Open Source Joystick FFB / DIY FFB Joystick

just found very impressive "unclassified" document on this subject from 1948;)

http://www.dtic.mil/dtic/tr/fulltext/u2/639028.pdf

just found very impressive "unclassified" document on this subject from 1948;)

 

http://www.dtic.mil/dtic/tr/fulltext/u2/639028.pdf

Wow, what a find, and it's even had an OCR tool run over it which is fantastic for keyword searches!

@Slartibartfast

...

BTW - did you get my reply on the programming subject?

Yes, I did, sorry for not responding. I've been busy of late but will hopefully be able to give it my full attention shortly.

Briliant, thanks VO101_MMaister

 

One more question though. I'm currently drawing the mechanism and it occurred to me that 30 degree motion range (as on the small stick) is probably way too much for 500mm long shaft..

Any suggestions on the travel range on top of the stick?

 

Jay

For my part when I went seeking these design limits I was quoted:

10° forward
25° backward
17° left and right

The actual length of the stick from pivot point to tip of the grip is 300mm for pitch, and 250mm for roll (the pitch axis being 50mm lower than the roll). Also, the grip, though vertical, is also swept back 120mm:

These specific values are from a glider but should be typical for most stick controlled aircraft. I don't recall exactly what the full limits in my design are but I did add adjustable bolts such that one can tailor individual limits for the amount of throw in each direction and I have the bolts set to achieve exactly the dimensions specified above.

So if you design is 30° is from full lock to full lock it might actually be a little constricted (though probably fine), however if your talking about 30° from center to one side, then yeah, that's way toooo much :)

good evening, everyone.

I was busy for my certificate…

Do you need information about cockpit design?

in the 1970's, stick's range of movement is defined by MS spec

roll max ±7 inch

pitch push max 5, pull max 7 inch

and stick length max 13.5 inch

I don`t think that we are gonna find a common ground on this one, but why should we? Everyone can build a stick with the preferred deflections and forces, that is the beauty of this.:)

VR it is a very good document, thank you for sharing!

I can’t help with regards to motor choice. Backlash will always be existent with spur/straight cut gear mechanisms and any gearing system in some magnitude. Helical gears would be more ideal.

 

Another topic I haven’t seen discussed here is force sensing in addition to a force feedback. Without force sensing it will never be a proper closed loop system. Strain gauges are very cheap and a bridge can be placed almost anywhere on the Control stock and calibrated. Then programmatically the force being applied to the stick would always be known. This has the obvious benefit of detect hands on or off stick but also when interpolated with stick position can deliver a proper moment reaction force in a continuous loop.

Force sensing is only used where super expensive/complicated 3ph ac systems are being used in FAA level D sims or other specific purposes where extreme precision is critical or the precise force is actually relevant to the goal of the simulation. Stuff like this however is done via calculating and correlating the motor output torque curve plus whatever mechanical advantages are inherent in the system and everything is extrapolated via position+drive state, which is plenty precise and reliable for the goals of what is being done. This is already complicated enough without bogging it down with unnecessary considerations.

Gears are not ideal due to backlash, cogging, the need for expensive minimal backlash gears, sealed mechanisms/grease/precision machining req, gear wear making things worse over time, and helical gears are impractical which are even more expensive/less accessible and req even higher precision machining to incorporate and unless you buy really expensive ones are still very much subject to backlash. To keep from reinventing wheels and falling short of what others have already established, anyone serious about this would really benefit from bookmarking Roland van Roy's site, simprojects.nl, particularly the http://simprojects.nl/diy_force_feedback.htm section and study what he has done over the years.

Real systems tend to use direct drive/pushrods/bellcranks with AC 3ph motors/drives but when they need to be compact tend to use 'shaft winding', where you use pulleys or a pulley/bellcrank combination. The drive transfer happens with a cable fed through a cross-drilled hole(s) on the drive side, then wrapped several times around either side of the hole (or two holes with their own wrapped sections) before it goes to the pulley or bellcrank it is driving, which it wraps around and is fixed to. Steel cable traditionally used but kevlar rope is better due to its very low stretch/creep. Timing belts have cogging issues (even expensive ones with v-pattern teeth), v-belts slip, gears are impractical/expensive/difficult to properly incorporate, direct drive way out of budget and massive in size...

Example of 'shaft winding' power transmission:

This provides a simple means for smooooooth power transfer at unlimited mechanical advantages at very low cost. Cogging effects are limited to that of the motors themselves which btw FF/control loading systems necessitate special (typically expensive) motors. For DC servomotors (definitely the best option for DIY) you should look for ones with a helical armature, which keeps the torque output smooth even at low rpm.

Keep in mind you can't simply overcome a regular motor's cogging by making the gear ratio (or pulley ratio) so high you don't feel it otherwise the system suffers inertial effects as the motor is spinning too fast to stop and reverse directions without some overrun so motors much be chosen carefully because frequency response of the entire systems is critical to performance. DC servomotors are the best option here because driving them is easy (cheap) as well as controlling the drives and their power output constant and well chosen ones have no cogging at low rpm's.

Example of a good DC servomotor for this application:

note the helical armature, that prevents the length of the polarized bar from coming in contact with it's drive element at the same time, preventing cogging

Not to sound discouraging, but hardware is honestly the easy part... interfacing is the real challenge which is why the option of hacking an MSFFII for its brain/drive signals is so enticing because it was actually done well and responds beautifully to many sims already and I say this having already figured out a good way to interface.

Source: Years of my own dabbling on the subject plus I work for a company that makes high end professional motion platforms and control loading systems and I've spent a lot of time picking my boss's brain and talking shop... he did the control loading systems on the Air Force One Blackhawk some years back for example

Force sensing is only used where super expensive/complicated 3ph ac systems are being used in FAA level D sims or other specific purposes where extreme precision is critical or the precise force is actually relevant to the goal of the simulation. Stuff like this however is done via calculating and correlating the motor output torque curve plus whatever mechanical advantages are inherent in the system and everything is extrapolated via position+drive state, which is plenty precise and reliable for the goals of what is being done. This is already complicated enough without bogging it down with unnecessary considerations.

Thanks for the detailed response. I thought it seemed like a rational consideration but you're obviously much more knowledgeable on this subject.

Gears are not ideal due to backlash, cogging, the need for expensive minimal backlash gears, sealed mechanisms/grease/precision machining req, gear wear making things worse over time, and helical gears are impractical which are even more expensive/less accessible and req even higher precision machining to incorporate and unless you buy really expensive ones are still very much subject to backlash. To keep from reinventing wheels and falling short of what others have already established, anyone serious about this would really benefit from bookmarking Roland van Roy's site, simprojects.nl, particularly the http://simprojects.nl/diy_force_feedback.htm section and study what he has done over the years.

I was only trying to say helical is more ideal than straight cut in response to the post above mine. I wasn't saying that geared power transmission is the ideal solution.

Real systems tend to use direct drive/pushrods/bellcranks with AC 3ph motors/drives but when they need to be compact tend to use 'shaft winding', where you use pulleys or a pulley/bellcrank combination.

If you say my previous post, I mentioned bellcranks. Good to know I was on the right track.

The drive transfer happens with a cable fed through a cross-drilled hole(s) on the drive side, then wrapped several times around either side of the hole (or two holes with their own wrapped sections) before it goes to the pulley or bellcrank it is driving, which it wraps around and is fixed to. Steel cable traditionally used but kevlar rope is better due to its very low stretch/creep. Timing belts have cogging issues (even expensive ones with v-pattern teeth), v-belts slip, gears are impractical/expensive/difficult to properly incorporate, direct drive way out of budget and massive in size...

 

Example of 'shaft winding' power transmission:

 

This provides a simple means for smooooooth power transfer at unlimited mechanical advantages at very low cost. Cogging effects are limited to that of the motors themselves which btw FF/control loading systems necessitate special (typically expensive) motors. For DC servomotors (definitely the best option for DIY) you should look for ones with a helical armature, which keeps the torque output smooth even at low rpm.

 

Keep in mind you can't simply overcome a regular motor's cogging by making the gear ratio (or pulley ratio) so high you don't feel it otherwise the system suffers inertial effects as the motor is spinning too fast to stop and reverse directions without some overrun so motors much be chosen carefully because frequency response of the entire systems is critical to performance. DC servomotors are the best option here because driving them is easy (cheap) as well as controlling the drives and their power output constant and well chosen ones have no cogging at low rpm's.

 

Example of a good DC servomotor for this application:

note the helical armature, that prevents the length of the polarized bar from coming in contact with it's drive element at the same time, preventing cogging

 

Not to sound discouraging, but hardware is honestly the easy part... interfacing is the real challenge which is why the option of hacking an MSFFII for its brain/drive signals is so enticing because it was actually done well and responds beautifully to many sims already and I say this having already figured out a good way to interface.

 

 

 

Source: Years of my own dabbling on the subject plus I work for a company that makes high end professional motion platforms and control loading systems and I've spent a lot of time picking my boss's brain and talking shop... he did the control loading systems on the Air Force One Blackhawk some years back for example

Thanks, this is all great information. Could you share more images of the overall mechanism? It'd be much appreciated.

Thanks for the detailed response. I thought it seemed like a rational consideration but you're obviously much more knowledgeable on this subject.

 

 

 

I was only trying to say helical is more ideal than straight cut in response to the post above mine. I wasn't saying that geared power transmission is the ideal solution.

 

 

 

If you say my previous post, I mentioned bellcranks. Good to know I was on the right track.

 

 

 

Thanks, this is all great information. Could you share more images of the overall mechanism? It'd be much appreciated.

Thanks Sinusoid. BTW, I'm cavortingweasties on reddit, I used to only lurk this forum till you encouraged me to join and start posting last year. I am already on so many forums that I was hesitant, but I've always liked this one so it was an easy sell.

The hardware in the pic is Roland van Roy's yoke, which used to run by his/Ian's FF interface but has since been updated to be powered by an MSFFII (yes, the big Glentek motor has an MSFFII telling it what to do) by modifying the board to handle 400% the current and using his own 24v power supplies in the form of power bricks, and the mods were really simple consisting of some resister changes and adding some diodes. He meticulously documents it all here, where you can see it evolve from:

http://simprojects.nl/flight_yoke_FF_mechanics.htm version 1

http://simprojects.nl/forcefeedback_yoke_ii.htm version 2

http://simprojects.nl/ms_siderwinder_ff2_hack.htm -hacking MSFFII

For giggles, here's a schematic Roland made from the FF interfacing idea I came up with in 2008, which was tested and found to be reliable and straight forward. It's underlying function is based off extracting the airspeed and assigning it a 0-255 value (if using an 8 bit DAC) and using that to define a force potential. That goes through a DAC to convert to voltage, in this case I was using 0-10vdc cause because with that I could 'jog' the 3ph AC drives I had, bypassing their super complicated functions involving encoders and like 36 wires and a bunch of irrelevant features. An inverted signal is also needed, so a separate part of the circuit generates 0 to -10vdc to match the positive side. That way I could drive the motor with either left or right variable force in a very straightforward manner. (compact Japanese 3ph ac servomechanisms I got on ebay)

I say potential because if the stick is centered the output is 0vdc, but it will ramp up to 100% of the current output as you deviate from 'home'. The positive and negative potential is fed into the end taps of a position pot, and resulting deviations from the zero position will range in outputs from 0 to +/-10vdc depending on airspeed and whether the deviation is right or left.

It got nicknamed 'driven pot' because it's using actively variable +/- 0 to 10vdc to define the end max outputs of a potentiometer, and the wiper puts out whatever appropriate fraction of that depending on displacement which is being fed into the drive to jog it left or right at whatever force it's calibrated to. It might sound complicated, but it's a very simple way to turn stick position into variable +/- voltage output which can then be used to inform powerful cw/ccw motor commands.

...easy is a relative term though. This setup would only allow variable forces based on airspeed as well as easily manipulated offsets like trim. Everything else beyond that requires extracted 'trigger events', so things like stall shakes/stall making controls floppy, landing bumps, machine gun vibration etc req a pre-scripted/tuned library of effects. Every single game/sim it is intended to use with would req parameter extraction of airspeed at minimum, but you would need a lot more cues to get the full range of effects, though some sims make that data easier to get than others. Each effect needs tuning as well as variations in strength that would also need variable event triggers and each plane its own tuning and as you can see this quickly spirals out of control with busy work.

It's a great solution for a setup that is only going to use a single aircraft/sim or at least a limited range, but a nightmare if you want something that 'just works' with a lot of stuff.

This is why imo the smart money is on using MSFFII's brain, because all that busy work was already done by really smart people and many sims/games still make use of it. MSFFII in particular was done really well and and is well tuned. There are also 3rd party software programs to gain further control of it for sims that lack it and because of this the heavy lifting is already done, leaving the easy and fun part -making hardware that makes better use of it than the stock arrangement.

Thanks Sinusoid. BTW, I'm cavortingweasties on reddit, I used to only lurk this forum till you encouraged me to join and start posting last year. I am already on so many forums that I was hesitant, but I've always liked this one so it was an easy sell.

Oh, wow! Well I'm glad my attempt to persuade you worked :lol:

The hardware in the pic is Roland van Roy's yoke, which used to run by his/Ian's FF interface but has since been updated to be powered by an MSFFII (yes, the big Glentek motor has an MSFFII telling it what to do) by modifying the board to handle 400% the current and using his own 24v power supplies in the form of power bricks, and the mods were really simple consisting of some resister changes and adding some diodes. He meticulously documents it all here, where you can see it evolve from:

 

http://simprojects.nl/flight_yoke_FF_mechanics.htm version 1

 

http://simprojects.nl/forcefeedback_yoke_ii.htm version 2

 

http://simprojects.nl/ms_siderwinder_ff2_hack.htm -hacking MSFFII

Gotcha, I didn't realize it was from the book! I'll dig into it.

For giggles, here's a schematic Roland made from the FF interfacing idea I came up with in 2008, which was tested and found to be reliable and straight forward. It's underlying function is based off extracting the airspeed and assigning it a 0-255 value (if using an 8 bit DAC) and using that to define a force potential. That goes through a DAC to convert to voltage, in this case I was using 0-10vdc cause because with that I could 'jog' the 3ph AC drives I had, bypassing their super complicated functions involving encoders and like 36 wires and a bunch of irrelevant features. An inverted signal is also needed, so a separate part of the circuit generates 0 to -10vdc to match the positive side. That way I could drive the motor with either left or right variable force in a very straightforward manner. (compact Japanese 3ph ac servomechanisms I got on ebay)

 

I say potential because if the stick is centered the output is 0vdc, but it will ramp up to 100% of the current output as you deviate from 'home'. The positive and negative potential is fed into the end taps of a position pot, and resulting deviations from the zero position will range in outputs from 0 to +/-10vdc depending on airspeed and whether the deviation is right or left.

 

It got nicknamed 'driven pot' because it's using actively variable +/- 0 to 10vdc to define the end max outputs of a potentiometer, and the wiper puts out whatever appropriate fraction of that depending on displacement which is being fed into the drive to jog it left or right at whatever force it's calibrated to. It might sound complicated, but it's a very simple way to turn stick position into variable +/- voltage output which can then be used to inform powerful cw/ccw motor commands.

 

 

...easy is a relative term though. This setup would only allow variable forces based on airspeed as well as easily manipulated offsets like trim. Everything else beyond that requires extracted 'trigger events', so things like stall shakes/stall making controls floppy, landing bumps, machine gun vibration etc req a pre-scripted/tuned library of effects. Every single game/sim it is intended to use with would req parameter extraction of airspeed at minimum, but you would need a lot more queues to get the full range, though some sims make that data easier to get than others. Each effect needs tuning as well as variations in strength that would also need variable event triggers and each plane its own tuning and as you can see this quickly spirals out of control with busy work.

 

It's a great solution for a setup that is only going to use a single aircraft/sim or at least a limited rance, but a nightmare if you want something that 'just works' with a lot of stuff.

 

This is why the smart money is on using MSFFII's brain, because all that busy work was already done and many sims/games still make use of it. MSFFII in particular was done really well and and is well tuned. There are also 3rd party software programs to gain further control of it for sims that lack it and because of this the heavy lifting is already done, leaving the easy and fun part -making hardware that makes better use of it than the stock arrangement.

Wow! This is wonderful stuff! Airspeed data export shouldn't be a problem with export.lua. As you said though this is a minimum. For the F-15C (my first love) airspeed wouldn't be required, the FCS blends airframe load factor with pitch rate, and aircraft response is held constant throughout the envelope. For S&G's here's what the actual F-15 force feel characteristics look like from a NASA TP (You may have seen this before)

Longitudnal:

Lateral:

For the Su-27, FFB is already supported, even stick shaker which has its parameters defined su27fmoptions.lua.

IMHO, the modern jets would be the easier ones in terms of control loading

Aircraft with "reversible" flight control systems (WWII aircraft basically) could be quite a challenge aside from a basic airspeed schedule like you said.

Again, thanks for the details. This is a really good contribution.

Has anyone seen this:

https://www.xsimulator.net/community/threads/diy-ffb-steering-wheel-mmosffb-in-progress.7769/

Seems like a lot of the work has already been done, I think the software even has a DCS profile.

Hey boys

It's been little quiet around here, I hope everybody's busy working on OSFFB:smilewink:

I was quite ill last week so I've done ton of research on DC and BLDC motors, PSU's, drivers, torques, transmission techniques and so on and on...

Finally I managed to draft concept design for my "floor mounted" control stick. It's a work in progress however half past one in the morning here so I wanted to post at least one teaser shot before I fall into my bed.

More coming soon!:pilotfly:

Jay

Finally I managed to draft concept design for my "floor mounted" control stick.

Oow, oow oow!

That is absolutely gorgeous!!

(you may need to leave me alone with that rendering for a moment :p)

Seriously though, that relay looks absolutely perfect!

Thanks :)

I need to share few numbers with all of you involved.

There is one crucial ratio that I was grinding my teeth about last couple of weeks.

"Money vs Torque".

In my opinion, project like this should not exceed £300 budget. It wouldn't make sense if it was any more expensive because there are already other options available for "higher price" e.g. BFF driver cards for BLDC motors.

However building 300 quit FFB device has it's limitations. Considering prices of DC motors, suitable power supply unit and driver cards, I think we are looking at max possible torque of our DC motors about 0.5Nm.

With reasonable gear reduction, keeping in mind acceptable inertia of the motors, the system like the one I'm drawing, will be able to generate no more than 4-5Kgf on top of 500mm long stick.

To achieve mentioned 10Kgf, we would have to stretch the budget way over £500.

As for my self I'm gonna live happily with 5Kgf limit;)

Let me know what u think gang...

Jay

Yes 5kg for 300€ would be great.

Thanks

"Money vs Torque".

In my opinion, project like this should not exceed £300 budget.

Might it be possible to spec out multiple options?

For me I think £300 for a 0.25 Nm system sounds fantastic! (in fact I would actually be quite surprised if I get out of this project for as little as £300) but for other enthusiasts even 1000 USD wouldn't be all that expensive.

Most heli cyclic controls seem to start around the 1000 USD price (such as Flight Link's G-Stick III). Could a single design be fitted out with either mid-priced or heavy-duty components, or would the physical design itself need to me significantly different? If the same design can be fitted out with 0.25 Nm components for 500 USD or 0.5 Nm components for 1000 USD this would allow builders to construct either a 500mm flight stick or a 1000mm cyclic.

It would also be good to offer a desk mount or side-stick option as well, essentially a MS Sidewinder FFB 2 replacement, though I understand this would require a significantly different design.

For me though 0.25Nm sounds perfect and if it can be done for £300, that would be absolutely fantastic!

When it comes to reasonable gearing it is important to find out what is the minimum output RPM which is responsive enough. Maybe the much bigger DIY FFB steering wheel community has some numbers on this.

Hi all, just found this thread last night, amazing work, hope the progress is going well!

I currently have about 5 MS FF2 sticks, which I plan to mod, but I'd also love to build one myself.

I have favor to ask, there is little need for Force Feedback stick if the force feedback not implemented in the game.

In the wishlist section of upcoming F 18 module, I've made a request to support force feedback:

https://forums.eagle.ru/showthread.php?t=195393

Please can you chime in, a huge part of reviving FF joysticks in flight sims will be to explain why force feedback is needed, and show that there is an interest from the community.

Thanks.

Hey all

I think that the physical design between mid-range and heavy-duty needs to be significantly different . Each version will have very different motors in size, gearing, PSU, etc. Overall size of the device will change accordingly as we want to have all the components neatly packed to smallest possible space.

The brain however (Arduino+control loading software), should be identical for any individual mod. Everybody can choose theirs own H bridge, motors and psu.

I have finally chosen my pulleys that I think are gonna work best, DC motors and belts so I will change this in the CAD model and show you the whole thing soon.

I'm also ordering these parts now to make simplified version of the transmission first and test the feel of it.

Will keep you posted.

Jay

BTW

anybody in UK with descent metal workshop?

I just started using my shop again and I have to finish fulfilling the pnp Cougar hall mod orders I have. Once those are completed I want to focus on a FFB base again. A few years ago I did a lot of experimenting with many different motors and controllers. I have built a few CNC machines for specific tasks so I had a lot of hardware laying around. After a lot of time and frustration I abandoned the project because I could not get a smooth feel to the stick (stiction and cogging). I recently revisited the project but took a different approach using load cells with conventional 2 phase steppers as well as servo motors. I finally got close to the feel I was looking for. I need to reassemble my test rig and I will post some pics. These are not new ideas to anyone, this is just what I am experimenting with now.

2 arduinos in the system

Pro Micro as HID (reading thrustmaster shift register and X, Y, axis)

Nano for interpreting force sensors and providing direction and pulse to stepper driver.( move to 48mhz DUE eventually)

Parallel beam load cell with HX711,

nema 23 stepper and driver

Obviously there are a lot of problems that go along with the steppers. Mainly the torque varies significantly with microstepping and heat. The force sensors were very easy to implement and are used to drive the joystick to a specific deflection based on the force applied. This completely eliminated the stiction and allowed a very soft but consistent center. My first test was with a parallel beam force sensor mounted to a large rc servo and I was really surprised how well it worked with minimal code.

I mostly fly Helicopters so my intent was to make a force centering and trim stick without FFB but have it be highly adjustable (deflection and force). Eventually adding FFB as projects produce viable code. VR Flistsim your design is similar to my single axis test rig and I found it the simplest to implement. I tried ball screws, worked great, but difficult to implement. For me I think I'm going to try a 3 phase closed loop stepper next. With a closed loop system you don't need to have position sensors on the stick and I found the steppers smoother feel than a BLDC motor. You can just read the current step as your axis position. I will start posting code, specs, pics, and results of load cell setups with various motors.

My long term intent was to make a universal and upgradable FFB gimbal base kit that different systems could be added to based on the consumer's budget and goal.

Miles

Hi Miles

All that sound very impressive mate. Although I'm not completely convinced that the steppers are the best choice, I would love to see more about your setup, especially the implementation of force sensors.

I may be wrong with the choice of brushed DCs however building a test rig is the only way to find out:joystick:

The feel of my FFB Yoke powered by DCs is still quite satisfying though.

I've got very good feeling about where this is going :thumbup:

All right boys..

This picture will give you an idea about the final size of my design. All the "power" components are modeled after real goods that are on my "buy to test" list and everything was compacted as much as possible.

Still need to finish belts and some bits and bobs but it's looking neat already:D

MetalGear Honk, how is the code coming along?:smartass:

@ VR flightsim,

That looks really great. The only thing I would change is to make the stick 180 degree reversible. That would get pivot axis as close to the seat as possible if the user does not have the height to slide the bulk of the unit under their seat. Again great design!!. I am not convinced that steppers are the best choice either. The only way they work for this application at all is with load cells or a hybrid closed loop system. I don't think you can go wrong with brushed motors. It is what they use in current systems and its the most economical solution. I find the new hybrid closed loop stepper interesting and I just want to experiment with them. Especially the 3 phase ones which act like a standard servo motor except with a lot of torque. I think I still have a load cell attached to a servo and I will try and post a video of that operating tonight.

Miles

Hey guys, just a short question concerning modern aircraft and a simmer's FFB joystick. I a modern Fly-by-Wire airframe, would you feel in RL increasing resistance at higher speeds? Since there is no direct connection, do real FBW aircraft simulate that for the pilot as sensory feedback?

Have any interfacing solutions been worked out yet or is the plan to make the hardware first and then try to figure that out?

Also to anyone stumped on how to create -smooth- power transition, read or reread the post I made about shaft winding... https://forums.eagle.ru/showpost.php?p=3274336&postcount=255 this is a viable and accessible option and also used in RL systems for its performance. It's feasible/cheap enough to diy, yet good enough for the Air Force One Blackhawk sim and many others. No belts or gears in this shoestring budget will be able to provide anything near a smooth output.

You can't compensate for cogging effects by making the gear ratio dramatic enough to mask it -this causes the motor to spin way too fast and then suffers inertia effects where the stick is commended to stop or reverse but the motor can only sluggishly obey because it's spinning so fast. You lose the ability to have higher frequency effects as well.

And a brushless motor:

https://www.ebay.com/itm/High-torque-3-8N-m-12V-DC-motor-DC-high-power-600w-resolver-brushless-servomotor/222596428686?hash=item33d3c7f38e:g:-NkAAOSwJLJZfAnF