The following is my summing up, just posted in response to a chap on
http://www.rcuniverse.com/forum/m_10...m.htm#11151185
who found when he dismantled his units that much of the problem arises from inadequate clearances on the moving parts, causing them to stick and jam.
"for information.
Thanks for your most helpful post, confirming as it does - and more - my findings. What follows is intended to be my final assessment of these units:
As before, the fundamental design error is the use of a lead-screw, notoriously inefficient at transmitting power because of high frictional losses inherent in rubbing the two threaded surfaces one on the other. Unless of course a ball-bearing lead screw were used, though unlikely to be available in this size or suitable for the forces involved if it were. Nor does the small size of the motor help efficiency.
Having now removed the drive motor and electronics (using a cutting disc as I cannot find a Hex key to remove the screws) I can confirm that what appears to be just a motor driving the in-line lead screw does in fact have a tiny 3 stage gearbox built in to the end. I suspect that in those small sizes with a small number of teeth efficiency is not too good, It is at least arguable that were the overall gear ratio and the thread pitch both twice as great the inefficiency would be significantly less and available pull significantly greater. In my view the speed could be halved to improve force, and still be fast enough (though as doubling the ratio would also double the force, speed would not fall by that same factor of 2)
This might go some way to reducing the several problems which arise from such low efficiency:
1/ The system struggles to overcome friction in the lead screw and (as you mention) in the locking pin etc. My own test on a complete unit, done by holding a finger lightly against the wheel when it is being raised, shows that available pull before stalling fluctuates a great deal - a graph would be a saw-tooth. This is because of the less than perfect fits and surface finishes moving one on another. In my case the margin for error - the difference between worst running current and electronic trip - was so small as to be unusable, the slightest touch would stop the wheel moving.
2/ 5 cells instead of 4 made marginal improvements but nothing like enough to make it reliable.
3/ One time in 6 or so, one leg or the other would refuse to unlock and extend.
4/ Unwanted cycling of the controls, presumably because of the electronic trip being activated, occurred quite frequently. At times merely the metal to metal contact of a screw touched against the mounting frame would cause one or other unit to cycle.
5/ On 4 cells the maximum current when the legs continued to move, for 2 units, was close to 1 Amp. When stalled 2 Amps with a buzzing sound until the trip activated after 2 seconds or so. On 5 cells it was 2.7 Amps.
6/ Because such currents would cause terminal voltage of normal Rx batteries to drop to dangerous levels - loss of decoding, delayed recovery, and slow throttle fail safe activating - I would not dream of operating them from the Rx battery even if I could make them work reliably - unless perhaps in the smallest size - and in my view and those of others I have read, using the Rx battery is just asking for trouble.
Having removed the motor and cut off its lead screw I could operate the unit by hand. It immediately became obvious that the transverse locking pin was marginally tight in it slot, on one side. This caused it to tilt, making things worse.
It was then also clear that the sharp corners in the frames between the locking slots and the transition slots were too tight. There is no need for the corner to be sharp, so careful work with a needle file freed the whole thing up. Before doing so, even without the undercart leg fitted, pulling the linkage free from the locks took significant force, afterwards no force to speak of.
On the second unit I found another problem - movement was stiff even without the u/c leg fitted. Slackening even very slightly the screws that hold the frames together eliminated the problem. This turned out to be because the two plain brass bearings set into the side plates were not quite flush, reducing the effective spacing by about 0.008" which was just enough to squeeze the pivoting trunnion. I linished them flush and solved the problem.
That the fit of these parts is so critical helps explain why some have problems and some do not - all depending on the clearances (which of course have to be minimal because of the 30 to 1 magnification seen at the wheel)
Having removed the drive system and freed up the mechanism I decided that marginal operation and unwanted cycling were simply not acceptable, and the obvious solution would be a more efficient motor with conventional gears. Fortunately I had precisely such a mechanism immediately to hand - a servo!
Initial tests showed that even with a 7" leg and 4 ounce wheel, a bog standard 60 degree Futaba ball race servo would handle the forces involved in one unit. A 17 kg cm 160 degree retract servo would of course do so with torque to spare. Because the aircraft in question is a TopFlite Thunderbolt 64" one, already built, it would have been very difficult to retro-fit the long linkages that would have to pass between the wheel wells and the upper skin to reach the centre servo bay, and bench tests showed that such long linkages need to be very stiff to prevent bowing that would prejudice secure locking at one end. And of course tricky to set up because the locking pins are not visible when installed in the wing.
Plan B was therefore to cut off one of the unwanted end of one of the frames and bolt directly to it one of the two retract servos I had to hand, so that the output arm of the servo was on the centre line of the retract unit. (A single nut and bolt will suffice as long as the far end of the servo is then supported in the wing.) By a stroke of luck I had a steel rod with a threaded end that matches the transverse pin and installed it as a push rod about 2" in length between the locking pin and the servo arm. These being proportional servos it was easy to match the travel to that of the locking positions before installing in the wing.
I had to cut away enough wing skin to allow the servo to fit, but not that much and not difficult to rework. I could have used the frame and servo to reinforce the strucure but on this wing it does not seem to be necessary.
I now have - at considerable overall expense in parts, time and frustration - undercarts which work every time, do not jam up, which lock precisely in place at either end and which have considerably more servo power available than they need - so much so that the servo does not noticeably slow down when lifting the leg. This incidentally confirms a point I made earlier, that the efficiency of the standard unit is of the order of single figure percentage points - because the servo arrangement is so much more efficient it wll not result in excessive drain from the Rx battery.
As for future retract models - at least one of the small cheap units, working on much the same principle as the mechanics of the E Flites, are entirely satisfactory in models up to say 56" span. At least that same unit can have grub screws fitted to adjust minutely the play at each end.
I cannot envisage paying E Flite prices for units that I then have to convert to servo operation after fine-tuning the fit of the parts - I did so on this one occasion because it was the easiest way out of where I had found myself.
I have taken a few photographs that I could put on my web site if anyone wishes.
If on the other hand E Flite decide to offer just the mechanisms without the drive system - as Robart do as an alternative to their pneumatic retracts - and at a significantly lower price, then they would be an attractive option. Especially if the frames were arranged to accept servos bolted on.
All of this and previous postings are of course only my opinion, others are free to disagree, and I have no commercial reasons whateever for making these observations.
Sorry if anyone thinks this long-winded, but I tried to make it complete in itself and as helpful as possible. I will put the same on the few other web sites where I have seen this subject raised.
Idris
The following is my summing up, just posted in response to a chap on
http://www.rcuniverse.com/forum/m_10...m.htm#11151185
who found when he dismantled his units that much of the problem arises from inadequate clearances on the moving parts, causing them to stick and jam.
"for information.
Thanks for your most helpful post, confirming as it does - and more - my findings. What follows is intended to be my final assessment of these units:
As before, the fundamental design error is the use of a lead-screw, notoriously inefficient at transmitting power because of high frictional losses inherent in rubbing the two threaded surfaces one on the other. Unless of course a ball-bearing lead screw were used, though unlikely to be available in this size or suitable for the forces involved if it were. Nor does the small size of the motor help efficiency.
Having now removed the drive motor and electronics (using a cutting disc as I cannot find a Hex key to remove the screws) I can confirm that what appears to be just a motor driving the in-line lead screw does in fact have a tiny 3 stage gearbox built in to the end. I suspect that in those small sizes with a small number of teeth efficiency is not too good, It is at least arguable that were the overall gear ratio and the thread pitch both twice as great the inefficiency would be significantly less and available pull significantly greater. In my view the speed could be halved to improve force, and still be fast enough (though as doubling the ratio would also double the force, speed would not fall by that same factor of 2)
This might go some way to reducing the several problems which arise from such low efficiency:
1/ The system struggles to overcome friction in the lead screw and (as you mention) in the locking pin etc. My own test on a complete unit, done by holding a finger lightly against the wheel when it is being raised, shows that available pull before stalling fluctuates a great deal - a graph would be a saw-tooth. This is because of the less than perfect fits and surface finishes moving one on another. In my case the margin for error - the difference between worst running current and electronic trip - was so small as to be unusable, the slightest touch would stop the wheel moving.
2/ 5 cells instead of 4 made marginal improvements but nothing like enough to make it reliable.
3/ One time in 6 or so, one leg or the other would refuse to unlock and extend.
4/ Unwanted cycling of the controls, presumably because of the electronic trip being activated, occurred quite frequently. At times merely the metal to metal contact of a screw touched against the mounting frame would cause one or other unit to cycle.
5/ On 4 cells the maximum current when the legs continued to move, for 2 units, was close to 1 Amp. When stalled 2 Amps with a buzzing sound until the trip activated after 2 seconds or so. On 5 cells it was 2.7 Amps.
6/ Because such currents would cause terminal voltage of normal Rx batteries to drop to dangerous levels - loss of decoding, delayed recovery, and slow throttle fail safe activating - I would not dream of operating them from the Rx battery even if I could make them work reliably - unless perhaps in the smallest size - and in my view and those of others I have read, using the Rx battery is just asking for trouble.
Having removed the motor and cut off its lead screw I could operate the unit by hand. It immediately became obvious that the transverse locking pin was marginally tight in it slot, on one side. This caused it to tilt, making things worse.
It was then also clear that the sharp corners in the frames between the locking slots and the transition slots were too tight. There is no need for the corner to be sharp, so careful work with a needle file freed the whole thing up. Before doing so, even without the undercart leg fitted, pulling the linkage free from the locks took significant force, afterwards no force to speak of.
On the second unit I found another problem - movement was stiff even without the u/c leg fitted. Slackening even very slightly the screws that hold the frames together eliminated the problem. This turned out to be because the two plain brass bearings set into the side plates were not quite flush, reducing the effective spacing by about 0.008" which was just enough to squeeze the pivoting trunnion. I linished them flush and solved the problem.
That the fit of these parts is so critical helps explain why some have problems and some do not - all depending on the clearances (which of course have to be minimal because of the 30 to 1 magnification seen at the wheel)
Having removed the drive system and freed up the mechanism I decided that marginal operation and unwanted cycling were simply not acceptable, and the obvious solution would be a more efficient motor with conventional gears. Fortunately I had precisely such a mechanism immediately to hand - a servo!
Initial tests showed that even with a 7" leg and 4 ounce wheel, a bog standard 60 degree Futaba ball race servo would handle the forces involved in one unit. A 17 kg cm 160 degree retract servo would of course do so with torque to spare. Because the aircraft in question is a TopFlite Thunderbolt 64" one, already built, it would have been very difficult to retro-fit the long linkages that would have to pass between the wheel wells and the upper skin to reach the centre servo bay, and bench tests showed that such long linkages need to be very stiff to prevent bowing that would prejudice secure locking at one end. And of course tricky to set up because the locking pins are not visible when installed in the wing.
Plan B was therefore to cut off one of the unwanted end of one of the frames and bolt directly to it one of the two retract servos I had to hand, so that the output arm of the servo was on the centre line of the retract unit. (A single nut and bolt will suffice as long as the far end of the servo is then supported in the wing.) By a stroke of luck I had a steel rod with a threaded end that matches the transverse pin and installed it as a push rod about 2" in length between the locking pin and the servo arm. These being proportional servos it was easy to match the travel to that of the locking positions before installing in the wing.
I had to cut away enough wing skin to allow the servo to fit, but not that much and not difficult to rework. I could have used the frame and servo to reinforce the strucure but on this wing it does not seem to be necessary.
I now have - at considerable overall expense in parts, time and frustration - undercarts which work every time, do not jam up, which lock precisely in place at either end and which have considerably more servo power available than they need - so much so that the servo does not noticeably slow down when lifting the leg. This incidentally confirms a point I made earlier, that the efficiency of the standard unit is of the order of single figure percentage points - because the servo arrangement is so much more efficient it wll not result in excessive drain from the Rx battery.
As for future retract models - at least one of the small cheap units, working on much the same principle as the mechanics of the E Flites, are entirely satisfactory in models up to say 56" span. At least that same unit can have grub screws fitted to adjust minutely the play at each end.
I cannot envisage paying E Flite prices for units that I then have to convert to servo operation after fine-tuning the fit of the parts - I did so on this one occasion because it was the easiest way out of where I had found myself.
I have taken a few photographs that I could put on my web site if anyone wishes.
If on the other hand E Flite decide to offer just the mechanisms without the drive system - as Robart do as an alternative to their pneumatic retracts - and at a significantly lower price, then they would be an attractive option. Especially if the frames were arranged to accept servos bolted on.
All of this and previous postings are of course only my opinion, others are free to disagree, and I have no commercial reasons whateever for making these observations.
Sorry if anyone thinks this long-winded, but I tried to make it complete in itself and as helpful as possible. I will put the same on the few other web sites where I have seen this subject raised.
Idris
and another problem found after posting the above:
and one more problem! When fettling the 2nd unit this evening I realised that movement was quite stiff even without an U/C leg attached. As others have done I loosened the frame screws and the stiffness disappeared.
Using digital calipers I established that the distance between the two frame parts provided just enough clearance for the trunnion and its to think plastic shims - but the brass bearings force fitted into the frame parts were about 4 thou each proud of the surfaces, reducing the spacing by 8 though - enough to make movement noticeably stiff. The plastic shims spacing the trunnion from the brass bearing would not have worn so the stiffness would not ease with use.
I fixed it by using linishing the brass bearings flush with the frame.
What it comes down to I suppose is a lack of adequate quality control when making items with fine tolerances - I suspect though it is too late now to check, that the operating currents were significantly higher than they should have been.
Idris