Cosie Convert

I, like many others, have been keen to engineer 'out' the overboost peaks associated with running higher boost pressures.

I'm now having a re-think

Points for consideration/comment -

1 - Momentary overboost of say 3 psi for 1 - 2 seconds gives the car a much more powerfull 'feel' through the gears or when starting to overtake.

2 - If the overboost is only held for 1 - 2 seconds then heat build up in the charge air is minimal due to the heat soak effect of the intercooler.

3 - Engine internal temperatures and EGTs will not have time to increase assuming adequate fueling (not normally a problem)

4 - Some manufacturers build in this facility (Escort Cosworth)

I am now running with 24 psi overboost and 21 held. The car feels faster and has less requrement for the water injection ( read less knock) than it did running 22 psi peak and held.

I wonder if we have become blinkered to accomodating the standard boost cut out ? Perhaps an extension to the timed overboost allowance (As per later MYs and PPP) is the way to go ?

Dare I say even engineer 'in' the peaks in boost

Andy

john banks

I have been coming to the same conclusion, and the EBC lends itself to a so-called "scramble boost" approach or shorter peaks, and the favourable effect on temperatures appeals.

The only reason I have wanted it spike free is as you say to pander to the fuel cut. For road use a MY00 PPP decat seems happy at 18.5 PSI held, but over 19.3 held and you get fuel cut. Sometimes the length of peaks is variable, and it seems a bit unpredictable as to where/when it cuts.

I may experiment a bit when I put the fuel cut lifter and EBC on. When setting the EBC up before with 18 PSI target, putting in a bit "too much" proportional gain made spool up very brisk indeed but it seemed to fuel cut with only a momentary 20 PSI which puzzled me.

[Edited by john banks - 3/23/2002 11:44:45 AM]

Cosie Convert

Damn ! I was looking for an arguement

Perhaps the later MYs boost cut out is a combination boost/time function ?
I know on my 95WRX the fuel cut is instant !
The needle on my boost gauge never even got time to climb above 10 psi before cutting when my FCD valve stuck shut recently !

SCOSaltire

assuming that the controller is PID then...

the best setup for this type of controller is a small overshoot and then smaller undershoot.. and then level..

PID controllers have been in use for along time... ive used them on controlling pendulums upside down (to stop falling over), on model crane to stop the load swinging, on pneumatic valves to control pistons... and some more..

so.. i can see that a slight peak is boost is desirable.
controlling the size of this peak, the wavelength of the peak (how long its there and takes to decay) and making it fit in with the fuel cut limit is the key...

you want a slight overshoot.. because this leads to a more rapid increase in boost...

to stop the overshoot.. ud have to be approaching the peak slower and hence wont reach your boost level in the same time...

but... i dont know about cars that much... just the science and engineering behind PID controllers...



[Edited by SCOSaltire - 3/23/2002 12:19:52 PM]

john banks

Don't laugh, but there is a nice Excel spreadsheet that has an "Oven simulator" into which you can put different PID constants and it plots the result. It is part of a University of Exeter PID course. http://newton.ex.ac.uk/teaching/CDHW...ck/OVENCTL.XLW

Go to the simulation tab, change Start Ts to 0 deg C, Pmax to 40000, and then the setpoint is 175 which we can imagine to be say 17.5 PSI.

This is how I learned about PID by fiddling with this so I could recognise what to change with the boost controller on the road. Sounds daft, but it was surprisingly intuitive once I got to know this model.

For instance for a really well controlled response put the Proportion to 300 in the model and see the result.

Try 750 for a peaky response. You can also fiddle with D. More D damps it, less D makes it peaky.

In fact, fiddling with this again I think I will just set D and I and allow P to be changed with a pot. Changing from 100 to 1000 gives quite a nice range of variation. Another pot will control duty cycle/target boost from 0-80%/8-20 PSI held or thereabouts.

[Edited by john banks - 3/23/2002 2:31:17 PM]

Cosie Convert

John

I've not played with the PID sim but I have control systems at work which I experiment on regularly
If I was designing in overboost I would use a very low 'P', this will minimise the boost build time but will cause an overshoot. Then adjust 'P' to achieve desired overshoot.

'I' should then be set to adjust the time taken to pull back the overshoot.

The hunting, typical of low 'P', would probably be best controlled by an orifice in the supply line.

As the main purpose of 'D' is to anticipate and control overshoot I don't think it's required for this type of control.


Of course, you could always just fit a cheapo bleed valve instead !

john banks

I have been worrying my local backroads tonight with the EBC in-situ

The way I have the equations a high P give more peak and faster spool. I tried 17 held with a 19-20 peak and it felt at least as good as 18 held.

I have just posted some findings in the EBC thread.

I have found that by putting D in I can actually improve the speed of spool up by using more P with the same overshoot, and the code is done, so it stays. I found the same on the plotter too - time to target is shorter with P and D than just P alone with the same peak.

Now I have felt this in action I wouldn't touch a bleed valve with a barge pole. The quality of the drive and the sheer speed of spool up is vastly superior. I can emulate a bleed valve by turning PID off - there is a definite push when it comes in at WOT and that is beyond running a duty cycle that would get you up to held boost with a small peak.

I am using a fairly small orifice in the supply line - even then it only adjusts from 8 to 18.5 PSI from 0 to 86% duty cycle, so I think it is required with the limited flow of the factory solenoid.

[Edited by john banks - 3/23/2002 11:27:17 PM]

Cosie Convert

Strange as it may sound I think we are saying the same thing when you increase P and I say lower P !! I mean P operational over a smaller range of measured value and I assume you mean increased, as in effect/output, which is the same thing. (Please don't quote this statement anyone )
I can see now how D would assist by allowing you to adjust P further.
Do you use any 'I' at all for fine trimming ?

[Edited by Cosie Convert - 3/23/2002 11:41:29 PM]

R19KET

Good topic, and part of the answer depends on how the ecu reacts to the spike: EG, is it retarding the ignition, and by how much, to compensate for the spiking.

Some of us have been "mapping" in some overboost for some time now, and just how good the results are, vary depending on the ecu.

The stock ecu seems to cope pretty well, within it's limitations, and MY's. The later ecu's are better, because they would look for the knock limit, whilst the earlier ecu's took forever to re-advance. You guys could check this with a "select monitor".

Same goes for the earlier 6 row Links(top row is 1bar +). If you map the ignition to a "peak held" boost level, but then induce overboost, the ecu will retard the ignition. The alternative was to reduce the mapped ignition to avoid getting det'.

The later 10 row Links are better, depending on what boost you're running, how much overboost you run, and where it falls in the map. If you're lucky, the "overboost" would take you into the next mapped row, so you could run the max ignition in the "boost held" row, and then map the ignition in the higher row to suit the overboost. Same applies for the fueling.

With the more advanced ecu's, such as the Motec, and Pectel, the whole map is user defined, so you can optomise each row, to suit the boost you want to run.

CC: I'm curious to know if your engine is stock, or not. Or are you just a "lunatic"

Mark. another lunatic, running 2bar +

SCOSaltire

the new ford TDci engines (diesil i know) have a built in boost spiking mechanism...

transient overboost i think.. or something like that...

Cosie Convert

Mark

My car has stock internals, turbo and ECU. It is a MY95 WRX RA and therefore 'perhaps' has a closed deck block ?

I have installed my own parallel feed fuel rail with FPR and uprated fuel pump, 3" exhaust system, bigger bore turbo intake pipework and K&N cone. Twin injector water/Methanol injection system, Boost control valve and FCD. Big bore turbo to I/C hose.

I have run up to 26psi (1.8bar) midrange for the 1/4 mile runs, a bit less for the road (25.9 )

Good results, no knock, max EGT 800C. Unconventionaly low CO of 5.5% Best 1/4 time = 12.3 @ 115mph. 0 - 100 = 9.6sec from AP22 (backed up by drag strip timing gear)

Increased breathing and more fuel mods in progress !

So, yes I guess I'm a lunatic too

Any performance figures for your car ?

Edited to say I don't allow the knock sensor to communicate with the ECU, the FCD also stops the ECU seeing more than 15 psi boost. So I always run with maximum advance........ which is nice

Water injection requires more advance so it all fits together rather nicely


[Edited by Cosie Convert - 3/24/2002 6:24:52 PM]

R19KET

CC,

26psi + 5.5%Co + 800C don't really add up, from my experiences, even running WI.

Where are you measuring the EGT's, and personally, I wouldn't run as low as 5.5%, that's pretty lean.

Have you actually checked your car out using det' cans, or do you run a Knock Link. Even then, I'd double check it.

Without looking at it, it's hard to say if it's closed deck or not. Subaru are pretty inconsistant.

Anyway, your figures are pretty impressive. Any RR figures for it.

Mark.

john banks

5.0-5.5% CO or about 12.6-12.7:1 is supposed to be the theoretical point for maximum power if you can control temperatures?

CC's Knocklink shuts up when the WI comes on. We are working on triggering the WI in between the last green first yellow Knocklink LEDs.

Our local rolling road's TMIC fan seems woefully inadequate to be able to safely run highly tuned cars IMHO.

R19KET

John,

Erring on the side of caution, on turbo cars, I was advised to run 1%CO per 1k rpm, +/- a little.

Couple of things to look out for, are, the lambda sensors have a habit of going wrong, often still reading, but inaccurately. I'm on my 7th, I think...

If you're using the stainless steel K sensors, such as those available from RS, they don't like the heat cycling, associated with cars, and have a tendency to break down. Again, often they will still read something.

I don't know where you're taking your egt's from, but under high boost, there's roughly a 200C drop, across the turbo.

I know you guys are pretty hot, and may know these points, but just incase.

Mark.

Cosie Convert

Thanks for the info Mark

I use a Knocklink bolted on to the recognised position. It's set at max sensitivity. At low boost, I get one or two greens, at high boost I also get one or two greens until the charge temp builds then I can get a fairly quick transition up through the ambers. Triggering the Water injection takes the signal back to normally one green I have not used det cans but agree that it would be a good idea just for piece of mind.

The Methanol content in the W/Inj helps richen the mixture but this is not detected by the Lamda probe, in fact it shows weaker.

My EGT probe (K Type thermocouple) is inserted in a hole drilled into the turbo inlet casting. My experience of these probes in an industrial combustion process is that they either emit the voltage (dictated by the dissimilar metals) perfectly or fail on open circuit. Nothing in between.
The displays can sometimes show random temperatures when the probes fails on open circuit, this is due to the very low voltages being measured and is normally very obvious.

My 80 MPH cruise EGT has remained constant throughout my modifications at 625 deg C so I'm fairly confident my readings are representative.

As John states, I have not run on the rollers due to poor TMIC cooling at our local RR but once I get my custom charge cooler built I'll get it checked

GavinP

CC,

I have a Knocklink on my car and I rarely see amber.... ?
Installed in the same place at max sensitivity - at much lower boost (1.2bar max).

Running with the Dawes and Link, even 1 bar @ 3000rpm doesn't light it up and you see nothing beyond the first green until 5500rpm+ where engine noise is a factor.

AFAIK all WRX engines before 96 were closed deck. Being a RA, it should be even stronger but 26psi would make me nervous personally.

Let me know how you get on with your chargecooler project - I'd be interested to hear how it compares to mine.

Thanks

Gavin

R19KET

There is a big difference between a "WRX", and a "WRX STi", regardless of it being an RA.

Jap WRX's, have "cast" pistons, and a 7500rpm red line (7750 cut).

STi's have "forged" pistons, an 8000rpm red line (8250 cut).

"Most" of the early STi RA's were closed deck, but not all. I have a '98 WRX RA engine (used for spares), and it was open deck, with cast pistons.

The surface texture of a closed deck block is pretty rough, because it's sand cast. Open deck blocks have a much smoother texture.

Subaru are notorious for having "mixed parts bins" on their earlier cars, and regardless of what the spec' should be, it's often pot luck.

IIRC, Pat's STi 2, was OD, with cast pistons !!!!!!! but then again, he's destroyed SO MANY engines, it may have been from one of the others

Mark.

Cosie Convert

Gavin

That's why I'm fitting a chargecooler, the std intercooler is totally inadequate for my level of boost, it survives for about 10 seconds WOT due to it's heat soak abilities but then it all gets too hot and det sets in. Your findings were part of my research on this topic I'll let you know how I get on

Mark

Is it possible to tell from engine numbers if it is OD or CD ? Need to go and check out my block texture now Hope nobody see's me stroking it

R19KET

CC,

I've yet to find any definitive way to check, other than removing a head, oh, and of course a "stroke", but be gentle

I'm also interested to know why you chose the CC (charge cooler) route, over a FM. Was it purely cost, or is it considered to be more efficient ?. I've spoken to a few people about it, and some rate them very highly, whilst others say they are fine up until when the water gets very hot.

I'm curious, because Subaru used a CC on the old Legacy's, but dropped them in favour of an IC.

Mark.

Cosie Convert

I don't like the excessive pipework associated with a FMIC. Note I didn't say anything about lag don't want to go upsetting peeps Oh and then there's cost £££££££ of the FMIC and I have a donor I/C waiting to be butchered and I like a challenge and I like trying new ideas .....blah blah blah

Charge coolers are popular in the USA/Auz on high boost cars. As you say the water temp control is critical. IMHO most factory set ups have inadequate pumping rates and radiators, relying on the engine not being on boost for too long in order to catch up with the cooling, or storing a large volume of water in the system to compensate.

That's why I'm intending fitting a high flow pump, large diameter water lines (32mm) and a small car radiator (typically the size of a FMIC ie nissan or corsa).
My aim is to keep the circulating water as cool as possible, target ambient +3 or 4 degrees.

From an engine airflow point of view the standard TMIC is very efficient.
The pressure drop is barely measurable. I have seen pressure drops of .08 bar quoted for an APS FMIC and that is probably the best one on the market.
I'd need to get my calc out but 1psi is worth approx 8 deg C in turbo discharge temp. Additional benefits come from reduced exhaust back pressure.

That's the theory anyway.

ps when it doesn't work I'll rotate the throttle body and make a FMIC

[Edited by Cosie Convert - 3/26/2002 1:52:32 AM]

dowser

Dragging this back on topic a bit () - with a my00 decat'ed PPP & OE boost control I used to peak to 1.3 and then oscillate wildly while the ecu tried to get things under control.

There was a noticeable drop in performance at the 1.3bar peak, which I always took for the ignition being retarded. But I thought ignition was retarded only when the ecu detects knock from it's sensor (but my knocklink registered nothing)?

However, I now run a Dawes (2mm bleed) which also hits 1.3bar in the cold (1.2 when ambient is above about 8 degrees) - performance is not affected at 1.3 (except the obvious ).

I'd like to understand how the ecu deals with fuel cut and ignition retardation - it seems to me to that the ecu must be using some form of 'rate of boost change' parameter to decide when to retard ignition as well as input from knock sensor?

Or was my OE boost control 1.3bar performance drop caused by something else? Engine still ran perfectly smooth - it just lost a bit of it's urgency for a split-second as the boost hit 1.3 (afterwards the oscillation stuffed everything up anyway - but there was a definite drop before the oscillations, while the boost was still rising).

Another my99 standard ecu car I know uses (used, he's now on a Dawes too) bleed and FCD - he was getting 1.5 peak and 1.2 held. He says there were no problems with this...but also that the Dawes gives stronger performance set at 1.2bar with no peak. Prior to this thread I would have said he was suffering the same as my PPP'ed car...now I'm more confused than usual?!

Richard

Cosie Convert

Richard

There will be an optimum boost level from your turbo, exceeding it will cause the exhaust backpressure to rise and reduce the efficiency of combustion. With the small exhaust turbine on the TD04 I think it is quite easy to start causing a restriction, dropping power, even although boost is higher.
I wonder if 1.2 bar is the 'sweet spot' for MY00's ?

It could be that you had excessive backpressure whilst the boost was still rising as the wastegate would be closed.

I have a noticable 'dead zone' on my car, if I apply full throttle in 5th at low revs (2000) as the boost climbs, the power only increases slightly. Once the boost hits target and the wastegate opens, dumping some of the exhaust pressure (which had been held back to spool the turbo)the power comes in hard.

This is the price I pay for a MBC. If i had an EBC I could set it to start opening the wastegate earlier, turbo spool up would be slightly slower but power delivery would be more progressive.

Does this sound similar your problem ?

dowser

CC - I kind of agree with the theory (a recent rolling road operator also suggested boost was in the 'sweet-spot' for the td04 and that any more might increase power but lose torque). However, in cold weather with the Dawes I also hit 1.3bar and there's no noticeable performance drop.

It's even held a bit longer with the Dawes before tailing off (from 2.8k up to 4k). With OE boost control (but with slightly oversize restrictor after lots of experimentation) a steady throttle opening would stop the peak of 1.3 and oscillations. It would still hit 1.25 briefly, but then wouldn't undershoot b4 hitting target (about 1.15>1.2 bar).

The performance hit only appeared when the oscillations were imminent - but they occured just b4 the 1.3 peak. Either the ecu was detecting a rising boost rate that was too high and was trying to protect the engine....or the performance hit was somehow causing the oscillations.

Someone's suggested in another thread that fuel-cut on 99/00 cars may also be monitoring 'speed of boost increase', rather than boost level....dunno', but would like to!

Even stranger, while testing with a view to altering the ecu map (), I removed the Dawes (had been on 4 weeks) and refitted my OE boost control pipework *with* a standard sized restrictor. This was on a cold day & I would have bet money on it causing wild oscillations. It didn't - boost control was very good with a slight peak b4 hitting target. Better than b4 fitting my Dawes, with the 'optimum' sized restrictor.

I should have tried resetting the ecu just to see if it made any difference - but didn't think about it until afterwards!

My understanding (learned from others!) is that the ecu map guesses a duty rate required to meet a fixed target. I wonder what parameters it bases it's guesses on (throttle/load/maf/map)? And why would running an MBC with the OE solenoid blanked off for a month change boost characteristics when reconnecting the solenoid? Even if the ecu is able to 'learn' the best duty rates - any info it has with the MBC on is totally fake...... I'm going to have to try again after resetting the ecu I think...

Richard

FOZ STiV5

Good thread all,

CC,

Just wanted to make a comment about the Knock Link. You and others have said you get green leds ON. I have just had a new ECU mapped up and during mapping my mapper said he wanted to retard the timing. I said WHY???????? there were no leds lit (not even green) then he said pop the headphones on so I did. Sure enough there was det... faint but still det. This maybe transitional .... but personally I would rather have no leds lit at all.

My KL was mounted on the block and was working fine.... it used to light up with my std STiV ECU.

P.S I had a max inlet temp of 32 deg C with a STiTMIC & new splitter I designed. (Range on day 14 ~ 32 Deg C)

Later all,

Julian

R19KET

Julian,

Out of curiosity, was the sensitivity set to max ?

Mark.

Cosie Convert

Mmmm I still get one or two greens even after i come off the power as long as the revs are 5k+. Can't be det then

Can you make det cans ? (penny pinching again ) What do they comprise of ? Why cans and not just can (single) ?

john banks

http://home.netcom.com/~bsundahl/kno...nockSounds.htm

Cosie Convert

And all for less than $10 again

thanks

FOZ STiV5

Yep sensitivity set to max!!! and I tried a couple of mounting positions. One mounted directly under IC. However onto the inlet RHS bolts as commonly used was exactly the same in terms of sensitivity.

Julian

FOZ STiV5

Great LINK John,

Well worth doing guys if you are unsure of Knock. I was until my mapper helped me out. Still feel a little sad that I lost 1.5 Deg of advance as a result. Still ...I console myself that my engine is happy even I am not. Don't get me wrong the motor goes well...but it could be even sharper

Never happy enough !!! how much power is enough ????????????

Later all

Julian