Additional TPA breakpoint for low Throttle #13006

pichim · 2023-08-04T09:35:33Z

Extension of the current TPA curve to lower the TPA factor for low throttle.

I have observed that the quadcopter can produce rough sounds up to jumping around when a higher loop gain is applied. During flight, these PID and filter settings behave fine and provide sufficient robustness margins, and the motor traces appear acceptable.

I attempted to address this issue through the use of dynamic filters and dynamic D gain; however, I discovered that using static filter and static D term settings result in better flight performance, particularly for handling prop wash.

During testing it was beneficial to extend the TPA curve by reducing the TPA factor for low throttle levels to resolve the quads issues after arming.

It's just a proposal
Needs testing
Default values need TBC

Update 13.11.2023

According to @haslinghuis proposal the name of all *_vanish parameters where changed to *_fade.

Parameters

Name	Default Value	Range	Short Explanation
tpa_rate_lower	20	0 - 100	Percent reduction in P or D at zero throttle
tpa_breakpoint_lower	1050	1000 - 2000	Breakpoint where lower TPA is deactivated
tpa_breakpoint_lower_fade	ON	ON, OFF	off, on - if on lower TPA is only active until tpa_breakpoint_lower is reached the first time

Update 19.08.2023:

Included tpa_breakpoint_lower_vanish option so that the lower TPA curve vanishes with air mode. If this is set to ON the the TPA curve starts at 0 with the value 100 - tpa_rate_lower and increases linear to 100 until 'airmode_start_throttle_percent * 10 + 1000' is reached. TPA lower is only active if air mode is disabled. This means after throttle is over airmode_start_throttle_percent once TPA lower will no longer be active and the user has full PID strength at low throttle.

With the vanishing option ON it just prevents the quad from beeing jumpy at zero throttle until you activate air mode via switch or increase throttle and reach 'airmode_start_throttle_percent`
With the vanishing option OFF PID gains are reduces accordingly over the hole flight

Also want to thank @SupaflyFPV and @KarateBrot for the discussions, the testing and helping with with the code itself.

Update 31.08.2023:

Removed dependency from TPA lower with the vanish option ON and airmode_start_throttle_percent. The behavior of TPA is now dependent only on the three parameters listed below.

Parameters

Name	Default Value	Range	Short Explanation
tpa_rate_lower	20	0 - 100	Percent reduction in P or D at zero throttle
tpa_breakpoint_lower	1050	1000 - 2000	Breakpoint where lower TPA is deactivated
tpa_breakpoint_lower_vanish	ON	ON, OFF	off, on - if on lower TPA is only active until tpa_breakpoint_lower is reached the first time

Propposed Default TPA Curve

Measured Sensitivity Function Roll-Axis

github-actions · 2023-08-04T09:35:45Z

Do you want to test this code? You can flash it directly from Betaflight Configurator:

Simply put #13006 (this pull request number) in the Select commit field of the Configurator firmware flasher tab (you need to Enable expert mode, Show release candidates and Development).

WARNING: It may be unstable. Use only for testing!

blckmn · 2023-08-04T11:17:22Z

AUTOMERGE: (FAIL)

github identifies PR as mergeable -> FAIL
assigned to a milestone -> PASS
cooling off period lapsed -> PASS
commit count less or equal to three -> PASS
Don't merge label NOT found -> PASS
at least one RN: label found -> PASS
Tested label found -> FAIL
assigned to an approver -> FAIL
approver count at least three -> FAIL

src/main/msp/msp.c

src/main/flight/pid.h

SupaflyFPV · 2023-08-18T21:38:12Z

this is a great idea, due to the high PIDs which pilots are now using with modern tuning methods, and trying to chase the last 5% propwash with highest PIDs possible. With PIDs at full strength at rest/0 throttle this yeilds a lot of noise on arming, and a Tasmanian devil of a touchy quad which requires some distance and care....

This is an example of the PR being tested with 0% low TPA, 20% TPA and 50% TPA

The 50% was used with a 'Vanish' option which switched off the TPA after the airmode activation point was reached - a great feature allowing more docile low TPA settings, whilst in flight the full PID strength was returned...

I have also flight tested with 0, 20, 30 and 40% TPA active in flight - with hard stop moves at 0 throttle. 40% was noticable but lower levels may not be depending on the PIDs of the quad.

pichim · 2023-08-19T10:28:31Z

@SupaflyFPV thanks for testing and sharing your findings :-)

Rebase on master and update comment Make TPA lower independent from air mode Included tpa_breakpoint_lower_vanish option Changes according to PR comments Corrected comment for API version Bugfix in msp.c Additional TPA breakpoint for low throttle

src/main/blackbox/blackbox.c

ctzsnooze · 2023-12-04T01:55:18Z

It may be easier to enable/disable via CLI, and perhaps more functionally clear, to re-name the CLI values:

TPA_LOW_RATE = Percent cut when throttle is below TPA_LOW_BREAKPOINT
TPA_LOW_BREAKPOINT = Breakpoint for the low throttle TPA reduction
TPA_LOW_ARM_ONLY= Whether TPA_LOW effect is only active only until takeoff

Not sure about TPA_LOW_ARM_ONLY. Tough to think of a good name. TPA_LOW_UNTIL_TAKEOFF is most intuitive, but is a bit long.

TPA_LOW_MODE is easy, but doesn't explain it.
Other options... TPA_LOW_ALWAYS or TPA_LOW_AFTER_TAKEOFF or TPA_LOW_IN_FLIGHT with reversed logic.

I tend to like TPA_LOW_IN_FLIGHT since when true we are confirming we want it to continue while we fly.

PS: Why do we need a separate threshold, other than just using the existing air mode threshold, to disable TPA_LOW if the user wants it only briefly at arming?

SupaflyFPV · 2023-12-04T02:09:01Z

It may be easier to enable/disable via CLI, and perhaps more functionally clear, to re-name the CLI values:

TPA_LOW_RATE = Percent cut when throttle is below TPA_LOW_BREAKPOINT

TPA_LOW_BREAKPOINT = Breakpoint for the low throttle TPA reduction

TPA_LOW_ARM_ONLY= Whether TPA_LOW effect is only active only until takeoff

Not sure about TPA_LOW_ARM_ONLY. Tough to think of a good name. TPA_LOW_UNTIL_TAKEOFF is most intuitive, but is a bit long.

TPA_LOW_MODE is easy, but doesn't explain it. Other options... TPA_LOW_ALWAYS or TPA_LOW_AFTER_TAKEOFF or TPA_LOW_IN_FLIGHT with reversed logic.

I tend to like TPA_LOW_IN_FLIGHT since when true we are confirming we want it to continue while we fly.

PS: Why do we need a separate threshold, other than just using the existing air mode threshold, to disable TPA_LOW if the user wants it only briefly at arming?

I see what you mean

Could be

TPA_LOW_FOR_ARM
TPA_LOW_ON_ARM
?
TPA_LOW_UNTIL_ARM - is ok IMO too and clearest

The seperate threshold was created (used airmode initially) in order to remove the dependency for the feature on airmode. There may be situations where airmode is on a switch or off which complicates...if I remember correctly...

haslinghuis · 2023-12-04T02:12:22Z

I am using AIRMODE on a switch (as some might be using Idle Up settings)

pichim · 2023-12-04T11:19:02Z

hey guys, thx for the feedback.

given names:
following a tip from @KarateBrot I named the parameters:

tpa_rate_lower, tpa_breakpoint_lower
so that when you use get tpa_rate in CLI or search for tpa_rate in your dev environment you find both parameters, so tpa_rate and tpa_rate_lower (same idea for tpa_breakpoint) which i found practical.
tpa_breakpoint_lower_fade @haslinghuis proposed the name fade for the switch logic. i don't care that much about the name but found tpa_breakpoint_lower_fade a descriptive name. i also like "TPA_LOW_FOR_ARM" or "TPA_LOW_ARM_ONLY". just tell me what you want : - ) and i will change it.

dependency to air mode:
as @SupaflyFPV statet this is what we experimented initially. but i thought its maybe a good idea not to create a dependency.

i assume honestly this PR will mainly be used to tame the quad after arming until take-off for aggressive gains and / or filter settings.

src/main/cms/cms_menu_imu.c

src/main/flight/pid_init.c

Co-authored-by: Jan Post <Rm2k-Freak@web.de>

ctzsnooze · 2023-12-06T21:12:21Z

@pichim please clarify what exactly tpa_rate_lower achieves, in the table above.
Does a value of 20 mean that TPA reduces D to 20% of normal, or does it mean a 20% reduction from normal.
Please edit the table at the top to clarify this, and maybe tidy up the on/off note.

ctzsnooze · 2023-12-06T21:15:32Z

src/main/flight/pid.h

@@ -237,6 +237,9 @@ typedef struct pidProfile_s {
    uint8_t angle_feedforward_smoothing_ms; // Smoothing factor for angle feedforward as time constant in milliseconds
    uint8_t angle_earth_ref;         // Control amount of "co-ordination" from yaw into roll while pitched forward in angle mode
    uint16_t horizon_delay_ms;           // delay when Horizon Strength increases, 50 = 500ms time constant
+    uint8_t tpa_rate_lower;                 // Percent reduction in P or D at zero throttle


Please ensure the comment is absolutely clear. For example, does a value of 20 mean that the TPA effect will cut D to 20% of normal, or result in D that is 80% normal, at zero throttle?
Currently, TPA high of 80 will cut the D to 20% of normal.

hey chris, since tpa lower is just an extension of tpa I basically copied the logic of how it is applied and adjusted the description according to tpa_rate and tpa_breakpoint. default values in the table above are now correct and the red line in the figure represents the defaults. maybe @SupaflyFPV has some clearer explanations?

Pichi is very clear in his comment here. "Percent reduction" means that with a value of 20 you will get 20% reduction/attenuation, or 80% of your original value.

yeah its the same as TPA_RATE. It follows the same logic, just for the LOWER breakpoint. 👌

* Removed white spaces and everything that is not new Rebase on master and update comment Make TPA lower independent from air mode Included tpa_breakpoint_lower_vanish option Changes according to PR comments Corrected comment for API version Bugfix in msp.c Additional TPA breakpoint for low throttle * Changes according to PR comments * Update src/main/cms/cms_menu_imu.c Co-authored-by: Jan Post <Rm2k-Freak@web.de> * Update src/main/flight/pid_init.c

limonspb · 2024-01-28T02:46:14Z

Hey @pichim
i was going through the code and noticed this:
{ "TPA RATE LOW", OME_FLOAT, NULL, &(OSD_FLOAT_t) { &cmsx_tpa_rate_lower, 0, 100, 1, 10} },
is there a reason why its FLOAT here?
Because it is declared like this:
static uint8_t cmsx_tpa_rate_lower;
what am i missing? And what means the last "10" before the brackets close?

pichim · 2024-01-28T08:43:28Z

@limonspb tbh, i have no idea : - ) i just copied how it was already done for tpa

haslinghuis · 2024-01-28T22:56:05Z

@limonspb please fix it :)

nerdCopter · 2024-01-29T14:37:50Z

i think 10 is the increment size @limonspb. needs verification, however.

limonspb · 2024-01-30T02:47:05Z

@limonspb please fix it :)

:)
#13337

* Removed white spaces and everything that is not new Rebase on master and update comment Make TPA lower independent from air mode Included tpa_breakpoint_lower_vanish option Changes according to PR comments Corrected comment for API version Bugfix in msp.c Additional TPA breakpoint for low throttle * Changes according to PR comments * Update src/main/cms/cms_menu_imu.c Co-authored-by: Jan Post <Rm2k-Freak@web.de> * Update src/main/flight/pid_init.c

KarateBrot assigned pichim Aug 4, 2023

KarateBrot added the RN: MINOR FEATURE label Aug 4, 2023

KarateBrot added this to the 4.5 milestone Aug 4, 2023

KarateBrot added the Needs coordination with betaflight-configurator label Aug 4, 2023

haslinghuis reviewed Aug 6, 2023

View reviewed changes

src/main/msp/msp.c Outdated Show resolved Hide resolved

klutvott123 reviewed Aug 6, 2023

View reviewed changes

src/main/msp/msp.c Outdated Show resolved Hide resolved

klutvott123 reviewed Aug 6, 2023

View reviewed changes

src/main/flight/pid.h Show resolved Hide resolved

haslinghuis added the Tests failing label Aug 22, 2023

pichim force-pushed the dev_lowerTPA branch from 65209d6 to dbeacd2 Compare August 31, 2023 09:38

haslinghuis removed Tests failing Needs coordination with betaflight-configurator labels Aug 31, 2023

pichim mentioned this pull request Oct 4, 2023

Added tpa lower parameters to header betaflight/blackbox-log-viewer#657

Merged

nerdCopter added the Needs coordination with blackbox-log-viewer label Oct 4, 2023

SupaflyFPV approved these changes Oct 20, 2023

View reviewed changes

pichim force-pushed the dev_lowerTPA branch from dbeacd2 to d9efa62 Compare November 4, 2023 12:06

pichim force-pushed the dev_lowerTPA branch from d9efa62 to 97405c8 Compare November 4, 2023 13:00

haslinghuis reviewed Nov 12, 2023

View reviewed changes

src/main/blackbox/blackbox.c Outdated Show resolved Hide resolved

Changes according to PR comments

4b1001b

haslinghuis approved these changes Nov 17, 2023

View reviewed changes

sugaarK approved these changes Nov 26, 2023

View reviewed changes

sugaarK requested review from klutvott123 and KarateBrot November 26, 2023 04:07

haslinghuis requested review from blckmn and ctzsnooze December 3, 2023 21:00

KarateBrot requested changes Dec 4, 2023

View reviewed changes

src/main/cms/cms_menu_imu.c Outdated Show resolved Hide resolved

src/main/flight/pid_init.c Outdated Show resolved Hide resolved

pichim and others added 2 commits December 4, 2023 16:16

Update src/main/cms/cms_menu_imu.c

218ca0a

Co-authored-by: Jan Post <Rm2k-Freak@web.de>

Update src/main/flight/pid_init.c

2044889

Co-authored-by: Jan Post <Rm2k-Freak@web.de>

KarateBrot approved these changes Dec 4, 2023

View reviewed changes

KarateBrot added the Tested label Dec 4, 2023

KarateBrot merged commit 13d1dc8 into betaflight:master Dec 4, 2023

ctzsnooze reviewed Dec 6, 2023

View reviewed changes

Uh oh!

Additional TPA breakpoint for low Throttle #13006

Additional TPA breakpoint for low Throttle #13006

Uh oh!

Conversation

pichim commented Aug 4, 2023 • edited by KarateBrot Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Update 13.11.2023

Parameters

Update 19.08.2023:

Update 31.08.2023:

Parameters

Propposed Default TPA Curve

Measured Sensitivity Function Roll-Axis

Uh oh!

github-actions bot commented Aug 4, 2023

Uh oh!

blckmn commented Aug 4, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

SupaflyFPV commented Aug 18, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pichim commented Aug 19, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Uh oh!

ctzsnooze commented Dec 4, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

SupaflyFPV commented Dec 4, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

haslinghuis commented Dec 4, 2023

Uh oh!

pichim commented Dec 4, 2023

Uh oh!

Uh oh!

Uh oh!

ctzsnooze commented Dec 6, 2023

Uh oh!

ctzsnooze Dec 6, 2023

Choose a reason for hiding this comment

Uh oh!

pichim Dec 6, 2023

Choose a reason for hiding this comment

Uh oh!

KarateBrot Dec 6, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

SupaflyFPV Dec 6, 2023

Choose a reason for hiding this comment

Uh oh!

limonspb commented Jan 28, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

pichim commented Jan 28, 2024

Uh oh!

haslinghuis commented Jan 28, 2024

Uh oh!

nerdCopter commented Jan 29, 2024

Uh oh!

limonspb commented Jan 30, 2024

Uh oh!

Uh oh!

pichim commented Aug 4, 2023 •

edited by KarateBrot

Loading

blckmn commented Aug 4, 2023 •

edited

Loading

SupaflyFPV commented Aug 18, 2023 •

edited

Loading

pichim commented Aug 19, 2023 •

edited

Loading

ctzsnooze commented Dec 4, 2023 •

edited

Loading

SupaflyFPV commented Dec 4, 2023 •

edited

Loading

KarateBrot Dec 6, 2023 •

edited

Loading

limonspb commented Jan 28, 2024 •

edited

Loading