Design Log
The purpose of this document is to log the progress, track issues, contemplate the design for future members who work on this.
Current State of Affairs
- The 3D printed prototype has been made and tested. tldr; adaptive grip works, now going to figure out the allen key thing and move into a high fidelity design (metal)
- There is a preliminary higher fidelity cad model, but it needs some work before manufacturing
- Napkin math has been performed to get a rough estimate of the required force (120 N squeezing force)
The goal is to finish the CAD of the metal pieces and sort out the pin joints
Motor to be used will be a Polulu motor, or a similar 12V gear-motor for testing purposes.
Task List
- Figure out the pin joints
try to eliminate any folded sheet metal parts- look into fattening the front plate (most adaptive grippers in this category have a much larger distance between the front pins and the base plate)
Investigate if the tip lengths should be shortened (tip lengths may interfere with each other when adaptively gripping smaller itemsImprove the tip design - Milled- Consolidate mounting options for rubber pads
- Figure out the Allen key attachment
- More precise calculations for the output speed/torque of the jaws as a function of input speed/torque and relative position
Adaptive spring placement
Current State of Affairs
- The CAD has been created and reviewed by two mentors
- The slides prepared before the meeting were hugely advantageous.
- Helped organize thoughts before the meeting
- Allowed the meeting to efficiently flow through all the topics without stalling
- Structured meeting
- one downside is it takes time to make good slides. The time is worth it
- The design is sound except for a few minor changes to further optimize the design primarily for manufacture
- There will be no Allen key attachment - our mentor Cory suggested the tooling bit is the simplest solution and should be used.
- Effort is now focused on fixing the cad CAD and preparing for manufacture
- Effort is also focused on creating a complete BOM
Constrains & Criteria
Constrains and criteria were outlined in the slides to help define the task, but the constrains and criteria were backwards. for clarity these are defined here. Constrains must be met for a design to be considered. Criteria are used to compare different designs. A design matrix can be used in the future to help compare different concepts and designs against each other.
Below is a list of everything mentioned in both design reviews. Some of the items on the list will not be performed they are crossed out.
Task List
L-Bracket for motor mount, milled component for motor mount
This is final design type of issue, for now 3d printed is fine, milled is better because it can hold tighter tolerancesTake out chamfers of the front and back milled components
Cost little to nothing, some chamfers are essentialMake the nut carriage simpler to manufacture
costs 100 on fictiv, similar to the other brackets- Ream holes on laser cut pieces
- Add anti-seize to BOM for pin joints
- Add 3/16 id shims to bom
- specify spot welds instead of weld fillet (help retail temper on the sheet metal parts)
Misumi shaftmight not be able to get the steps required, or will require a post op -see if this can be machined in house first
a tight tolerance step at the minor diameter of the lead screw is required to interface with the f-loc gear, might bot beable to post op machine that out of an existing screw- Adhesive backed rubber
- Try different rubbers - sorbethane
- add loctite 660 to bom - mcmaster
- redo the spring location - interface is fine but location is wack
- spring calculations and iteration
- make drawings for all of the components
Rubber bands will be used. The reasoning for choosing rubber bands for the compliant portion is that they can extend way more than springs. They can also stack much easier than springs, in case the force of a single rubber band is in sufficient. another reason to choosing rubber bands over springs is that the rubber bands worked very well in the preliminary prototype. Again, this this current design is still a prototype, if the rubber bands are insufficient, springs can be used in the final design - want to dive into the prototype as soon as possible. A few possible disadvantages to using rubber bands compared to springs is their lower chemical resistance to oils, may have to end up replacing these springs more often than springs. - good corrosion resistance though. another disadvantage may be that rubber bands to not follow hooke's law. That being said, rubber band are still easier and cheaper to work with than springs.
Demtool package sent out today with all of the sheet metal parts and the carriage nut
Things to be done still: hardware bom this includes stuff from mcmaster, KHK and one part from misumi. The front and end bracket also need to be machined and the lead screw needs to be modified once it arrives.
Alot has happened since the last log, parts arrived, the mechanism was assembled, and the mechanism went through some preliminary testing.
Manufacturing
Due to timelines, we had to remove the anodization from the order. The parts received were impeccable. One issue encountered (which was my fault - not the manufacturer) was the washers did not fit the clevis pins out of the packaging, and each of the spacers had to be filled in order to fit. This is because these spacers were the only parts not specified to be reamed, and when cut, the parts had a massive bur. The ID specified for this part was 5 thou larger, but this was not enough clearance.
The front and end lead-screw mount plates that were manufactured in house were incorrectly machined. The screw holes are 1mm too close and the width is 1mm to thin. This did not cause any issues during assembly thankfully but in the next revision this should be corrected. It also shows that the parts can be further optimized for weight
Assembly
Assembly went quite smoothly, although there were a few bumps which are now described. First off, the loctite 660 and loctite 380 both worked phenomenally to bond the desired components together, I accidentally bonded the sorbothane to the weldment whilst it was assembled, causing it to seize... thankfully i was able to take it apart and scrape out the adhesive without causing permanent damage
One thing that was frustrating, was that all of the clevis pins have about 1mm of play in the axial direction. These are specced correctly on McMaster, but the parts have a 1mm extra usable length compared to the drawing given in the catalog. This does not give us any issues, although the mechanism appears to be loose. the OD of the pin fits perfectly with the ID of the laser cut parts, but the length has play. This can easily be corrected for with some 1mm spacers, but should be noted when using clevis pins in future designs.
The spring pins were also a massive pain in the ass to insert, maybe they are always like this, or maybe the hole was too small, but it was not an enjoyable experience. The holes given in the drawing for the part matched the specifications on the McMaster catalog, so it could be that the holes were incorrectly machined but this was not validated.
When the mechanism was initially assembled, there was a lot of binding in the lead-screw drive portion. What I suspected, was that the front and back support plates screwed down before they were correctly aligned. The solution to this was to loosen all 8 fasteners, align the plates such that there was no binding, then slowly fasten all the fasteners ensuring there was no binding throughout the process. This worked like a charm
Testing
Overall the mechanism works great... with the adaptive links locked. the main selling point of this mechanism was found out to not work, so for the next iteration, the adaptive capabilities will be removed from the mechanism
First thing that was noticed was the snails pace of the end effector. This is actually a good sign as the mechanism was calculated to move slowly. From the testing we can actually see how slow is slow. For the next iteration, we will design for a faster mechanism by changing the gear reduction, as the mechanism is over design with respect to torque.
Nothing broke during my limited testing - gears all survived, links all survived rubber survived so this is all good news.
The biggest issue → the adaptive links. Note that the rubber bands/springs used to control the adaptive features, were the only piece of the puzzle not calculated for... The plan was to just test different spring elements during the testing phase, but no rubber band seems to provide sufficient force. The temporary fix for this was to solder a loop of wire tightly around the adaptive links where the rubber bands would have gone. This worked surprisingly well. A replacement part is in the process of being manufactured.
The main star of this entire assembly was really the sorbothane pads used for grip. The pads are so sticky and squishy that it can pick up literally anything. It also has extremely good performance for moment loads (picking up the end of a hammer, not the head) the coefficient of friction used in calculation was 0.6, but based on feel it was easily more than 1. This means that the torque exerted is most likely more than required. More testing will have to be done to confirm this. This info can be used help reduce the weight as well as increase the speed of the mechanism. More testing should be done.
Test footage of the end effector, the adaptive links were locked, and the gripper was able to pick up anything and everything.