Hey, it's me again! 😄

My pedals are cheap and for airplanes, it hasn't been a problem, but for helicopters, they are sending erratic signals to the chopper.

I tried using different filter values but I don't understand what I'm doing because that changes the overall axis center point somehow... and doesn't smooth erratic signals.

Is there an option to smooth out noisy control axis?

You can see the settings here... https://imgur.com/a/r42EzaW

- First image is with no filter

- Second image is showing setting filter to 2

- Third image shows how center has shifted to 2448 with filter set to 2

 

Edited August 9, 2025Aug 9 by Virtual-Chris

Maybe that is caused by the deadzone? Deadzone has two different settings, "linear" and "non-linear" - use the "RAW" button to get to the selection checkbox.

17 hours ago, Lorby_SI said:

Maybe that is caused by the deadzone? Deadzone has two different settings, "linear" and "non-linear" - use the "RAW" button to get to the selection checkbox.

Thanks... To simplify trouble shooting, I've deleted the dead zone.  Trying this on Collective now which is a simpler up/down axis.

Filter 4 seems to tame the noise, but the issue is that the center has shifted about 1855 so when the axis is fully left, it's only at -14000, and at center it's 1855.  So I can't get the collective fully down now. Seems very strange... See screen shots here (first is center with filter, next is full left with filter, third is settings)... https://imgur.com/a/4mK3i5X

Edited August 11, 2025Aug 11 by Virtual-Chris

I guess that the filter is just too strong and doesn't register any more changes at that point. Does it make a difference when you are you using the progressive or linear filter?

I suppose that you would have to set the axis to update "continuously" instead of "on change". But that has other downsides.

The alternative is to create a script and implement your own filter. The easiest way to do that is to reduce the number of data points. By default, an event based axis has 32K discreet values (-16384 to 16384). By reducing the number of data points, you would leave more room for the potentiometer to "jiggle" unnoticed. 
For example, you would set Axis Min to -100 and Axis Max to +100, Rounding Int. Then create a script "MyGroup-MyScript" with this code (Tail rotor pitch is bound to the rudder axis)

(L:MyGroup-MyScript) 163.84 * (>K:AXIS_RUDDER_SET)

This would reduce the number of "steps" to 200 (from originally 32000), and thus also reduce any jiggling.

The LVar at the beginning is the so-called "script variable". It contains the value transmitted by the control that is calling the script. The name is always "(L:scriptgroup-scripttitle)"

 

Yes, this also reduces accuracy. But every type of filter will do that to some extent.

Edited August 11, 2025Aug 11 by Lorby_SI

21 hours ago, Lorby_SI said:

I guess that the filter is just too strong and doesn't register any more changes at that point. Does it make a difference when you are you using the progressive or linear filter?

I suppose that you would have to set the axis to update "continuously" instead of "on change". But that has other downsides.

The alternative is to create a script and implement your own filter. The easiest way to do that is to reduce the number of data points. By default, an event based axis has 32K discreet values (-16384 to 16384). By reducing the number of data points, you would leave more room for the potentiometer to "jiggle" unnoticed. 
For example, you would set Axis Min to -100 and Axis Max to +100, Rounding Int. Then create a script "MyGroup-MyScript" with this code (Tail rotor pitch is bound to the rudder axis)

(L:MyGroup-MyScript) 163.84 * (>K:AXIS_RUDDER_SET)

This would reduce the number of "steps" to 200 (from originally 32000), and thus also reduce any jiggling.

The LVar at the beginning is the so-called "script variable". It contains the value transmitted by the control that is calling the script. The name is always "(L:scriptgroup-scripttitle)"

 

Yes, this also reduces accuracy. But every type of filter will do that to some extent.

My only issue now is that when I add a filter to an axis it somehow “shifts” the axis by like 1800. So rather than being centered at zero it’s now 1800 when centered. And the max on one end is about -14000 instead of -16000. This seems like a bug. It’s simple to reproduce… add an axis. Open it up and set the filter to 4. Watch the center shift. 

4 hours ago, Virtual-Chris said:

My only issue now is that when I add a filter to an axis it somehow “shifts” the axis by like 1800. So rather than being centered at zero it’s now 1800 when centered. And the max on one end is about -14000 instead of -16000. This seems like a bug. It’s simple to reproduce… add an axis. Open it up and set the filter to 4. Watch the center shift. 

Right. Not a lot that I can do about that at the moment.

When you don't want to use scripts, try setting the Fliter to "linear". Then, if 1800 is your offset, set AxisMax to (16384 + 1800) and Shift to -1800 

You can send an email to the support address (last page of the manual) with your proof of purchase, and I will send you a fixed pre-release variant of AAO. But be mindful that this will already have the latest SDK libraries applied (not a problem with 2020 but might be a problem with 2024, since the latest is the beta-state)

 

Edited August 12, 2025Aug 12 by Lorby_SI

7 hours ago, Lorby_SI said:

Right. Not a lot that I can do about that at the moment.

When you don't want to use scripts, try setting the Fliter to "linear". Then, if 1800 is your offset, set AxisMax to (16384 + 1800) and Shift to -1800 

You can send an email to the support address (last page of the manual) with your proof of purchase, and I will send you a fixed pre-release variant of AAO. But be mindful that this will already have the latest SDK libraries applied (not a problem with 2020 but might be a problem with 2024, since the latest is the beta-state)

 

Thanks. I have a new collective control on order that hopefully is less noisy so I’ll wait until that gets here before chasing this further.