Implementing a form of closed loop boost control on Wolf v500 ECUs

[OnAll-FOUR]

It deosn't seem like many people run v500 ECUs on here, but I thought this might be handy info for anybody using one.

I also wanted to get comments from anyone that can remember their University control theory to make sure I haven't implemented something with an inherent instability.

Situation:
Wolf in their wisdom provides boost control mapped against RPM but it is not a closed loop control. While this is fine if configured right, any change to exhaust, or other engine factors will effect the boot performance. It is also harder to get the turbo spooling quicker because your just opening the wastegate based on RPM not Boost.

While doing some data logging on my daily commute to work, I had a light bulb moment and realised that a type of closed loop system could be implemented in the features that Wolf provides.

Implementation:
Below is a screen shot of the boost control configuration screen, with duty cycle Vs RPM. This is the base configuration which in my case was setup to provide Approx 14psi.

On the left you can see a range of modifiers they provide such as air temp, engine temp and throttle position which are all handy. Then there is the "Special Trim" where you can choose what signal you map in the last modifier. One of the options is the Boost, psi value.



Then with boost PSI selected, you can map a +/- percentage to modify the base table by based on the boost. Since I'm aiming for 14psi I have it giving and extra 10% until it reaches 10psi, then it ramps down to 0% at 14psi and then through to -10% at 18PSI.

To give an idea, previous testing showed a decrease of 15% duty, dropped the boost to 8-9 Psi from 13-14.



Result:
The seat of my pants dyno tells me that this made a massive difference to the turbo spool, and comparing these data logs you can see the closed loop control adjusting the boost solenoid duty cycle in the second graph.

BEFORE - Open Loop Log: You can see the duty cycle move with RPM as mapped


AFTER - Closed Loop Log: here you can see the duty cycle start to decrease as it approaches the preset 12PSI. Closed Loop Control!!


This is were it all gets a little control theory.

I think what I have implemented here is a form of differential controller (integrator ?).
The greater slope I put on the special trim, the greater I increase the gain on the output of the differential controller. How far should I go with this? The greater the gain the faster the boost response but I believe greater chance of instability and boost overshoot? I guess trial and error.

By limiting this to +/-10% for now, I hope to avoid some of this instability but that now make is it a non linear system. Again is this going to give me stability issues?

Might be time to crack out the Uni text books

[timbosaurus]

I switch off engineering mode when I play with cars

I'm using a V500 though, so might have a think about it when I'm bored

[OnAll-FOUR]

I switch off engineering mode when I play with cars

I'm using a V500 though, so might have a think about it when I'm bored

Okay fair enough. the write up was a little wordy, that is also why I included pictures.

This is from this morning. Changed the set point to 12PSI and you can see it operate arround that point on the last part of the graph so it is working. But..

Itis kinda hard to know how well it is working. It is adjusting the boost control solenoid, but I would not sat that the boost is holding smooth. Comparing it with previous logs I don't think it has incresed the noise, but I don't think it has decreased it either. There also still appears to be a dead point arround 4500 which I was hoping the closed loop wuold filter out but it doesn't appear so.

[RAAAH23]

im not educated nor have had any experience although i do understand the concept
very interesting to someone that hasnt had any experience.

[Hen]

To me it sounds like you are on the right track. The steeper the slope of your MAP based "special trim" line the faster the control. Too steep and you'll get oscillations, too shallow and the response will be slow. I think it'll take some time to get the +/- % values right and the slope right, and hopefully it's not the sort of thing which'll suit one day but not the next.

But I'm also amazed that the v500 doesn't just let you control the wastegate based on MAP just like an EBC. Maybe there's some advantages I'm not clued up enough to know.

Hen

[CHe]

Talk to Robbie at Wolf. You get free support on there current products.

They helped me heaps with setting up/troubleshooting my v500

[OnAll-FOUR]

Talk to Robbie at Wolf. You get free support on there current products.

They helped me heaps with setting up/troubleshooting my v500

I have heard some people say that but when ever I have tried calling them they have been very brief with me on the phone and their email replies have been less than desirable.

I have sent another email with similar screen shots and questions to them, but no reply as yet. I might try another phone call as well, but I'm off overseas in a few days so I will follow it up further when I get back.

To me it sounds like you are on the right track. The steeper the slope of your MAP based "special trim" line the faster the control. Too steep and you'll get oscillations, too shallow and the response will be slow. I think it'll take some time to get the +/- % values right and the slope right, and hopefully it's not the sort of thing which'll suit one day but not the next.

But I'm also amazed that the v500 doesn't just let you control the wastegate based on MAP just like an EBC. Maybe there's some advantages I'm not clued up enough to know.

Hen

cheers, I was a little surprised as well. I bought it thinking it had "Boost Control" I just didn't realise it was open loop. Maybe there is a benefit to the open loop design but I'm not aware of them either. Anybody care to enlighten us?

[MWP]

Its not open loop (like the Jaycar item), but its also not a good closed loop controller.

There is feedback to the boost controller (the Wolf ECU) as it can monitor pressure, but its just relying on a dumb map to make the corrections (PWM to the bleed valve).
If conditions change, like a clogged bleed valve, weaker wastegate spring from heat, etc, the Wolf wont be able to compensate for this properly... a proper boost controller will.

[OnAll-FOUR]

Bugger, I think I have found a fundemental flaw which will cause oscillations in this setup. It will only work when the initial RPMvsDuty cycle table is properly setup.

For example when it reaches the desired boost level, the special trim table will apply 0 trim to the RPMvsDuty table. But if this table is incorrect for any reason then the boost will start to increase/decrease because of it. As it does this then the special trim table will start to reapply a trim to the boost level and correct for this again. As it again approaches the desired boost level the trim will go back to 0 and the whole process will repeat. You might be able to make this all happen so fast that you wont see it(put a larg gain in), but I don't think it is really achieving proper control of the boost level.

As MWP was saying if any factors such as the waste gate spring change then this will not be able to accommodate it like a proper EBC.

What I think it can do though is improve on just the basic RPMVsDutyVsDuty table performance by still using the table, by having a flat region arround the desired boost. That way it will help the spool when it is starting to build but when it gets closer to the desired boost level, the flat part to the Special trim will mean that in tha region it is just working off the RPMvsDuty table and hopefully wont induce any oscillations.

Either way, looks like a proper EBC is on the christmas shopping list

Cheers guys

[Hen]

Bugger, I think I have found a fundemental flaw...

For example when it reaches the desired boost level, the special trim table will apply 0 trim to the RPMvsDuty table. But if this table is incorrect for any reason then the boost will start to increase/decrease because of it. As it does this then the special trim table will start to reapply a trim to the boost level and correct for this again. As it again approaches the desired boost level the trim will go back to 0 and the whole process will repeat. You might be able to make this all happen so fast that you wont see it(put a larg gain in), but I don't think it is really achieving proper control of the boost level.

Nope, what you've described there is no problem. It's simply lag control (P, PI, PID, etc), which a vast majority of CNC machines, and probably most other (not hugely advanced) control systems use.

If you were aiming for 14psi, maybe given the RPM vs duty cycle it'd wander up to 14.5psi before the "special trim" table reduced duty enough so that it'd continue to produce 14.5psi for ever. Bear in mind though that there is a big difference between steady state stuff (what we're talking about here) and what actually happens in a car, where all these things change quickly.

And remember it is too high a gain which will cause oscillations, not too low.

Hen