There are several configurations they run at one of my local tracks, but they almost all end up requiring the same gearing. There is one configuration however, which is very different than the others in terms of gearing because it results in the straight being greatly extended, and which for me means that with the standard gearing I end being on the limiter for about 75% of the straight vs the normal 50% (which I normally view as a worthwhile trade off in regard to the other configurations).
I can and might just start throwing sprockets at this and watch the lap times, but I wonder if there isnāt some baseline rule that can help make sense of situations like this. (e.g., if youāre done winding out the engine, before x% of the straight you might want to look at different gearing).
fwiw, this lap around here was pretty decent in terms of outright time, but itās hard to look at that early-maxed-out rpm and not feel like there should be something better. (CVT anyone?)
Do a histogram of rpm and compare that to the powerband (hp, not torque). Youāll want most of your time in the powerband, so it may or may not be obvious from that which way you need to shift your gearing.
Hereās the histogram. My interpretation (which would be great to get thoughts on), is that even with going down the straight āon the limiterā for 75% of it, Iām still spending the majority of overall runtime in fairly low rpm ranges, which is not something thatās generally desirable to increase on the face of it.
I guess in non-qualifying situations the decision would come down to whether or not gearing for a higher speed down the straight could provide passing advantages or not (assuming one didnāt think they would start losing positions in the tighter stuff simultaneously).
In a qualifying situation, I dunno. I guess I would have to just try different gearing to see if the outright best lap time dropped or not by going with a higher gear ratio. In my roadracing days and early studies I was taught you always want to concentrate the most on going fast in the fastest sections, because thatās where the biggest gains can come, but that was based on having a gearbox. Single speed really can change that equation depending on the track, which I guess still just blows my mind, lol.
In addition to the histogram, I find it can be eye-opening to find the rpm window that gives 90% peak power and plot those horizontal lines over an rpm trace. From a quick google, thatās from 8k to 12k for a KA100. You can make two math channels (min_rpm=8000 and max_rpm=12000) to plot. Then set the x-axis to time instead of distance, and you basically visualize the histogram a different way.
I think you still want to bias towards the high speed stuff, so in your case that means dropping some teeth from the rear to gain more power at the top end. Thereās only so much you can do to defend against a kart with higher speed on the straights (typically one maneuver and now you have to stay off-line), but, in the slow stuff like exiting a tight corner, it is easier to defend (within reason and small gearing changes).
In the end though, there is no single right ratio when you consider a real environment. Especially if drafting plays a role, you wonāt know before the race how much boost youāll get (or not get) from that. It sucks to gear for a draft you donāt get, and youāll be sitting out in the wind if you make a pass. āCloseā is about as good as youāll get in reality. Looking back, itāll be easier to say āI shouldāve dropped a tooth or notā, but thereās no way to have that level of foresight.
Egad. Iām a programmer by profession and I lost patience trying to figure out how to deal with āmath channelsā. Sparse documentation and their webinars are long and seem to include a custom channels that most of their examples are based on.
If instead of a clunky UI this same functionality was done as an API:
a) it would be 1000% more consistent with modern day tech norms
b) people could document and share snippets of helpful code easily
c) in addition to just pulling up the desired information (as it does using math channels), it would also open up the possibility of using those results to calculate theoretical results on the fly (or against a target value(s)) to see what changing things would do, all without ever having to turn another wheel on the track.
d) sigh
I find the videos are way way too long - they could convey the information in about ten minutes instead of the hour. Agreed that for documentation would be very helpful.
In lieu of an API, they do give out a DLL for reading the XRKs. Not nearly the same thing, but itās something and it is the kind of awful solution you expect from AIM.
Iāve written a python script around it that I use for everything now, and the general use of it came from anotherās work. Iāll see if I can dig up the source I used and Iāll post it. You still need RS3 for downloading the data, but then everything else is in python (and Iām sure you could call the DLL with most other languages, too).
Wow @dodo this is very interesting. I looked things over and it does seem like you do just about anything you needed to with it. That said itās as raw as raw gets (what are you even rendering the output to/with?!), and makes most open source projects Iāve used look like commercially finished products, lol.
In all seriousness, respect for diving so deep with your data.
Iāve been pulled into the swamp. Been playing a little bit with creating custom math channels, based on finally finding useable information I cobbled together between the following three sources. One the two āuser manualsā I found was on the AIM site, the other Iāve never seen before and itās kind of better in some ways. Also, scrubbing the video below was very helpful just in knowing how to construct the functions/formulas. I donāt know if Iāll create anything useful anytime soon, but Iām glad to have the knowledge and the resources to know that I can if I want to.