Today's mechanical conundrum

A reader writes:

As soon as I heard about "Steve Durnin's D-Drive, [possibly] the holy grail of infinitely variable transmissions", my BS meter activated and the needle swung to "Possible thermodynamics violation".

But in his favor he's got an actual physical prototype...

...and is attempting to have a metal model made so its input and output power can be tested.

What do you think of the concept, and can you tell how on earth it works? I'm still trying to figure out how this is too different from CVT, other than maybe a wider range.

I'm still wondering if this is somehow impossible, but personally I'm open to the possibility that it's a similar step such as CVT and the in-article claims are typical science-journalism overestimations.


Oh no - it's another New Inventors prize-winner!

Fortunately, though, an infinitely-variable transmission (IVT) is not actually in any way related to perpetual motion. All it is, is a continuously-variable transmission (CVT) that has some way to run its variable "gear ratio" all the way down to infinity-to-one, also known as a "driven neutral".

(This is, by the way, not the same as just running the gear ratio up so much, billions or trillions to one, that the final gear in the train is functionally immobile, and could be embedded in concrete without having any effect on the load of the driving motor for some years. A true "driven neutral" could be driven at a trillion RPM for eleventy frajillion years, and never turn the output at all. A transmission that bottoms out at zillion-to-one gearing would, however, be perfectly usable as a real-world infinitely-variable transmission.)

Because it can gear down to infinity-to-one, this does indeed mean that this transmission doesn't need a clutch, which does indeed reduce complexity. Whether a real-world version of the D-Drive would be too big or too heavy or inadequate in some other more complex way for real-world duty, though, I don't know. But there's nothing crackpot-y about the basic idea.

As the video makes clear, the big deal here is making an IVT - actually, a mere CVT, that still needed a clutch, would do - that uses standard gearbox-y sorts of components, or can in some other way handle lots of power and torque without being unmanageably big, expensive and/or quick to wear out.

Normal CVTs have been available in low-torque machinery like motor-scooters for some time, and are now showing up in some mainstream, full-sized cars as well. But they're still a fair distance from ideal.

It's easy to make a CVT, you see. Here's one made out of Lego. It's hard to make a CVT that can handle lots of power. And yes, the fact that most CVTs contain some sort of friction-drive device is a big part of the reason for this.

Note, however, that there's a big difference between dynamic-friction CVTs like this one or the Lego one, in which friction between moving parts transfers power, and static-friction CVTs like this one, in which friction locks components together (as in a clutch!), and they don't wear against each other.

But even here, real-world elements muddy the water and make it hard for someone who doesn't actually work at the engineering coalface to tell whether they're looking at something genuinely new and useful, or something that's not new at all, and/or won't work. Here, for instance, is the NuVinci transmission, a friction-based CVT that spreads the friction stress between numerous relatively lightly-clamped spheres - it's related to the "ball differential" with which R/C car racers are familiar. The NuVinci's makers claim it's useful for high-power, high-torque applications. And maybe they're right. I don't know.

For an excellent example of the ugliness that can happen when somewhat specialised knowledge is repurposed by people who, at best, don't know what they're talking about, look at this particular piece of "water-powered car" nonsense, where the well-known-to-jewelers electric oxyhydrogen torch is claimed to be some sort of incredible over-unity breakthrough. This sort of thing happens all the time - it's just, usually, not quite such a blatant scam.

As the Gizmag article mentions, many commercial CVTs are also deliberately hobbled by car manufacturers. They force the transmission to stick to only a few distinct ratios, and also to want to creep forward when at rest, just like a normal automatic transmission. This isn't a limitation of existing CVT technology, though; it's just deliberately bad implementations of it.

(The manufacturers do this so that people who're used to normal autos won't be freaked out by a CVT. Those of us who'd like the superior technology we pay for to be allowed to actually be superior just throw up our hands, and cross those cars off the worth-buying list.)

I think one trap for the D-Drive could be the second motor that handles the ratio-changing - that might need to spin really, really fast in certain circumstances.

There's also the fact that this is only really an infinitely-variable transmission at one end of the ratio scale. The D-Drive can gear down an infinite amount, and right on through zero to negative (reverse) ratios. But unless I'm missing something, I don't think it can gear up at all. So the output shaft can't ever turn faster than the input shaft. This is a problem if you want to do low-power flat-highway cruising, when the engine's turning quite slowly but the wheels are turning very fast.

Normal cars have significant gear reduction in the differential, though - the "final drive ratio". Perhaps if you make the diff a 1:1 device, which shouldn't make it that much bigger, the D-Drive's output-ratio limitation won't matter.

The reason why I'm saying "might" and "perhaps" so often is that I, like the New Inventors judges, am not actually an expert on the very large number of mechanisms that the human race has invented over the centuries. The simplicity of the D-Drive makes me particularly suspicious. The D-Drive's mode of operation may be a little difficult for people who don't work with mechanisms all day to intuitively grasp, but there aren't many components in there, and none of them are under 100 years old. Actually, that's probably a considerable understatement; I'm not sure when epicyclic gearing became common knowledge among cunning artificers, but I can't help but suspect that a master clockmaker in 1650 wouldn't find any of the D-Drive's components surprising.

Sometimes someone really does invent some quite simple mechanical device, like the D-Drive, that nobody thought of before. But overwhelmingly more often, modern inventors just accidentally re-invent something that was old when James Watt used it.

To get an idea of the diversity of mechanical movements and mechanisms, I suggest you check out one of several long-out-of-copyright books full of the darn things. I think Henry T Brown's 507 Mechanical Movements, Mechanisms and Devices is the most straightforward introduction; it's a slim volume available for free from here.

(If you'd like a paper edition, which I assure you makes excellent toilet reading, you can get the one I have for eight US bucks from Amazon. Here's a version of it for four dollars.)

And then there's Gardner Dexter Hiscox's Mechanical movements, powers, devices, and appliances, whose full title would take a couple more paragraphs, which is also available for free.

Both of those books carry publication dates in the early twentieth century, but many of the mechanisms in them were already very, very old. Like, "older than metalworking" old. But several of them are still, today, unknown to practically everybody who's not able to give an impromptu lecture about the complementary merits of the cycloidal and Harmonic drives.

(You may, by the way, notice rather a lot of mechanisms in those old books that do the work of a crank. That's because one James Pickard patented the crank in 1780 - plus ├ža change. This forced James Watt, and many other early-Age-Of-Steam engineers, to find variably practical Heath-Robinson alternatives to that most elegant of mechanisms to get the power of their pistons to bloody turn something. Watt's colleague William Murdoch came up with a kind of basic planetary gearing to replace the crank. Planetary gears have, in the intervening 230-odd years, found countless applications - including the D-Drive!)

Getting back to Mr Durnin and The New Inventors, they both currently allege that the D-Drive is a "completely new method of utilising the forces generated in a gearbox". According to this Metafilter commenter and this patent application, that may not actually be the case, since 18 of the 19 formal Claims made in the application appear to have been turned down. But, again, I could be getting this wrong, because somewhere behind the impenetrable thicket of legalese I suspect the "Written Opinion" may be saying that the final Claim actually is patentable as a separate worthwhile thing. (See also this forum thread.)

This all has me thinking, again, about the repeatedly-demonstrated gullibility of The New Inventors. When I can bring myself to watch the show, I keep thinking - OK, actually sometimes shouting - about how I'd spoil the party by asking at least one out of every four inventors "would you be willing to make a small wager that your device is not fundamentally worthless, or a duplicate of something that's been in production for years?"

(Sometimes, I'd just say "Have you always dreamed of being a rip-off artist, or is it a recent career development?")

The New Inventors seem to not have much of a peer-review system to keep the show free of crackpots, scammers and ignorant inventors who're unaware that their baby was independently invented in 1775. Or maybe there's just a shortage of interesting inventions, like unto Atomic magazine's shortage of interesting letters, so they let even the dodgy ones onto the show as long as they look impressive.

Perhaps the people on the judging panel just studiously avoid saying anything that might attract legal action from an inventor outraged that someone dared to point out that his magic spark plugs strongly resemble 87 previous magic spark plugs out of which the magic appeared to leak rather quickly.

Personally, I suspect that some insight into the newness or otherwise of the D-Drive may lurk in the various kinds of differential steering used in tanks. (Many of those have also been implemented, needless to say, in Lego.) And don't even ask about differential analysers.

It doesn't even take a lot of searching to find other IVTs. Here's one that, like the D-Drive, has no friction (or hydraulic) components. Its highest input-to-output gear ratio is quoted as "five to one", which is weirdly low; perhaps it's meant to be the other way around.

I hope, I really do hope, that the D-Drive turns out to be a proper new and useful device. We can always use another one of those.

But I remain very unconvinced that something this simple, aiming to do this straightforward a task, really is useful, let alone new.

UPDATE: As mentioned in the comments, Gizmag have a new post about this.

To summarise: The D-Drive does not remove all friction components from the drivetrain, because it can only ever be a part of that drivetrain, and needs supporting stuff that'll probably need friction components. And yes, it would need a motor just as powerful as the "main" one to drive the control shaft.

And Steve Durnin is apparently proud of independently coming up with a system similar to Toyota's Hybrid Synergy Drive "Power Split Device". I must be missing something, there, seeing as if this is the case then the D-Drive probably isn't patentable, and probably wouldn't even be particularly marketable.

22 Responses to “Today's mechanical conundrum”

  1. Mohonri Says:

    I'd like to see a complete diagram of the device as well, since it seems to be lacking.

    The one hesitation I have about the device is this: when there is an actual load on the output shaft, how much load is placed on the motor which controls the speed or gear ratio? If it's anything significant (which I suspect it will turn out to be) then we're back to square one--how do we efficiently control the speed of this (high-powered) motor, including reverse? Heck, I could make an IVT with a single set of planetary gears, where the input shaft (let's say the sun gear, though it could be any of the three) spins at a constant speed and I control the output shaft speed (say the ring gear, but again, it could be either of the remaining two) by rotating the remaining gears (in this case, the planetary gears). This would let me do forward, reverse, and neutral, but requires significant power on the speed-controlling shaft.

  2. demiller9 Says:

    Yes, planetary gears convert torque as well as rotational speed. I think the inventor is overlooking this because he drives his demo with a small motor, and the counter-drive is using a similar size motor. The video makes the statement that the counter shaft does not need a powerful motor, but I believe they are wrong.

    Toyota's Hybrid Synergy Drive has a similar system (a planetary gear transmission) and uses a powerful electric motor/generator to turn the sun gear, with the ICE (internal combustion engine) driving the planetary gear carrier. The ring gear drives an output shaft which has other reduction gears before it gets to the wheels. The engine torque is split by a ratio determined by the gear sizes, part to the motor/generator, part to the output shaft. When the Toyota needs more torque on the output shaft they extract less from the generator. It's explained well on this page.

    I think Steve Durnin's prototype would fail if he connected the input shaft to a large motor or ICE but kept a small one on the balancing counter shaft. The small motor wouldn't be able to take the torque the large one would give it.

  3. me3 Says:

    It's a different layout of the Toyota Hybrid Synergy Drive.
    See Prius etc.

    2 or 3 motors, and an output shaft.

    Very clever computer balances the above to blend power from motor and batteries to wheels.

    Essentially it's just a differential.


  4. TwoHedWlf Says:

    Yeah, the actual power required for the control shaft would be a concern, it does look like it would take a good bit of power. But this looks like a workable enough system I find it hard to believe the guy has not made some estimates of the power requirement on that shaft. It seems like such an obvious thing and it doesn't set off my scam alert enough to think he's totally ignoring it. So it may be it's just a fraction of the total power, and when it's in top gear it would just be locked in place requiring zero power.

  5. Kagato Says:

    Gizmag's got a follow-up...
    Sounds like it might still be useful and clever, but not the complete picture.

  6. TwoHedWlf Says:

    And, I'm not sure if there is any torque multiplication going on there as the output shaft speed drops.

  7. trouserlord Says:

    The previous commenters have nailed it. You can certainly get the same characteristics from a regular differential. Here's a little thought experiment (if you want to do it for real use Lego and you'll get to keep your arms):

    Take any Volvo 240 that happens to be lying around, no limited slip diff and make it have a manual gearbox so as not to confuse people with torque converters and such..

    Jack up the back so the rear (driving) wheels are off the ground. Put it in 1st gear with the engine stopped.

    Now rotate one of the rear wheels. You'll find that the opposite rear wheel will rotate in the opposite direction at the same speed. No surprises there if you know what a differential looks like (if you don't, look at a Lego one). The outer cage is locked by the stationary engine so the rotation of your wheel and its attached sun gear transfers through a planet gear to the other sun gear and wheel.

    Now the fun part. Start the engine and let it run at a constant speed. With no load on either rear wheel, they will both turn together at the same speed. However, if you slow one of the wheels down by say, sticking your shoe against it, the other wheel will speed up. If you completely stop your wheel, the other will go twice as fast as the no load case. If you turn your wheel in the reverse direction, the other wheel will turn faster still. What you have here is a variable speed transmission with the engine as the power input and you as the control input. This is what the D-Drive is demonstrating.

    Now here's the problem: As soon as you load up the output (other wheel) with any torque load, that same torque is going to be applied back to your (control) wheel. You can get your driven neutral by letting your wheel spin, or a fixed gear by locking it completely, but anything else is going to need either a another large variable speed engine to drive the control wheel at the desired speed, or a brake to regulate the speed that it naturally wants to turn, which of course wastes energy and you've lost any reasonable level of efficiency.

    I would never wish to discourage anyone from tinkering with stuff for fun, but for those not specialist in their particular field of endeavour I offer this gift of two words: literature review.

  8. Red October Says:

    That's quite interesting.
    Dan, I wasn't aware that many commercial CVT cars limit the CVT to certain preselected ratios. Do all do this? When I sold Fords, I remember the CVT models were rather unusual to drive but not shockingly so. A bit sluggish, if anything, which made me wonder just why anyone was interested in this technology.
    That practice does remind me of some years of I think Jeep Grand Cherokees that actually had six-speed automatic gearboxes, with a controller set up to function as effectively a four-speed gearbox, although a different "3rd" was used when upshifting from second than when downshifting from overdrive. One gear was completely ignored. A different Transmission controller can give them the full six speeds, though.

  9. dynamitedave Says:

    Fendt and Massey Ferguson have a similar system for their CVTs.

    Using planetary gearsets, at the front of the gearbox input power is split into 2 directions, and at the rear the two paths are recombined. One path is a straight mechanical connection between the gearsets. The other path is via a hydraulics. A variable displacement pump is driven by the second path, which then drives a hydraulic motor connected to the output gearset.

    By altering the pump displacement, the mechanical path shaft speed is controlled (path one). Zero displacement results in zero speed in path one as it is easier to spin the unloaded pump than try to move the tractor via path one. Full displacement gives full speed with power going via both paths.

    I'm told that at slow travel speeds, most of the power is transmitted via the hydraulic path. I don't know the ratio at full speed but a high percentage is via path one.

    These gearboxes manage 30 metres/hour at full engine speed (2200 rpm) to 60 km/h (local laws depending) at 1600 rpm (eco mode). These gearboxes do have a high/low ratio after the CVT to increase their efficiency, which gives a max 25ish km/h in low. I think the biggest engine is 380 hp. Fendt do have their Trisix with 600? hp, but it runs 2 gearboxes to power the 3 axles.

    The advantage the Fendt system has is that the speed control of the output gearset doesn't waste very much power, unlike the impression I get from the above invention. The Fendt system recombines the two paths with both paths applying their force in the same direction, with losses from the gearsets and hydraulics as you would expect.

    For some reason, US farmers can't handle CVTs. For that market, MF has had to put a "gearbox mode" option in, so there are 19 or so distinct ratios. Sort of defeats the purpose of CVT.

  10. RichVR Says:

    Sorry to threadjack but I have a problem and with all of the engineering knowledge here I thought someone could help me. How do I go about removing the last AA battery of three (two out already, second one was tough)from a three cell LED Maglite? I'm afraid to bang it any harder than I already have. If it makes a difference its a Duracell alkaline. The battery was brand new, as were all three, when it was put in.

    It saw moderate use and then the voltage dropped to the point that it just stopped working. I tested the LED by putting two brand new batteries in (just for a quick flash) I never mix old and new cells.

    Sorry for the interruption.

  11. TwoHedWlf Says:

    Hmm...I don't think they should be stuck to begin with?

  12. Popup Says:

    Apparently you're not alone, Rich.
    Quoth Maglite:

    When this happens, it almost certainly means that the batteries have leaked and are stuck inside the barrel. Note: Batteries normally swell before leaking causing them to get stuck inside the barrel. Mag Instrument does not warrant against battery leakage, however, we do have agreements with Duracell, Eveready and Ray-O-Vac that allow us to replace your flashlight. If one of these brands of batteries is stuck (by leaking in your flashlight), send the flashlight to us and we will repair or replace the entire flashlight at no charge to you.

  13. Popup Says:

    And while on the subject of LED lights...

    I have a couple of cheap crappy LED lights from our favourite crap purveyors that seem to have an electromechanical problem. Fiddling with the screw joins often makes them work, which has led me to believe that it's corrosion/oxidation on the threads. Can someone recommend me an electrically conductive gloop to put on the threads? You can buy 7g of 'silver grease' on ebay for USD 50 (with shipping), or half a kilo of copper anti-seize paste for a fraction of that. Would either work? Or should one stick with graphite?

  14. RichVR Says:

    Heh. I'm rather embarrassed that I didn't think to go to their website myself. Thanks for that. For an $8.00 handling fee I might as well but a new one and cut open the stuck one with my Dremel just to take a look.

    Strange that I can't remember the last time I had an alkaline cell leak on me. Oh well.

    Thanks again.

  15. Jens Says:

    Popup, stick with graphite. Silver or copper could lead to galvanic corrosion, if your light is made of aluminium.
    You coulde use plain vaseline or another grease, something usually makes contact anyway.

  16. human909 Says:

    After seeing new inverters and the differential based CVT here on dans data I had a good look at this issue.

    I quickly realised that differentials and planetary gears add a degree of freedom to the rotation that if appropriately controlled could be used to provide an IVT. Torque and RPM can easily be traded off for each other.
    Mathematically this second input needs to be analogue in order to provide an IVT either variable brake or a variable motor are obvious solutions.

    Having two powerful input motors seems that oppose each other doesn't seem efficient especially when you consider that ONE of them has to be variable in speed anyway. Brakes either waste too much power OR (if you gear it down and reduce power) require high torque components.

    My conclusion was the best bet was to 'brake' the motor using a generator whose electromagnets are controlled to yield braking effect. In that way most of the braking energy is recovered while providing an IVT. Ideally suited to hybrids. I thought that I was clever little inventor. :-)

    Though after reading the comments today I discovered (unsurprisingly really) that this is essentially what the Toyota's Hybrid Synergy Drive does.

    But all this comes down to the fact, IS THERE A NEED? The answer is probably not for most applicatios. When you have now have 7 speed dual clutch gear boxes what need is there for IVT? Good efficiencies for vehicles (and most other applications) as only needed over a range rarely exceeding 10x. With 7 gears this can easily keep the engine withing 25% of its optimal range. Simple gear boxes do the job very well without added complication or efficiency losses.

    IVTs have already found their place in farm machinery where there is need for extremely low speeds and high torque. In the case of Hybrids it fits in well in the existing needs and requirements of the vehicle.

  17. -AD- Says:

    Another variable transmission exploiting the properties of planetary differential gears comes to mind, the prototype Paxman-Fell railway locomotive.

    It involved four different engines, four unidirectional clutches, three differentials and four hydraulic couplings.

    Unsurprisingly, it proved somewhat less than reliable.

    Excellent description of it here:

  18. Mohonri Says:

    Personally, I think the whole point of an IVT will eventually become moot. Why? Because I think series hybrids (the Chevy Volt being the first example) have a good chance of being the future. Think about it: you can run the engine (and it can be any fuel-driven engine, from a fuel cell to a turbine to a plain ol' internal cumbustion engine) at the high-efficiency speed, and let the electric motor (and its drive circuitry) do all the fancy footwork on speed control. With the addition of batteries, the fuel-driven engine can be much smaller (yay peak-shaving!) and efficient--no running a 250HP engine at 50HP most of the time.

  19. human909 Says:

    Combining the last two posts and you have what has been standard for decades in heavy vehicles with extremely high torque requirements. Diesel-electic is common place in trains and heavy mining industry.

  20. Red October Says:

    Quite true; I understand that GM brought people from the Electro-Motive division (responsible for their locomotives) in to develop the drivetrain for the Volt. It promises really exciting things for motoring; I can't wait to see how it develops.

  21. PhilArmstrong Says:

    Speaking of random inventions on the internet, has anyone else been spammed by Steorn recently? I just got hit with an invitation to join their *new!* *shiny!* *forum! * (which I won't do them the favour of linking to) via some low-life spammer in the US. Following the various keywords on google leads to a nether world of free energy true believers...

  22. Joshua Says:

    When I first read this post I had alarm bells going off in my head about "PERPETUAL MOTION RUN YOUR CAR ON WATER HARNESS THE POWER OF THE SUN" type bullshit, but after reading your post, it does makes sense that this is possible, but perhaps less than practical with the advent of electronic cars.

    Love the name of your blog btw.


Leave a Reply