Two reservation for now... i'm thinking the weights can't perform work unless they physically get lower in the gravity field, and that angular displacements of the wheels which don't result in a change in weight height are probably not going to work.
But i need to test this with certainty - i still don't fully comprehend this system yet.
I also note this evening that i'm getting very flaky static torque readings from WM2D, and it's probably not consistent enough for the way i'm trying to use it. I'd been trying to plot the torques across a range of configs to see the trends, but it looks like i need to devise a more concrete means of measuring the forces, eliminating any sim quirks.
All very tentative for now.. i suspect though that if the force differential IS real, any displacements attempting to exploit it will be impossible for other reasons.
I've just completed a working transmission system - as the weights fall, rotating clockwise, they torque the wheel they're attached to anticlockwise, lifting themselves due to the force difference; basically levering themselves up and around.
So the transmission IS possible, and works exactly as intended. Which is nice. :)
But seeing it working has now given me better clarity on the potential torque differential: it IS confirmed - there's more than enough force from the paired armature torques, to lift the whole wheel anti-clockwise against (against! Who'd've thunk it!? AGAINST!!) the overbalancing torque, however seeing it working, it seems almost trivial..
- it seems all that is needed is a regulator mechanism, to limit the relative speeds of the wheel and armatures. If this relationship can be governed reliably then PM seems all but guaranteed. This would be an obvious use for the mysterious pendulums..
- the system currently depends upon a stator. I was hoping Bessler's solution was truly stator-less, as he's strongly intimated ("everything must, of necessity, go around together, there can be nothing about it that remains stationary on the axle") however this is little more than an aesthetic - the stator wheel displayed for clarity in the attached sim can be centered on the axle, and the pendulums likewise easily concealed. In other words, the entire stator can be hidden inside the wheel, giving the outward appearance of his "peritrochium". Maybe his claim to the contrary was just chaff - a little white lie? Or maybe this isn't it, and i'll find the pendulums can't work as intended?
- Bad: as i suspected, the output work by the wheel seems constrained by the actual physical lowering of the weights - no matter how the gear ratios are tweaked, once the weights reach vertical the acceleration will cease, and it'll eventually oscillate to a halt in that position.
- Good. Maybe. Perhaps biased pendulums can transfer the reduced armature angle to the wheel, resetting the drop? Momentum, and the force difference, would seem to be in our favour...
Tomorrow i'll see if i can get the weights back up to their 90° starting angles, using only the available rotor energy, and a swingy thing. No doubt, this'll just turn out to be tugging at bootstraps, like so many others, but it's a nice 'unchecked lottery ticket' to sleep on....
Yep cheers mate, and i'm sure you're right... i actually built one myself last summer, from Mecanno and roller pulleys from RS components - it wasn't up to much as a test rig, but was enough to confirm that with the requisite 1:1 gearing ratio (to keep the weights suspended at 90°) , there's zero torque. However, it also confirmed that there's a net torque on the entire system.
And i have a suspicion that this is the point of MT143, too - the net torque would be the key, if there's a way to use it...
I've just got back from work, so when i've finished reading the news i'll have another go at it - first off, i intend to run the sim a while, let the rotor get up to speed, then pause it, lock the rotor to the weights temporarily and see if there's enough energy on the rotor to re-lift the weights back up to 90°.
Presumably this experiment will have a negative result... Because if it's positive, then we're OU already..!
Edit: in fact, since it only took a second, i've done the test just now....
- using the last sim i posted last night, i let it run up to speed, then disconnected the transmission system and locked the armature to the wheel...
- the momentum of the wheel easily lifted the weights back up to the 90° point, and up beyond it even...
So, this certainly indicates an anomalous gain of energy - if i rig it to auto-reset without intervention it'll be true PM. Of course, i'll play around first and see if i can get the gain bigger or smaller, tighten the accuracy etc etc.
I also have other ideas for it - i want to try it the other direction, as i originally envisaged, but this'll require a condition in which the torque on the wheel exceeds the net torque from the armatures - which in turn will require the armatures to be counterbalanced, adding slightly more complexity (because the counterbalances need to rotate with the wheel, rather than hang from the axle, yet without cancelling the wheel's overbalance)....
If this can be achieved, the overbalancing torque will exceed the armature torque required to generate the overbalance. Again, this will keep the arms raised at 90° only this time the OB moment will work with, rather than against, the system.
And i still need to experiment with varying the respective radii - the previous results of which led me back to this config after giving up on it last year.
Also the pendulum idea i mentioned last night - lots of variables to tune there, too.
So lots of things yet to try, it's a simple system but with many possible permutations..
This took ages, had to let the sim run thru many cycles until the weights were hanging vertically, at which point the momentum on the wheel had topped out at 6,840 kg-m^2/s.
I removed the transmission to the stator wheel, and pinned the small wheel interconnecting the armatures to the main wheel, locking the arms.
The attached model is a snapshot of the sim at this moment.
The momentum is only sufficient to restore the weights back up to 83°, so we've lost 7 degrees of drop arc - and that's without factoring losses either..
Hence there's no point trying the pendulum idea i had last night - that won't help.
I will however try rigging it to run in the other direction tho, going with the OB moment this time. If i can figure a way to counter-balance the arms, negating their cost, but while still maintaining the over-balance, then there's still a chance of success..
I downloaded your twotorques. I have simply changed the appearance and added the measurements of the torques. Did you notice the torque of the left and the right motors add to the centre? The cycling variance is due to inertia I think.
Tried clockwise, output marginally higher due to going with rather than against the OB torque, but still sub-unity.
So i'm moving on to the next problem; counter-balancing the arms to reduce their anti-torque, but while still maintaining the overbalanced condition.
I suspected this may be impossible - that the counter-balance will cancel the OB torque, however the early indications are that this isn't the case, so this may be an alternative route to an asymmetry....
Hi Tarsier79 - that first model was just an animation, not an actual attempt at a sim! As i stated several times, it's run via motors, just to illustrate the intended motion.
The diverging torques i was on about arise when the armature radii extend beyond the wheel's center, so that both weights are effectively on the same side of the wheel.
Measuring the two torques in that case independently, by first pinning the arms to the wheel and letting it find it's center of gravity, and then subsequently unpinning them and pinning the wheel instead, allowing the arms to find their CoG, reveals a significant split, and it was this i was intending to use initially..
In fact, i'm glad you reminded me of this because i forgot to include this detail in the last round of sims - so i've just re-done one, including the larger radius arms.... and got exactly the same result as before..
I've also come to realise that any such force difference is probably trivial and incidental, and can likewise be achieved via gear ratios.
But there may yet be merit to the approach, if i can get it to work another way... right now i'm experimenting with counter-balancing the arms, reducing their cost of operation while hopefully preserving the full OB torque. I half-expected these to be conflicting requirements, but now i'm not so sure, still having fun here..
Right this seems crazy, but it appears i can have two of these wheels in parallel, in mirror image config..!
The thing is, these rigs can be configured to rotate in either direction by tweaking the gearing...
Hence i can have a second wheel, with inverted weight angles - this time suspended at 270° - that can turn CW or CCW...
Can you see where this is going?
The second wheel can turn in the same direction as the former, even though the weights are oriented 180° out of phase... this means the weights can counter-balance each other, cancelling their cost of operation, while the two main rotor torques will add together...
If this works out as planned the results could be very interesting..!
You have now made a connection with the "Hammer toy" found on MT #138, one raises as the other falls. Only energy or force required is to overcome friction loss.
Yes, i think you may be right; there IS no energy or work interaction in the classical sense - the only work is that produced by the force asymmetry.. but principally, it is a force asymmetry, rather than an energy one - the latter almost an epiphenomenon of the former.
This is quirky and unexpected - i'd strongly anticipated clearly defined input-output FxD integrals, yet here there's just a net unbalanced force.
Here's what i've done - as mentioned previously, i've attempted to decouple "shift energy" from RKE, concluding that i can't boost the latter, thus can only attenuate the former.
So i've counter-balanced the raised weights, so that there's two pairs; two left-pointing, and two right-pointing, and these are mounted on two separate but parallel / coaxial wheels.
However, instead of having one left and one right on each wheel, i've put the lefts on one wheel, and the rights on the other.
So one wheel's always OB clockwise, and the other, anti-clockwise.
Then, via further axially-mounted transmission wheels, i've connected each of the left and right weights together, still on their respective wheels, such that they counter-balance one another.
And this is the bit i was skeptical about... surely, if the intended OB load is supported by an equal and opposite load, the OB torque in nullified, no?
And apparently, this ISN'T the case - the OB torque remains, independent of the counterbalance!
This means the OB torque can drive the re-orientation of the weights, which are effectively free to rotate - nothing changes in height, hence no work has been done.. nonetheless, the machine is always overbalanced!
The first ones i made above, without counter-balancing, were all unity - it's possible to get them up to high and sustained speeds, but they eventually slow, change direction, re-accelerate the other direction then rinse and repeat, as the weights gradually tend towards vertical. Obviously, the net work done on the wheel can only be presumed to be equal to the inevitable drop in height of the driving weights.
With the doubled-up counter-balance method however, there's no possibility - no need - for the weights to progressively get lower.
They can even climb higher - a sim i ran a little while back gained GPE while running.
Currently i'm getting very low torques - from tomorrow i'll start trying to optimise the power ratios.. power conversion's the name of the game though (incidentally i've long thought this is the meaning of the scissorjack glyph).
In a nutshell, the enclosed model demonstrates four torques at play - two opposing lever torques, and two corresponding OB torques on their respective wheels. The two opposing lever torques cancel each other out, leaving just the OB torques to do their thing with minimal resistance.
Finally, at this stage i'd only classify this as an intrinsically unstable system (of which i've now seen many), compared to one that makes a clear per-cycle gain. Whether it can be coaxed to the next stage remains to be seen...
Tried raising the accuracies by a factor of 10... only result is the sim runs 10x slower, but the unbalanced torque persists.
Something i want to try later (gotta go to work now) is to add horizontal beams extending the weights further to the sides - so the lefties are positioned further to the left, and the rights, further rightwards..
Also, running thru many "out of memory" sim cycles (ie. enabling editing without resetting) i notice now that the outer weights do start to sag slightly, however then i notice that the inner weights start to lift slightly.. The thing is, they're connected to rigidly support one another, so it shouldn't be possible for either the lefties or righties to get lower without their opposites getting higher. Thus any such unreciprocated variation in height would have to indicate slack in the transmission, i think...
Edit: my point there is, the wheel's motion can't be proportional to the change in weight height, if the lefts are getting higher while the rights get lower... they're all the same mass, so the net change in GPE is zero... which would seem to leave an actual unbalanced torque as the culprit.
