PID tuning/setting on the actual transmitter

It's just something I have been trying to get my head around if there would be any benefits. In some situations I think it could be, for both speeding up responses to stick movement and in others dulling out depending on the aircraft in use and pilot proficiency. Would it be something that is even possible to implement on a tx? To be quite honest if it were possible I think I would prefer this to expo hands down. Probably take a significant amount of time and patience to set up and beginners could end up trashing many planes if set up wrong and could get lost in settings. I am thinking more for aircraft usage not multirotor or heli, and yeah I know you could always stick a flight controller in the aircraft, but still just interested to know if it were possible.

Steve,
I know the term "PID" only in some scopes which do not seem to fit here. Can you explain what it means for you?

PID is a control system abbreviation and describes a 3 element (law) control configuration or algorithm:

P - proportion control law
I - Integral Control law
D - Differential control law



When flying an RC model there are a large number of closed-loop control loops that in fact may already be use PID control algorithms, eg. inside every servo on the model, inside the gyro and FBL controllers. On the highest (outer) level the pilot is also part of the closed control loop and the pilot is in fact is the learning control element closing the loop to keep the model in the air.

Deviation and other programmable TX's all provide a programmable P term (loop gain or scale/range). The I and D elements however require high speed, fixed period or continuous analogue feedback to work properly (the y(t) above in the diagram) otherwise it dynamically affects the D and I gain terms. The low cost telemetry we have cannot provide this level of feedback nor provide the reliability required for safety purposes.

This is where Drones and UAV comes in, their sensors, control algorithms and control/feedback loops are all local on the model and you have a remote command feed to initiate the required actions - the mission critical control functions are not distributed across the TX and Model.

PID is not the right words sorry! What I am aiming for is a 'dynamic expo' of say 30%, then that 30% would follow the position of the stick, not be fixed at 0% point. If I am holding the stick at 50% movement, at that point I still have 30% 'dynamic expo' in effect before and after that physical stick position. Currently the expo is calculated at the fixed point of 0, for an expo setting of 30%, stick movements between 50-80% have no expo effect applied. My method you would have the 'dynamic expo' available over the entire range of movement of the sticks. The tx would constantly have to recalculate the 'dynamic expo' surrounding the stick depending on the position of the stick. A fixed value that 'follows' the sticks. In the picture I think you can see what I mean.

Just to be sure: you want the expo zero point moving along with your stick, e.g. if you have your stick at 80% you want a small change when the stick moves not around the stick's zero but around 80%.
Ok. But how should be determined when the new zero point has to be set new? If the stick was not moved for a constant time? Or when the direction changes?
I think it might be possible to realize such a function (we would have to extrapolate the expo function to +/-200% and apply it shifted), but I am not imaginative enough to see how to get it working in a meaningful way.

Hi Steve,

I am trying to visualize what effect this would have on the control of the model. Adding to the issues that RBE mentioned of how to implement it, that same issues will affect the aircraft's control.

Lets explain:

If you apply the dynamic expo immediately then you will have the 30% dynamic expo applied the whole time - which means it just reduce the scale/range by 30%.

If you apply it after a period, say one second at the same 80% stick position, suddenly the movement required to stop or maintain the roll will change. I think this will confuse the pilot as the gain will change dynamically and perhaps in a non predictive way.

Yeah thats where I was getting confused with the PID mention, including things like if the stick was at 80% the rate at which it got there vs reaction time of the output to 80% and then from that 80% point say a movement back to 20% and the resulting rate of change and stick distance movement to that position and the output change to there. With current expo you can 'numb' the sticks around zero only. Say a scenario for a manouvre, maybe in the wind a knife edge, where you are holding more rudder and elevator than your set expo in the current method, ie you are out of the expo zone, you loose the advantage of having expo at that point to control your manouvre which I am trying to get my head around, I think may be nice having expo available. Kind of a 'numbing' round the sticks at any position. Getting technical this could incorporate the rate at which the stick is moving to the distance traveled over time and have a 'numbing' based on that. Although I can see benefits in some scenario's it may prove worse in others, was mainly aiming to see if anyone else had been thinking about this. What may be nice is an extra setting that would allow how 'hard' the current expo setup could be applied to a stick movement depending on the rate/distance at which the sick moves, calculated from the 0. But anyway, at the end of the day, its good like it is and perhaps don't fix anything that aint broke!

Would it help to have a switch-triggered setting (means: the expo zero is set new when you flip the switch, applied to the actual stick position)? So you could use it if you want to and reset it too. No issue with "long-time" measurement...

PhracturedBlue wrote: ... Instead, what we'd need to do is to use the rate of change to compute the curve. The problem is that we are sampling at a very high rate, so detecting rate of change isn't so easy.....

Thinking about it we already have the PI components in the RC control loop and it might just benefit to an open-loop feed-forward D term, eg. the gain increase the higher the rate of stick movement.

• The P-term provided by the TX Scale/Range and Expo settings
• The I-term provided by the Pilot - based on the movement history/trend of the model, corrections are applied and on a plane the trim buttons are used to remove the constant offset corrections
• The D-term based on the stick rate of change.

The D term would thus reduce the gain (scale) at slow stick movements, i.e. you would have the normal 100% scale and anything slower (than a threshold to set by each pilot) the gain is proportionally reduced

vlad_vy wrote: I wonder how you can get used to the transmitter, which dynamically changes the response. I think it's impossible.

Hi Vlad

That is true, also my first thought. But if it's based on the stick rate of change (and not position as Steve originally suggested) it would be a defined and predictable gain change that the Pilot would become use to.

It is not that unusual to have D-terms in HID devices control loops to assist the pilots of full scale aircraft.

It would be a nice feature to experiment with but as a necessary feature I am not sure, it would have to be trial and error to see the real life effects.

Aside from this unrelated but related ... I think the current implementation can perhaps already do this with some careful tx setup? Can you have the 'position' of the throttle stick say between the stick movement of 40%->100% used as a reference to the amount of expo being applied to a different channel. In this scenario of 30% expo, having throttle of 0%-40% having an effect of 0% expo on the channel, then as the throttle is increased from 40% through to 100% have the expo weight increase from 0% expo through to 30% expo at 100% throttle stick movement on a different assigned channel?

I have no tried it yet, but I am sure one can implement it with a few virtual channels to generate the conditions to be used as input to set the channel values via the complex mixer pages.