can linear actuator clutch be glued/removed?

[glowytheglowbug]

can i make the clutch in a linear actuator stop slipping totally? its for my linear actuator engine which needs high precision 

[2GodBDGlory]

Here's some pictures from the Technicbricks blog I used to frequent before it was shut down (accessed through the Wayback Machine):

Based on those, it looks to me like it'd be a pretty simple matter to superglue that orange piece to the white piece, but according to the article, it's quite challenging to actually open the actuator in the first place, because of the glue on the DBG part in the base.

If you want to see the articles, I'll link them here:

http://web.archive.org/web/20181115020450/http://www.technicbricks.com/2008/07/tbs-techreview-02-technic-pf-line

http://web.archive.org/web/20181115020448/http://www.technicbricks.com/2008/08/tbs-techreview-05-technic-pf-linear

Edited September 15, 2023 by 2GodBDGlory

[glowytheglowbug]

11 minutes ago, 2GodBDGlory said:

Here's some pictures from the Technicbricks blog I used to frequent before it was shut down (accessed through the Wayback Machine):

Damn seems doable, i forgot but does the orange piece have a hole all the way through? or could i drill out a tiny hole, use a strynge to insert a bit of superglue and maybe? :/ not sure tho

[2GodBDGlory]

I think it has a hole all the way through, but the issue is that there's a screw inserted into that hole from the other side, so you're going to have a pretty hard time drilling it out, I think.

You'd need a pretty small syringe too, but it might be feasible!

I'd just make sure to have it extended when you insert the glue, so you don't glue the shaft into place as well

Since (as I recall?) you've already broken an actuator on this model, you could always try tearing down the already-broken one first to see if it's possible to do it without destroying stuff too badly

[glowytheglowbug]

11 minutes ago, 2GodBDGlory said:

I think it has a hole all the way through, but the issue is that there's a screw inserted into that hole from the other side, so you're going to have a pretty hard time drilling it out, I think.

You'd need a pretty small syringe too, but it might be feasible!

I'd just make sure to have it extended when you insert the glue, so you don't glue the shaft into place as well

Since (as I recall?) you've already broken an actuator on this model, you could always try tearing down the already-broken one first to see if it's possible to do it without destroying stuff too badly

yeah i have a broken ish linear actuator (not rlly broken tho since its just the plastic bit on the end that didnt want to live and just cracked in a few pieces) 

 

i think i can use those injection needles and really thin super glue


are there any kinds of superglue that wont melt plastic? or react with the grease and stuff to melt everything

Edited September 15, 2023 by glowytheglowbug

[2GodBDGlory]

12 hours ago, glowytheglowbug said:

yeah i have a broken ish linear actuator (not rlly broken tho since its just the plastic bit on the end that didnt want to live and just cracked in a few pieces) 

 

i think i can use those injection needles and really thin super glue


are there any kinds of superglue that wont melt plastic? or react with the grease and stuff to melt everything

Sounds good, so long as you can drill out the metal screw successfully!

I've always just used generic superglue when gluing Lego, without issues, but I'm not an expert, and who knows what effect the grease might have

[aeh5040]

I'm a bit confused why you would want to do that. The "clutch" only slips if you try to drive the actuator beyond its end stops (I believe). If you are doing that then without the clutch you will just stall the motor or break something.

[Lok24]

5 minutes ago, aeh5040 said:

you will just stall the motor

correct. And doing this and detected in a program enables enables you to learn how many rotations/degrees you need from one end to the other of the LA.

Or how many to move to a desired position.

Like with steering a car.

It's very interesting question.
Could'nt you replace the "clutch" with a 3D-printed part?

[aeh5040]

5 hours ago, Lok24 said:

correct. And doing this and detected in a program enables enables you to learn how many rotations/degrees you need from one end to the other of the LA.

You can also just measure this directly without programming, so again I'm not sure what the purpose would be.

[Davidz90]

8 hours ago, aeh5040 said:

You can also just measure this directly without programming, so again I'm not sure what the purpose would be.

Yes, you are right, this alone is simple to do manually. What is the issue is that the motor is driving the LA with variable speed, given by a pretty complicated formula, and no desynchronization between motor and LA is allowed. A slip at any point in motion would desynchronize the system.

I think it is an open question whether using deliberately excessive inputs and allowing for some flex in the system can be done in a way that self-corrects instead of accumulating errors.

[Lok24]

8 hours ago, aeh5040 said:

You can also just measure this directly without programming,

How could that be done?
How could I tell the LA to move exactly to 12 mm from bottom?

[aeh5040]

5 minutes ago, Davidz90 said:

Yes, you are right, this alone is simple to do manually. What is the issue is that the motor is driving the LA with variable speed, given by a pretty complicated formula, and no desynchronization between motor and LA is allowed. A slip at any point in motion would desynchronize the system.

I think it is an open question whether using deliberately excessive inputs and allowing for some flex in the system can be done in a way that self-corrects instead of accumulating errors.

I see, I think. But I'm really not sure whether locking the clutches will help. If a LA gets driven past its endpoint then something has gone very wrong already. Better to aim to keep them well within their range.

[Lok24]

3 minutes ago, aeh5040 said:

If a LA gets driven past its endpoint then something has gone very wrong already.

No, it's the normal way of calibrating something, therefore you can select the force the motor runs with.
I'm pretty sure all the technic sets use that. But then you would not have to modify the LA.

 

 

[Davidz90]

6 minutes ago, Lok24 said:

How could that be done?
How could I tell the LA to move exactly to 12 mm from bottom?

LA has a specified thread pitch and 1 rotation equals to a specific distance. From my measurements, for small LA, 12.5 rotations = 15.8 mm. I used that to build a working micrometer:

 

[Lok24]

5 minutes ago, Davidz90 said:

LA has a specified thread pitch and 1 rotation equals to a specific distance.

Sure. But when driving with a motor and the LA not being on one specified end when starting programm?

Edited September 16, 2023 by Lok24

[Lego Tom]

Super glue probably isn't the best choice. It has strong pull bonding but poor shear strength. A better alternative would be an epoxy, such as JB Weld, especially if melting the original plastic isn't desired, which epoxy will not do. I have used it to repair a plethora of things and always have it on hand. Cure time is 24 hours but when cured, it's permanent.

Not needed in this application, but you can also build up broken parts with it to make them stronger than original and when set up, you can work it with files/sandpaper. Yes, I sound like a commercial for the stuff, but it's really that good.

[Davidz90]

18 minutes ago, Lok24 said:

Sure. But when driving with a motor and the LA not being on one specified end when starting programm?

Sure, but this does not apply here.

[glowytheglowbug]

heya woah theres a ton of replies, i dont have any formula for calculating speed or anything and my current code is quite simple, it uses david's formula to set all motor's relative position and then i make motors go to a certain relative position based on the output shaft angle + 5 degrees

3 minutes ago, glowytheglowbug said:

heya woah theres a ton of replies, i dont have any formula for calculating speed or anything and my current code is quite simple, it uses david's formula to set all motor's relative position and then i make motors go to a certain relative position based on the output shaft angle + 5 degrees

main reason for gluing the la clutches shut is if they slip too much they basically will not go over the dead spot in the engine and thus try to oppose motion, breaking itself

[Davidz90]

Yes, the lengths of all 4 actuators are determined by the crankshaft angle. I agree, if we allow for no slip, then the starting angle has to be specified very precisely.

 

Still, I'm wondering if allowing a slip is not a better option? You see, if the LA motor is in some specific phase of motion (specific speed), there is only exactly one crankshaft position that matches it. Allowing motors to catch up depending on feedback from crankshaft seems like a good idea?

[Lok24]

7 minutes ago, Davidz90 said:

Sure, but this does not apply here.

That sounds like programming.....

43 minutes ago, Davidz90 said:

What is the issue is that the motor is driving the LA with variable speed, given by a pretty complicated formula,

To move to a defined relative position you once need and absolute position.
Perhaps I missed the problem, sorry.

[Davidz90]

1 minute ago, Lok24 said:

To move to a defined relative position you once need and absolute position.

Yes, the starting, absolute position is known because it is tied to starting crankshaft angle, which I admit, has to be known very precisely.

[Lok24]

Then you have to move the LA manually back to zero before starting program.

a usual way is
- set force to a desired value (here: when LA stops, before clutch slips)
- start driving back to zero
- when motor stalls : stop motor
- set current angle to 0

Edited September 16, 2023 by Lok24

[Davidz90]

16 minutes ago, Lok24 said:

Then you have to move the LA manually back to zero before starting program.

No, they are tied to a common crankshaft, you cannot move them to 0 separately. You only need to manually check that LAs are not pushing/pulling on the crankshaft (so their positions are exactly matching the intended one) and determine the exact angle of crankshaft.

[glowytheglowbug]

Just now, Davidz90 said:

No, they are tied to a common crankshaft, you cannot move them to 0 separately. You only need to manually check that LAs are not pushing/pulling on the crankshaft (so their positions are exactly matching the intended one) and determine the exact angle of crankshaft.

yep only problem is that lego is plastic and plastic bends easily

[Davidz90]

9 minutes ago, glowytheglowbug said:

yep only problem is that lego is plastic and plastic bends easily

Yes, and that's why I think that the only way forward is to figure out a driving algorithm that actually alows for some slip/elasticity and doesn't accumulate error but instead uses that elastic feedback to self-correct.