The usual answer to this question seems to be never, or rather once at the beginning and then leave it in second. But what if you're on a course which is too fast or too slow for that. I've got some data, and now some theory that changing down is not advantageous, changing up maybe if you're riding the limiter for more than 2-3 seconds.

The theory behind this is the seemingly blindingly obvious fact that to go faster you have to use more throttle. There is a rule of thumb that in order to go one second faster on a run, you have to be on the throttle for 2 seconds more. I tested this on some data from our last event, where my co-driver was going several seconds faster than I was. On one day he was on the throttle for 4.4s more than I was and went 4.4s faster. On the next day he was on the throttle 4.8s more than I was and went 2.7 seconds faster. So the observation is to go 1 second faster, you have to be in the throttle somewhere between 1 and 2 seconds more. Or conversely, if you're off the throttle for 1 second, you'll go something like 0.5-1s slower.

Changing gear of course involves being off the throttle. Looking at the data, I see that we're changing gear in something like 0.4-0.5s. So in order to make changing gears worth it, you have to find about quarter of a second from somewhere to make up for it. I've seen this in the data, it seems to cost you around 0.2s to change up, compared to a run where you didn't change at that point.

This would explain my observation that changing down for a slow corner never seemed to be advantageous. At an autocross school I was at last year, there was a exercise which seemed to be tailor made for changing down. It was a triangle, so each corner was a slow 120 degree corner. Taking the corner in second the car never seemed happy, it fell below its power band and felt dead coming out of the corner. But when I tried chainging down, I went slower still. So I'm assuming the time gained powering out of the corner never made up for the interruption in power when I changed up again.

This is reinforced by the fact that you never spend much time in first, even at the start. You very quickly get to a point where you have to change up. I have data which shows that from 10mph until you have to change up at about 30mph is about 2.2 seconds and is roughly independent of the engine power. (Ie its the same for Bertie (a JustaCooper) as Algy (a JCW).) Its limited by the grip the tires can produce. You can only spend 2 seconds on the throttle in first and you're losing a quarter of a second by changing up. Its difficult to make that worthwhile.

I've been looking at data I have and seeing how fast a MINI can accelerate in each gear. You can express this in mph gained per second, its an easy measurement to make on the graphs that Harry's Lap timer gives me. If you want to do calculations on this, I'd change that to meters/second. (2.2mph is about 1m/s.) I see that I can get about 10mph/s (4.5m/s, or 0.5g) in first. In second its 7.5mph (3.4m/s or .3g). Once you get to third, I see a big difference between our cars, Bertie will only do 2.2mph (1m/s 0r 0.1g) where as Algy can manage 4.7mph (2.1m/s or 0.2g). Engine power is now obviously the limiting factor, not tires.

So in first 10-30 takes you 2s, in second it takes you 2.6. So you might be able to make up the time spent changing gears, but I was never able to show that it was an advantage.

At the other end of second the question is when is it worth changing up to third. I got a lot of interesting data about this at the last two events we were at. These were a Pro Solo and a National tour event. Both of them were fast courses. National events are notorious for being fast, I've hardly ever seen courses that fast at local events. Also I had a co-driver (Alex) who was quite a bit faster than me, so I could compare the data between our runs.

The power curves of both cars (JustaCooper or JCW) are such that there is no rev range where you're better off in third than second. That is until you get to redline. At redline the limiter stops you accelerating. This handily negates the penalty for lifting on when changing up. Compared to riding the limiter, my data seems to suggest changing up only loses you about 0.02s. So you don't have to get very much of a win to make changing up worthwhile. The question might be more is there a convenient place to change down. If later there's a hard deceleration (i.e. going into a sharp corner) you can take advantage of that as an opportunity to change down without losing time. If there's no such handy time, you might find you lose time changing down again, or just being in the wrong gear.

The next question is how much time to you gain by changing up rather than riding the limiter. This depends on how long you're ringing the limiter, which is pretty obvious from looking at the data. With the acceleration in gear you've already worked out, its some relatively simple math (and Newton's laws of motion) to work it out. For Bertie its approximately 0.02s gained in 1s, 0.2s gained in 3s and half a second gained in 3s. These agree well with the data I was looking at at the tour. Riding the limiter for 3s was where changing up seemed worthwhile, for 5s it was big win.

For the JCW, the figures are about 0.04s gained in 1s, 1/3rds in 3s and almost a whole second in 5 seconds. As your car gets more powerful, the gain from changing up grows.

 

Very interesting read. Being a novice to the AutoX world I can already agree with you that down shifting will actually lose time. For me this happened one time specifically and against better my judgement. I knew with the course I ran, I would be staying in 2nd. But on a hairpin turn on my 5th run I slowed down faster/harder than planned, so in a scramble i attempted to shift into 1st gear and botched it. I ended up going past the gate and almost tried to push it into reverse. That cost me at least 4 seconds right there. And it was shaping up for a good run too.

Its nice to have some hard data showing that, in most cases, shifting to be a hindrance on quick run. (Overly simplified I know)

Thanks for the write-up!

 

I made three runs with my Lotus Elan on a course were I was hitting the rev limiter near the finish of the course. I compared that to shifting to third and back down, and shifting to third and then missing the down shift and coasting through the finish. There wasn't a significant difference in the times.

There was one (and only one) course where shifting down to first paid off when I did it right in my NA MINI. There was a pretty long pull after exiting that corner. I think that is covered by the road racing rule that corner exit speed is all important when the corner leads onto a straight.

And there was one course where shifting the Elan to third and leaving it there paid off but it took Andy McKee's skills to make it happen.

 

Good point, I hadn't thought of that. A long enough straight following a slow corner will allow you to make up the time lost. Long straights isn't something you often come across in autocross.

 

That prompted me to do a little theoretical modeling of coming out of a corner in first and changing up or just being in second. The results depend a bit on how fast you can change gear, but basically show there's an exit speed which gives you a break even point, pretty much regardless of how long the following straight is. If the following straight is short enough that you never get out of first, then you do win, but the question shifts to the straight following that next corner. Short enough is like 20-30m, with corner exit speed up to 22mph.

The break even speed is somewhere around 10-15mph depending on how fast you change gear (a faster change gives a higher break even speed). The time saved never gets above about 0.2s, unelss you get down to ridiculous corner exit speeds like 2.5mph. I don't think I've ever met a corner exit speed of less than 15mph so changing down isn't going to buy you much if anything.

If you can change gear in 0.4s, a corner exit speed of 20mph might buy you 0.1s by changing down. So I think I'll stick to not changing down, unless I get down to 15mph or less.

 

late and slightly not to do with the topic but,

i've changed into 3rd and had difficulty getting back into 2nd from being to aggressive with moving the shifter to the left (not used to cars with the "push harder to the left to get into reverse" setup). when you've got only 3 runs (at nats) that is a big risk to take. i've never tried downshifting into 1st for fear of putting it into R. I wish this car had a lockout system

 

Only one person has ever driven our car and had trouble with the 1-2 shift. They were pulling towards them as they were pulling down and ended up in the reverse gate while going from first to second. With the turbo there is no advantage to going to first so I've never had to shift into first during an auto-x run unless I've spun it.

Moving from 3rd to 2nd really isn't too bad. The shifter in these cars doesn't take much force at all so you just have to not get all worked up and try and jam the car into 2nd. Just a smooth gentle action with the tips of the fingers is more than sufficient.

If you've never had to shift from 3rd to 2nd then Nats probably isn't the place to try it. Chances are with a HS car you won't be riding the limiter too long as your speeds will be lower so not shifting to 3rd probably won't hurt you much. In the STX car I fully anticipate needing 3rd gear. We needed it at the Wilmington match tour in one spot and were on the limiter in 3 other spots. National level courses just seem to get faster and faster...

 

gotcha. first i've heard of someone having trouble going from 1-2... hope i don't catch it now lol. for my evo, i've had numerous spots where a 2-1 downshift really helped.. but i dont LFB to keep the boost up, so that might be it.

and yeah, 3-2 isn't bad at all, it is just me getting worked up during autox. i do it on the street all the time and there's no remote chance of getting to/through the reverse gate, just autox. and as btwyx said, for HS cars there's not much to be gained. at spring nats i'm not even sure i got to the rev limiter

 

bumping this - can I confirm: there is no reverse lockout? I know the further left portion isn't locked out, what about the gate to R itself? locked out when moving forward, or no?