Rectangular HSS with Block Properties Table

[j_rosler_vcat]

I've been trying to make an HSS block that is adjustable solely via block properties table. However I'm struggling with the radii. The problem is there are two radii per corner (inside radius and outside radius), and two parameters that can alter them (length of flat wall and wall thickness). I can't get both parameters to work together, only one or the other works no matter how I setup the block (using chained parameters, constraints, an arc dimension instead of polyline, etc.)

Is what I'm trying to do possible within AutoCAD's limitations? I've seen other similar blocks that use a visibility to change the wall thickness, or ones that simply don't modify the length of the flat wall at all, but I want to be able to control the entire block with just the block properties table if possible.

I'm far from an expert with dynamic blocks, if someone could give me hand or just let me know if I'm wasting my time and this cannot be done, it would be appreciated.

I've attached my closest attempts so far (they are both very similar). 

[Libbya]

Yup, definitely possible.  Complicated but at the same time easy to accomplish.

The issues are caused by scaling in a way that cannot match your scale base locations, or from incorrect stretch windows/selection sets. 

Be careful with chaining.  Don't use chained parameters unless necessary.  I don't see a situation where chaining would be required for this block.  

Similarly, adding actions to the selection sets of other actions can be problematic.  Don't do so unless necessary.  For example, you have a lot of stretch actions within the selection sets of other stretch actions.  Only the independent scale actions need to be placed within the selection sets of specific stretch actions in order to move the base position appropriately.  While the stretch actions added to other stretch actions aren't causing issues in this specific block, it also isn't helpful and is bad practice.  If you get into the habit of whimsically adding actions to the selection sets of other actions then there will come a time when that causes issues (e.g. adding stretch actions to the selection sets of rotate, polar stretch, or flip actions, etc...).

To fix the first version, I added 4 stretch actions to the R2 parameter, changed its base position, and changed the base position of each of its scale actions.  I adjusted the stretch windows and selection sets of the other stretch actions in the block.  

I didn't bother with diagnosing/fixing the second version.  Attached is the first one updated.  

[j.palmeL29YX · Accepted solution]

I would build a block as shown in the attached dwg, using Parametric Constraints and Dimensions.

The Block Properties Table only needs 4 columns: _Depth, _Width, _t and _r1. 

These parameters define the geometry complete and unambiguous. 

[j_rosler_vcat]

This is perfect! I need to get better at using constraints because I didn't realize this could be done so easily. Thanks for the help.

[j_rosler_vcat]

Thanks for the advice and taking the time to correct my block. The only problem is that the inner radius (r2) still doesn't follow when the outer radius (r1) changes, which was the main issue I was struggling with. I'm assuming I can fix that by calculating what r2 should be for any given HSS size and adding that value to the block properties table. However, the other person who replied changed my block design to one using constraints, which seems to work exactly as I'd hoped, so I'll just use that one instead. 

[Libbya]

Sounds good.  

[Libbya]

There are a variety of reasons that I would not use constraints for this block. 

Here's your block back for educational purposes and in case anyone else in the future would like to know how do it without constraints.  As you said, one option is just to fill in the block properties table with the appropriate values for R1 and R2 separately.  If your original post had adequately described what behavior you were seeking (linked R1 and R2 VS. unlinked) I would have linked them, but because you didn't, I chose the option that gave more freedom rather than less.  Linking R1 and R2 is very easy.  Just make R2 chained, turn off its grip (the R2 grip being ON was a clear indication that you wanted them independent and NOT linked!), and stretch R2 appropriately with R1 and with the thickness.

[j_rosler_vcat]

My bad for not explaining properly. I guess I wasn't even sure how I expected R1 and R2 to behave myself, whether they needed to be linked or not, or if it was even necessary to have an R2 at all... At one point during the troubleshooting process I even had an R3 parameter which was identical to R1 but with a different base point. Needless to say, I was very lost and making things overly complicated for nothing.

Looking at your updated block, I'm starting to understand some of the mistakes I was making. I plan on making similar blocks for all the various steel sections, so I appreciate the assistance.