Single Vanos BMW S50 (Euro)

[Johan E]

Hello,

Is this a reasonable result when it comes to controlling the Vanos? It seems like the control works fine, but does not exactly reach the target at low rpm, and then seems to be a bit over the target after 4000 rpm. Is there anything anything i could try to change in the configuration to get closer?

[atlex]

Have you dared to touch the PID setup ? I'd bump the P up a bit (say 20% of what it is now) and see how it behaves. It's also worth logging oil pressure over this time as there are really three factors - ECU measuring the sensor/applying PID to drive drive the solenoid - Solenoid itself and the Oil Pressure being let through to drive the VVT sprocket. Obviously oil pressure starts low and goes up which may account for some factor of the undershoot/overshoot, but a higher P in the PID may help correct it quicker...

[Adamw]

@atlex just for your interest, the BMW VANOS systems that have the 2 solenoids per cam are what BMW call "high pressure VANOS", they have their own separate oil pump driven from the cam chain that runs at something like 100bar.  I've never seen anyone test how much the pressure varies with RPM or temperature, but I suspect it is probably much less variable than a more typical VVT system that uses normal engine oil pressure.  

@Johan E I have a similar suggestion to Atlex, I think this just needs some PID tuning.  The first thing I would suggest is to go down the bottom of the ECU settings tree, go to the PID Setup item and set this to Inlet bank 1, we can then log the PID terms that are being used.  In the VVT settings, change the PID Setup to custom, go to Inlet bank 1 settings and increase the filter to about 4-6, this should improve the noise in the position signal. 

For tuning the PID, this is how I would usually do it:

1. Assuming CL idle is set up, change the idle target to about 1500 so it is running smooth.
2. We want to do "step testing", by this I mean abruptly stepping between 2 VVT targets, I suggest 10 & 25deg as it will usually still idle happy at that.  I like to do this by temporarily setting up some sort of switch condition on the VVT table axis, it depends what the car has available but for example if I have a high/low boost switch on the dash I set up like below (you can export your existing table to save it)  
3. Now while watching a timeplot of cam position Vs target (similar to what you have setup in your screenshot above), you can step between the 2 different targets, adjusting P, I or D one at a time to see the effect on response.  Im not going to explain PID theory here in detail as there are hundreds of videos on youtube etc.  But the basic idea is generally to increase P as much as you can, when you have a fast oscilation you have gone too far.  Generally if position does just one or two bounces over/under target when you do a step change in target your P is about in the ballpark.  You can increase D to dampen that bounce out.  Increase I so any small error is removed as quick as poss.  If you have too much I you will usually get a slow rolling oscillation.  The tricky part is P, I & D all do influence each other, and they can cause similar effects in some cases, for example too much I, too much P or not enough D can all cause a similar looking overshoot, so sometimes it helps to look at those logged PID terms to see what is causing what.

Looking at your pic above I would say you need more P and more I, D looks ok but often when you increase P you will also need to increase D.      

    

[Johan E]

Hello,

Thanks both for your replies, really appreciate it. I'll do proper testing as soon as I can and we'll see what this leads to.

FYI: I just switched from an old G4 to a G4X and can see how much more precisely everything seems to be controlled. It probably depends on several things, but overall you can really see the development of the products and software. When it comes to the Vanos control, I got everything working with the old box, but once I get everything right in the new one everything will surely be even better.

[Johan E]

...as I said, I'll do proper testing soon, but could see that the default settings - when I enabled "custom" - looked like this. Maybe it's because no default values are entered? I mean, it looked a little weird with zeros everywhere. Should I start from this and increase to, for example, "1" for "P" etc.?

 

 

 

[Adamw]

I would say you are looking at a map offline so any new settings that need to be pulled from the firmware will be vacant.  Change it to custom when online and you will see the value that is currently being used in the firmware. 

[Johan E]

Yes Adam, that was from my most recent map offline, but when online the settings shown are the same (below) when i turn to "custom".

[Vaughan]

P is 5.0, I is 0.3 and D is 9 in the single Vanos default PID

[Johan E]

Thanks Vaughan, I did some testing last night av ended up with 9, 0.3 and 4. Everything works much better now, but will do som final tuning and testing with the the default values to. This is what my last log showed:

[Adamw]

Did you increase the filter?  It still looks a bit more erratic than I would usually expect.  And I think it will still take more Integral yet.  

[Johan E]

Adam, forgot to increase the filtering - will test that too. Agree that the regulation looks a bit erratic and will also work further with Integral. Although it feels a bit difficult to tune in PID, it feels interesting and important to understand for other purposes as well.

Just FYI, I got Long Term to work really well with the logic rules via Math. The knock system also works very well now and has provided the opportunity to fine-tune via the individual ignition correction. 

Thanks for the feedback.