Final Drive Ratio's-Determinates

[drivetrainUW-Platt]

What are your determinates for your final drive ratios.
Thinks I can think of:
Gear avaiability, transmission in particular:

You really cant find anything less then 11 tooth transmission gears, and with a 520 chain you wouldnt want to run anything smaller, thats just too tight of a corner for the chain to run around. The rear gear can be whatever tooth count, but the more teeth the bigger the gear gets. Packaging the drivetrain is greatly affected by the gear on the differential, and the brakes, but we run outboards at the wheels so no concern.

Acceleration and shifting.

I was talking to an Alum from Mich Tech and he said back in the 90's when he was on the team they ran a kit that put an 8 speed transmission in a Honda F2 (He also claimed over 100hp NA, after blowing up 8 engines in testing) The guy knows his shit, and knows it good so I dont doubt he knew what he was taking about.
Shifting takes time and there is a lot of room for driver/component error in a shift, especially with bad linkage and inexperienced drivers. Most people dont use more then 3 gears in a formula car on a FSAE track, so if the ratio is too narrow, those 3 shifts might not be enough. I always wonder if its better to use 2-3 gears around the track, or if 4th even 5th should be used. They are there and are adding weight, assuming someone can shift consistantly, why not use them.

Wheel/tire size.

This is the first year we are using 10" rims. I think the overall diameter of the tires is 18", only 2" dia difference in the overall height, so it wont make a huge difference, but its definitely a factor.

Engine

Where the powerband and where the operating RPM are will be what limites how much power goes to the wheels. A driver can shift more often, but if they arent getting into the powerband, they arent using the full potential of the gear. We have no dyno data on our engine, so I have nothing to go by for power curves for each gear and where they overlap.

What else am I forgetting? Any insite on the above mentioned would be great.

Last years setup: Honda F2 12T front 45 T rear with 13" rims, 20" Dia tires. We used 2nd and a little bit of 3rd gear at competition. The engine had a big deadspot at half throttle so the car was slow. Inexperienced drives as well, no real good data if the gearing was good.

[oz_olly]

Charlie Ping wrote a really good paper on final drive ratio titled "Shift-time Limited Acceleration Final Drive Ratios in Formula SAE". It describes how the acceleration time (0-75m) is very sensitive to both final drive ratio and shift time (latency). The lower your latency the more shifts you can do. So say you have a latency of 0.15 seconds with three shifts and reduced it to 0.12 seconds you can now shift once more therefore making your total shift times 0.45 seconds and 0.48 respectively. Depending on the numbers and the final drive ratio you can actually be faster by shifting slower.
In my thesis I have looked at this effect over an entire lap, not just the acceleration event. I am using two commercial simulation programs but I have only done the sims with one so far. My initial conclusion is that a lap around FSAE-A enduro circuit is far more sensitive to shift time than it is to final drive ratio. There are 'dead' spots where the final drive ratio puts you in a region of the gear box where the ratios pourly match the track, but if you increase final drive ratio you will quickly move to another region of the gearbox which results in fatser lap times. Based on our recorded data and the simulations we only need to use three gears. As I explained earlier the shift ratio vs lap time curve is not linear as the number of gear shifts per lap makes a big difference on the overall lap time. If any of you would like to see the surface and contour plots of latency vs final drive ratio vs laptime let me know.
I also plan to do the same thing swapping final drive ratio for transmission efficiency in a bid to prove that a dual clutch transmission is superior to a CVT.

Cheers

Olly

Team Leader
ACME Racing
UNSW@ADFA

[murpia]

Check these previous threads too:

http://fsae.com/eve/forums/a/tpc/f/332600868/m/48510378821

http://fsae.com/eve/forums/a/tpc/f/125607348/m/71810543821

Regards, Ian

[drivetrainUW-Platt]

Thanks guys, good start. I know shifting is a big factor, but I shy away from electric/pneumatic shifting setups. They are expesive, unreliable in many cases, and complex. Hard to beat a lever and a cable or rod.

I am more worried about lap speed then straight line acceleration. Acceleration you are driving for 4-5 seconds, autocross/endurance is a lot longer and the wrong setup will just compound the problem over and over.

[csquinn]

Its actually pretty easy to beat the rod and cable. With our old solid rod we couldn't get under 250ms (power to power). With pneumatics, we can consistently get <80ms. And the pneumatics involved are not very expensive. Just my opinion.

[drivetrainUW-Platt]

What about reliability. I cant speak directly for pneumatic, but may teams I have talked to over the years of FSAE have had at least some time on the track with the electric shifting not working (madtown a few years back as well).

Shifting is a whole can of worms its self. Does an electric/pneumatic setup varry the drive ratios vs a mechanical? Yes you can shift faster, but will you shift more and use more gears or will it just remove some of the "down time" between gears.

[csquinn]

Yeah, the electrical system we used in 2003 had problems. The pneumatic system we developed for the 2006 car was very reliable after the first couple weekends of troubleshooting.

The shifting method doesn't effect your final drive ratio. Thats only a matter of your sprocket sizing and such. A pnuematic (or any other fast system) just allows for less downtime. You can choose what final drive ratio (how many times you want to shift) independently. And its pretty huge for the driver to not have to take his hands off the wheel.

[drivetrainUW-Platt]

We ran mechanical paddle shifters last year, and although we didnt shift much, when we did it was RIGHT there. Foot clutch as well.
PPL cried and said there wasnt enough room for a thrid pedal but I never had a problem with it. THe shifter was kind of a last minute piece and didnt turn out to be as good as it could. As of right now I think they are planning on going back to a stick with hand clutch, we will see....

But back to drivetrain, out of curiousity how many gears were you using throughout competition CS?

[csquinn]

1st for skidpad only.
2nd through 5th for everything else.

[Welfares]

We've been running enduros in 2nd for several years with an R6 13-52 drive at the back.

The tracks have been getting increasingly tighter so we'll have to gear it down a bit more and use two gears this year.

Have a look at your tractable torque vs output torque at the wheels as low speeds, how much power do you need?

I think in 2003, Chalmers came over with something like a 3:1 final drive and ran acceleration in one gear.

[drivetrainUW-Platt]

Were they fast in accel though.
Without a time comparision, that statement means nothing.

I could say we used 15 gears in accel, but could have got last place.

[csquinn]

We got 3rd in 2005.

[drivetrainUW-Platt]

Thanks, how about the one gear run buy Charlmers?

[Charlie]

I don't feel like a big long post, because I wrote a big long paper that answers most of these questions already (or at least gives one numbers-based take on some possible answers). So I would recommend reading it, it isn't terribly long, complicated, or expensive (most schools can get it for free).

I wrote the paper because there was not much information on this subject, I went to the trouble to put it out there and get it published (it was not a thesis or any part of a coursework), and I still get emails from people who don't bother to read it but want a quick answer.

Here's a couple thoughts:

1) If you don't know your engine's power numbers than you can't do the calcs. This is kind of like building an upright from a material that you don't know it's properties. Not recommended. But, if you must do it, the answer is to physically test. Just like you might physically test an upright to see it's breaking point, you can just build different ratio sets and test them on the real car. Simple and very effective. Use some basic logic to come up with three or four different combinations. It's pretty easy with a chain drive.

2) I doubt the only front sprocket you can get is an 11 tooth, I'm certain you can get something other than that with a fairly common engine!

3) I would beg to differ that the shift time does not have an effect on the final drive ratio. Logic kind of dictates it ( a 1 second shift time would make it preferable to have a final drive that requires less shifting). See my paper for some numbers if you are skeptical.

4) I'm tired of people who call systems 'unreliable'. Failures occur because of design flaws. No system is inherently unreliable. If you don't want to take (or just don't have) the time to understand a more complex system that's an acceptable reason for not doing it. But don't call a system unreliable; that is not an engineering outlook, it is an outlook of someone who does not understand how design process works. There have been lots of failed mechanical shifters through the years with bent linkages, etc. If you don't put enough thought into a design, it fails. It doesn't matter what the concept is.

5) Looking at results from accel vs. the configuration will just leave you scratching your head; I think a lot of the basis for a good accel time is traction, weight split, etc. So benchmarking a shift strategy on this is pretty frustrating and I would not recommend it.

For example, Chalmers ran accel with 1 shift at FSAE-A (they did have a run with zero shifts but I was there and I saw the fast time with 1 shift, though both times were similar). We ran the same event with 3 shifts. Our times were very similar, with Chalmers being slightly faster (but also about 20 kilos lighter). They also had different tires and a different weight split. So many variables; don't try and back out just one and use that as justificaton. Used a numbers based approach to start and verify it with testing.

Hope this helps.

[Christopher Catto]

I am using two commercial simulation programs but I have only done the sims with one so far.

Olly, I'm not sure if I remember, did you mention the software in another post not long ago? Was it Tesis or Lapsim? I'd be interested to know what software you use.

I'm trying Lapsim for some stuff (not gearchange related just yet) and although it's ok I kind of wonder how much these sims relate to real life without the complexity I would expect them to have. I think for acceleration they should be very accurate though. Just when corners come it's obvious that drivers and tyres are very very complex to model.

Anybody tried putting their FSAE car for 1/4 mile runs in some videogame like GTR2? This game gives the possibility to change params and some of them reasonably accurately too (for straight line accel the Pacejka coefficients would be sufficient to define the tyres). Dunno if these games have a 1/4 mile track. Dont have much time for gaming now days. Man, what I would do with a bit more time and a fully funded Uni... http://fsae.com/groupee_common/emoticons/icon_biggrin.gif (time = money, money = money, time+money = a helluva lot of money)