Gear Reduction

[PHogan]

Hi everyone,

We will be competing in the Formula Hybrid competition with an all-electric car. Our motor spins up to 15,000 rpm at normal operating voltages (capped by the competition at around 300 V). Clearly this is much too high. In looking for a balance between the endurance and acceleration runs, we've decided that a gear reduction of 8:1 is in order.

Now, this is much too tall to be had all in one go. One thought we had was to incorporate an intermediate shaft mounted on two pillow block bearings between the output of the electric motor and the final drive/diff. This could give us the desired gear reduction by going 2.8:1 twice, or something similar in an effort to reduce sprocket wear.

Am I overlooking something here? Or does this sound like a reasonable idea to you all? Thanks in advance.

[Racer-X]

I'm curious about your total reduction. It is higher (numerically lower)than that on our car in 6th gear and we're powered by an f4i with a 13,500 RPM redline. You might want to check your math or reconsider your top speed.

Personally I would do something similar to the setups most baja cars have, with a "gear box" and a cvt. Your way would work IMO. Might want to look into the various gear drive FSAE cars and see how it worked for them.

[Jon Burford]

I don't recommend a CVT if your working with electric since your motor will have great torque low down.
I would maybe consider a gear reduction.

[Racer-X]

The only reason I recommend the cvt is to maximize the torque available at low speeds. Given that a motor only has so much torque, even though it is a flat curve, it won't have enough to be traction limited if its geared for a reasonable top speed. A two speed would work too.

[PHogan]

We're not thinking top speed so much as spending as much time as possible in the high torque, high power region. The torque curve starts high and falls off at 7000. The power curve starts low and peaks at 7000. The point where the two curves cross paths is around 9000 rpm, which corresponds to just under 60 mph in a 9:1 reduction (~20" tires).

Gearing the car faster might give us a higher top speed, but we'd be spending more time outside of the optimal power/torque region.

[Racer-X]

If you make peak power at 7k your gearing makes sense. I was expecting peak power closer to the 15k you mentioned.

[Owen Thomas]

I'm with Racer-x; re check your maths and your assumptions. I don't know how experienced your team is, but remember that the onset of wheel spin is way before a slip ratio of 1.0, and that the highest speed you will see is not at the end of an accel run.

Your way should work, but it will add a lot of extra parts and more importantly length. You might consider using a lower ratio and just not using the high part of your rpm range and sacrificing the potential torque. Also don't forget to give the guy who specced your motor a proper thump on the skull for not thinking about packaging http://fsae.com/groupee_common/emoticons/icon_wink.gif.

[PHogan]

Well, to get back to my original question, does the intermediate shaft seem like the right way to reduce gears by this much? Are there any other options we should look into, such as planetary gears followed by a reduction to the final drive? Thanks again everyone.

[Owen Thomas]

Probably the "right" way to do it is a gearbox for some of the reduction, then a chain drive for adjustability. Also probably not what you wanted to hear though. Normal spur gears should work fine; relatively low losses and very compact.

[Paul Achard]

Our hybrid team did a two-stage 520 chain reduction two years ago. It was bulky but it worked. Last year they downsized the chain ( I was surprised it survived!) and pulled off a single-stage (approx 7:1) reduction ratio.

I would suggest running some OptimumLap runs to get a good feel for how the FD affects lap times.

Also do not be too harsh on EV powertrain designers, guys! From my experience simplicity of EV packaging is directly proportional to price...

[Kirk Feldkamp]

What's so crazy about 7:1 or higher single-reduction? It's done on karts all the time. It's mostly just an issue of chain/sprocket load capacities and packaging space, right? If it wasn't such a prohibitive package, you'd see more of them in current EV's. The vast majority of current "state of the art" EV transmissions are only in existance in the form they are because that's the package that Tesla needed in the Roadster. Truth be told, these transmissions actually have two reductions inside them. Most everyone else seems to have gravitated to Borg's solution "because Tesla uses it." Ok, and because it has off-the-shelf availability, a proven reliability record (now), and there aren't a lot of other options in that market at the moment. Aptera used it. Coda uses it. Fisker uses an, uh, "interesting" solution that sorta makes you wonder "why?!" (It's two bigass motors with pinions feeding into opposite sides of a gigantic ring gear mounted to the rear diff). It's no wonder how the car ended up at ~6500 pounds. Weight begets more weight! http://fsae.com/groupee_common/emoticons/icon_eek.gif

From what I gather from most EV powertrain discussions, most people figure the "best" solution is to have a motor sized so that, with gearing, the forces generated during the constant torque portion of the RPM range are near the traction limit. The power level is sized based on the Marketing Department's acceleration claims. Same goes with the RPM range requirement, which is based on the desired mechanical top speed of the vehicle. While I appreciate the simplicity and robustness of this approach, it's far from "ideal". First off, it means that the motor you end up with is much larger than it "needs" to be. This is where the big, bulky traction motors you find on the market have come from. They're sized around basic, single-speed transmissions. As a result, the electrical to mechanical efficiency is solely determined by the instantaneous operating point (load and RPM).

Two speeds had a chance in the beginning... they really did! Tesla initially designed their transmission for the Roadster with two speeds. There were a *bunch* of reports that leaked about significant durability issues related to the two-speed trans. Keep in mind, we're talking about production vehicle reliability, not "FSAE reliability" (which is sort of a joke in "real world" terms). From what I can gather (through a bunch of analysis), the problem was they had sized the motor and chosen the transmission ratios to hit their acceleration targets... but hadn't taken into account the realities of what that meant when they translated that into real parts, with real inertia, and real wear. Sure, they could just meet the claimed acceleration specs, but they were beating the snot out of the transmission to do it. I think they had to shift it really fast (and thus *really* hard) to get the job done. Faced with the unfortunate situation where they would release a vehicle with dubious long-term reliability, they opted instead to redesign the motor (+30% torque, with similar weight and volume gain) and design a new, more reliable single-speed transmission. There were some law suits in there along the way for transmission suppliers that couldn't meet their claimed durability requirements. Fun times, lessons learned. Personally, I think Tesla's decision in the long run was smarter. Start simple, make it reliable... and if in the future your R&D people can work out a reliable solution that can provide you gains ($$$, performance, weight, etc.), then so be it. Until then, keep it simple and sell cars! Unfortunately, I think this has scared off other EV companies from more efficient solutions, at least in the near term. Fortunately, that doesn't mean you, the new generation of FSAE Electric designer, have to follow their example! Oy, I'm starting to sound like Z. Haha.

So what does two gears buy you? How about "N" gears? The first thing to look at is what gearing does for you. Gearing changes the forces generated. Gearing changes the RPM of the motor at a given vehicle speed. Gearing does not change how much power is transmitted at any given operating point. Most electric motors have a distinct full load lug line. They generate (relatively) constant torque until they hit their maximum power level, at which point the power remains (relatively) constant until they hit maximum RPM's. So let's say, for example, that you have two gears to play with. Now you don't have to generate as much torque (smaller motor possible) to generate the same peak tractive forces. You have a low gear! When you run out of RPM, you shift into high gear. By that point you're probably into the constant-power portion of the powerband, so you're simply extending the vehicle speed range at that point.

Just like IC engines, electric motors also have efficiency "islands" that can be mapped out. The analogy is the BSFC map that you'll develop for an IC engine. Through gearing, this means the tractive forces and RPM's are scaled proportionally. This also means that the efficiency islands scale around according to gearing as well. As gear count increases, the range of speeds and loads that the driver (or transmission controller) can achieve higher efficiency also grows. Two speeds is better in this regard, "N" is best. "N" gears is the elusive continuously variable transmission (CVT) that everyone talks about so often, usually in hushed tones. There are a handful of CVT varieties, all with various benefits and drawbacks (mechanical efficiency, packaging, durability, cost, complexity, etc.). Long story short, if you overlay everything... my analysis shows you can at least break even (if not improve slightly to moderately) on total efficiency with certain types of CVT's (full and half toroidal) on certain types of drive cycles. I've done this analysis with actual DOT drive cycles as well as a number of other representative real-world drive cycles. The difference is you can downsize the overall mass (and volume) of the powertrain package by roughly 40% compared to a similarly performing single speed motor+transmission combo. If you're clever about how you assemble things, you can dramatically improve the packaging of the powertrain. The general single speed layout *sucks* in my opinion because you have to pass a long driveshaft through the engine bay to the other side of the car. This volume becomes relatively uninhabitable due to the rotating and translating nature of the spinning shaft. I guess the point is, it may not be "simple", but there are definitely potential benefits to be had for those that choose to look for them.

PHulkster: To answer your question, it doesn't really matter how you get there. To be quite frank, there's probably no "optimum" way to get to the ratio you want. The solution you choose should fit within the resource allocation strategy (time, money, manpower) your team has decided upon. A lot of it will be determined by the packaging space available. Likewise, it also depends on how you plan to send the power to the wheels. Live axle? Differential? With the point that you're at now (based on the question), I would tend to error on the side of simplicity and low cost. Testing and development time, as opposed to designing and building some super fancy-shmancy solution, are going to help you place higher at comp.

-Kirk

[DougMilliken]

Originally posted by Kirk Feldkamp:
What's so crazy about 7:1 or higher single-reduction? It's done on karts all the time.
...
So what does two gears buy you? How about "N" gears?
My memory of karts (on dirt track) is that 9T sprockets don't last very long--but it's easy to watch the wear and change as needed.
...
Q. for Kirk (or anyone) -- does any electric drivetrain use a planetary two speed like the drag race Lenco? These shift with a clutch and are extremely strong, and compact.

[Kirk Feldkamp]

A 63T rear sprocket with small-ish pitch chain is downright tiny compared to 13" tires. Karts these days seem to bottom out around 11T drivers in applications with some power (on a 219 chain).

Yes, I have worked on the periphery of a vehicle that used a custom 3 speed Lenco-style transmission on a electric Class 4/5 application. The execution of the custom transmission they were using was awful, so it's hard to know how well it could have worked with proper development. When it did work, it worked just fine.

-Kirk

[mech5496]

http://www.yasamotors.com/sites/default/files/products/supporting_image/YASA-750%20Motor%20Efficiency%2020.03.12_0.jpg

This is the efficiency map of our motor, available via the manufacturers' website. As per my calculations, using any kind of reduction would lower the overall system efficiency (CVT's even more) and would add complexity and weight. We would love to have a reduction of about 1.4:1, but for the aforementioned reasons we opted for a direct drive. So, yes, there is potential in using a smaller motor for weight reasons, but IMHO it is offset by the weight/complexity/effort needed for a reduction system.

[Kirk Feldkamp]

In a FSAE type application, the benefits could be tough to justify. There simply isn't a lot of potential benefit for the risks (both programatic and performance). In a production vehicle application, the benefits could be very real. Downsizing and cost reduction are major factors. This extends beyond the powertrain itself, as battery packs (the primary cost driver in most EV's) can also end up being downsized. Ultimately, it comes down to the drive cycles you expec to perform. In something like a race, you're rarely part load for very long, and most of your time is either on or off the throttle. Due to the lightweight nature of the overall vehicle, even relatively small motors available have little problem providing enough juice even with zero or light gearing to rip the tires loose. In a heavier vehicle, however, if you want better acceleration with a smaller, cheaper, lighter motor, then you would need to approach it from another angle.

-Kirk

*edited for grammer*

[Pete Marsh]

I'm not sure I'm following you correctly here? Motor efficiency map shows reduction lowers overall efficiency?

By my reckoning, at 60kph (typical 13") and 750 Nm, a 3:1 reduction would only need to be better than 75% efficient to better this unit direct drive.

I would have thought it would be pretty easy to achieve this level of performance from many different designs.

Pete

[mech5496]

Pete,

We use 10". Given the fact that we need a top speed of about 130kph, we need a reduction of 1.3:1. According to our calculations, the torque figure of 750nm is already more than enough (we are traction limited anyway). So, yes, a single gear reduction may have helped overall efficiency a little, but would also mean more weight, cost, complexity (therefore less reliability) and manhours (which were not available). So, direct drive for us! http://fsae.com/groupee_common/emoticons/icon_smile.gif

[Charles Kaneb]

From kart experience, a really good (RK, CZ, or HTK Panther) #219 chain, will have a 10-50 hour lifespan on a 25-40 horsepower engine.

10-tooth 219 direct-drive sprockets last well - the confusion is that 10 and 11 tooth sprockets do not last long on clutched-drive karts because the bearing's undersized.

A 10-9x reduction should not be problematic, will use all COTS parts, and will come in under 10" OD. If you still need more you may have to accept the limited longevity of 9-tooth sprockets.

I doubt you will get a cheaper or lighter system than a single chain reduction - even if you burn up $40 in 9-tooth sprockets every hour of operation for a 4WD car.

[Charles Kaneb]

Here's the chart with the dimensions of #219 sprockets:

Asuza Engineering Sprockets (http://www.azusaeng.com/Sprockets/AzSDno219.pdf)

If your balljoints are 8.5" apart top-to-bottom, you should be able to fit a #219 sprocket in the sixties on the inboard side; to get a 7:1 reduction you'll need a 9-tooth driver. Good luck with eight or seven tooth drivers; it might get fun.

[PHogan]

Our motor manufacturer recommended that we not go with anything lower than 16 tooth. In one of their own applications, they were running a 12T and it was really pushing the chain. We'll be running with a fair bit more power (and especially torque) than the 25-40 hp engine you're talking about there. If we were to jump up 8:1 from a 16, it's a 128T final drive, which would bottom out at the very least on acceleration and probably at static as well.

[Racer-X]

I feel the team that ran into issues with the 12 tooth had other problems than the size.

We used to run a 12 tooth on our car with a 38 tooth rear, 520 chain, and we have the F4i and 7 inch wide tires. We never had to replace a chain on a car once it was assembled.

12 tooth sprockets are also fairly common on bikes.

[Kirk Feldkamp]

PHulkster,

You never said what size chain you're using...

-Kirk

[mech5496]

I assume that all the reported sprocket issues are due to limited chain wrap around the small sprocket. As the (drive) sprocket tooth count decreases, so does its' diameter, and that results into fewer teeth being loaded from the chain. If that's the case, a simple mechanism with two idlers (teflon maybe?) placed on the outer sides of the chain near the front sprocket would increase chain wrap, while providing a really simple (and much lighter than traditional) chain tensioning mechanism. I think Auckland did something like that in their last 4cyl car (do not remember the year) and also our team did so back in 2008, only with a single, chassis mounted idler on the lower side only. Multiple holes on its' mount allowed for tension adjustability. The solution was really simple and reliable, but I prefer the Auckland system...