Case study – screw up Vol. 1

So I just figured out a new series that will be called Case study – screw up Vol. X. 🙂 this series will be dedicated to machine design fail.

In order for this series to be constructive I will share examples where I have screwed up something obvious or not and have figured out the solution after the fact. So here is the most recent one. Enjoy.

I had to make a servomotor holder that needed to align the motor with a roller that this motor is rotating and provide a housing for the flexible coupling.

The part was made out of solid block of Aluminum.

Some of the design details:

1. A bore that was in H8 tolerance to center the servomotor
2. 4 x M8 threaded holes for attaching the bracket to the plate that housed a bearing for the roller
3. A through hole for the roller shaft
4. “Window” for tightening the elastic coupling
5. Indentation because of the sliding mechanism that was on the plate holding the roller bearing ( to be sure that it will not interfere )

So when it came to the assembly process I was happy with this part.

We assembled the housing, the roller, the coupling and the motor and started testing it. There was a rough sound that cyclically repeated when we ran the motor. After some inspection we figured the motor was not aligned with the roller. It was not a small misalignment, it was around 0,4mm of run-out. That is enough to seriously shorten the life span of the particular elastic coupling… The problem? I have designed the alignment of the servomotor with the housing, by tolerating the bore and its position relative to the base that was mated to the plate. I did not, however, designed the alignment of the housing to the plate that housed the bearing for the roller. So yea… It had to be reworked.

I added two dowel pin holes to the housing and added a pair of dowel pin holes on the mounting plate and dimension everything to align the roller shaft to he servomotor perfectly.

Then we mounted everything back together…. and there was a similar noise once again… I started measuring and testing everything. Every part was perfectly within its tolerance. I changed the roller bearings on the roller. Then we tried to measure the run-out of the roller shaft in respect to the bore for centering the servomotor (1). And the run-out was 0,2-0,3 mm. How was this possible? When looking at the documentation the run-out was supposed to be 0.03mm max, not 10x as much!!!

Accidentally we noticed that when we started loosening the 4 bolts that mate the housing to the bearing plate the noise stopped almost altogether. What? Why? …..

The screw up. The part was deforming during the assembly process. The indentation on the part (No.5) was the issue.

Four M8 bolts that were mating the housing to the bearing plate were deforming the part as shown in the picture, Because of this deformation in the indentation section the whole part deformed. This produced an oval bore for centering of the servomotor, and a few other slight variations. This was the reason for that runout.

Aluminum that is used as a material for this part also played a role. If steel would have been used the deformation would not have been so significant. The choice of aluminum over steel was bad for another reason. During the problem-finding process we disassembled and assembled the system numerous times. This caused the threads to strip. So again the steel would have been a better choice.

The final nail in the coffin was the design i.e. the design for manufacturing. This part was expensive to machine and was difficult to keep all the tolerances because of the numerous clamping positions it required. So a completely different design was needed that had to incorporate everything learned from this one.

This was not so obvious screw up but I can be sure to remember a few things for the future:

1. Always provide a thread on the flat mating surface, never on the indentation like on my first part (5) – so not to cause deformation when tightening
2. If you know that a part will be assembled multiple times it is safer to make it out of steel – so not to strip the thread
3. Look at the whole assembly and try to visualize the part that had to be in tolerance and then consider every part that is influencing that tolerance – so not to miss something like I did
4. Design for manufacturing. If a part needs to align two parts and house a rotating component, isn’t it more obvious to design it so it is machined on a lathe?

If you are interested in the part that I have designed as a replacement for this mistake I have provided you with a FREE 3D pdf. Enter your email to gain access to the download link bellow.