Quote:
Originally Posted by Wads
No problem but whats wrong with taking the flat spot in the acceleration curve as the shift time? the acceleration will drop off when pushing in the clutch and the curve will only rise again when the next gear is engaged.....
Unfortunately I think it will be a while before anyone here will be able to test the DCT with any accuracy..... 100Hz sampling won't pick up a claimed sub 8 millisecond shift, I don't think even the industy standard VBOXIII could resolve that (the measuring gear I use is made by the same company and is within -/+ 1% as accurate according to their own testing)
|
Do you mean the flat spot in the velocity vs. time curve (as it seems you used originally). The problem is that acceleration is the directly measured parameter, velocity is then derived numerically introducing inaccuracy. Furthemore there is simply a larger and more clear change to work with when using acceleration directly. Reading the knee of the velocity curve accurately is a lot harder than finding the drop in the acceleration curve.
On top of this you clearly must agree that the time estimated from both curves disagree fairly dramatically. That is the biggest problem I see with this data and that was about the first point I made. Two conflicting curves, they give totally different estimates and do not match in the time domain. Who knows what the best conclusion is.
Myself as well as most folks "in the know" seem to agree that 8ms shift times are not possible. There is a lot of claims about VW/Audi DSG accomplishing this, but more cerdible sources claim about 30ms. I think M-DCT will surely be in this same general neighborhood of times.
Lastly on the accuracy of the VBOX. You have to be careful on your very definition of "accuracy". Sure I bet the system is 1% accurate for some measurements but similarly there is no way it is 1% accurate for all measurements. As far as requirements go 200Hz should get us to withing 10 ms (don't forget Nyquist) accuracy and that should be about the minimum. 500 Hz - 2 kHz would be much better for really figuring out the differences down to 1 - 5 ms which would be required to see differences between individual drivelogic modes.