Replies to This Discussion

DeveloperPermalink Reply by R_Lefebvre on August 21, 2012 at 5:46pm

Can hydraulic motors be efficient enough?  I always thought they were pretty bad.

Permalink Reply by Dwgsparky on August 21, 2012 at 6:28pm

Hi, we spent many months working on a real world system to spin brushes on series hyd motors, it does not work as the pressure drop across each motor is related to the torque produced. So a loaded motor will slow down and take more pressure so all the rest will slow down. The only way is indepenant speed control for each motor. See you at maker fair on Sat.?

Permalink Reply by Will Snodgrass on August 21, 2012 at 7:23pm

I just don't see how that could even approach the kind of instantaneous that we get out of an electric motor....not to mention the RPM's.... Or am I missing something?

DeveloperPermalink Reply by R_Lefebvre on August 21, 2012 at 8:03pm

Link to said efficient, high speed motors?

Any chance they would be easy to keep in phase?  (thinking of tandem rotor tradheli).

DeveloperPermalink Reply by R_Lefebvre on August 22, 2012 at 3:29pm

Does that include the plumbing though?  Flow will have to be high to achieve RPM.  Normally the flow in a pump is low which is probably how those are calculated.

Don't want to rain on your parade, just some healthy skepticism.

Permalink Reply by Vernon Barry on August 22, 2012 at 5:57am

Sorry, but I must jump in here. Being that I had to learn a ton from real world use of variable speed hydralics to build this http://www.youtube.com/watch?v=zPTvoabcU-k&feature=g-upl.

Yes, that's right, I play with turbines and hydraulics when I'm messing with an Arduino.

Point being, there are tons of gotchas whith hydro.

You know thinking about it with the proper fluid you might not need a tank and just have a closed loop pump running

This won't work, I've tried it. Here's why, First in your scheme, you should be using a variable displacment pump for speed, never a ball valve. All you would do with a ball valve is make the pressure relief valve at the pump spill like mad to take the fixed volume of fluid not going to the motor. This basically just heats the fluid till it boils from turbulence. Next, you cannot run a pure closed loop. You must cool the fluid and remove air before going back into the pump. The reason again is heat causes the fluid to expand and it will get air bubbles even in a closed loop. Trust me, I know this because I used the exact system on the first drivetrain. I used an Eaton model 6 pump motor combo back to back pump and motor. http://www.eaton.com/Eaton/ProductsServices/Hydraulics/Transmission...

Point being, I now run a model 11 with a 4 gallon external tank. The model 6 being a fuild volume of less than a quart and no charge pump to force air evacuation,the system would run only for about 10 minutes before I lost power due to air which is compressible. More info on the comparison http://www.google.com/url?sa=t&rct=j&q=&esrc=s&frm=...

Permalink Reply by Helitrasher on August 22, 2012 at 5:01am

Would the wieght and mechanical complication of the fluid, lines, actuators, tank and pumps (or pump and servo's) be less than four direct servos ?

You could possibly get away with three servos if you use a swash plate setup, and the movement required for -15º to +15º of  pitch would need at least 10mm of smooth and precise movement with very little force. 

 

Permalink Reply by Vernon Barry on August 22, 2012 at 7:48am

Nothing wrong with throwing ideas around. It's a little like Edison, he just started brute forcing the lightbulb and eventually found the one that worked.

The jetcart was the same way. Simple thought, I had an engine that has an ouput shaft that spins at 4,400 rpm governed just like a 3,600 RPM Typical engine. Sure, it's a little faster but nothing a belt drive cannot fix. Thus the basic concept was if the trans works on a mower, it should work here. So on the model 6, I read the specs saw that 4 HP curve and thought, "that's just a derating for longevity". Wrong, that trans will only transmit 4 Hp at best on a cold day for 10 minutes before the oil boils, even with cooling. Internally, the short path between the motor and pump is not enough. No charge pump to pressurize the entire system (high and low side) the oil boils on the low side and then that's if for power transfer. The model 11 has the charge pump which pressurizes the semi closed loop an prevents the oil from boiling inside the loop. It's still not perfect as after a couple of hard runs, I loose power even though the oil tank, cooler, and pressure loop seem to be a reasonable temp.

So Again, I tried no less than 7 different setups, multiple pumps, motors and combinations, broke lots of expensive parts and to this day, still want to upgrade to a better setup. I have several years of work into it.

Permalink Reply by Vernon Barry on August 22, 2012 at 8:05am

That said, back to your idea. I've has a very simple idea related to yours. Take 4 standard heli belt driven tail rotors and a central motor. The parts would be off the shelf, the tail rotors are already variable pitch, have the piano wire to put the servo near the center hub, simple reliable belt drive. Again, relatively cheap, off the shelf parts. I even think the only change to the arducopter software is to slow the update rate we normally use for the ESCs to a slower rate servos can use. Mechanically set the servos so the minimum position is neutral pitch. You might need to adjust the mechanical arm advantage on the servo or linkage such that then max servo (aka throttle) matches a reasonable pitch. A lot of helis use a small gear (for the belt drive) driven by a large spur and thus a large center gear, driving 4 (or more) belt shafts, could be very easy to laser cut the center plates for the bearing holders etc.

This is on my list to try one day and lends itself to a central gasoline engine as the power plant. The big question is, why bother, just use a single central rotor and swash and run the traditional heli model code. I think the loss of efficiency of 4 or more small rotors, and belts, doe'nt make up for the larger clean disk area provided by a single rotor. I think tuning wise and code wise, a bit easier for the multi-rotor, but the questions behind is what is the final goal. Single engine, long flight, heavy payload seems to be the goal, but I think single rotor is the better answer in that case?

I honestly think though that my idea is very easy to test and would work out of the box, with the only real change in software being the update rate and we already know how to do that. Cheap tailbooms and tail rotor parts could be had for around $25 -$30 each. Maybe $200 in hardware all up.