Felix

OK, I finally got my finger out and wired up a boost sensor to my DL1 data logger (big brother to the AX22; see http://www.race-technology.com/WebPa...1/DL1Home.html ) and gathered some data. Below is a graph of road speed as measured by the DL1's GPS (haven't hooked up an RPM feed yet), boost as measured by the sensor (MPX4250A) and longitudinal g as measured by the DL1's accelerometer. The run was done trundling along at about 30mph in third gear (~2000 rpm) and then foot to the floor, shifting into forth at 6000rpm or thereabouts and backing off because of traffic. I need to look through the rest of the data to see if there are any full rev range pulls but I doubt it; there was too much traffic. I think some early weekend morning runs might be called for.

I'm surprised by a number of things:

1. At 3000 rpm the boost is less than 0.5 bar; the factory 993tt boost curve shows 0.8 bar. Why so little boost down low? The run didn't start until 2000 rpm; is 1000 rpm not enough to fully spool up the turbos?
2. There are hiccups in the longitudinal g figures and the car's speed even though the run feels very smooth. Is this to be expected?
3. After 5000rpm the boost climbs from 0.6 to just over 1.0 bar very rapidly; this kick in the back can very much be felt. But why only what seems like a lethargic 0.6 before that point?

Some background: this is a 97 993tt with recent new MAF, plugs, plug wires, cap and rotor and runs a factory 430bhp ECU.

A full rev range (1000-6900rpm) run may be more revealing.

Also attached is the factory 408bhp engine curve for comparison purposes.

Jean

Phelix, you don't post much , but when you do, dang! This needs some serious thinking

Your acceleration numbers seem in line with the engine setup that you have, therefore the only thing I can think of is that your boost guage is not properly placed or setup or has an issue, it seems to be showing 0.2bar less than normal across.

Your long Gs also seem a bit off when I look at your peak Gs , I would expect to see a 0.5-0.6G spike during the run at full boost to get these numbers, and also by looking at how little your Gs drop during your gear change, unless your shift has been tremendously fast, have you been practicing a lot at the 'Ring lately? You might want to do another run while ensuring level position of the accelerometer (The DL1 does not have GPS does it?)

As far as the lines not being smooth, this is very normal, I don't know the DL1, but I am sure you can dial-in an averaging factor of 0.2 seconds up to 0.5 seconds sampling even, it will be more accurate and show the graph as a smoother curve.

Felix

Hi Jean,

Always happy to provide you with a challenge!

There is a single upshift at about 80mph where the boost drops off dramatically, as does the acceleration. This would have been around 6000 rpm. Other than that no shifts take place. I'm no speed shifter in the 993tt; it's too expensive a gearbox and clutch for me to do that. Haven't been back the Ring either; I miss that place.

At the moment there is a fair bit of tubing in the plumbing for the gauge, probably a 3 meter piece of hose from the factory measuring point just in front of the intercooler to the sensor currently located in the car itself. I'll shorten it when I do a more permanent install.

The DL1 has a GPS sampling at 5 times per second; the DL1 itself samples 100 times per second. I've since had a look at some runs when ACE_parts borrowed the DL1 and his curves have similar fluctuations. The DL1 was taped to the floor on the passenger side of the car.

Jean

Hi Phelix,

I think my previous post is confusing.. to check whether the boost level that you see is correct, I wanted to :

1- See how does your acceleration measure from 30-80mph and how it compares to other similar cars, you are in line... Therefore boost seems not to be a problem, other than the reading.

2- However, I noticed that the long Gs (acceleration rate) are a bit soft for such a run, and you should see 0.55-0.6 G, so it made me wonder two things, a) if you did indeed have a boost problem, since you only have a peak at about 0.45G or b) if you have a calibration issue.

3- I therefore looked at the drop in Gs during your 3-4th shift at 80mph/6kRPMs, I noticed that you only drop to close to zero Gs while shifting, and to do that, you must have shifted VERY aggressively, and I know you did not.. In a typical quick shift, I see Gs dropping to -0.2 , - 0.3G between gears, while yours is close to zero.

So I now think that your Gs are not being properly read, they are too low in acceleration, and too high in deceleration, typical calibration problem that I see with accelerometers.

Since your acceleration numbers are good, the long Gs are not, and boost is not, that is my opinion. It would be great of course if you could post a longer run.

Net net, I have a headache, but I think your car and boost are doing great

The settings on the logger let you change the smoothening of the curve while drawing a graph, you need to go to Data/variable manager/standard variables/long acceleration/smoothing and set a higher value...same for speed if you want to see a smoother line. The smoother graph is in fact more meaningful to look at than the little spikes that you get with quick sampling, since it averages the readings somewhat.

Kevin

Phelix, try the same run but use 4th or 5th.. The engine loading vs time will be greater. You will see different boost results..

Felix

OK, here's more data. Speed and boost vs time for a 60-130 run. Note the high boost levels in the 4th gear section and the initial boost spike in 5th gear.

Felix

Here's a second run. Not sure why the dip in boost early on in fifth gear.

Felix

And here's a third run, this one starting off in 2nd. V strange 5th gear boost curve.

Felix

and here's boost vs rpm in 5th. ignore the ends of the curves doubling back on itself; this is because it's pure rpm vs boost graph ignoring time.

TB993tt

Phelix

Thanks for sharing the graphs - I can't really say anything constructive, they look fine to me

The varying boost is fascinating and when one considers that the fluctuations are caused by the ECU balancing the boost/timing/fueling according to the conditions of each individual run to give you the optimum output, it makes you wonder how it can be beneficial to go fixed boost control ?

People using tuners whose "solution" is the fixed boost/chassis dyno method should take a good look at these graphs - the fixed boost car may run great on the chassis dyno and give some big numbers for THAT run, but as is illustrated vividly above, each run is different and requires constant trimming of the boost.

Depriving the ECU of this important function is IMO a recipe for serious engine trouble and it is only thanks to the other built in safeguards that we don't hear of many engine failures.

Rickard 993 Turbo

Tb

Can you exaplain to me why a Porsche turbo need this more then other cars, all other turbo cars out there run fine with BC and a fixed boost value..?

TB993tt

I guess it depends on how you define "running fine" ?

Porsches ran "fine" for years with fixed boost courtesy of the ubiquitous wastegate spring giving a fixed boost value.

The game moved on with the 993tt Motronic and its ability to control boost, timing and fueling very accurately.

On the road this results in a 408hp Porsche having better and more sustained performance than other brands (with similar quoted peak power) with the power being achieved over a greater percentage of the rev range - the "area under the curve".
The ability of the Motronic to reduce the boost when the engine gets too hot (compared to fixed boost where only timing is pulled severely, leading to a much greater deterioration in performance) and increase it when the air density is low all lead to a better on the road performance.
The problem is, very few tuners have the tools and knowledge to program the ECU (to perform as Porsche intended) for a new set of components and hence why most "tuners" either have to buy in generic tuned maps or resort to the old technology.

Obviously fixed boost cars can be made to be very fast and reliable - Protomotive monsters for example, but they are mapped by a very clever guy with a lot of experience and the right equipment - It still must be true that adding the extra dimension (of boost variation) to the ECU must always be an advantage

Jean

Phelix,

It looks like in 3rd at 5.5K RPMs or so, is when boost starts to go strong, the same approx in 4th., and then there is a failed attempt in 5th gear at those same RPMs.. There are many factors that could be affecting this, I suggest that you set the frequency to 0.3-0.4 seconds or so for the boost readings to have a more smooth curve and eliminate the "noise" factor. You can do this from the software of the DL1 if I am not mistaken.

As far as variable boost control, I am with TB on this one, it sounds obvious to me that a car having capability to boost up to 1.5 Bar depending on air density, timing etc.. for a few seconds will perform better than one with a fixed 1 Bar limit. I know of broken engines and melted heads running fixed boost, namely with MAF setups causing damaging erratic boost and spikes, these typically are happening in countries with cold temperatures...An experienced tuner is key in these setups.

Felix

Jean,

The data seems to gathered at 0.01 second intervals by the DL1; I exported the data at 0.05 second intervals to make th volume more manageable.

TB,

Assuming everything is working as it should these graphs are (as you've already implied) an insight into the complexity of the engine management system and shows how many parameters the ECU is juggling to do what the driver is asking.

Felix

Here's a graph comparing the two ECU's. Admittedly it's not a 100% correct comparison in that the steady state speed of the cars was different for the 4th gear runs (430 ECU at 20 mph and RS ECU at 46 mph) so the turbos were spooling up for the 430 ECU at the point where the turbos for the RS ECU were not under load. The acceleration curves are virtually parallel bar the time delay.