Don't worry I don't take it as a criticism - it is interesting to discuss.

There are basic mathematical approaches for finding PID constants - the most basic being Zeigger Nicolls (sic?) but most process control engineers do it by trial and error I understand, but estimate first what they think will be required - this is my approach also. If I find it does not work on track I will need to adjust the constants.

The lowest time to reach target will always be achieved with an underdamped resonse where it overshoots undershoots then stabilises. I don't want significant overshoots or fluctuations so I set accordingly - there are undoubtedly small overshoots which the gauge is too slow to see - I will datalog eventually to find them.

It is all a compromise. I agree about the static + integral design, but it spools up too slowly, and when the system changes too much can easily cause fluctuations as the integral component destabilises a system easily.

Adding proportional gain improves the response time by ramping the duty cycle during spool up. Too much and you get bigger and bigger overboost spikes. Too little and it is slow. However, I found that by adding differential control to static + PI you can reduce spiking for the same rate of spool up. I wanted to keep it simple, but PID plus static gives the best results. I have tried static, static + I, static + P (spiky) +-I, static + D (slow!) +-I, static with PID. The latter is by far the best.

A maximum duty cycle spools you up quickly which is attained by P, the spike is controlled by D but preserving most of the response. Adding I removes offsets. D also keeps fluctuations caused by I in check.

You end up with a responsive but slickly damped system which drives really well. As brisk as a ball spring, but with better control than the OEM system.

There is a really good simulator of PID as an Excel spreadsheet which is actually an oven controller simulator (!) but will plot response curves and you can change PID and watch the results, and also see how well your system reacts to unexpected changes. PID is by far the best.

http://newton.ex.ac.uk/teaching/CDHW/Feedback/

I found it very useful to learn about PID and how real world problems can be solved. By the time I tried the controller on the car it was very easy by observing the behaviour of the system to work out what to change.

The other thing I found was the more control methods used (PID + static being four) the more robust the system is to unexpected changes whilst preserving control bandwidth.

I think the Apexi AVC-R uses static plus proportional gain, with adjustments based on gear. I don't know whether it uses I and D as well. For an easy to setup system the fewer the better, and if I did not think I needed four methods I would not use them.

[Edited by john banks - 2/25/2002 12:56:52 PM]

 

From what CC is saying it sounds like spool up and spiking will be more pronounced when the car is being caned around a track.

I can see why a bleed valve setup would really struggle - on many Subarus the held boost is closed loop (probably integration?) but the spool up isn't. On these cars unless the MAP sensor is modified a la Superchip, all you attain with a bleed valve is increased spiking which can give a modest improvement in performance. On track with heat this spiking could be worse resulting in detonation in an already hot engine with high intake temperatures. Is this a reasonable explanation?

If so, the solution with an EBC is to turn P down or D up. With a hyrbid MBC it would be to increase the bleed component and turn the ball-spring component down. With a bleed valve only producing overboost it is probably to take it off!

So I would estimate that from worst to best spike behaviour under high load we have:

1. bleed valve
2= Unichip boost controller
2= Hybrid or ball-spring MBC
4. factory boost controller
5. EBC with at least static, I and D, which would probably need P to be fast.

Am I talking garbage or does this make sense?

It also explains why when you setup any boost device you see the biggest spikes after a few full boost runs.

 

John/Dowser/CC,

this was one of the points I was trying to make - you are merely more articulate than I.

On track the boost does seem to rise anyway - with the additional factor of having the engine run at high rpm and being constantly booted - ergo overboost spikes in places you don't normally get on the road - certainly not for any sustained period of time -whereas on track this can be every press of the throttle. It is interesting having the ability to datalog and see the difference in behaviour of the engine on track - it will do things after five laps that you wouldn't see on the road - and after ten laps will behave differently again.

Of course if you don't really floor it on track then a) why be there, and b) you are driving around these effects.

Theo clearly understands this and had tuned his car appropriately!

I am not claiming in any way to understand the theory - I am reflecting what I have seen empirically. The worst combination probably being bleed valve/Superchip, but I have seen it with Unichip set ups with a couple of cars having to cool down after say five laps as the boost is suddenly spiking to 1.4bar, instead of 1.1bar - fortunately these guys have had boost guages. I saw it with my Link before I worked out what was going on.

It took me a long time to optimise a map for road and track - I 'think' I have done it now - the improvement being to significantly slow the speed the boost rises on track - this has effectively stopped overboost were I don't need it.

I guess this could be done with a Bleed Valve or a larger restrictor - but we are back in the land of knowing how to do it and what to look for before it is too late. Also, with the ECU or EBC approach this can be done relatively easily from the driving seat.

Finally - Dowser - scaremongering - no - just trying to take a reality check - if you look back over three years of Scoobynet Technical this debate, and the rather sad consquences has come up again and again - normally 'after the event'.

Trout

 

Richard, just to make myself a bit clearer....

What I noticed is:

Drive at 160 km/h for a long time : EGT 700°
Give it full throttle from there : EGT goes up (of course)
Go back to constant 160 km/h after that : EGT 650°

First few times I saw that I was a bit miffed ... but it makes sense in the way that at 160, the ECU is still in "closed loop" and will run reasonably lean (all things considered). A few good throttle applications will then dump in a lot of fuel, and the EGT's afterwards are actually lower.

Don't know if you noticed that my little theory is 180° opposed to what CC is saying ... not I'm not saying he's wrong, but the above I have seen quite a few times on my car.

Maybe the turbine needs to see quite a lot of higher EGT's before it actually gets to it's optimal temperature, and it's not getting there with just a few squirts (even though the actual EGT's will temporarily be 850°)

Your PPP will still be on the overfuelling side of things at full throttle, so *relatively* engine temps will certainly not suffer if you give full boost here & there. To make things even more complicated, it might well be that whilst giving half throttle, you are actually stressing the turbo more, as the boost in the intercooler will be a higher than in the manifold (due to closed throttle butterfly) with may make the turbo run way outside it's effenciency map ... confused ?

The Ring was not my best example BTW, as you have longer hi speed stretches which will bring down intercooler temps. A short circuit with a lot of braking/accelerating (like Zolder) will be much harder on the engine.

Trout, to be fair to John, he has warned (and so have I) quite a few times that you should know what you are doing. Personally, I think, despite the attention now being with bleed valves, that the information in Drivetrain these last few months is a lot more *balanced* than it ever was before. In fact, quite a lot of myths have been uncovered ('your UK car will run lean if you up the boost' etc...), which can only be good.

To me, going on track without at least boost gauge, oil temp, and knocklink (or similar), whatever your tuning, is - partly hindsight - less clever than using a sensibly setup Dawes.

Good discussion though.

Theo

 

I have this problem in medicine all the time. It is knowing that a problem is multifactorial and not understanding what all the factors are or suggesting some factors are in the equation which aren't and not knowing the balance between them all. Some can be measured, some can be theorised. At the end you are left with what some would call an "art" as it is not easily reducible to logic. Possibly the same with tuning hence those with experience sometimes intuitively "know" the right way to do things, even if it is based on a complex network of experience. So experience does count. So does someone to come and challenge all your sacred assumptions. It's the fun of any applied science/engineering I suppose.

 

nice to see a doctor who can peer into the realms of real science and ditch this biology crap they hold in such esteem.

 

[Tongue in cheek mode]Expressing uncertainty and saying something is multifactorial is a way of achieving two things in my book - saying you don't have a f**king clue whilst sounding clever. If done with confidence you usually get away with it. Unless there is a lawyer involved. Reductionist species that they are wanting everything black and white so that you win or lose[/Tongue in cheek mode]

Seriously though, depending on the relative proportion of factors involved the outcomes can be the complete opposite of what you expect if you have not included all the variables and there is another rogue variable which you haven't spotted which may have a complex interaction with the ones you can observe. Associations that look spurious are not and things which you think are true are not.

Sorry, how on earth did we get on to amateur philosophy of science when all we were doing was talking about bleed valves? Sorry my fault and Adam didn't help

[Edited by john banks - 2/25/2002 3:54:35 PM]

 

LOL !

Damn, rumbled

Theo

 

Adam M said, "my pectel has PID plus static boost control" in another thread.

Can you tell me any more about it and any info you have on how it does it and how to set it up (ie what variables there are) would be good inspiration for me! Or any pointers to finding out more MOST appreciated Adam. Glad to see others do it the same as me or is that the other way around (even though I got there not knowing any of them actually used this method) I take it it must be the best method then "officially" if Pectel do it? Intrigued...

This is actually quite a productive thread and it useful for me anyway. Thanks Trout. One of the more interesting threads recently...

 

John,

as I said it was started to provoke a debate that didn't look like flocking sheep! Look the sheep are flocking - run for the hills!

I am finding it very interesting - why do cars perform differently on track - and some set-ups differently to others. Where is Theo's EGT I ask myself if he gets 700 deg at 160kph! Or only 850 deg max - in his tailpipe

Anyway I am learning a lot - and being a geneticist by training I think that all multifactorial problems have a solutions (cos geneticists are like that) it is just that one brain isn't big enough to store it all. Need some form of interconnect!

Anyway, before we all head off into the muppet forum and lose the sensible dimension - the other comment that did occur to me was that for a mechanical device, bleed valve or restrictor to work for an all round car then it will be suboptimal. Which is the same for any form of BC of course - what I am alluding to is that a sharp boost rise on road which is *yummy* is totally unsuitable on track - and vice versa. I prefer to change this from the driving seat - but hey, personal preferences!

Trout

 

......off to soak my ECU in cats eyes, newts tails and the polished bones of an ancient dragon!

 

I think "yummy" is a fabulous description of good spool up.

 

>>Where is Theo's EGT I ask myself if he gets 700 deg at 160kph! Or only 850 deg max - in his tailpipe

I'm not sure if you were really asking ? If you were, in the uppipe, so you would need to add 20 to 30°s maybe for the max number. The car has a pretty rich setup.

Are your temps very different ?

If you weren't asking, I just made a fool out of myself

Theo

 

Theo,

those are fairly conservative figures - it much run fairly rich as you say. I run on the lean side on the road - much richer on track. Max egt 925deg on track - it is an STi however and from memory the max safe for an Sti is 955deg - need to check.

At 160kph on the road I am probably 750-775 - again in the uppipe.

Other downpipe EGT cars are getting up to 900deg on track.

Trout

 

Interesting temperatures, I use a K type thermocouple in the actual turbo volute chamber, high speed cruise gives 650 C and extended full throttle use, a max peak of 800 C. I don't think I'm running rich as lamda (downpipe) is steady at 840mv
My water/methanol injection drops max T by 50 deg C in addition to extinguishing the first amber on my Knocklink. Which can only be a good thing !

cc

 

CC - yeah part of me thinks the guages I am using are not that accurate - it is a DEFI as the temps are high in some areas - higher than I would expect or have seen elsewhere.

Anyway - this is a thread about bleed valves!

Trout

 

It is much more interesting than bleed valves

 

Guys, great thread!!

Slightly off subject ..pls forgive me but... someone mentioned LEDS's on their knock link..... assuming it is mounted under the intercooler on the block rising to bell housing.... what are the norms in terms of LED light up during hard acceleration and full spool in 5th at high speed???

I have been modifying my resrictor to get ideal peak with very little over shoot i.e +.1 Bar and rearely get even the first green LED lit. (Current 16.8psi peak 15.4 Held)

Sorry for changing tone
Cheers..

Julian

 

It depends where you have the sensitivity set. Also each engine/installation had it's own basic sound profile.
I set mine so that under known safe conditions (full throttle, cool charge, high rpm, 0.9 bar boost) the two greens are on.
When I 'up' the boost or the charge temp rises I see the first amber flicker, that's just my cars normal profile, anything above this and I get concerned.
On the odd occassion during set up, when it has knocked, all 5 lights came on momentarily
If you set the sensitivity as per the manual you will not miss a Knock !!

 

Thanks CC…

I have set max sensitivity and a week ago when I was having problematic over boost I was getting the full disco … transitional knock only ……. So no need for heart failure. Still … unwanted…. So I modified the bleed back to turbo and have a boost curve/profile I am happy with. As an added benefit only the first GREEN LED lights up occasionally under hard acceleration. No point changing the orifice further as the ecu is not hunting for its desired boost … I guess looking at the boost characteristics that it is 1.1 Bar so the peak under full throttle of 1.2 is close to ideal.

I just wanted to know what norms other saw on the KL…. Thanks anyway

 

Well Trout this has been interesting reading!
i bet theres alot of people running the dawes set up having second thoughts about running the "Yummy" high boost on a trackday.
its been a good pointer for me as i had considered the Dawes but have never been 100% with it, bit of a worrier methinks!
but to be honest i would rather have safety out on track for a good few laps than be enjoying the yummy spool up out on the road occasionaly.
im only hoping for 3 dyas on track this year but i know the car could go pop at anytime but anything to prevent that must be good. my car is very standard compared to you guys, just breathing mods, i.e filter and exhaust, but reading this makes me think i will wait till i have enough pennys and tune the car so i have a little more piece of mind as the Dawes thing is getting like a snowball rolling down a hill at the moment.
Yep it sounds great but i suppose once the trackday season gets underway in full swing i suppose it wont be long before we hear about a few letting go with this set up this summer. hope not but this thread has certainly made me think!

Lee