h a l f b a k e r yNumber one on the no-fly list
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
This idea is closely based on Constantinescu's torque converter, but with some changes that should make it more durable. Also, since my idea uses more conventional parts, it's easier to see and understand how it works.
The engine drives a crank shaft with two cranks on it, one 90 degrees ahead of
the other.
Each crankpin has a connecting rod linking it to a pin on the planet carrier of an epicycle gear system.
The distance from the sun gear to the pin on the planet carrier is greater than the lengths of the cranks, so that while the crankshaft makes a 360 degree rotation, each planet carrier makes an oscillation spanning less than 180 degrees. (If the carrier pin is twice as far from the sun as the crankpin is from the crankshaft, then the oscillation should span 90 degrees, I think).
The sun gear of each epicycle gear set connects to a flywheel.
The ring gear of each epicycle gear set connects to a double acting ratchet, so that forward motion of the ring gear produces forward motion of the final output shaft, and backwards motion of the ring gear produces forward motion of the final output shaft. As a (usually beneficial) side effect, double acting ratchets also act as overrunning clutches, so that the output shaft can spin faster than the engine without forcing the engine to spin faster. A side effect that's always beneficial is that the output shaft is incapable of spinning backwards, so that if the car is started facing up a steep hill, it won't roll backwards when shifted out of park (or when the parking brake is disengaged).
It should be easy to see that when the output shaft is locked, the ring gears of the two epicycle gear sets are locked. When this is the case, the engine drives the planet carrier in an oscillating motion forwards and backwards, and the sun gears (and their flywheels) spin forwards and backwards (faster than the planet carrier). As the flywheels change speed, their momentum applies torque to the planets gears held by the planet carrier, which transfer that torque to the ring gears.
Thus, when the ring gears are blocked from moving, and the engine is running, torque is being applied to the ring gears. Furthermore, the faster the engine runs, the faster the planet carrier oscillates, and the faster the flywheels change direction... which naturally means that the faster the engine runs, the more torque is applied to the ring gears.
When the output shaft is able to spin freely, the ring gears are able to osscilate freely. Less of the energy of the engine goes into spinning the flywheels, and more of it into accellerating the output shaft.
In other words, it has the exact same type of functionality as we expect of a hydraulic torque converter, but without the hydraulics.
There are two important differences between my idea an the original Constantinescu torque converter.
One is that since I use two momentums (the flywheels) instead of one (the pendulum) the engine speed can be much more steady.
The second difference is that since I use rotational instead of linear momentums (flywheels instead of pendulums), the vibration inherent in the design is expressed as an oscillatory vibration instead of a linear vibration, which produces less noise and is probably less noticed by passengers.
If the two momentums were 180 degrees out of phase, instead of 90, the vibration would be eliminated, but there would no longer be a transfer of kinetic energy from one momentum to the other; instead kinetic energy would go to then from the engine, forcibly varying it's speed. To get a steady engine speed *and* no vibrations, 4 momentums 90 degrees out of phase with each other would be needed, increasing cost, weight, and complexity.
Original Constantinescu Torque Converter
http://fluid.power....const/const006.html
[goldbb, Feb 23 2009]
[link]
|
| |
Sounds potentially workable. |
|
| |
Let me just clarify... You're producing reciprocating motion in the planet gears , and using the inertia of a flywheel attached to the sun gear to provide a reciprocating torque to the ring gear. The ring gear then provides 'AC' torque that is rectified (via. ratchets or other) to provide the 'DC' torque and rotation for an output shaft. |
|
| |
I get the feeling that, if built strongly enough, this idea would work. However, I doubt it will be very quiet or reliable, or very kind to the engine, as you'll be loading it more or less in time with it's own pistons, meaning you're likely to enhance the harmonics going on in the crankshaft, which will then snap! |
|
| |
An add-on could be variable inertia flywheels (radially moving masses) that would allow for more control, reducing idle creep but improving low rev response. |
|
| |
I'm real sorry, but I just can't follow this without
some sort of picture. I know what is connected
to what, but that doesn't limit the possibilities
sufficiently to convey a design. Maybe I'm just
too lazy. Ok, probably I'm just too lazy. |
|
| |
[ColourClocks] I didn't get it from the description given (not that I can think of a better way to describe it), but more from the figures in the link. |
|
| |
Imagine an epicyclic gear train, with a large 'Sun' gear in the centre, with a ring gear on the same centre, but larger, and the gap between them taken up by a planet gear. |
|
| |
The ring gear (outside) is connected via. a rectifying system (forwards -> forwards, backwards -> forwads) to the car's geartrain. |
|
| |
The Sun (inside) gear has a flywheel attached to it, giving it inertia (resistance to change of rotational velocity). |
|
| |
Using a crank system, the car's engine pushes and pulls on the planet (connecting) gears and twists them around the axis if the other two gears. |
|
| |
This means that there is a backwards/forwards movement between the flywheel and the car's geartrain. If the movement is slow (idle engine) the flywheel will simply oscillate in time with the engine, but as the speed increaces, the flywheel will be less willing to move and therefore the torque transmitted to the drivetrain will rise. |
|
| |
The more simple analogy... Imagine you're stood in front of a kiddies roundabout. You have to move one of the handles the full range of your reach. If you move it at low frequency, then it's ok. As you try and move it faster (imagine the engine providing the force to your shoulders), remembering it has to move the full distance, think of your feet as being connected to something that will absorb energy by sliding sideways, the movement of your feet gives the crivetrain it's energy. |
|
| |
Skrewloose, thanks, your summary was clearer than my description :) |
|
| |
/I get the feeling that, if built strongly enough, this idea would work. However, I doubt it will be very quiet or reliable, or very kind to the engine, as you'll be loading it more or less in time with it's own pistons, meaning you're likely to enhance the harmonics going on in the crankshaft, which will then snap!/ |
|
| |
If I used only one crankshaft, connecting rod, epicycle gear, and rectifyer, then it definitely would be loading up the engine in time with its own pistons, and producing lots of vibration. |
|
| |
However, with two crankshafts, a quarter turn out of phase from each other, as one flywheel is slowing down, the other is speeding up. |
|
| |
I've done a small amount of mathematical analysis of the idea, and if the engine is running at constant speed, and the output shaft is not moving (because it's too complex for me to analyse if it is moving:)), then each flywheel's angle will be a sinusoidal function of time; similarly, each flywheel's angular speed will be a sinusoidal function of time, and each flywheel's rotational energy is a sine squared function of time. Since the two flywheels are 90 degrees out of phase, this means that the sum of the rotational kinetic energies is, in effect, some constant times ( sin^2(t) + cos^2(t) )... in other words, a constant. |
|
| |
This means if the engine turns at constant speed, and the output shaft isn't turning, energy is *not* transfered between the engine and flywheels; energy is only transfered back and forth *between* the flywheels. Thus, no work is done on the engine or by the engine, and it would indeed be /very kind to the engine./ |
|
| |
At least, it will be kind to the engine, when the geartrain is so heavily loaded that it can't move. |
|
| |
I don't have the math skills needed to analyse the system when the geartrain is lightly loaded. |
|
| |
I'm not particularly worried about the amount of power transferred to the drivetrain, that's what the engine was designed for! It's more the short timescale forces produced as everything changes direction. Given your reference to the maths, suggests to me that you're probably right, and the nasty transients are kept within the confines of the torque converter (assuming perfect sinusoidal movement, although it's probably pretty close). |
|
| |
Given enough time (which I really don't have) and either a C compiler or VB, I could probably write something to simulate the system for you. |
|
| |
My gut feeling is that there are gonna be some huge forces going on as things change direction (think of the con rod for a piston in an engine, and bear in mind that it's a small chink of Aluminium, then think of a flywheel, after being geared up by the effective ratio of the device!), meaning the con-rods and crank within the TC are gonna have to be bulky and expensive. |
|
| |
Assuming, however, that those aren't an issue, and that the weight and cost can be brought down compared to a fluid based TC, I think you may have them on efficiency (assuming some good bearings and correct choice of gear tooth profiles). |
|
| |
That's why my idea is to use an epicycle gear when splitting power between the flywheel and freewheels. |
|
| |
In all of the drawings I've seen of Constantinesco's torque converter, power is split by pushing and pulling on the middle of a rod, with the forces applied perpendicular to the rod's direction. This is a sure route to metal fatigue. |
|
| |
But a planetary gear not only is sturdier (being more of a plate or disk), it spreads the power across several planet gears. |
|
| |
Now, I'll admit that there will be lots of back and forth force on the teeth of these planet gears, and on the teeth of the sun wheel, and on the teeth of the ring gear... but there are lots of teeth for the load to be spread amongst. |
|
| |
So not only will there be a little less fatigue, the most likely type of failure is for a gear tooth to break off. This isn't precicely good, as it will produce horrible noises and partial loss of power, but at least it's not like the loss of a connecting rod, which would result in complete loss of power. |
|
| |
I understand that the use of a planetary geartrain will reduce the chances of catastrophic failure of the inertia system, but you're gonna need a pretty sturdy crank and con-rod assembly, as they're gonna be taking all the cyclical loads that exist internal to the system. It's this 'flywheel energy transfer' bit that I'm worried about. |
|
| |
The stress on each part can be reduced by increasing
the number of cranks, con-rods, planetary gears, and
flywheel, and freewheels, and decreasing the mass of
each flywheel. |
|
| |
This increases the cost, but splits the power
transmitted across more pathways, decreasing strain on
each mechanical path. |
|
| |
We can also decrease strain by increasing the size of
the sun gears, resulting in slower movement of the
flywheels for any particular engine speed. This results
in less torque transmitted to the output... there's
always a tradeoff somewhere. |
|
| |
PS: I think that the system would be quieter and more
efficient if each of the ring gears were to use a pair of
freewheels (aka overrunning clutches) instead of a
ratchet. |
|
| |