[karla]

The problem
In my experience the most sensitive part of a Heli is the tail rotor.
I started out with belt driven tail rotors, and they were forgiving if I happened to land a bit on rock etc, but still the delicate pitch arm mechanisms or the tail rotor shaft would get hurt and after that create vibrations. Later, came the torque tube solution, it claimed to be more efficient, maybe true but SOOO much more unforgiving if you even touch some high grass while landing. Some of the small gears would get very small indents that would create vibrations and you really had to change all torque gears to get back to operation.
After having both my 250 size Heli’s grounded, one belt and the other torque tube tails, due to tail rotor problems I just took a very long rest from this tiring situation.
They were just sitting there for a long time, all intact and ready to fly if not the tail issues.

My solution
I was inspired to see so many new Helis working fine with motor driven tails I thought I would give it a try to convert the torque tube 250. I think the result is very promising.
So, want to share my learning if any one like to walk down this path.
Librepilot Heli is my preferred FC for sport 3D flying, not going waypoints or GPS stuff. If you want that, Ardupilot is my choice.
 
Starting point
The trex 250 has physical tail mechanism for pitch control and a tail servo to drive it, the power rotation comes from the toque tube feeding off the main rotor motor.



This will all be changed into a digital mechanism controlling the RPM of a tail motor.
 
What we can throw away is
. the tail mechanism for pitch control
. the blades and shaft housing etc
. the toque tube and its drivers
. the tail pitch servo



What we need to add is
. a motor
. an electronic speed controller
. a propeller



I was using half of the tail housing to mount the motor on to it.



This is the new tail rotor motor set up.



The ESC connects with the power wires running inside the tail boom and is mounted where the tail servo previously was mounted. The control wires for the ESC are connected to the OpenPilot Revo Nano flight controller’s Yaw output.

. props size specs:
GEMFAN 3052 Three-Blade Propeller
Rotation diameter 76.2mm
Center hole thickness: 8mm

. esc specs:



I mounted the motor on the left side of the tail boom. The standard way to do it is on the right side. The reason is the weight of the motor on the left side will help the Heli to have a closer to zero roll attitude in hover. If, place it on the right side the weight would have increased the tilt angle even more.

For the rotation direction of the tail prop, I wanted it to have the inner side of the prop to rotate upwards to meet the downwash from the main rotor to be more effective in normal flight. That means its turning in a clockwise rotation.  That defined I needed to choose a clockwise angled prop.

The difficulties
The weight taken out was around 30 grams. But with the motor mounted far out on the tail boom meant the CoG was moved aft, a lot. So needed to ad ballast in the nose of the cyanope to compensate. Needed around 30 grams. This could really have been avoided by using a much smaller tail motor. I don’t even need more than 40-60% of the thrust this one produces. Better smaller motor, lower weight, less or no need of ballast in the nose. I think this motor could be good to convert my 450 size Helis.

Two other problems, to avoid the rotor prop to spin up dangerously when connecting the LiPo, the Idle up setting not only cut the main rotor motor but also cut the yaw input to zero. The other problem was how to select the neutral point in the radio channel input so it will have enough yaw authority when flying. At neutral the tail motor will have to run at some speed, defined by the neutral point. Now I think I have got it set okay, its flyable but likely can be tweaked much better.

I will add my settings file here.

So far, I have resurrected one of my two 250 Helis
I feel good about this, its promising 

[TheOtherCliff]

Have you considered using bi-directional ESC setup, or are you OK with just zero tail thrust and let main rotor torque the tail in that direction?  Maybe a bit ineffective when main rotor is at zero pitch?  Maybe bi-directional is only needed for extreme stunts.  I wonder if a brushed tail motor would be better because it works better and zero speed and doing quick reverses.  I've got a couple really old small helis with brushed tail rotors.

I am honestly interested in your preference of Ardupilot for GPS flight modes.  I've seen a lot of videos about it.  In particular, I'm interested in the sub 250 gram autonomous airplanes.  I guess one thing that keeps me from trying other firmwares is that LP is about the only firmware that supports the OpLink built into Revo/Sparky2.  I really like LP's range and "no receiver needed".  I don't like LP's sometimes inadequately documented GPS flight mode (setup/) tuning.

What do you think about Ardupilot GPS setup, GCS GPS maps, and waypoint flight?

[karla]

Yes, I considered a reversing tail, briefly, and concluded a normal one way motor will be good enough for my level of flying for a good while.

Below is a video of a pilot flying mild 3D with a 450 Heli, the FlyWing FW450 version2, and I know for a fact that tail motor does not reverse, since I have one 



I added another video, a larger Heli and with more hard core 3D, especially how he is bashing the tail around. That tail do reverse, however in the end the pilot is not happy with it since it limits his flying 



I do not know the difference between brushed and brushless motors, but I am very interested to find a lighter motor alternative and suitable ESC for my other 250 Heli. I just took what I had lying around.


Yes, I know the built in OpLink radio in Revo and Sparky is very sweet. In some of my fixed wings I use Ardupilot but with a pair of OPLM units. That works both with telemetry and control (ppm). You only need to set the data link to Serial on the two units then you can keep your transmitter and still have it work with Ardupilot on the plane and MissionPlanner at the ground station. You can even try it out just flashing Ardupliot plane firmware to your Revos and Sparkys.

I use Librepilot for my Helis, and Traxxas Speedboat and Rover.
Quads do not interest me anymore but for fixed wings and sailboat I use Ardupilot where robust GPS flights with waypoint missions, auto launch, vtol's, soaring thermals are important. For those applications it's handy to have onboard SD card to get massive log data to analyse flights, Revo/Sparky don't have that.
For mild 3D helicopters I think Librepilot is holding out great with the Vbar flight mode.
Ardupilot is just on another level for robustness of functions, comprehensive Wiki and soo much development and new technology adoption.

Just an effort to learn a new software

[TheOtherCliff]

Does Ardupilot support the RF in any FC's that have on-board RF/telemetry/control link (that you know of, offhand)?  If so, it sounds possible to enhance Ardupilot to use the Revo built-in OpLink.

Maybe another way to ask that is "Are there any other (non-OpenPilot) FC boards that have on-board RF (or was OP the only one for some reason)?"

I think that LP has a smallish issue when doing FC with on-board RF.  There are two pieces of code that really need high real time priority to work right: Stabilization and RF packet protocol sync.  I don't think it is handled as well as it could be.  It could be made synchronous...  It may be that other firmwares avoid this issue by avoiding having RF built in...  For that matter, making the FC the coordinator, which used to be possible, can avoid this issue by moving the real-time constraint to the OpLink.

I have plenty of OpLinks.  For that matter, before Revo was born, I had used some 2.4GHz RF links in serial mode with CC3D.  There were some issues, mainly when you have a non-intelligent serial link via RF, the RF doesn't know when to send the data.  It waits till either a whole block of serial data (about 256 bytes, minus header) is accumulated or till there is a delay between one incoming byte and the next (a timeout).  Either way, there is a delay that doesn't have to be there if it is only a bit smarter.  The delay causes unnecessary control delay if you are using this RF link for control.  You also get slow telemetry because every packet must time out to be sent before another packet can be requested and sent.

When using separate OpLink and FC, even if the RF serial link is smart (e.g. using OpLinks that are separate from the FC with LP firmware), it actually converts the received packet into real PPM; transmits that PPM to the FC.  That adds time that is bypassed when the RF link and FC are in the same board and don't actually have to convert to real PPM to send from one to the other.  (I assume that) Revo's built in OpLink sends the "virtual PPM" to the receiver code as soon as it gets the packet, bypassing the typically 22ms per PPM frame.

Brushed motors have brushes that are a mechanical rotary switch.  They switch from using this coil to that coil several times per revolution.  The brushes wear out eventually.  I have a Dremel tool where the brushes finally wore out after several decades of light use, so they can last a good long time if well built.

The bad thing about brushless motors is that they don't like it when they are close to zero speed, like when they are changing direction from forward to reverse.  It takes a small fraction of a second.  The good thing about brushed motors is that they change from forward to reverse basically instantly.

[karla]

Top of mind I only know of 3 FC's that do have built in RF, Revolution full size board, Revolution mini and Sparky2.
For all of them the RF is disabled in Ardupilots hardware definition when flashing Ardupilot on them.

Here is a landing page for all the FC's that are supported by Ardupilot, it also describe the hardware definition on Ardupilot as well as have links to the 'manufacturers' original definition.
https://ardupilot.org/plane/docs/common-autopilots.html

When I think of it, have never come across boards with RF builtin, but a lot of other functions are now built in like SD card, FPV overlay, very good BEC support. Compass and RF seems not popular to integrate, what I have seen at least.

Thanks for the explanation of brushed and brushless motors relative to tail rotors on helis!
Best!

[jcg1541]

The problem
In my experience the most sensitive part of a Heli is the tail rotor.

My experience with FPV with the 250 rotors: Tail can be very sensitive to PIDs change, but it depends on how light you need the tail to be.
I use Algin 250's main grips and DFC. My parts listing is https://nocomputerbutphone.blogspot.com/2017/12/converged-drone-developers-platform.html .

For 250-gram takeoff weight, the tail should be at least 7 grams, 1106 sized and above with 2700kv , which needs good D gain tune.
If you can allow 10.5 grams(excluding nuts and lead wires) 1306 size of the tail motor alone, the tuning is easier.
My PIDs for the tail is <field name="YawRatePID" values="0.0013000000035390258,0.003000000026077032,2.1999999242136255e-5,0.30000001192092896"/> , for 7 gram motor with 4 inch diameter props.
Depending on temperature, add/subtract 10% on the PIDs.
My last adventure with CC3D ,

, it flips and dives with 250 grams with 35 minute endurance.

[karla]

My experience with FPV with the 250 rotors:

That was useful!
Thanks a lot

[jcg1541]

The bad thing about brushless motors is that they don't like it when they are close to zero speed, like when they are changing direction from forward to reverse.  It takes a small fraction of a second.  The good thing about brushed motors is that they change from forward to reverse basically instantly.

Yes, and it gets worse that the BLHeli speed controllers can not be configured to continuously spin indefinitely. CC3D doesn't have idle-up for tail. When I dive vertically down for more than 6 seconds, the tail stops completely and can not spin up fast enough. The result is spin crash. The workaround is to use over sized tail ESC to over build the tail's torque.
Here is the crash video after 6+ seconds dive.

And my part listing says to use the same sized ESC for both the main motor and the tail motor.
https://nocomputerbutphone.blogspot.com/2017/12/converged-drone-developers-platform.html

[TheOtherCliff]

Yes, and it gets worse that the BLHeli speed controllers can not be configured to continuously spin indefinitely. CC3D doesn't have idle-up for tail. When I dive vertically down for more than 6 seconds, the tail stops completely and can not spin up fast enough. The result is spin crash. The workaround is to use over sized tail ESC to over build the tail's torque.

Something there sounds a little fishy...  The FC should maintain the neutral pulse position for the tail.  Neutral collective has the least drag and will run the tail motor more slowly.  I wonder if the fast air over the tail just momentarily stops it (like if the tail drifts a little and air blows backwards through it) and then it won't start back up, or simple left right wiggle will slow it down.  (Careful, dangerous of course) You might test this by throwing a rag into the tail while it is idling.  When you remove the rag how quickly does it start up?  I know that setting the neutral (idle) for quadcopters, if I set it very low and stop a motor, it will not start up again at that very low setting (I usually use SimonK because I got many ESCs for less than $4 each years ago).

I also know that some motor/ESC combinations have a hard time starting up instantly with a lot of power.  They will start with a small amount of power, but not a large amount.  I had a quad with very old BLHeli on it.  With motors stopped (Attitude mode) and the quad very slightly tipped, adding power, the lowest motor would not start, which ever motor was the lowest (because of the tilt), because it was trying to jump to a higher speed than the others.  The solution was to tilt the quad, to tilt that motor up slightly by hand while trying to restart the motors (GPS mast in the middle made this fairly safe).  Then it would start, and once running it would fly well.  I upgraded to latest BLHeli (that was perhaps 5 or more years ago) and the problem went away.  A jump from off to high power may be failing for you like it did for me.  You could test by setting tail motor on a 2 position switch (tail ESC directly connected to RC receiver), with one switch position being motor off and the other being medium power.  See if it will start.  Warning, in the old days, I have seen motors burn a wire if it is hard to start and you try to start suddenly with too much throttle.  If you find a combination (motor, ESC, settings like timing or 1 notch slower than the fastest startup (slowest being for heli throttle)) that will start the tail with this switch, then I bet it will start after a dive. 

Does BLHeli etc have an ESC governor mode that responds quickly to stick changes?  That could work because the ESC will see the motor slowing down and automatically add power?

There are other brands of firmware (Betaflight) that support bi-directional DShot ESC protocol where the RPM gets sent back to the FC.  As I recall they use this to program a vibration notch filter (RPM Filtering) for the accels or gyros.  I wonder if they can use this for a governor mode?

[jcg1541]

Something there sounds a little fishy...  The FC should maintain the neutral pulse position for the tail.  Neutral collective has the least drag and will run the tail motor more slowly.  I wonder if the fast air over the tail just momentarily stops it (like if the tail drifts a little and air blows backwards through it) and then it won't start back up, or simple left right wiggle will slow it down.  (Careful, dangerous of course) You might test this by throwing a rag into the tail while it is idling.  When you remove the rag how quickly does it start up?  I know that setting the neutral (idle) for quadcopters, if I set it very low and stop a motor, it will not start up again at that very low setting (I usually use SimonK because I got many ESCs for less than $4 each years ago).

"momentarily stops it " is whats happening. It doesn't happen indoors without wind blowing and without hard maneuvers. It has to be outdoors. When the tail is thrown around at 100kmph, everything is possible. I gain thinks that even with RPM 1, the yaw is still too strong because the tail is thrown around simulating an over yawing, and so the only option is to go down to RPM 0.
It gets very complicated that ESC can technically be scaled to spin at a minimum at something like RPM 5. But, if you want your heli to be covert, sitting without drawing attention, minimum RPM 5 is not an option. Anduril Ghost 4 is one example. And yes, 250 heli can go over 100kmph routinely, if see the last 3 seconds of my video here, it shows the horizontal speed alone is 82kmph.

[TheOtherCliff]

To make my long story much shorter...

My experience with my quad with very old BLHeli says that ESC firmware version would not start up from zero when there was a jump from signal throttle zero to medium.  This very old firmware _would_ start with a much smaller jump from signal throttle zero to low speed.  And the problem was corrected by simply flashing newer ESC firmware.  I think this may be what is happening to your tail.  And it can maybe be corrected entirely in the ESC as my firmware change showed for my problem.

Step 1: Reliably recreate the problem without crashes.    I think "throttle on tx switch" is a good way to instantly change from throttle off to some other value, and this may be related to your problem.
Step 2: Try some ESC changes to see what might fix the recreated problem.  I wonder about ESC timing, ESC governor mode, ESC fast startup (not fastest).  You have said that using a larger ESC fixes the problem; you could test that too.  I wonder if the larger ESC just has different firmware, or better (not necessariy larger) electronics.

[jcg1541]

To make my long story much shorter...

My experience with my quad with very old BLHeli says that ESC firmware version would not start up from zero when there was a jump from signal throttle zero to medium.  This very old firmware _would_ start with a much smaller jump from signal throttle zero to low speed.  And the problem was corrected by simply flashing newer ESC firmware.  I think this may be what is happening to your tail.  And it can maybe be corrected entirely in the ESC as my firmware change showed for my problem.

Step 1: Reliably recreate the problem without crashes.    I think "throttle on tx switch" is a good way to instantly change from throttle off to some other value, and this may be related to your problem.
Step 2: Try some ESC changes to see what might fix the recreated problem.  I wonder about ESC timing, ESC governor mode, ESC fast startup (not fastest).  You have said that using a larger ESC fixes the problem; you could test that too.  I wonder if the larger ESC just has different firmware, or better (not necessariy larger) electronics.

The small ESC that caused my crashes was 6A 3S BLHeli_S , which was much newer than BLHeli . But these 6A BLHeli_S were discontinued and non-existing these days in the market. And actually, after my last fly-away, I do not have any of those 6A ESC any more. The problem is ESC for sure, nothing to see in CC3D.

The most enduring ESC in the market is the 20A BLHeli , and that is all we can test with in this 250 category these days.
For the tail, to keep the motor cool, I turn off the active breaking, and I turn off all the governor. The governors have p gains and I gains that I assume heats up the motor.
I also have a very harrowing video of tail ESC-motor combination overheating after full collective pitch punching up 1 mile into sky. The ESC cuts its own power to 5% during overheating.
The 20A BLHeli needs to be naked if you are doing what most people do these days. The ESCs are mostly installed on a stack, naked.
I do use governor on the main rotor to have constant main rotor RPM.

[jcg1541]

My latest good tail with 7 gram motor is in this video,

The tail PIDs are 0.0013/0.003/3.0e-5, seen in Stabilization/Advanced tab