Subaru Impreza GC8 & RS Forum & Community banner

Homebuilt GC8 for Time Attack

63K views 269 replies 24 participants last post by  bue car 
#1 · (Edited)
Hi!
Always wanted to start a build thread but never had enough time or interesting things to write about. Now looking back I thought that I have something to share, some ideas for this season in local racing series, and it will be good to start a journal just for my own record and as a motivation to move forward.

Short background story.

It all started in 2013, by that time I already owned Impreza STi type-R for couple of years and found myself a bit confused with it. Here I need to mention that I live in Cyprus. It's rather small island in Mediterranean Sea :) Cars are RHD here and before country joined EU lots of JDM cars were imported. So, sometimes you can spot quite rare japaneese cars here. When I saw my Type-R I was amazed because in Europe you rarely could see 2-door Impreza. At that point I wasn't even aware of existence of these models, I thought that type-R sticker is just a bad joke of some honda fan. Also, I wasn't big subaru fan at that time. Imprezas were just much cheaper than similar vintage EVOs.
So, after owning this car for some time I was confused. It was old (1997), poor market value, not very reliable but eating money really fast. Insurance was expensive, fuel costs were crazy... Then I ran into head gasket problem while trying to sort out mess with vac routing. I had engine rebuilt only to get a dropped valve 6 month later (the next week after I got car mapped after prolonged break-in period). So, I just put in used engine from salvage yard and thought "what to do next?"
I wanted to mod the car but I kept questioning myself, why do I need to do that? Well, at that point I ran onto local time attack series and it was perfect. Owning this car become full of sense in one moment and participation in "Street" class with strict regulations promised to save me from expensive mods. Basically I had to keep engine stock and no lightweight parts were allowed, so I could only get suspension. I was happy with this because I wasn't excited in modifying good and quite a rare car extensively. I even bought DZ03G tyres in size that fits without any modifications like arches rolling.
On local track where all time attack events are held I got lap times around 1:18 after I changed some shitty lowering springs for stock sti ones and I managed to get nto 1:15.xx on the first event I took part with coilovers and group N bush kit installed. It was mid-season 2013 and I even managed to win that event in my class although I knew that there's one EVO 6 lapping the track in low 1:14.xx, he just wasn't taking part in that event.

Then my Dunlops arrived. That was the day when I realized that tires are probably the most important single thing that will define your lap time. This time fast EVO came to the event and did low 1:14.xx lap again. I managed get to 1:13.xx. (he had same tires btw ;)) It was a shock, for me in the first place. The next event was on the next day and some guys came to check the car asking if turbo was stock and so on, because such time for street class car was rather surprising :)

Long story short, this car took me to second place overall in 2013, and win in 2014 in AWD (Unmodified) Street class. The car won every race it participated in. In 2014 I managed to set fastest unmodified street awd class time ever and if I recall it right - second fastest time among all street cars time loosing more than 0.5 sec to GDB from senior "Street Modied" class (engine mods allowed + removing aircon and stuff) and winning just 0.03 sec from Corvette ZR-1 in third place.

This is how it looks like




And this is one of the most fun videos - free ptactice session before final 2014 event, in wet

2014 Time Attack 5 Practice:


At the end of 2014 season I decided that keeping a track dedicated car in good looking condition capable for driving around city is hard, expensive and not worth it. I promised myself to stop tearing good rare car in competition.

So, this journal will be about this one:


I don't plan anything extraordinary, it's not going to be record breaking project, rather story full of pain and suffering about fixing old cars :)
hope you like it.
 

Attachments

See less See more
1 5
#248 ·
I also have the rear wheel lifting issue and higher spring rates do not help. My car is 2260 lbs and I am way oversprung at 12k F 10k R. I have a suretrac rear so she spins the inside rear like crazy on the long high speed turns. I have no idea what to do.
 
#249 · (Edited)
What sway bars do you have?
I think good plated diff should help in any case.
I tuned mapdccd to go almost full lock very early on TPS values, so it helps to transfer torque to the front as soon as possible.
Regarding suspension settings - going softer on rear dampers settings helped me a lot. Further rasing front damping didn't helped much, but I wonder if it helped to raise fronts+rear damping together. My best time was on 15 of 30 clicks in fronts and full soft in the rear. And I would say that while car was definitely lifting wheel it wasn't a problem for a acceleration at this point.
Other than that, probably it's worth to try different spring/sway bar combos with main idea of keeping rears soft so wheels stay on the ground but hard enough to balance front roll stiffness and not to cause massive understeer. Maybe try higher tyre pressures in the rear to loose some rear grip as well..
I'm thinking that source of the problem is weight distribution. Car is front heavy, so you either keep it from rolling with high front roll stiffness but have understeer as the result, or going on three wheels but then you have leak for torque if there are no good diffs to deal with it...
 
#256 · (Edited)
So you were running with the wrong timing for the last 2 years ?
No, I just had second event of skipping couple of teeth during the past 6 months. And I only been on track about three times, so it happens alarmingly often. And now I remembered that originally when I removed old timing belt 2 years ago it also was off by a little bit.
Looking again at the screenshot of the log I posted I started questioning if it may be something related to fuel or tune? There is abnormally high spike of injectors duty just before the event, although AFR seems to be normal.
The tune now is still a work in progress an I hadn't touched it in a while due to recent mess with this virus. I remember that before the lockdown I was working with boost control PID, so target boost is conservatively low at 1 bar and boost cut is low as well at 1.4 bar. It may be that I hit a boost cut and timing skipped because of rough cut implementation. I need to open log on my PC and check it out... But previous time I strongly believe that problem with timing was due to fuel starvation in turns and misfires due to fuel pressure drop. That time I felt car lost power and started misfiring right after such thing happened. Fuel starvation starts when I have less than half tank of fuel in right turns. I need fuel surge tank or sonehow move pickup lower, but since it can be fixed by adding fuel this it constantly being delayed. As an easy support until this is done I was supposed to wire fuel pressure sensor to ECU to compensate for small pressure drops and display warnings, but I'm still waiting for crimp tools for expansion connector :(
Initially I didn't thought about it since this time it happened only on third lap in with almost full tank, but looking at the boost log makes me thinking that it might be a boost cut...

Shopping for timing components appears to be tricky :) newer style tensioner with additional pulley can be easily installed with just a bracket swap and sometimes is marketed as an upgrade for older cars. While older style of tensioner with its respective bracket is also sold as an upgrade for newer cars as proven WRC-style solution :)))
Will probably go for new oem timing belt and newer style tensioner to have both setups on hand to choose from.
I also might have newer style bracket on old broken engine.
Next Sunday there will be first time attack event this year, so probably for now I will just reset the timing and check the tune to make sure boost and rpm cuts are safe enough...

Update:
Nope, map limit was not reached and there were no cuts. In fact I can see that boost began to drop moments before rpm value went from 4000 to about 13000 due to sync issue and about second before that there was high noise/knock level on cyl 3-4. (It's wasted spark for now, so knock levels are paired). Unfortunately knock strategy is in early stages of development, so can't even tell if it's real knock or not. Although it matches MAP spike in this instance.
 
#257 ·
Having separate fuel and oil pressure sensor will help you save your engine.
I do have Defi gauges and since I have a Defi oil and fuel pressure I am feeding the signal into Link ECU.

Its relatively easy to scale it into the ECU with volt meter.

I can take a look at your map and log.
Msg me with links if you want.

Also I've created a engine protection based on AFR's .. so if your fuel pressure drops, AFR's drop and that removes ignition in my case. (5D ignition table)
Its slow.. slower than knock but just another layer of protection.
 
#258 · (Edited)
Thanks! Yes, the plan is to have all the sensors in ECU, when I run out of IO I plan to use some CAN extension solution. I already have fuel pressure sensor wired to external datalogger and Bosch combined pressure+temp sensor installed in place of dash oil pressure switch. Now it's matter of getting my order with wiring supplies to get it finished.

I still have old oil pump without bolt holes for timing belt guide on crank pulley. I think of getting a new pump with the guide, is it worth it? Also newer engines have guides on cam gears - are there any ways to install those on older engine?
 
#259 ·
So, now I'm stuck choosing oil pump.. Don't know what is in there right now, but thinking of ordering 11mm.
Greddy t518z oil cooled turbo is waiting, I have a suspicion that this engine has oil squirters as well, so I hope that 11 mm will be fine...
 
#261 ·
So, update time. Softer limiter settings with progressive fuel cuts seemingly helped. No timing issues so far, but full timing kit is arrived with new style tensioner and extra pulley and new 11 mm oil pump to use belt guide on crankshaft sprocket.


Since last post there was first 2020 time attack event and it went fairly uneventful. My time was a huge disappointment though. 1:15.4 is over 2 sec from my 2014 time and even worse than some of my earlier tests. My current best time for this car in this form is 1:14 and my plan is to get into mid-low 1:11.xx.

I replaced 24mm rear whiteline sway bar for original 18 mm one and it helped with lifting of inside rear and traction on corner exits, but… car struggled from boost cuts because I had MAP limit set very conservatively and close to boost target of 1.1 bar (to make sure that it's under control) and for some reason I can't control boost. Even with WG duty 0% pressure is rising after 5500 rpm. I got MAC-style 3-port solenoid instead of stock one but testing at home (although with db-killer installed) proved that stock one works fine. Not sure what is there problem.


The last (of three) 15-min race session I raised the map limit to 1.4 bar and it appeared to be the limit of my VF28, but that made car working in untuned concervative cells on maps with rich afrs and "safe" timing…

Screenshot of last session datalog illustrating attempts to control boost with 0 WG duty:




And to end any hopes for good time, fuel cuts start to appear in that session, so for the whole event I couldn't put together a single lap without any issues.

Also I was too excited an almost killed left front (most loaded) tyre with way too much heat and pressure… need to rotate tyres now :(

Some pics from the event:

2020 Time attack 1

Since my car wasn't among the top runners I barely got any pics from local photographers and one that got is a bit embarrassing:



Full album:



Managed to put car on a scales at the event - 1195 kg (2634 pounds) with still a lot of stuff to remove (even speakers are still installed). Not bad! In my class 1150 kg is the limit - below that you have to use 235 tyres max (265 over 1150 kg) I'm using 235s now, but still, if I will remove sound insulation, speakers and heater - I might need to watch this if I plan to use more than 235s.


Next time attack event is on Sunday next week and I have a lot of issues to resolve before that. Probably will not bother with timing and oil pump, will just check once again if it still ok.

I'm in the process of re-doing intercooler shroud in aluminium instead my 3d printed one. Need to switch to frontmount and t518z, but first I want to validate my tuning abilities and all control systems :)

On intake side of things I also replaced pipercross foam filter with enclosure with bigger K&N. Will try to proove that more air > cold air :)))



Fuel cuts are also something I need to take care of. This is my second car where I got fuel starvation after installation of Walbro pump :( strainer / filter arrangement just sits a bit higher than stock. I can't see other differences that can cause issues… tried to figure out setup from Honda installation kit that I thought will be better, but no luck. Easy fix will be to keep the tank full, but after 1-hour highway drive, practice session and one race session I already end up with around half of rank and surge issues.


But mainly I will concentrate on tuning efforts, this time during the day and without silencer in muffler to simulate hot midday track environment. Road tuning during much colder evening with DB killer proved to leave too many surprises.


Also I need to find a way to solve VSS noise issue. Car has cable driven speedo and while I can install gearbox-mounted vss and feed signal to ECU, I will loose speed on dash and odometer which I'd rather wanted to keep. Anyone had experience fixing such issues? I heard that it's common issue for cars of this era with reed switch-based VSS in dash...
 
#262 ·
Your PID are showing values so it would seem that you are running some WG.
3 port mac does not operate below 10% duty.. not sure if you know that.

On top of that.. there is knock in your screenshot..

Not sure if you've done it.. but If you have a fuel pressure sensor you can enable safety in the Link ECU.
 
#263 ·
Fuel pressure sensor is installed, but not yet connected to ECU.
Knock signal is somewhat functional, but I'm not sure that this signal is reliable at high rpm. If I see high levels at mid rpm it's probably knock, but on high eng speed I can't confirm that knock is real, can't hear anything in headphones.. too much noise.
 
#265 · (Edited)
I bought a Link knock block.. I log all my runs and record the audio for analysis.
Then I use WavePad sound editor to run through visual filters.

This allows you to confirm the knock both visually and via audio.
I'd be afraid to tune it otherwise.
Oh yes, I know the feeling... That's why I got my fuel and timing rather conservative for now. But I don't feel like getting additional knock-listening devices though... They are quite expensive and basically in my understanding they do the same as dialed ECU settings should do anyway. It's just tuned system with all components provided as a kit and ready to work together out of the box...
I just threw in some basic settings and confirmed that noise level responds in some realistic-ish way to RPM and load, tweaked gain to level out individual cylinders and this is how it is now. If you have a look at my log screenshot once again - at 6000 rpm it is in fact now working due to RPM lockout. Too much noise up top to see real knock...
So, I'm not sure how I will approach this in a way that will give me enough confidence to push it and rely on it. I have one external Bosch wideband sensor that is hooked to mic input to logging PC, so I can listen the noise and while the sound is very clear at low-mid rpm, I'd lied if I told you that I can hear anything useful at high rpm :(
Also contributing to this problem is that I do it all myself and on the road, so it's usually quite a lot of stuff I need to keep track of...
/I really like Link ECU, but I'm still learning things and the more I learn, the more I see risks and problems of this problems :)
And since I'm doing completely different stuff as my day job, it's rather difficult to switch and keep track of all important aspects when you have time for twice a month or something like that...
 
#266 ·
Update time!

To spice things up before the next time attack I decided to swap turbo during the week. And as I'm busy on a day job from 9 till 7 every day I decided to do it on my parking spot at home. On Monday I checked the layout in general to make sure that there are no major problems in the surface and cleaned up things a bit, removed intercooler an prepared for the job.

Next day I removed the turbo which was fairly easy. All nuts were fresh and clean and some were even a bit on the loose side. V-band on downpipe allowed me to disconnect it without even jacking the car up and I used a vacuum pump to get some coolant out so I even managed to loose only couple of drops of it not making any mess on the parking spot.

The process:







It seems that I had some exhaust leaks due to loose nuts and/or warped flanges…




Next day was the install.

Old vs new:





Greddy T518Z basically TD05-18G as I understood, got it for a fair price from a friend with a rebuild kit and serviced it in local shop.

Old VF28 still seems to be in good shape and will probably be retired from racing to serve daily duties in my SG5 Forester instead of stock small td04..


Install was a bit tricky and it took several attempts to allign oil drain hose, turbo inlet and uppipe studs. What worked for me is to install rubber hose on lower drain tube first, put turbo on studs making sure that oil drain slid into the hose and then stretch the inlet. Luckily silicone inlet was flexible enough to make to possible.

Also this turbo has no water cooling, so I just put hose from the coolant supply straight to the coolant tank.

It was oil supply that appeared to be tricky… I saved banjo bolt from old td05 that came originally on this car specifically for this occasion because journal bearing turbo might need different restrictor. But to my surprise not only it was basically the same that I got out of vf28, the thread in Greddy turbo was different. I should thave done my homework, this turbo was laying ready to install around for several months already… so next day I went to find correct banjo bolt and could find only one without restrictor at all.



At that point I was sure that turbo has restrictor internally, but checked once again and found that except for restrictor-ish looking hole in the center it also has holes on the side of the oil port that I didn't noticed :(



Finding the bolt with correct restrictor was not possible before the event, so I got this bolt welded and drilled to 1.4 mm which seems to be ok for the application.



Water this everything went pretty smooth and everything was ready for testing on Saturday.

To be сontinued...

 
#267 ·
Update time!

To spice things up before the next time attack I decided to swap turbo during the week. And as I'm busy on a day job from 9 till 7 every day I decided to do it on my parking spot at home. On Monday I checked the layout in general to make sure that there are no major problems on the surface and cleaned up things a bit, removed intercooler an prepared for the job.

Next day I removed the turbo which was fairly easy. All nuts were fresh and clean and some were even a bit on the loose side. V-band on downpipe allowed me to disconnect it without even jacking the car up and I used a vacuum pump to get some coolant out so I even managed to loose only couple of drops of it not making any mess on the parking spot.

The process:







It seems that I had some exhaust leaks due to loose nuts and/or warped flanges…




Next day was the install.

Old vs new:





Greddy T518Z basically TD05-18G as I understood, got it for a fair price from a friend with a rebuild kit and serviced it in local shop.

Old VF28 still seems to be in good shape and will probably be retired from racing to serve daily duties in my SG5 Forester instead of stock small td04..


Install was a bit tricky and it took several attempts to allign oil drain hose, turbo inlet and uppipe studs. What worked for me is to install rubber hose on lower drain tube first, put turbo on studs making sure that oil drain slid into the hose and then put on the inlet. Luckily silicone inlet was flexible enough to make to possible.

Also this turbo has no water cooling, so I just put hose from the coolant supply straight to the coolant tank.

It was the oil supply that turned out to be tricky… I saved banjo bolt from old td05 that came originally on this car specifically for this occasion because journal bearing turbo might need different restrictor. But to my surprise not only it was basically the same that I got out of vf28, the thread in Greddy turbo was different. I should thave done my homework, this turbo was laying around ready to install for several months already… so next day I went to find correct banjo bolt and could find only the one without restrictor.



At that point I was sure that turbo has restrictor internally, but checking once again and found that except for restrictor-ish looking hole in the center it also has holes on the sides of the oil port that I didn't noticed at first :(. Don't me like me, do your homework, people!



Finding the bolt with correct restrictor was not possible before the event, so I got this bolt welded and drilled to 1.4 mm which seems to be ok for the application.



The rest went pretty smooth and everything was ready for testing on Saturday.

To be сontinued...

 
#269 · (Edited)
Saturday was the first test of the new setup. Car was finished late in the night on Thursday, so on Friday morning I primed oil system by disconnecting the crank sensor and cranking it until I saw good oil pressure for several seconds. Then I relaxed a little bit and didn't touched the car untill Saturday morning when I packed everything and went to the track for testing. I drove the car all the way to the track never even coming into the boost, just stopped to check for leaks and issues shortly after departure.

On track I lowered WG duty to be on safe side, made sure that fuel, ignition and safeguards were appropriate and went for a test. I tested WG spring on new turbo and it starts to open at about 0.9 bar that is not far from my 1 bar "low boost" goal. I did several tests but boost was oscillating quite badly with closed loop control. It also was first test of our joint RX8 project where I participated and was involved in datalogging side of things and there were some issues to solve, but generally it went not bad...



My car took much more effort. Doing all myself is interesting, but that's a lot of work. I tried to stick to workflow plan but constantly was distracted and sidetracked by different issues.

I saw that injector duty is high and there is no headroom (although mixtures were quite rich), so I raised base fuel pressure to about 3.5 bar (51 psi) that theoretically should make my 440 cc injectors flow in low 500s..

Tweaked fuel a bit and returned to boost but it kept oscillating… with trackday time running out I returned to open loop boost control and it kept boost at around 1 bar from the first try. That was a big relief, so I made minor updates and decided to race next day as is.

Also I completely did knock detection setting from the scratch making it less sensitive because it was picking up too much noise on high rpm.. still not 100% functional, but not cutting too much timing now :)

Next morning I came to the track again for the second time attack event of the season. Got my stickers and a warning from track officials for my off in Mazda the day before.. it turned out that it was late in the day and nobody noticed that, so they found dirt and stones on the track only next morning on the race day… not good..

Traditionally there are no championship races on hot summer months here but due to covid stuff this year was different. Many participants came, but some of the top runners in my class decided to skip the event. Still around 30 cars competed in total. In my class there were Evo 6, wide-bodied hawkeye Impreza, and mk6 Golf GTI. All really strong competitors with stable times around or below my PB from 2014 in old car (1:12.7), even GTI :)

I unloaded all the stuff from the car an changed tires for 235s dunlops.. the day before I had car on older set that I used originally in 2013 and 2014. I knew that it was finishe and didn't cared too much but I wish I checked them more often.. you know, just in case...



Still managed to get couple of mid 1:15 laps from them. I don't know how...

I was expecting to improve my personal best in this car which was somewhere near low 1:14s, and…. was really close to do it :(

The even went fairly... uneventful. I tried to analyse logs after each time on track and tweaked fuel and WG duty to dial it in and turn up the boost a little bit, although I wasn't doing any significant changes because except from initial practice run I had only two 15-min race sessions and any unexpected behavior in the session ment that I had to go to the pits and loose the time on the track.

As usual I was overloaded with information and saw major problems and opportunities only analyzing logs after the event.

I spent quite lot of time designing and making new intercooler shroud, changed turbo and intake setup hoping to get lower IAT and more effective turbo setup. WIP pic:



So, how it turned up? Well I don't really know. On the day of the event I was feeling that new turbo is doing great. Spool time was almost the same, response was good and car felt stronger on the same boost which matched my expectations from bigger, more effective turbo and intercooler ducting updates. What I saw in the logs the next day?



This is couple of laps from the race day compared to log of first event from mid June. As you can see, I use different gears here and there but overall it matches really close.

Here is closup of the same two logs overlapped.



What I noticed immediately that my IAT was higher on second even after all the changes. But my ECT was significantly higher too!

The reason for that is that I forgot to remove front licence plate hanging in front of radiator and covering over 25% of the area. I noticed higher coolant temps at the end of the sessions but car wasn't overheating so I noticed the licence plate on the car only at the end of the event.

How it contributed to the IAT increase? Who knows.. air filter is currently not isolated from engine bay, so I think it played significant role… at least I see that when IAT goes down it seems to go down faster, so my ducting sorta works…

As for my time. Top speeds were good, on par with my PB from 2014, but speed in shower corners are significantly lower. And I brake too much before them because I have no confidence that I will have enough grip. I previously blamed tires that are produced in 2014 an probably well passed their peak performance, but speed in fast corners is in fact not that bad! It is on par with my PB lap and even better in one of the turns.

I finally managed to refresh my video-making skills and. Put together video with data. I was to lazy to trim it down so feel free to skip all cooldown laps - interesting ones are #10, 12 and 15.

Fastest one is #10. I had opportunity to improve it but fuel cuts started to interfere on the later two...



Frankly second practice session was a bit more fun (because I thought it was final session of the day:))) but best time was in this one, so it went in processing first. I remembered and figured out my workflow for these videos so will probably do it later as well.

Data comes from two separate logs and visualized in race render. Some of the data is in real units (kpa, deg C) and these mostly came from ECU. Others that make less sense are from data logger and not scaled yet (values are millivolts).

Looking at the video I saw that I use the map with a lot of DCCD lock that was meant to fix spinning inside rear wheel issue which is now mostly fixed with weaker sway bar. Although I still can't put power down without DCCD lock I probably need much less of it now, and this is probably what affects my grip on slow corners a lot. I already tweaked map a bit and will test this next time on track.

That's all for now. 4th place (and last) place in my class about 2 seconds behind the competition and my own PB… Car still has issues with voltage drop on high loads (can be seen in logs above), it goes down below 12v. Alternator was replaced and issue is still there, so probably I need to look for problems in wiring. Air intake needs access to fresh air, I need bigger injectors to raise the boost and use the potential of new turbo, need to wire several sensors to ECU (crimping tool is arrived!)

AEM meth kit needs to be installed and integrated with ECU. And probably I need to get fresh set of tires as well. Thinking about getting Hankook Ventus TD (z221) in 235 width…

Next event is in September and there's not too much time left for this list :)

For now I got week off from track-car related stuff and went to mountains for some hiking and spent some time this weekend in the mountains. There are a lot of hiking trails here and it's weather much bearable in the summer.

And local mountain roars are really nice too ;)





Frankly this is one of the main reasons why I still stick to street class in TA and keep my car street legal and insured :)

Fast driving is obviously for the track but you can still have lots of fun driving on these roads safely with proper car.

Also, hill climbs are really a thing here. On these exact roads! A lot of local events, nice cars and strong championship every year. Should really try it someday. But it's much more serious compared participation in street class TA.
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top