Virtual Test Build Results

[jedroach]

Hello,

I completed a virtual test build of the MicroHouse in OnShape. See the results here:

https://cad.onshape.com/documents/05ee4ed8debb791b540e74de/w/b34eb3abf186ae8e86ecf274/e/0e3aef5ccf0cf29e489f504f

This build was done using Microhouse_WRENchassis_v0.5.dxf as the source data. The assembly was done using the manual firstly and then referring to the SketchUp model to fill in the blanks. I only created unique parts in the part studios and then used patterning features within the assembly to create the rest of the parts. So, I did not do a strict accounting of part quantities, but I did find some quantity errors which I noted below.

Issues:
I could not open MicroHouse_othercomponents_v0.5.dxf with any of the half dozen programs I tried.
Step 0: What size should the rails be?
Step 2 and 3: F2 and F6 labels are swapped. ( I think F1 should be changed and F2 and F6 should be the same part.)
Step 4: SKI/B, C, and D are left out of the manual and SKI/B needs to be installed on the first frame you raise.
F3 = F5, R1 = R2, R3 = R5
Step 7: Panel 009: ROOF/4 should be labeled EXWALL/4
Step 9: The UNDER-X pieces need labels. There should be a FLOOR/4 for where UNDER/4 goes.
Step 10: I think CEIL/2 and CEIL/4 are the same as ROOF/5, but quantity is 2 short.
Add notes for interior and exterior trim pieces.
Step 11b: END/2/F1 is missing from DXF. It is the same as END/1/F1, but quantity is 4 short.
END/2/F2 is missing label
END/4/F5 is missing label
END/3/F4 is missing label, (same as END/4/F5)
END/3/S1 is incorrectly labeled as END/1/S1
There is a little bit of interference between END/4/F3 and END/4/F4. I confirmed that it exists in the SketchUp model as well.
END/5/S2 and END/5/S3 are identical. END/5/F2 and END/5/F3 should be identical, but they aren't.
Step 19: INT-A/2/F5 is missing label
Panel 032 should have INT-B/6/P2 label
INT-B/3/F1 is missing and INT-B/3/F2 is missing label
INT-B/4/F2 and INT-B/4/P2 are missing
INT-B/5/F2 INT-B/5/P2 are missing. The castellation is asymmetric by about 2.9mm which makes assembly difficult and means that INT-B/5/F1 isn't the same as INT-B/3/F1
INT-B/6/F3 is missing and there are 2 of INT-B/6/F2
Step 20: INT-A/PK should be quantity of 4. Two parts are stacked together. This can be seen on the pdf if you zoom in really close, but isn't made clear otherwise.
SKI/C and SKI/D: The spacing between the notches is off by about 10.5mm.

Comments:
Please let me know if I should post these to another section of the git.

I noted a few instances where parts with different labels are actually identical. Whether or not this is appropriate is probably a matter of opinion. My inclination would be to make them the same part number so that there is more flexibility when nesting parts and is more in line with the idea of modularity. A counter argument is that it might make assembly more intuitive if parts in a sub-assembly all have the same prefix.

Because of the way the PEGs are installed, the length of the connectors determines the center to center spacing of the frames. This results in a spacing of 1199.5mm I suspect this should be 1200mm. Additionally, The length of "CONNECTOR" disagrees with the upper connectors "C-03"

Perhaps we can identify the locations to pre-drill holes for screws. This might require a tool change on the router, but for some routers, this is trivial. If the holes are on a separate layer, then it should be easy to not drill them on the CNC if that is the preference. This would at least give some guidance as to the proper number of screws to use per panel.

I would like to see some support around the perimeter of all the skin panels so that the joint doesn't separate under load. This was done very nicely where FLOOR/1 meets FLOOR/2, but this detail was left out of the walls, ceiling, roof and corners. Because of the globally applied tolerance, I'm doubtful that the dovetails will transfer any meaningful load before some other part of the structure yields. I think laminating reinforcers, like is done at the joints of the super box, is a more robust approach for shear and tensile loads.

Best Regards,
-jed

[claytonprest]

Hi Jed,

This is absolutely the right place to post these issues. This is really excellent feedback and incredibly helpful that you've picked up on these errors and inconsistencies. We are building dedicated Wikihouse design software (to be freely available on the web) which should reduce human error like this in the near future. In the meantime thought, Github repos and CAD remain the platform - so we really value all your assistance here.

Enjoying looking around your Onshape model too - fantastic work! I've posted responses to your comments below, and now updated the cutting files and Assembly manual in the repo as a result.

I could not open MicroHouse_othercomponents_v0.5.dxf with any of the half dozen programs I tried.

That is very odd. I haven't had any issues with this, and haven't had anyone report this before. It should open successfully in the latest version of draftsight, Autocad or Rhino or any CAD software that can read DXF.

Step 0: What size should the rails be?

We nominally suggest 75mm wide for this model. You'll need to work with an structural engineer to determine the correct depth. I've added this note to the assembly guide.

Step 2 and 3: F2 and F6 labels are swapped. ( I think F1 should be changed and F2 and F6 should be the same part.)

I see your notes below but we prefer to keep them as unique parts. It might not make sense here on a perfectly symmetrical design like this, but the cyclical numbering is better logic for the Wikihouse software we're developing as well as for assembly by humans. Why do you thing F1 should be changed?

Step 4: SKI/B, C, and D are left out of the manual and SKI/B needs to be installed on the first frame you raise.

Ski-B does need to be installed on the first frame and last frames only. I've added a note in Assembly Manual for SKI-C and D installation.

F3 = F5, R1 = R2, R3 = R5

After checking, I believe these are correct and match the assembly manual. What's your reasoning for this? Is it possible you've been assembling back to front?

Step 7: Panel 009: ROOF/4 should be labeled EXWALL/4

Well spotted. Thanks I have corrected this in the repo file commit.

Step 9: The UNDER-X pieces need labels. There should be a FLOOR/4 for where UNDER/4 goes.

You're right there should be a variant FLOOR/4 to accommodate the Under/4 component. It seems there is a sheet missing as only 8x FLOOR/1 parts are included. As a result I have now added Sheet 035 to the cutting set.

Step 10: I think CEIL/2 and CEIL/4 are the same as ROOF/5, but quantity is 2 short.

Yes, this is clearly an error on our part. I have added two more components (on new sheet 035) and corrected cutting set to CEIL/4 for all these. Assembly manual has been updated too.

Add notes for interior and exterior trim pieces.

Fair point. We have now been doing this for subsequent projects, showing an additional step to install the trim.

Step 11b: END/2/F1 is missing from DXF. It is the same as END/1/F1, but quantity is 4 short.

Yes END/2/F1 is missing, I have now added it on Sheet 026. Labels were also swapped on END/1/F parts and did not match the Assembly manual. The quantity is 2 short as you say. Fixed this shortage by copying part over to Sheet 023 as well.

END/2/F2 is missing label

Label now added.

END/4/F5 is missing label. END/3/F4 is missing label, (same as END/4/F5)

These parts may be too small to label / engrave. But I have added it anyway. They're actually not essentially required.

END/3/S1 is incorrectly labeled as END/1/S1

Corrected in the cutting set.

There is a little bit of interference between END/4/F3 and END/4/F4. I confirmed that it exists in the SketchUp model as well.

Adjusted the length of END/4/F4 to fix this.

END/5/S2 and END/5/S3 are identical. END/5/F2 and END/5/F3 should be identical, but they aren't.

Thanks, this has been fixed.

Step 19: INT-A/2/F5 is missing label

Just too small to realistically engrave labels on that one I'm afraid.

Panel 032 should have INT-B/6/P2 label

Done.

INT-B/3/F1 is missing and INT-B/3/F2 is missing label

Added on Sheet 034.

INT-B/4/F2 and INT-B/4/P2 are missing

I think the cutting set incorrectly assumes INT-B/4/P2 and INT-B/4/P1 are identical, which they aren't. There are two copies of INT-B/4/F3 which is identical

INT-B/5/F2 INT-B/5/P2 are missing. The castellation is asymmetric by about 2.9mm which makes assembly difficult and means that INT-B/5/F1 isn't the same as INT-B/3/F1

Issue has been fixed now.

INT-B/6/F3 is missing and there are 2 of INT-B/6/F2

I see there are 2 of INT-B/6/F2, when they should be different. This has been corrected (see INT-B/6/F4). I have also added the missing INT-B/6/F3.

Step 20: INT-A/PK should be quantity of 4. Two parts are stacked together. This can be seen on the pdf if you zoom in really close, but isn't made clear otherwise.

Corrected in cutting files.

SKI/C and SKI/D: The spacing between the notches is off by about 10.5mm.

Very odd. Thanks for picking up on this. It's been fixed now

You make a valid observation about the different labels which are actually identical. For ease of assembly and (at a later point) machine readability, it's actually logical to label them all uniquely in a numerical sequence. Because future models will be parametric, it's unlikely all designs will be perfectly symmetrical, it needs a universal approach so it avoids using one system for one design and a different system for another - if that makes sense?

The centre-to-centre frame spacing of 1199.5mm you may have experienced in your virtual model is a result of pre-applied CNC manufacturing tolerance of 0.5mm inset on all components. In the reality of construction this doesn't really happen and this little bit of tolerance is very helpful during assembly.

It's a good suggestion to identify the locations to pre-drill holes for screws, and we actually are already doing this now on all current projects. They are on their own layer, and regularly set out at the recommend 150mm spacing, so can either be pre-drilled or just marked by 'dinks'.

Another great observational point regarding additional support around the perimeter of all the skin panels. Other engineers have made the same comment and on recent projects we are adding additional connectors in the walls and roof at the points where the two sheathing panels meet. I'll release some two storey model files shortly on this Github where you can see this.

Thanks for all your help here!

Cheers,
Clayton

[jedroach]

Hello Clayton

Thanks for the feedback.

The problem with the MicroHouse_othercomponents_v0.5.dxf appears to be a problem with how github is handling that dxf. I originally downloaded that file individually using the "raw" button and that resulted in a file that was 23MB and wouldn't open. When I downloaded the entire repo as a zip it worked fine. Now that I can access that file, I'll work on giving it the same treatment as the chassis components.

Step 2 and 3: F2 and F6 labels are swapped. ( I think F1 should be changed and F2 and F6 should be the same part.)
I see your notes below but we prefer to keep them as unique parts. It might not make sense here on a perfectly symmetrical design like this, but the cyclical numbering is better logic for the Wikihouse software we're developing as well as for assembly by humans. Why do you thing F1 should be changed?

This is just more of my theory about minimizing unique parts. If F1 had the same gender connector on both ends, then F2 and F6 would be identical parts. However, I accept your reasoning for unique parts, so you can ignore.

I'm looking forward to seeing more models.

BTW, I attempted to join the slack by going through the typeform, but nothing came of it. Any suggestions?

Thanks,
-jed