. Perhaps one could smooth the response of an FCA where the profile of the passageway has been modified? This may achieve smoother ramps in pressure and add driveability and consistency with the non linear area machined into the FCA groove. To my knowledge, this is not a trait highly modified pumps are known for.

What say you? (all)

 

[SHughes]

Phuck that. This thread makes me want to stick with 12 valves forever. Might even throw the stock lift pump back on LOL

Sent from my DROID3 using Tapatalk

it makes me wish that FAST would jump into this arena...or even have someone like TS (need a BIG budget and i want royalties Dennis. LMAO ) contract FAST to adapt this to the Diesel world.

 

[triton]

Wow. Thanks for sharing. I'm shocked that this is the first time PID's and EFI Live has been mentioned on the boards (that I've seen.) I gotta start clicking around with this stuff.

You can download that software for free at EFI's website. It's the full program, you just can save anything.

 

[DaveMatthews]

Holy chit my head hurts. And I'm gonna have to read it again to get any kind of understanding out of it...:bow: