Peter, if I understand what Lemon told us some years ago when we were preparing the instructions, that is incorrect.

There are two different things here:

1) The Lemon StabPlus has stronger filtering than the regular Stab and more aggressively rejects vibration. You are correct that a 6 axis stab can accumulate vibration by integrating the sensed disturbances and eventually that can even overload a storage register. That is why you have to be careful using stabilisers in IC powered aircraft with high vibration.

But that is not what is happening here.

2) Lemon deliberately built in a "Throttle" to elevator mix as an ersatz sensor of airspeed. It was intended to compensate for over controlling above 25% "throttle" in Autolevel mode only. It normally shows up as anomalous behaviour on the bench when testing Autolevel mode and is explained in some detail in Appendix 6 of the Stabilizer Plus Reference Guide on pages 40 and 41.

I put "Throttle" in quotes since I understand it is actually using the output of channel 1 from the receiver as its sensor input for the mix. Spektrum/Lemon systems always normally have the throttle on channel 1 so in most cases it is indeed the Throttle value. What you did was to take the actual throttle value from a different receiver output channel and at the same time force output channel 1 to always be minimum pulse width so the mix never gets applied since it never gets above 25%.

In other words the value of I in the stab is not altered; you by-passed and neutralised the mix by using a different channel controlled by the transmitter throttle stick.

It is not possible for normal users to alter the PID values in the Lemon stabilisers. They are set in the firmware. You would have to rewrite the firmware and reflash the receiver. The onboard pots and master gain simply scale whatever internal combo of PID gains the designer chose.

John

Originally Posted by pesp

Just note that the stab plus does not really have a throttle to elevator mix, although you may experience as if it does. What happens is that the stab plus does not manage to perfectly calculate the impact of the vibrations,. With channel 1 (THR) set to low (-100%), the stab plus does not add up all the movements from vibrations. With throttle high it adds up all the movements from vibrations over time. So the longer you fly, the more it deviates from level. Mathematically, all this adding up over time is called integrating - which is the famous I in PID as John described very well above.

Peter

Thanks for correcting me John. I did some (bench) testing and did indeed detect some throttle to elevator mix built-in. I hadn't noticed it before as it was so minimal that I hadn't noticed it by looking at the position of my servo's. But after hooking the receiver up to my millisecond meter, I could indeed see a small change in the ms pulse width when applying throttle. I will correct my previous post.

I understand the filtering of vibrations is a difficult one. I have to admit that may depend a lot more on other factors than the lemonrx firmware: the intensity of the vibrations, the frequency, the foam/glue used to glue the receiver in place, the strength and foam density of the plane the location in the plane where the receiver is mounted in, etc etc. So I am not going to say that lemonrx did a poor job here. I love their products and swapped out all my orange receivers for lemonrx and Spektrum. On the other hand, a lot of the above mentioned factors may be hard to change. I have for instance a small jet, and there simply isn't any place to put the receiver close to the CG.

My main point is that my launch switch capability (locking channel 1 THR to -100%) just gives us another option.

I have 4 stab plus receivers in different planes. All of them would level properly after half a minute of flight, but only one of them was capable of properly returning my plane level after 5 minutes of flight. The deviation was not only the nose pointing up, but also the wings not being level. Now I have the option to make it work for all 4 planes. Ironically enough, I can fly well enough now that I hardly ever use auto-level. But others may not be there yet and may want to use this new option.

In any case, I will definitely use this new option for 'hands of the sticks' hand launching - something I could only do with some BNF models before. I like flying even when it is bursty wind and this is exactly what I need for less glueing and more flying ;-) .

One thing I have not tried is: does my launch switch setting also helps for take off from ground. I am just thinking that if your plane would not be perfectly level on the runway and you turn on the stab plus auto-level, you may get an unexpected twist/turn when you take off. Any-one ready to try the new launch switch setting for take off from ground and share their experience?

Pretty much always ground launch be it pavement, gravel or grass. My tests on fms j3 cub from ground and take off was fine. Thing is I may not be the best for feedback on that as I can fly just fine without autolevel on. My use of it is more due to vision issues (macular degeneration). To give the workaround a real test I need to change it (self level) to beginner mode and really lock down how much it can maneuver and test from take off to landing. Near future I think, weather here shifting to fall like pattern right now.

Did some more flights this eve before sunset using the simple cub (less to risk vs fms cub) I made with the Lemon plus. Played with turning the main channel 1 throttle cut on off. Ran 3 battery's 6-7 mins each with mixed flying. I'm done testing, if you want to fly self level use this bypass setup if at all possible. It's night day difference in handling. Low or high rates, either beginner or advanced modes hands down this is better. Overall it just is predictable, with that mix in play it constantly climbs granted not steeply but your always fighting it. Off, using channel 6 it stays put. Hand radio to buddy and no worry about what plane is doing.

The only short fall I see is having the option to use the setup. If running flaps on aux1 your basically forced to leaving self level on always so as to use channel 5 (gear) instead. My case I tend to toggle between gyro and self level but this plane is basic with just throttle, ailerons, elevator and rudder so not a problem. I wish Lemon would address this mix of throttle to elevator so as to make it adjustable or at least be disabled without this work around. As is I find it too aggressive and it exponentially gets worse as flight time increases to the point of annoyance in trying to counteract it with the stick.

Originally Posted by Trukntigger

Did some more flights this eve before sunset using the simple cub (less to risk vs fms cub) I made with the Lemon plus. Played with turning the main channel 1 throttle cut on off. Ran 3 battery's 6-7 mins each with mixed flying. I'm done testing, if you want to fly self level use this bypass setup if at all possible. It's night day difference in handling. Low or high rates, either beginner or advanced modes hands down this is better. Overall it just is predictable, with that mix in play it constantly climbs granted not steeply but your always fighting it. Off, using channel 6 it stays put. Hand radio to buddy and no worry about what plane is doing.

The only short fall I see is having the option to use the setup. If running flaps on aux1 your basically forced to leaving self level on always so as to use channel 5 (gear) instead. My case I tend to toggle between gyro and self level but this plane is basic with just throttle, ailerons, elevator and rudder so not a problem. I wish Lemon would address this mix of throttle to elevator so as to make it adjustable or at least be disabled without this work around. As is I find it too aggressive and it exponentially gets worse as flight time increases to the point of annoyance in trying to counteract it with the stick.

Thanks for testing and sharing your experience, truktigger. It confirms what I experience. The workaround of locking THR to -100% is better to me as well, in all cases: hand launch as well as during flight and for landing.

So you experience the same as I do: the deviation building up over time when you don't cut the Throttle. That means it is the integration control that is the most annoying as that make things less predictable. Yes, there is also a THR > elev mix. But I personally find that less annoying than the integration. Most of my planes actually climb when I apply throttle - also those that don't have a stab plus. So I guess I expect that behavior.

I actually do not only have the annoying integration on the ELE but also the AIL controls. That means my plane does not only climb more and more over time. Also the wings get more and more off level over time. My throttle cut takes all that away and makes it predictable.

I also did some testing with the spektrum SAFE receivers. I checked safe receivers from 2 BNF models: a warbird from 3 years ago and a jet from 6 months ago. They both are great for hand launching. With SAFE on, they both have significant THR > ELE mix built-in. But they don't have any integration active.

Regarding the AUX channel. If you want to use it for flaps, you should use the AUX2 for the ESC. In that case you need to change the mixes I listed to AUX2 instaed of AUX1 and you connect your ESC to the FIRST (so not the THR one but the bind one) instead of the last pins on the receiver. Off course, that requires a transmitter with at least 7 channels.

Peter

Ahh, forgot the bind slot is also channel 7/aux2. So used to Spektrum receivers like ar636 or the newer 630/637 and bind port is that and data but not useable for servos. Will keep that in mind, have another lemon plus for different build I'm doing and that will come in handy.

That's interesting, Peter.
It would be consistent with the observation that some people find the StabPlus perfectly OK for auto levelling performance and some do not.
It may be a matter of how well isolated the receiver is from vibration in the model?

I'm guessing there is a trade off in the design decisions in a fairly simple autolevelling stab like the Lemon between excessive filtering to minimise drift which reduces sensitivity and minimising the filtering which leads to susceptibility to Integration drift.
Just guesswork on my part, but it does seem a reasonable theory.

John

Originally Posted by pesp

Thanks for testing and sharing your experience, truktigger. It confirms what I experience. The workaround of locking THR to -100% is better to me as well, in all cases: hand launch as well as during flight and for landing.

So you experience the same as I do: the deviation building up over time when you don't cut the Throttle. That means it is the integration control that is the most annoying as that make things less predictable. Yes, there is also a THR > elev mix. But I personally find that less annoying than the integration. Most of my planes actually climb when I apply throttle - also those that don't have a stab plus. So I guess I expect that behavior.

I actually do not only have the annoying integration on the ELE but also the AIL controls. That means my plane does not only climb more and more over time. Also the wings get more and more off level over time. My throttle cut takes all that away and makes it predictable.

I also did some testing with the spektrum SAFE receivers. I checked safe receivers from 2 BNF models: a warbird from 3 years ago and a jet from 6 months ago. They both are great for hand launching. With SAFE on, they both have significant THR > ELE mix built-in. But they don't have any integration active.

Regarding the AUX channel. If you want to use it for flaps, you should use the AUX2 for the ESC. In that case you need to change the mixes I listed to AUX2 instaed of AUX1 and you connect your ESC to the FIRST (so not the THR one but the bind one) instead of the last pins on the receiver. Off course, that requires a transmitter with at least 7 channels.

Peter

Originally Posted by jj604

That's interesting, Peter.
It would be consistent with the observation that some people find the StabPlus perfectly OK for auto levelling performance and some do not.
It may be a matter of how well isolated the receiver is from vibration in the model?

I'm guessing there is a trade off in the design decisions in a fairly simple autolevelling stab like the Lemon between excessive filtering to minimise drift which reduces sensitivity and minimising the filtering which leads to susceptibility to Integration drift.
Just guesswork on my part, but it does seem a reasonable theory.

John

That is indeed my theory why locking the THR to -100% makes things better, John. I have no doubt that this is why I can now hand launch without my hands on the sticks and I could not before. You can easily see the effect of the integration with a bench test with THR high. Move your plane left and up and keep it there for 10 seconds. Then bring the plane back level and think what will happen if you would throw the plane now by looking at the control surfaces.

When the plane is flying, the integration control is helping to get it perfectly level. But that only works if vibrations are not significant. I also fly drones and I know from experience that vibrations will cause auto-level to drift over time. Every minute, the drone’s will be more inclined where it is supposed to be horizontal.

In my opinion, planes don’t really need the integration control. Drones need to hang still. Planes don’t. And a plane that is not perfectly level will still fly more or less straight. So locking THR to -100% makes the lemon stab plus much better for planes as the behavior when switching on auto level is predictable. For sure, it makes worry free hand launching possible.

I tested that when flying 6 minutes witout turning auto level on before or during this flight (under normal THR high), and then turning on auto level for landing, still gives you the suprize of auto-level being not level. I think that is why so many people (nearly) crashed and reported auto level not to be usefull. That was also my conclusion, untill I discovered the effect of myTHR cut off workaround. Now, I am all excited again about the possibilities of the auto level capability.

I know the recommendation is to use auto level function only for beginner planes. But after such a great experience with my flying wing, I am ready to try my newly configured stab plus out in my jet. Let’s see how that goes.

Peter

I flew 2 batteries with my jet.

Take off: I am very comfortable hand launching my other planes, but throwing my jet was always a hit or miss. So I was very curious to see if this would be another crash or not. I programmed some up elevator mix (launch switch > ELE) so auto-level would try keep the nose up about 15 degrees. With that I mean that holding the plane with the nose 15 degrees up results in my elevator to be at the same position as with auto-level off. I activated auto-level and threw the plane overhand. It took off OK without touching the transmitter. But the dive before the climb was more than I liked. Second launch, I increased this mix to about 20 degrees. Better, but still too much diving. I will increase it to 25 degrees for my next run.

During flight: I did a lot of sharp turns and would each time flip on auto-level to see what happens. Auto-level would correct the plane's attitude in the right direction, but it would never bring it completely level. I probably had the gains to low as it always kept turning (slowly) in the same direction it was turning before I put on auto-level. Well, the integration control that becomes active when THR is over 25% would have helped here I guess. But since I had bad experiences with that, I had locked throttle permanently off. Maybe I should have programmed it that THR goes high when I turn on auto-level during flight (but definitely NOT during launch).

I am happy with what I achieved:
- I can safely hand-launch my jet without needing my hands on the transmitter, by using my launch switch (I use switch A). This is with: THR cut always on, ESC on AUX1 channel, MIX A > GEAR for autolevel, MIX A > ELE for getting the nose to want to point 25 degrees up.
- the stabilisation of the gyro's (in non auto-level) corrects for bursts of wind. I can turn that on and off with the GEAR switch.

Peter

As far as I know, some drone (multi-rotor) controllers mainly rely on accelerometers (to detect which way is 'down') to do auto-leveling, which may leave them less vulnerable to the effects of integrating a lot of noise/vibration when detecting how much they have rotated from level.

The Lemon StabilizerPlus is a "6 axis" stabilizer.
It uses 3x angular acceleration gyros (the "Rate" part) and 3x accelerometers (the "Autolevel"part).

AFAIK The PID adjustments can apply to both. Integration and derivative functions can be applied to either.

Angular acceleration stabs (like the Spektrum A3SX and Lemon regular stab) have no way to know which way is down. You need accelerometers for that since gravity is a constant force causing acceleration in a direction toward the centre of the earth.

Integration of a rate gyro results in heading hold not auto levelling. Essentially it is approximating how far the stab has deflected from its original position by integrating the angular velocity to calculate angular movement. It has no way to know what an external reference like "down" or "level" is.

John

Originally Posted by flying-llama

As far as I know, some drone (multi-rotor) controllers mainly rely on accelerometers (to detect which way is 'down') to do auto-leveling, which may leave them less vulnerable to the effects of integrating a lot of noise/vibration when detecting how much they have rotated from level.

Originally Posted by jj604

The Lemon StabilizerPlus is a "6 axis" stabilizer.
It uses 3x angular acceleration gyros (the "Rate" part) and 3x accelerometers (the "Autolevel"part).
...

Now I know. Thank you.

Originally Posted by jj604

...
Integration of a rate gyro results in heading hold not auto levelling. Essentially it is approximating how far the stab has deflected from its original position by integrating the angular velocity to calculate angular movement. It has now way to know what an external reference like "down" or "level" is.
...

I have heard of some stabilizers that have you turn it on while the aircraft is level, and then it determines how far from level it is by integrating the angular velocities. I am guessing that such integration can easily get messed-up by noise/vibrations and was wondering if this is how the stab plus determines "level", but as you mentioned above, it is not.

Oddly confusing seeing most of these devices do rely on being mounted in a certain position or calibrated to know what position they are starting from. The stab plus I can say is rather funny to use vs Spektrum series with safe. I have mine set to experienced mode auto level which increases the flight envelope. Cool but it's not as accurate or sensitive as Spektrum because fun times on the flite test cub, had it spinning and going inverted along with some pretty steep spin dives that Spektrum setup would never let me do. Still let go sticks and level right out (close enough). Been running the bypass still, can flip "A" switch to enable but just much better without that mix climb effect, plane very predictable with it off.

Experienced mode in a three mode SAFE model doesn’t have any autolevel. It’s purely AS3X, which is a rate gyro setup, similar to what Lemon calls Gyro mode. There are no limits on the envelope in this mode. In Intermediate mode, there are limits but no AutoLevel. In Beginner mode you get full SAFE AutoLevel and limits.

In experienced mode, the model can loop, roll, spin, etc. but not in the other modes.

In a SAFE Select model, by contrast, SAFE can only be turned on and off. AS3X is on full time. When SAFE is on, you have a restricted envelope. When it’s off, you don’t.

Kinda disagree on there on the Spektrum settings. Your correct in essence but the envelope can be increased with safe auto level on both with forward programming versions or mod of the earlier 636 models via usb and pc which I've done and tinkered with. This stated I did find Spektrums limits seem tighter, let's say using a simple ar630 receiver as example.

Flight modes set to switch "B" on a dx8gen2 so 3 choices. Set standard gains 40/50/60, priority 160/160/160 which is a basic start point right? Master gain to rotary knob, base 2x. Next set modes, can make 1 & 2 with as3x plus auto level and 3rd just as3x. Next I go set safe values, in that you pick the flight envelope for roll, pitch, yaw and even another gain. If you set 1 more beginner like and 2 more advanced in flight can see and feel the difference so you know it's working. Thing is even at max limit in settings it never really lets me get inverted or completely roll. I tried investigating this some and it's has multiple factors figured into the whole setup. From what I could tell amount of throw, baseline of aircrafts capability, even the gains and priority combined with the safe auto level values all go into the mix making it hard to get right and still have it self level in a reasonable way. If you can get enough momentum it's possible to go pass the hard limit and hit other side to continue the action but not easy.

This lemon receiver I'm finding is way easier to setup for breaking past the limits and still have it self level in a reasonable fashion. Decent handling with it constantly on where it feels like pretty much no limit but there is one. Point is my vision not what it once was and I struggle at times with the models orientation in flight when doing more than simple oval pattern. Lost several planes recently just from background not contrasting to model enough where I loose sight and just dump it in fear of worse could happen.

Not ready to give up and quit but this auto level feature a great aid for people like myself. I loose sight, put plane in a banking circle with slight climb and watch and listen till I get visual lock again. The auto level on either setup keeps it airborne with a relatively decent pattern and no fear of over turning and dropping. Life saver for certain when vision impaired. Ideal, no admittedly but again I'm not blind (not yet) and along with trying to keep my models color schemes high contrast I'm doing better. Recent loss was a black flite test build went invisible against dark green tree background- lesson learned, black even with bright yellow stripes not good, red/white/black high contrast been best so far.

Anyway finding I'm preferring the lemon plus for ability to still auto level but give a larger window to fly in vs tighter or toggling auto level off on with Spektrum setup. The lemon just needs a way to turn off that thr/elev mix, for now using the bypass on ch6/aux1. If I can get my Spektrum receivers to open the window more would make several of my bnf planes more enjoyable to use but so far that has been hard to do.

Side note: usually fly over lake here (shallow bay no people or hazards), one at a time in part because I just don't see well enough so safer and unless good friends which are aware of my limitations don't want hard feelings if a collision did occur.