FlyPT Mover

@cfischer and @Avee ,

It will run in Linux, so you can have a dedicated machine to run it on the "cheap".
But in my laptop Lenovo Legion y740 with Intel(R) Core(TM) i7-9750H CPU @ 2.60GHz
with an RTX2070 MAx-Q, I get:

With window visible:



With window minimized:


But the window is OpenGL runing at 144hz.
The above values are for a more complex scheme than what we see in the video.

Not the ideal, but for now what I can do.
It's that or no nodes.
I might have a silent version without interface. But not sure it's worth.

3% is not bad with six cores. That should leave plenty of headroom to run the simulator and some other stuff.

Interesting. I will likely look into this as I would like to be able to tune things without having to minimize the game in some cases, especially while others are driving. Lots to do before then but great to hear this is a possibility.

It will be easier to comapre filters in 4.0:



Top left, original signal with a 333hz sine and noise (333hz is the horizontal line showing in all the graphs)
Top center: Bessel
Top right: Elliptic
Bottom left: Butterworth
Bottom center: Chebyshev1
Bottom right: Chebyshev2

All removing the 100hz to 300hz range with order 5.
I like the Elliptic, seems the best one.

And the EMALP, DEMALP and TEMALP


Elliptic wins:

Good definition in the cuttoff frequencies, keeping a good amplitude of the original value and an almost clean signal on the removed frequencies.


Almost perfect matching on the frequencies limits:




So try to use it in Mover 3.x (only up to order 4)

@pmvcda
hi
As I see, Xplane-11 always sends data via 127.0.0.1 and 49000 and Flypt mover always receives data of Xplane though I check or not check data output, change port and IP in Data Output Tab. Is that right? Is that bug of Xplane-11 UDP?

Not a bug, but a feature.

XPlane can send data in two ways.
In that screen, you are sending the data selected on the list of that screen through the selected port.

Mover uses the other solution, it asks for the data, you don't have to check anything.
The port for that is setup on the network screen:


It even asks you the port to use for the requests by external programs.

Since a recent update which corrected the acceleration bug, we are finally able to profile a coherent classical washout algorithm for 6-dof rigs.

Here a simple starting point quickly tuned for FS2020 Zlin Shock Ultra and some settings for VR with virtual tracker for MC and optimisations.

You can also save about 30s. launch time by creating an extra shortcut explained here:

https://www.reddit.com/r/MicrosoftF...c/psa_to_skip_the_intro_videos_and_logos_set/

IMPORTANT: Don't forget to adapt the rig geometry and range to your design.

Big hug to @pmvcda

Do we have any API documentation for connecting game by ourself?

For now, what you can do is use the custom memory hook or custom udp to grab data from a game.

For 4.0, it's on the plans to have plugins and an "API"

Hi @pmvcda
Since beginning with flight simulation (comming from race), I can see a strange behaveour with my rig.
During huge movements - for example heave when flying a plane. I can feel "steps" in the motion, that feels like driving over a small pumper. This is disturbing the immersion at flying and I want to remove it.
In the beginning, I thougt it was broken axial ballbearings of my actuator, but I replaced them with expensive and quality ones and nothing changed. The last two days I was playing around with filters and settings and sometimes I think there is a tiny impact - but not sure.
So I started to look to the graphic viewer and it looks far away from smooth. The steps I can feel seem to match the steps you can se - but I am not 100% sure.
Do you have any Ideas, how I could smooth the movement?
Thanks a lot!!!

Wickie

Recording only heave, whith huge movements (made for recording)
LOGISTIC(EMALP(EMALP(VALUE;500);500);220;2)




With "high precision" - setting it looks even worse, but seems to feel tiny better:

With FS2020?

The UI is updated independantlly of the calculations.
The steps you see on the graphics can be caused by two reasons:
-Getting data from game at low rate
-The UI is updated slowwlly and some values are not updateda nd missed in the graphic (for speed reasons)

When you use high precision, the calculation thread is more intensive and that causes the ui thread to update slower, showing more steps.

Thi is not explaining waht you feel, just to say that the graphics are for guidance, not really precise.
It's going to change in 4.0 where I store all the calculations.

Tell me:
Are you testing live data or recorded one?
And if that's FS2020 , other game or in all of them and large movements.

Doesn't seem a filter problem.

Thanks a lot for the fast answer!!
It really wasnt any setting in the mover!!! I was able to change some motor settings and now everything is butter-smooth!!
Wickie

Thanks.


For connecting to P3D we are now using Simtools. Is there any ways for us going without Simtools? We have one pc for P3D and another for FlyPT. Can we use the built in MMF + DLL for P3D when we have two different computers?

Thanks
Victor

Hello, sorry for the late answer but I got some issues with the servo driver which has a firmware bug if used in pulse and direction mode but the Chinese factory doesn't want to admit his issue so I must change supplier for the future platforms.
From this video, you can see that when the pitch or roll position reaches the maximum angle, all the rig gets blocked until I release the pitch stick, which is the right on the joypad. But that does not always happen however when the rig gets in singularity position seems it doesn't such but in extreme singularity, the burst happens. Keep in mind that the ring never loses its position just got stuck for a while then the mover release all the packets buffered all in one time.

Thanks
Marco

I understand you want to do a stress test but your pose limits and range are not set in accordance to your actuator course.
I would reduce them to something more reasonable like 200 for translations and 10-20-30 for yaw roll and pitch

I agree with you but it was just to understand the reason for the bursts, the actuator gets stopped when out of range and that's an amazing feature, but I would understand the reason for the burst, I don't think is a mathematical reason

Keep in mind that the actuator has about 850 mm of stroke and would be nice to exploit all and I have invested all my resources to create an adrenaline motion platform so the extreme positions are not all as a stress test

Hello dear Pmvcda
take a look at my video where you can see that when I disable the Pose in the graphic view the diagram gets more smooth. Keep in mind my CPU is not stressed in both conditions so I don't think would be limited computer resources.

Thank you a lot as always, your work is essential.

Marco

Hi,

I'm really interested to test FlyPT Mover on my diy 6 dof rig. So far, I've been using a custom IK algorithm and code which has been working quite well, but will need some modification to take it where I want it to go.

So testing FlyPT Mover. I've set up the parameters and have it connecting to my ESP32 via USB serial port. An immediate issue has come to light. There appears to be a significant delay between moving the test sliders in Pose : From Motion to the actuators moving. I'm talking up to a minute delay.

IE, If I move the heave slider around nothing happens for between 10 and 60 seconds or even longer. After that delay the actuators start responding to all the movements, but all movement is delayed. It's as if the serial data has been stored in a buffer and is played back after a time delay.

Is the cause of this a FlyPT Mover setting or a Windows 10 hardware issue? I have no issues using my own Windows application to control the actuators via serial.

Anyone have any ideas?

That's a strange problem.
Are you using any filters in the pose or rig module?
Even so, it would need to bee some heavy filters.

Can you show the serial output module settings?

I haven't touched any filtering. The settings are below. I can send the whole file. I was about to code an application to redirect UDP from FlyPT to serial port. It's sending UDP just fine.

<OUTPUT>
<Type>Mover.Outputs.Serial</Type>
<Name>OUTPUT :: SERIAL</Name>
<Location>38,86</Location>
<Size>870,420</Size>
<Visible>False</Visible>
<CONTROLS>
<NUDInterval>1</NUDInterval>
<NUDStartTime>0</NUDStartTime>
<NUDStopTime>0</NUDStopTime>
<TXTStopString></TXTStopString> <TXTSendString>&lt;255&gt;&lt;255&gt;&lt;Axis1a&gt;&lt;Axis2a&gt;&lt;Axis3a&gt;&lt;Axis4a&gt;&lt;Axis5a&gt;&lt;Axis6a&gt;&lt;0&gt;</TXTSendString>
<TXTStartString></TXTStartString>
<TXTUniqueName>10</TXTUniqueName>
<TXTInputString></TXTInputString>
<TXTName>OUTPUT :: SERIAL</TXTName>
<SHDebug.chkShowHide>True</SHDebug.chkShowHide>
<SHInputSelection.chkShowHide>True</SHInputSelection.chkShowHide>
<CHKAutoConnect>False</CHKAutoConnect>
<DDLParity>None</DDLParity>
<DDLStopBits>1</DDLStopBits>
<DDLDataBits>8</DDLDataBits>
<DDLBaudRate>57600</DDLBaudRate>
<DDLPort>COM4</DDLPort>
<DDLTypeOfOutput>Binary</DDLTypeOfOutput>
<TVInputSelection />
</CONTROLS>
</OUTPUT>