New headers/exhaust for the '85/'86 people! Updated with baseline dyno charts.
#271
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
I'm old now and swimming against the current isn't all that attractive, but I have this great candidate S3 and I'm debating whether to scrap it for parts and join the 944 crowd or "go where no man has gone before" (except perhaps Mark
![Smilie](https://rennlist.com/forums/images/smilies/smile.gif)
It's projects like this one that keep me awake at night Greg...
#273
Former Sponsor
Thread Starter
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
While I have a great deal of respect for the more informed opinions already expressed in reply to this question, I have my own plebeian theory that I've held ever since Porsche introduced the S4 (I of course already owned an S3).
In short, I think they knew that if they simply made the engine breath better they'd see significant gains in rwhp. Porsche is famous for making incremental improvements during a model life and I think they *planned* to do what Greg and others have done, they just stopped. In the S4 they focused on the intake. They never got to tuning the exhaust.
Anyway, that's what I think.
In short, I think they knew that if they simply made the engine breath better they'd see significant gains in rwhp. Porsche is famous for making incremental improvements during a model life and I think they *planned* to do what Greg and others have done, they just stopped. In the S4 they focused on the intake. They never got to tuning the exhaust.
Anyway, that's what I think.
On the strokers I've built out of the S3 configuration, I've done that....and these engines make very good power, with the S3 intake....perhaps better than the S4....
#274
Nordschleife Master
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
A couple places in Orange sell 110, but its close to $10/gallon. Several of the smaller airports aren't too fussy about pumping 110 unleaded into a car either.
#275
Nordschleife Master
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
BTW you are almost always far better off to buy somebodies track project and put your good parts into it rather than start from a donor car and make it track ready.
#276
Rennlist Member
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
Having done this for over 15 years now, the 928 is the perfect club racer platform, as long as you got a decent engine to start. almost ALL of the engine failures have been suspect to begin with. with the stock stuff, headers, and some big wheels and tires (with a bolt on suspension), you will have many years of competitive racing in the "catch all" classes like ITE SCCA, or GT US sportscar championships. you cant find a more low maintenance platform out there, that can get 300 to 320rwhp reliably. AND, you don't need any fancy cage due to the stiffness of the amazing 928 chassis. this is NOT the case for the M3s, 944s, mustangs , etc. all need about 40 to 80k to get to the level of a very minimally prepared 928. Its wide, strong and nothing seems to break on it.
Unless you want to get in a fixed class, which pigeon holes your race options, and really want some 5 dollar trophy vs all the guys that didn't make it to as many races as you did, race a 928. you will have some thing unique, easy to work on and build into a race car and fast enough to compete with many of the fastest cars in any racing organization on any given weekend, AND have a lot of good fast cars to race against.
Unless you want to get in a fixed class, which pigeon holes your race options, and really want some 5 dollar trophy vs all the guys that didn't make it to as many races as you did, race a 928. you will have some thing unique, easy to work on and build into a race car and fast enough to compete with many of the fastest cars in any racing organization on any given weekend, AND have a lot of good fast cars to race against.
#277
Nordschleife Master
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
If it's true that the exhaust cross-over location doesn't change the power at all, then it may be that the gas velocity in the exhaust system up to the termination point is on the slow side. I think that if the dyno graph is about the same with and without any cross-over whatsoever, then that would in my opinion confirm it.
That's a very logical discussion and conclusion....but I'm not sure what any of it has to do with this discussion. I'm not sure that anyone, especially me, has stated that the location of the cross over pipe isn't important, on this engine. I've got a small mountain of various cross over pipes, made for this project....some of which worked well and some of which were terrible. I've been making "subtle" changes to the cross over, this week of testing. Changes in angles, changes in cross over contact area, etc.
You chose not to explain the details of your X-pipe experiment, so we don't know exactly what was changed between the two runs. This is not criticism, it's your data and you should do with it what you want.
However, the most interesting experiment that I would post if I'd be working on this project is just the "rear" X-pipe (which tunes to the second collector harmonic) vs. both the "front" H-pipe and the "rear" X-pipe (with the H-probably tuning to the fourth harmonic and the X-enhancing the sound.) This without the cats for the clearest experiment. If the gas velocity in the primaries would be optimally high, then I would expect the "just rear X" to produce meaningfully more torque at several points whereas the "both front H and rear X" to produce a flatter torque curve and probably higher overall peak power. If this experiment would instead produce torque curves that would plot right on top of each others, then I'd say that the exhaust diameters are too large for meaningful pulse tuning (and probably the cam overlap too small anyway.)
Again, I don't know what the experiment was so only you know whether this applies to this exhaust or not. All I can say is to make a general point that when the engine doesn't respond to change in the collector length then it's either that there's not enough camshaft overlap or the pipes are too large, or both. That's just general discussion.
#278
Former Sponsor
Thread Starter
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
Those were some general observations. I don't know enough about this exhaust to say whether those observations apply to this case or not.
You chose not to explain the details of your X-pipe experiment, so we don't know exactly what was changed between the two runs. This is not criticism, it's your data and you should do with it what you want.
However, the most interesting experiment that I would post if I'd be working on this project is just the "rear" X-pipe (which tunes to the second collector harmonic) vs. both the "front" H-pipe and the "rear" X-pipe (with the H-probably tuning to the fourth harmonic and the X-enhancing the sound.) This without the cats for the clearest experiment. If the gas velocity in the primaries would be optimally high, then I would expect the "just rear X" to produce meaningfully more torque at several points whereas the "both front H and rear X" to produce a flatter torque curve and probably higher overall peak power. If this experiment would instead produce torque curves that would plot right on top of each others, then I'd say that the exhaust diameters are too large for meaningful pulse tuning (and probably the cam overlap too small anyway.)
Again, I don't know what the experiment was so only you know whether this applies to this exhaust or not. All I can say is to make a general point that when the engine doesn't respond to change in the collector length then it's either that there's not enough camshaft overlap or the pipes are too large, or both. That's just general discussion.
You chose not to explain the details of your X-pipe experiment, so we don't know exactly what was changed between the two runs. This is not criticism, it's your data and you should do with it what you want.
However, the most interesting experiment that I would post if I'd be working on this project is just the "rear" X-pipe (which tunes to the second collector harmonic) vs. both the "front" H-pipe and the "rear" X-pipe (with the H-probably tuning to the fourth harmonic and the X-enhancing the sound.) This without the cats for the clearest experiment. If the gas velocity in the primaries would be optimally high, then I would expect the "just rear X" to produce meaningfully more torque at several points whereas the "both front H and rear X" to produce a flatter torque curve and probably higher overall peak power. If this experiment would instead produce torque curves that would plot right on top of each others, then I'd say that the exhaust diameters are too large for meaningful pulse tuning (and probably the cam overlap too small anyway.)
Again, I don't know what the experiment was so only you know whether this applies to this exhaust or not. All I can say is to make a general point that when the engine doesn't respond to change in the collector length then it's either that there's not enough camshaft overlap or the pipes are too large, or both. That's just general discussion.
I started tuning, today, in earnest and started adding timing, while keeping the fuel stable with Ken's LH chip and his potentiometer.
I'm not sure why....perhaps because the combustion chambers are scavenging better, but this thing loves additional timing. Way more timing than I could have imagined.
Instead of sounding flat, it is now starting to sing...you can hear the difference in the pulls on the dyno from the office in my second shop.
What does this have to do with your thoughts about the location of the X-pipe?
I'm just guessing, but I'd think once the engine starts "grunting" and making horsepower, the exhaust velocities are all going to change....making prior tests somewhat useless?
Would you agree that I'm probably going to have to retest the various x-pipes?
Last edited by GregBBRD; 07-01-2014 at 03:15 AM.
#279
Nordschleife Master
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
One cause of detonation, in my thinking is the hot exhaust gasses not emptying from the combustion chamber. On the one hand, mixing in the inert exhaust gas reduces the tendency to knock, much like with EGR systems. On the other hand, the heat of those exhaust gasses makes the unburnt charge more likely to knock. I believe that the second effect dominates, and well flushed combustion chamber is less likely to knock. If I am right, it makes sense that with improved exhaust the motor can tolerate more timing.
---
I don't know what the additional timing will do to the exhaust pulse strength in this case. From my turbo experiments, I am leaning towards thinking that within this range it probably doesn't do anything important to it. Retarding the spark to a silly level tends to strengthen the exhaust pulse (this I know for sure from trying) because the engine can harvest less useful energy from the later burn and it must therefore go out the exhaust (this is my theory of why it happens.)
---
One additional note on the dyno curves. Your stock vs. stock + exhaust graph shows about 10-20 ft-lbs improvement from the exhaust. Here's your relevant dyno graph:
![](http://i1033.photobucket.com/albums/a415/GregBBRD/STOCK_vs_HEADERS_WITH_X_PIPE_WTH_STOCK_CHIPS_zps3ae63937.jpg)
Note two things. First, the torque gain from the new exhaust side increases with rpm. Second, the new exhaust torque curve doesn't dip at higher rpms and instead stays up at relatively high levels.
It's my understanding that long-tube headers usually give the most improvement in the mid range, and gains taper to a lower level at higher rpms. The exhaust diameters then can tilt those gains, large diameters shifting the gains higher up in the rpm range.
Here's a generic "typical" graph from a web page that says things make sense to me:
![](http://www.sandersonheaders.com/~sanderso/images/tech/tubechart.gif)
http://www.sandersonheaders.com/Lets-Get-Technical.html
The gains increasing towards high rpms and the new exhaust torque curve not dipping at high rpms suggests to me that the exhaust pulses aren't as strong as they could be because the pipes are on the large side. This is educated guessing. This "on the larger side" might be just right if there are further plans to modify the engine. In any case, if my educated guessing is right, then when engine air flow is increased, the engine should start showing gains everywhere but disproportionate gains at the mid range rpms.
The above educated guess is not necessarily right, it's just a guess after all; it could be that the S3 exhaust manifold becomes a simple flow restriction at high rpms. It's not my guess that this is the only thing that is going on since the level (not gain) of the new exhaust torque curve doesn't dip much at all at high rpms.
---
If the engine currently responds strongly to changes in the collector length / x-pipe location, then I don't think any testing would necessarily need to be redone after evolutionary changes to improve cylinder filling or harvesting of power. The event wouldn't change dramatically. Now, more testing is always better, but I'd use the budget to test something else in that situation. If the engine _doesn't_ currently respond much to changes in collector length / x-pipe location, then as air flow is added by one method or another, it may make sense to allocate some of the budget to retesting the best collector length / x-pipe location. That's my generic view -- which case applies here is unknown to me.
---
Digressing to one more tuning issue: In terms of detonation tolerance, Ken's mechanism of dramatically shifting timing based on engine [intake air] temperature is a very nice feature (if I've understood it correctly.) This is what new cars do. It does lead to the dyno figures changing based on the conditions. For example, dynoing the new M5 will give you a different number every time unless you use a controller to run the fan speeds to simulate constant operating temperature conditions. Reasonable people can disagree what should be quoted as the "hp" number in such cases, but I think everyone should agree that the engine is maximized with ECU software that work that way. Hope you and Ken can work out some arrangement in which you can use this feature, or perhaps Niklas of JDS can make you a custom software that has this feature.
---
I don't know what the additional timing will do to the exhaust pulse strength in this case. From my turbo experiments, I am leaning towards thinking that within this range it probably doesn't do anything important to it. Retarding the spark to a silly level tends to strengthen the exhaust pulse (this I know for sure from trying) because the engine can harvest less useful energy from the later burn and it must therefore go out the exhaust (this is my theory of why it happens.)
---
One additional note on the dyno curves. Your stock vs. stock + exhaust graph shows about 10-20 ft-lbs improvement from the exhaust. Here's your relevant dyno graph:
![](http://i1033.photobucket.com/albums/a415/GregBBRD/STOCK_vs_HEADERS_WITH_X_PIPE_WTH_STOCK_CHIPS_zps3ae63937.jpg)
Note two things. First, the torque gain from the new exhaust side increases with rpm. Second, the new exhaust torque curve doesn't dip at higher rpms and instead stays up at relatively high levels.
It's my understanding that long-tube headers usually give the most improvement in the mid range, and gains taper to a lower level at higher rpms. The exhaust diameters then can tilt those gains, large diameters shifting the gains higher up in the rpm range.
Here's a generic "typical" graph from a web page that says things make sense to me:
![](http://www.sandersonheaders.com/~sanderso/images/tech/tubechart.gif)
http://www.sandersonheaders.com/Lets-Get-Technical.html
The gains increasing towards high rpms and the new exhaust torque curve not dipping at high rpms suggests to me that the exhaust pulses aren't as strong as they could be because the pipes are on the large side. This is educated guessing. This "on the larger side" might be just right if there are further plans to modify the engine. In any case, if my educated guessing is right, then when engine air flow is increased, the engine should start showing gains everywhere but disproportionate gains at the mid range rpms.
The above educated guess is not necessarily right, it's just a guess after all; it could be that the S3 exhaust manifold becomes a simple flow restriction at high rpms. It's not my guess that this is the only thing that is going on since the level (not gain) of the new exhaust torque curve doesn't dip much at all at high rpms.
---
If the engine currently responds strongly to changes in the collector length / x-pipe location, then I don't think any testing would necessarily need to be redone after evolutionary changes to improve cylinder filling or harvesting of power. The event wouldn't change dramatically. Now, more testing is always better, but I'd use the budget to test something else in that situation. If the engine _doesn't_ currently respond much to changes in collector length / x-pipe location, then as air flow is added by one method or another, it may make sense to allocate some of the budget to retesting the best collector length / x-pipe location. That's my generic view -- which case applies here is unknown to me.
---
Digressing to one more tuning issue: In terms of detonation tolerance, Ken's mechanism of dramatically shifting timing based on engine [intake air] temperature is a very nice feature (if I've understood it correctly.) This is what new cars do. It does lead to the dyno figures changing based on the conditions. For example, dynoing the new M5 will give you a different number every time unless you use a controller to run the fan speeds to simulate constant operating temperature conditions. Reasonable people can disagree what should be quoted as the "hp" number in such cases, but I think everyone should agree that the engine is maximized with ECU software that work that way. Hope you and Ken can work out some arrangement in which you can use this feature, or perhaps Niklas of JDS can make you a custom software that has this feature.
Last edited by ptuomov; 07-01-2014 at 01:33 PM. Reason: missing words [intake air]
#280
Rennlist Member
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
#281
Inventor
Rennlist Member
Rennlist Member
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
As I understand it, the S3's (comparably) long duration cams bleed off the intake charge. Even though high compression (10:1) is used to compensate for this effect, the S3 can still tolerate quite a bit of ignition advance.
In order to make peak power with his own 'tune', Greg will have to raise the advance level enough to bypass the stock air temp retard, but this could lead to knock.
Look for sudden divots in HP in the dyno graph for an indication of knock.
S300s chips were tuned with a knock sensor and the (generic Volvo/SAAB) maps were rescaled to allow more advance around the problem RPMs that were found.
Greg will not be able to make more average power than if he had just used the S300s EZ chip with the octane loop disconnected and provided for decent air.![Wink](https://rennlist.com/forums/images/smilies/wink.gif)
Adjusting the timing to the air temp is more effecient, adapting the burn time to the conditions. (Part throttle. Air density is a larger factor at full throttle.)
With my S4 chips, I use knock feedback to adjust overall timing to conditions, but it would be nicer/better to have an air temp input into the EZK.
Most of the stock map is uselessly cold. It's primary use is for a basic hot air retard and reacting to the air temp sensor being disconnected or shorted (octane loop).
S300s air temp maps are rescaled to engine operating temps. Additionally, there are five levels (new maps) determined by load.
In order to make peak power with his own 'tune', Greg will have to raise the advance level enough to bypass the stock air temp retard, but this could lead to knock.
Look for sudden divots in HP in the dyno graph for an indication of knock.
S300s chips were tuned with a knock sensor and the (generic Volvo/SAAB) maps were rescaled to allow more advance around the problem RPMs that were found.
Greg will not be able to make more average power than if he had just used the S300s EZ chip with the octane loop disconnected and provided for decent air.
![Wink](https://rennlist.com/forums/images/smilies/wink.gif)
With my S4 chips, I use knock feedback to adjust overall timing to conditions, but it would be nicer/better to have an air temp input into the EZK.
S300s air temp maps are rescaled to engine operating temps. Additionally, there are five levels (new maps) determined by load.
#282
Former Sponsor
Thread Starter
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
I am certainly no EZF expert but isn't that what the two "Airtemp compensation maps" are for? (High-load cruise and WOT, ST1 manual pg 59-60).
I've also made an effort to dyno test with "real world" intake air and manifold temperatures, instead of cooling these things below realistic temperatures.
We drove one of these vehicles for an extended period of time, on the freeway, last week, on an 85 degree day. The intake manifold temperature never dropped below 150 degrees.
Dyno testing on a 70 degree day, while keeping the engine cold (while the oil is hot), the manifold near 70 degrees, and the intake air at 70 degrees might yield "hero" dyno results, but isn't realistic for anything else. No one is ever going to experience those conditions, unless they heat up their oil with an electric heater, fire up their engine, and do WOT blasts before the engine reaches equilibrium.
From what I've been able to tell, testing in a "real world" condition and not doing "hero" runs reduces power output (on the dyno) by about 5%....which puts 305hp back down into the low 290hp range.
Like I've said several times.....time might have been better spent actually improving the power output, not spending years trying to figure out how to cheat the dyno results....
#283
Inventor
Rennlist Member
Rennlist Member
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
More relevant is the flowing intake air temp while the car is in motion. Unless there is a problem, it should be near ambient. Porsche designed it this way.
You will end up with a poorly tuned engine which has no relation to driving on the street.
#284
Nordschleife Master
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
Intake air temperature and engine coolant temperature are completely different when the engine is running at load. Intake air is much cooler. I agree that it's pretty close to ambient regardless of the engine coolant temperature, because the air mas flow is so high and the manifold has relatively little surface area.
It may still be the case that conditioning ignition timing on the coolant temperature is pretty good, because (intake air temperature - ambient air temperature) may be highly correlated with coolant temperature. It's not as good because of engine coolant temperature has "inertia" and the ambient temperature may vary a lot.
Basically, the three main variables that one would ideally want to control for in preventing knock with ignition timing at a given rpm are
1) octane of gasoline
2) any two of the following three: post-throttle intake charge pressure, temperature, and density
I think that for fueling, the coolant temperature is actually more important. I am not sure though. I think this because the intake port wall temperature is one of the more important determinants of how much vaporized fuel ready to be burned there actually is in the cylinder when the spark comes.
#285
Rennlist Member
![Default](https://rennlist.com/forums/images/icons/icon1.gif)
It's hard reading the snipe war between Ken and Greg for two reasons: First, they both contribute a lot to the 928 community, so I don't have a rooting interest in one side winning or losing.
Second, they're both right on the dyno topic. Running the dyno under unrealistic conditions produces results that may or may not be meaningful under real world conditions. It's also true that failure to control variables like air temperature will also produce results that are not meaningful in the real world. You'll always be unsure whether or not the latest change to the tune produced the new results or whether those results are due (in part) to the change in conditions. Making hundreds of runs under different conditions (while keeping the tune constant) would produce a data set from which it would be possible to ferret out from future runs which changes in performance were due to the conditions and which were due to the changes to the car. That would be expensive and would require sophisticated statistical analysis techniques. For those reasons, controlling conditions seems like a better approach.
So, the real debate should be about the proper conditions under which to run the dyno. I suspect there is no right answer. Ken runs his using a controlled 70 degree air temperature (IIRC). That may be fairly close to the real world in Oregon, but it's NOT real world in Arizona or Southern California. Running the dyno at sea level may produce results that are not replicable in Denver or Laramie. Ideally, we'd have reports on the results from these differing situations, too.
As long as Ken discloses what he's doing, knowledgeable readers should be able to translate his results to their conditions. I don't see that as cheating. That doesn't mean, as a resident of the desert Southwest, I wouldn't prefer to see runs at 100 degree air temps (I would!). It just means full disclosure is the opposite of cheating -- unless the variables used for the dyno would never exist in the real world.
Second, they're both right on the dyno topic. Running the dyno under unrealistic conditions produces results that may or may not be meaningful under real world conditions. It's also true that failure to control variables like air temperature will also produce results that are not meaningful in the real world. You'll always be unsure whether or not the latest change to the tune produced the new results or whether those results are due (in part) to the change in conditions. Making hundreds of runs under different conditions (while keeping the tune constant) would produce a data set from which it would be possible to ferret out from future runs which changes in performance were due to the conditions and which were due to the changes to the car. That would be expensive and would require sophisticated statistical analysis techniques. For those reasons, controlling conditions seems like a better approach.
So, the real debate should be about the proper conditions under which to run the dyno. I suspect there is no right answer. Ken runs his using a controlled 70 degree air temperature (IIRC). That may be fairly close to the real world in Oregon, but it's NOT real world in Arizona or Southern California. Running the dyno at sea level may produce results that are not replicable in Denver or Laramie. Ideally, we'd have reports on the results from these differing situations, too.
As long as Ken discloses what he's doing, knowledgeable readers should be able to translate his results to their conditions. I don't see that as cheating. That doesn't mean, as a resident of the desert Southwest, I wouldn't prefer to see runs at 100 degree air temps (I would!). It just means full disclosure is the opposite of cheating -- unless the variables used for the dyno would never exist in the real world.
Last edited by DKWalser; 07-01-2014 at 03:20 PM.