Performance

david_brophy

Check out the video below:

https://youtu.be/mTxOp-vzc0g

The drawer takes ~15 sec to start to move when I drag it. Here's the document:

https://cad.onshape.com/documents/690e602c0b08069eae458d07/w/f143d47164fd3143ad1a68bd/e/0b601c052c64e1b2fb635602?renderMode=0&uiState=6612cadf7dd0923fb3f0ebfd

Why is this? Can anyone else reproduce this long delay? Is this expected performance?

The model doesn't seem that complex to me... right? My long-term plans for this project would see a 10-20x more complex model... I'm really worried I'm investing hundreds of hours of design time into a platform that won't scale...

My system / connection are fine:

NeilCooke

Same here. I don't think it's unreasonable what you have done, but it is perhaps a little overboard to have the "drawer pairs" assembly and also have it assembled for each drawer configuration. Limits also slow down movement as there is more to calculate. However, like I said, it's not unreasonable the way you have done it. You will notice that it drags a little quicker if you use the manipulator rather than just "grab and drag". It's almost as if it is regenerating each assembly configuration before solving the mates, which is what it could well be doing. Support will know better - please open a ticket.

eric_pesty

The one thing I would is that slides pairs with limits on both slides are going to slow things down significantly, shouldn't be 10s but definitely is going to hurt.
You are mating both ends of the slides to two rigid things so you effectively have a bunch over-constraining happening. I'm also not clear on what is going on with the two sets of limits within the slides, shouldn't you have a limit controlling the total extension and a gear mate positioning the middle slide?

I would remove the limits from the slides, then mate the two sketches directly to each other with a slider with limits. You can then add the slides by fastening the fixed part and maybe use something like a planar between the end of the slide and sketch 2.
Basically turn the slides into a "cosmetic" component rather than use them to physically connect things and it will help...

A quick test would be suppress one of the slides in the "draws" assembly, and/or remove limits and seed if that helps.

david_brophy

OK I'll try to tweak the slide mates and see if that fixes things. 

eric_pesty said:

You are mating both ends of the slides to two rigid things so you effectively have a bunch over-constraining happening. 

Yeah, I needed a way to lock the two sides of the slide together, because I only want to connect it to the cabinet on one side... Originally the slide on the other side wasn't anchored to anything so it was all over the place. I was looking for a way to link the slider mates so one would always mirror the other. I haven't used the gear mate before, sounds like that will help...

david_brophy

OK so I changed two things: 

1) Changed the slide model so the inner part is mated directly to the outer part, and the middle part uses a linear relation to drive it's motion (with a 50% relation ratio).
2) Changed the slide pair assembly so the two slides are locked together using another linear relation.

That seems to fix it! The lag is only perhaps half a second now 🎉

Thanks again guys!

eric_pesty

Glad that helped, my experience has been to be careful to avoid having multiple mates constraining the same things. Even if they aren't "conflicting" they do slow things down significantly.

An example of this would be a shaft that is mated to different things at both ends with revolutes but the two end parts are also constrained to each other in the axial direction of the shaft. If the mates "agree" (i.e. the length of the shaft matches exactly where the two revolutes are located) it might work with no error but things will be smoother if one of the revolutes is changed to a cylindrical so they don't fight each other.