Fen

In reverse order - I have gained nothing at all from my standalone - 306bhp and a very disappointing figure, but that's a baseline and is on the back of around 2 hours set-up on the road with a tight engine. I would hope/expect to see much more out of it by the next time it's on the rollers.

Thanks for the link to the graph. Hopefully Transaxle will help out with some background, especially as my German is not great. I will post the link here to save flicking backward and forward:


I assume the blue figure is rear wheel and the black is calculated flywheel with the red being uncorrected calculated fly? That being the case the run produced 234.6PS (around 231 rear wheel BHP for those of us preferring imperial). The calculated flywheel figure is reported as 312.8PS, which is about 305 BHP. That's allowing for 25% transmission losses. When we have seen cars run in the UK we usually expect to see closer to 15% transmission losses from a 944 Turbo - I think you quoted the same for your own "422" car?

For an example of the difference that makes and more what I'd expect to see from 230 at the wheels one of the cars running at last year's UK dyno shootout made 230.1 BHP at the wheels and 273.6BHP calculated at the flywheel. That is a 250bhp car so with a K26/8, Wayne Schofield remap and dual port wastegate running 1 bar boost with the typical bleed-off of boost above 4,500 rpm seen on all cars with an MBC.

I'd be interested to hear Transaxle's comments on that (or anyone else's). Also 48 degrees Celcius air temp - that's hot isn't it for Germany in December?

Interestingly my car making a very similar 306.0BHP flywheel needed 258.8 at the wheels to achieve that.

Transaxle

There are actually two runs on one chart. The "dotted" run was done with a AFM and a (to small) banjo-bolt. The "solid" run was taken with a MAF kit without BB. The top red curve is calculated flywheel power, the middle red line is torque and the blue line is rear wheel power. The transmission losses were measured by the dyno automatically when the car rolls out and are displayed as a dark green line. Since both runs are on a single chart, I don't think the air temperature is real. We logged all DME/KLR data and I think to remember 17°C intake temperature for both runs.

The losses at this 88 944 turbo S seem to be to high. We had another 951 (K26.6) with the MAF on the dyno the same day.



Theses transmission losses are as expected.

Fen

Just for clarification, then. Which plot is the K26/6 car and which is the K26/8 car?

I make the second plot around 236 wheel HP and 292 flywheel. That's still slightly higher transmission losses than we are used to seeing on UK runs I have witnessed. Think you're saying that is the smaller turbo car, which does make it a quite impressive figure. Was it completely standard apart from chips and the MAF? What about the banjo bolt, FPR, fuel, wastegate, boost controller?

Regardless it's NOT 300+ bhp, so where's that plot, Andrew?

promax_motorsport

Hi Fen,

The plots I refer to are the ones shown above. 297ps to 313ps = 292bhp to 309bhp (/1.01387). The average is 300.5bhp

Different cars will achieve different results. As you well know different dynos will produce different results. Any published power figure can only be indicative of what might be achieved. Thankfully, most people appreciate that - especially on cars that are now over 15 years old.

Take your own car to the temp controlled cells used by Thor Racing or Parr Motorsport (they have same hub dyno - the tool of choice for Motorsport teams, TV and manufacturers) and you will see a higher figure. Run your car on the WRC dyno with a cold engine and an optimally placed inlet air temp sender - your readings will differ again.

As over 40 951 owners have now discovered - fit a SciVision MAF to your 951 and it will go faster. Simple. How much faster will depend on many things. However, a standard car with a shimmed wastegate, SciVision chips and MAF can expect approximately 300bhp (average - some more, some less)

The only real way of measuring engine output is under totally controlled circumstances on an engine dyno. That's never going to happen. Before and after tests are flawed becuase the physical envoronment changes (even though dyno software attempts to compensate). You can get different power figures on a rolling road dyno with different tyres!

Personally, I think published power figures are useful for demonstrating the potential of a performance enhancing product. However, there is no way this information can be relied upon as being the last word in measured accuracy. As a result, it can only be considered a guideline.

Taking all this into consideration, I'm satisfied the published claims are reasonable. However, we at ProMAX Motorsport will be carrying out before and after testing on more of our products to try and address this situation more scientifically and consistently. This I feel is a responsible approach which I hope will give customers better information and prove more specifically the performance enhancing gains one can reasonably expect from our products.

Regards,
Andrew
www.promaxmotorsport.com

Fen

Well of the plots above I think one is suspect due to the overestimation of transmission losses and one seems to be from a 250 car (the same one I think, to be fair), but I admit the other is a good result from a 220 car with the stock wastegate and no other mods if that's what it is. Is there a before plot from the same car to gauge the absolute accuracy of the rollers? I absolutely accept that the difference between 292bhp and 302bhp is way less than the discrepancy between dynos, but in mareketing terms there is a world of difference.

Regardless I am very pleased to hear that you will be doing some before and after testing on your products. That's all I've been chasing you to do for some time.

Transaxle

You are right...

Here is the whole story:

We got the K26.6 car onto the dyno and got this result:



A good start but then the car really breaks in. We looked at the data logging and found a massive knocking.



The car was equiped with a K26.6, a 3 bar FPR, a 3mm shimmed wastegate, a 1.6mm BB and chips. The KLR worked as perfect as expected. After delaying the ignition it drops the boost in 40 mbar steps (actually 39.0625 mbar steps ). You could not recognize the knocking from outside.

After reading the logged data, we filled the 95 ROZ fuel tank with 100 ROZ (the European classification is different from the US) and got the result from above (with an estimated 98ROZ fuel mix):



And the data logging looks fine too:



So you are right - it depends on the fuel....

Zero10

How did you tap into the output pin of the KLR for knock detection? Is there already a pin and wire there?
I plan to add a 'knock warning' light to my car, since I do so many stupid things with it.

Fen

Thanks for the clarification. There is no plot of the car totally standard to compare I take it (both in terms of peak numbers and where the power and torque are made?

From the datalogging plots the KLR is more capable at knock control than I had understood - I guess I had picked up some mis-information at some point on that. I don't like being wrong, but I'm quite prepared to admit it when I am.

Transaxle

The knock signal can be taken from #15 of the KLR. But you can't connect a light directly to it - the signal is much to short. In the diagrams we connected it simply to a fast ADC (1000 samples/sec). With this self developed ADC we recorded all relevant signal from the DME and the KLR:



The KLR is one of the biggest myths in automotive control units. Most of the information published in the web about this box is simply wrong.

mtnman82

Zero, assuming the scale on transaxle's data log is volts out, I think you might be able to do something simple like tap into the signal and run it into a digital 'one-shot' (I'd buffer it first with another piece of digital logic) that would turn on an LED for some amount of time (long enough to warn the user). An extension on this would be to use comparators set for different voltage levels (not sure if this is how the knock sensor works, or if it's only on/off) to trigger separate LED's for different levels of knock. A little microcontroller with several interrupts coming from different comparators programmed to set off the LED's (maybe even optional warning tones too ...) would make short work of this project, one would think. Hmmm, maybe I should get my patent attorney on this. Then again, things are almost never that easy.....

blodstrupmoen

knock sense

...

mtnman82

knock sense seems like a good start, but the duration (width) of the knock 'spikes' out of the knock sensor look to be very very short. You might not see the LED turning on until there's a lot of knocking (i.e. enough knock pulses to visually light up the LED, in which case it might be too late). I would think you'd want to light up the LED for maybe a second or something like that if you detected any knock signal over a certain threshold.

blodstrupmoen

Another option



Knock Link

Fen

Link now produce a newer product called knock block also.