Need to fully constrain sketches 8

[jaguar306]

I work in an environment that feels the need to force us to always fully constrain sketches. I feel that this practice is somewhat limiting and I find myself adding constraints just to get the sketch fully constrained which causes problems downstrean. I abhore this practice as I feel you only need to tie down what you require to constrain your geometry and would welcome the forums opinion on this subject.

 

[MSPBenson]

Jag,

It's better to fully constrain a sketch as it is more predictable when/if you update your model.
I came across a good example of this a while ago.
I use sometimes sketches to control studio splines. Just 2 lines either end and controling the angle and position of the lines to tweak the spline.
I never used to dimension the length of the lines. Who cares how long they are (or so I thought). A few days later I was editing the part and it kept failing on one of these sketches. Took me ages to work out that the length of the line was reducing down to zero and causing the spline to fall over as a zero length line hasn't got a tangential direction.
Something as simple as this caused a lot of wasted time.

Basically why spend time second guessing what will and won't cause update issues? If you want robust parts fully constrain sketches. And do this sensibly. Don't just throw lots of fixed constraints in there )

If you don't need a robust model then don't worry too much about the constriants.

Mark Benson
Aerodynamic Model Designer

 

[MCGNX]

I partially agree with both of the above. What I would like is in the instance when you throw a line in a sketch for reference of some sort and you find your self having to constrain the length or end points, to be able to just make that line a "construction line" that goes on forever in either direction....that is, it has no end points or length.

 

[jackley]

That is what constant length is for. Also turn it to a reference line.

Justin Ackley
Designer

 

[hudson888]

Jaguar,

No it isn't necessary to fully constrain sketches. The software allows for them to work effectively with NO constraints. Naturally few people would want to do that and some feel that you'll always be better served for stability of the model if things are locked down so to speak. On occasion needs will dictate that this would be the case. Sometimes you just don't need to.

Let's just point out that if you want to alter somebody's model using the parameters then you have the option to ADD constraints to their sketch or change existing ones so that it maintains its position relative to whatever you constrain it to and/or from.

Increasingly it seems we will have the capacity to use direct modeling and its descendants in NX-6 and beyond to add features as modifiers to earlier construction. In that sense I think again the insistence on fully constraining existing sketches that work without may not be all that important.

Lastly if you work with a mindset that is uptight about these sorts of things and the facts don't back up the need to concern yourself with fully constraining the sketches then what you have is a behavior that bears the hallmarks of ritual and superstition. I hope we're better than that at providing accurate and factual information.

Nil desperandum illegitum carborundum.

Hudson

 

[SiW979]

Jaguar

Also try and keep your sketches as simple as possible omit fillets from within the sketch and chamfers and use form features where possible as they take care of a great deal of work for you. I have people from all sorts of CAD back grounds coming to work for us and I train them not just to rely on sketches, but combine the full power of the system which many find hard to grasp especially Pro/E users where everything is sketch based. our company standards insist that all sketched are fully constrained and I feel once you become fully profficient in use of sketcher nothing is to difficult to achieve. The hardest part of sketcher is understanding the infered constraints I can guarentee that if one of our new users has a problem with an over constrained sketch or one that has fallen over drastically, it is because the system has infered a constraint without them knowing it and it's the ablilty to ba able to interorgate the sketch and find the offending consraint which makes the difference between and average and expert sketch user.

I also have problems with a lot of the 'old boys' who will not use sketcher and ceate models by using explicit curves only, they argue that the part is a one off and it will never need modifying once it's released, but it's the ability to modify the part during the developent process using named expressions from within the sketch and from the feaatures that make so very powerful have a look at the attached NX4 model, open the expression dialogue and change the radius to 111 and the width to 71. I have used lots of reference geometry within the sketch to control the position of form features and notice the extra bosses that appear upon creation of the edit. All postioned by fully constrained sketches and so very easy to edit compared to explicit curves or non constrained sketches.

All the best

Best regards

Simon

 

[SiW979]

Jaguar

and one more thing, remember that by clever use of things like exrude with offsets you can really simplify your sketches for example to extrude an L shape section you don't need create a sketch that consists of 6 lines to form a closed loop, you can sketch 2 lines to give you a match stick stlye sketch but add offsets to the extrude and you'll get a very simple fully constrained pparametric sketch less all the work of constraining the other 4 lines.

Best regards

Simon

 

[ewh]

Simon makes a very valid point about using expressions. Wether you are using the sketcher or not, and regardless if it is a one-off part, expressions make the development of the part much more fluid. My rule here is that if their are more than one of any feature, define it with expressions. You then only have to modify the expressions once, instead of editing every occurance of the feature. Arrays already accomplish this.

Believe it if you need it or leave it if you dare. - [small]Robert Hunter[/small]

 

[JohnRBaker]

Simon,

WHERE DID YOU GET THE INSPIRATION FOR THAT MODEL?

If you look at the image I've attached below you may see something familiar.

With each new version of our software, in this case NX 6, at some point during the early development phases, we get to where we can actually start modeling real parts. At that point I usually either go back and remodel some of my older demo models or I look for some new inspiration. This year I was in the office of one of our lead programmers and he had just gotten hold of an old Drafting Textbook from the 30's and it was full of these really neat illustrations of models that of course were intended as assignments for students to practice creating 2D engineering drawings. I love these old textbooks for several reasons. First off, I love the old hand-drawn illustrations, you could frame and sell them as 'art' in many cases, and they are also usually 'safe' to use, particularly if it's an old textbook as no one will complain if some image of the part ends up being used somewhere. Also they are just complex enough that it will usually take only an hour or so each to create the equivalent solid model.

Anyway, it would appear that we have come across the same old textbook (for the record, I didn't build in any fancy expressions or behavior like your model, which is probably how I managed to use only 25 features, but I did finish the model completely).

John R. Baker, P.E.
Product 'Evangelist'
NX Design
Siemens PLM Software Inc.
Cypress, CA

http://www.siemens.com/plm

http://www.plmworld.org/museum/

 

[JohnRBaker]

Simon,

And on the off chance that we have been working from the same old drafting textbook, here's another 'assignment' I used during my NX 6 testing. Of course this allowed me to test a much wider range of NX functionality.

John R. Baker, P.E.
Product 'Evangelist'
NX Design
Siemens PLM Software Inc.
Cypress, CA

http://www.siemens.com/plm

http://www.plmworld.org/museum/

 

[SiW979]

John

That image is from the back of one of your manuals, it is one of 22 additional projects for trainees to attempt after their weeks training. We always get a bunch of manuals everytime NX upgrades. The manual where that model comes from is the UGS practical applications NX4 manual if you like I can send it to you as I have it as a PDF as well.

Best regards

Simon

 

[SiW979]

John

By the way, I've been invited to come of to CA to help with development of NX7 so perhaps I'll get a chance to put a face to the name.

Best regards

Simon

 

[JohnRBaker]

Simon,

What project are you going to be helping with?

I assume from the 'name' that you work at J.C. Bamford. I visited the Rocester facility several years ago, but I don't get to the UK as often as I used to (now spend most of my international time in Asia). Anyway, let me know when you'll be here and we'll go to lunch or something.

John R. Baker, P.E.
Product 'Evangelist'
NX Design
Siemens PLM Software Inc.
Cypress, CA

http://www.siemens.com/plm

http://www.plmworld.org/museum/

 

[SiW979]

Hi John

Yes I do work for J.C.Bamford here at the world headquaters in sunny Staffordshire. The details are vague, but basically since we are your second biggest UK customer in the UK (after Rolls Royce)I have forged some very good relationships with some of your UK collegues and they said something about how you get some customers to come and have input towards the next versions of software and that they would see that I got an invite. Whether my boss will sanction it is another matter, but I'll try my best to build a solid case for a reason to come over to CA, I could try and combine it with a visit to our Savanah plant as we have a couple of CAD users there.

Best regards

Simon

 

[jackley]

I go case by case on adding radii and chamfers to the sketches. Depends on the size of the part.

But I agree 100%, the KISS theory (keep it simple stupid).

I love sketch. I love how I can tie sketches together to maintain wall thickness, etc.

Anyone take the sketch course offered from Siemens? I feel I'm pretty advanced with sketch, but wondering if this course would offer anything more for me.

Justin Ackley
Designer

 

[SiW979]

I went on advanced sketch course and to be honest I was teaching the instructor things he didn'd know so I was pleased that my sketching was obviously up to the highest standard. Sketching is like Marmite you either love it or hate it, but I always spend extra time on the courses I run making sure everyone goes back to their business unit fully conversant and enthused to drive the message home to the 'hate it' team.

Successful use of sketcher and all the associated expressions is key to successful use of CAD in my eyes, you have to have 2 trains of thought going, one about what it is your designing and the other how your going to design/build it in CAD. You wouldn't believe the amount of CAD models I see where people have spent ages fully constraining a sketch only to find that as soon as you modify 1 dimesion the whole thing falls over or they have to strip the model right back to basics in order to effect a change.
That's why I love my job now because I'm no longer tied up with all the day to red tape of front line product design, I just explore, play with, deal with and teach NX all day every day.

Best regards

Simon

 

[jaguar306]

Thanks to all of you for your responses. I can take a little bit from each of you and realize that our train of thought is similar. I also believe that using form features where possible is the thing to do as they are very powerful. i will make sure to pass this thread on to my peers as well as this forum as I see a lot of value in being able to open discussions on these types of day to day issues.

Thanks Again

Jaguar306

 

[hudson888]

I don't seem to have mentioned it in this post though I may have elsewhere, and I noticed there was mention made about whether to add radii into sketches. This was something that Justin mentioned that you do on a case by case basis and I tend to agree.

There is an argument that says that you can always add the radii or chamfers later, and that may as practical an approach as any. I'm certainly not one to be hung up on the minimum feature count as a way of keeping score as to the value of my models.

One situation that I find does make a good case for occasionally including more detail in a sketch profile is when that sketch lies on the split line of a casting or molding and will be drafted from that plane. This because especially where the amount of draft differs on either side of the split line the profile is better controlled where the blends are added before the draft. People who have done this before will see my reasoning I'm sure.

In all other cases we greatly prefer the four sided surface in the corner of a blend where the smaller radius goes over the larger convex or the opposite to avoid a similar three sided surface in the concave situation. We therefore try to maintain a certain amount of freedom to order our results favoring the the preferred outcome.

Best Regards

Hudson