How do you do rapid iteration good?
4 Likes
For reference, as of right now 3256 lathes all of its standoffs.
Between facing and drilling both ends, it takes us around 40-50 minutes each, per standoff, for a team who designs mostly with plates, standoffs, gussets, and tubes.
This is not very fast.
Just one example of many, but how do yâall not do that?
Why are you not 3d printing more standoffs? Or buying them as COTs parts?
But if those arenât viable options for your construction style or budget (Itâs hard to get a feel for what you mean when the last picture TBA has of your bot is 2017) consider what is taking so long for each part. What are the slow parts of the process? What makes them slow?
Process optimization is all about understanding your process, identifying what can be improved or removed, and then evaluating it.
3 Likes
We tend to thread our standoffs, so 3D printing is out. We have considered buying VexPro 3/8ths and 1/2 tube axle, and threading that, but since the IDs on both are so large we wouldnât be able to continue to use 10-32 screws, which allows us to standardize our designs to one screw size quite nicely. We have considered using threaded inserts, but became worried about the potential loss in fastening strength, and finding tube stock of the right size with an id that can be easily drilled out hasnât turned anything up. If we canât find a better solution, weâll likely end up using the tube axles, and just diversify our screw library.
I disagree. Iâve used threads in 3D printed plastic before, both threaded in AND by placing inserts (typically brass press-fit screw-install). 10-32 happens to be one of my favorite threads. I wonât say itâll be guaranteed to hold in FRC applications, but they arenât as weak as you might think. You may want to try doing it a few times and see how well it holds up.
Question: Have you heard of McMaster Carr?
3 Likes
We are planning on trying brass heat-set inserts on an Onyx print next time we get a chance, although until we have more experience with them weâll likely be a bit wary of using them in high load applications.
I honestly canât believe I didnât check McMaster. I legitimately went through like 10 tube stock suppliers, but didnât check McMaster. Either Iâm an idiot or I havenât slept in 28 hours. Or both. On second thought, definitely both. Iâll still be on the lookout for the right tube axle stock, just because being able to quickly and easily make custom length standoffs sounds pretty neat, but this is still a great find. Thanks!
Is the iteration youâre speaking of more like slightly modifying a subsystem or changing a subsystem out for a brand new one? Iâd say unless your manufacturing process is really fast and efficient (does not sound like it), avoid changing entire subsystems.
If youâre modifying an existing subsystem, try and reuse as many already machined parts as possible if your manufacturing process is a limiting factor (also use more COTS components).
Outside of that, consider the idea that iteration just may not be on the table for some things given your manufacturing circumstances and plan to not be able to do much. The ability to effectively iterate is its own resource, and not every team has that resource at the moment.
2 Likes
I donât disagree on being wary. I tend to use the screw-to-expand type myself, but even so it does sometimes slip.
Depending on what youâre doing, you may want to dig further into the standoff section on McMasterâyou might be able to use a male-female and use washers to adjust the length quickly. Or you might grab unthreaded and just through-bolt. Or you could get extra-long standoffs, part to length, and do a quick re-tap to finish the threads properly.
3 Likes
Just generally faster iteration throughout the seasonâs length. It took us until week three to get our drive base assembled in 2020, although not because we lack machining resources.
Our machining process is decently efficient. We have a portaban, a bandsaw, a table saw, two drill presses (one for deburring, the other for decoration) a Bridgeport mill, a Hardinge lathe, a Tongtai VTX-7A CNC Mill, two laser cutters, two Markforge Mark Twos, a sheet metal shear, a sheet metal brake, and a sheet metal roller.
We were going to get a Laguna Swift Vacuum router at the start of last semester, but that got delayed due to COVID. Weâll likely get that at the beginning of the 2021-2022 school year.
Edit: This is thread is mostly just me collecting information in preparation for the 2022 season, since Iâve accepted the fact that this season will likely see little to no actual robot design and iteration, hence the included tidbit about the CNC router, which should allow is to do things like manufacture a metal bellypan in house, and make a ton of gussets really fast, both of which I am definitely looking forward to.
Second edit: Also two U-Prints, although we try to do all of our final printed parts in Onyx, so theyâre relegated to prototyping for the most part. Print out ABS pretty quickly.
Not having to use a soldering iron for installation does sound quite nice, not gonna lie.
If youâre turning all your standoffs, why not mine your own aluminum?
In all seriousness, I strongly recommend thinking about how you design:
-
COTS spacers and standoffs are extraordinarily convenient. If you donât already, use a configurable part for each so that whoever does CAD doesnât have an excuse to not use them. If you use SolidWorks, I can send you my configurable standoffs and spacers.
-
Cut your total part count as much as possible. If you can do it with 1 part, why do it with 2? A big problem I see in a lot of ânoviceâ FRC designs are hacky mounting schemes requiring half a dozen custom parts, when a simpler method could use 1 part, even if that part has more features.
-
I noticed you mentioned having some Markforged printers - start thinking about how you can leverage a 3D printer to simplify things you would waste time making with a CNC mill, and vice versa. In essence, take advantage of your fabrication capabilities to cut those parts.
-
When you can, cut features. Pocketing a piece of plate is generally the most time intensive step in milling it, and usually reduces its mass by about 50% - instead, save yourself the time and just swap out your pocketed 1/4" plates for normal 1/8" plate.
As a final side note, it should not take 40 minutes to turn a standoff. You might want to look into how your students are trained on the machines, and see if there are underlying process issues in fabrication overall.
2 Likes
Could also consider something like Plastite Screws if it wonât be removed multiple times.
Also, if someone has a non McMaster source for the screw to expand inserts that would be great. McMaster is great for a lot of things but âI need this sometime in the next 1-2 months and am completely ok waitingâ isnât one of them. (Nor is it too great at âthis is my money and I like to keep as much of it for buying stupid joke domain names as possibleâ)
1 Like
Misumi has some, as long as you donât mind metric.
MSC and Grainger have some. Youâre on your own for most price comparisons.
For the 10-32 size, flanged, reverse (AKA try to pull it through the hole with the screw)âŚ
Grainger, 19.35 for 25
MSC, 12.65 for 50
McM, 8.50 for 25
I know MSC and I think Grainger will negotiate discounts, how big I donât know. McMaster wonât. MSC wins for this specific part. But if I went to the same-side version, McMaster wins, flanged or not (and Grainger apparently doesnât have it). YMMV.
As a general rule of thumb, Iâll generally check McMaster, Misumi, MSC, Grainger, in that order, for parts. If someoneâs on a âsave moneyâ kick Iâll price out all the stuff I want to get from at least 3 of them.
3 Likes
use tube nuts for standoffs
god tier
3 Likes
It there a reason youâre using standoffs instead of spacers and through-bolts? Itâs really easy to cut spacers on a lathe from round tube stock; just measure the length and cut it. If you have an organized list of lengths and diameters, you can bang out a whole robotâs worth in a few hours.
Or, like we often do, you can 3D print them all overnight and have them ready to go in the morning. And you donât have to worry about threads stripping or adding inserts to the print.
Plus, if I remember correctly from my mechanical design class, spacers are stronger than standoffs because the bolt preload goes all the way through the spacer rather than just clamping the plates to either side of it.
3 Likes
How are you normally doing these? We can normally hammer out one every couple minutes if itâs just a double tapped shaft.
A word of caution on the mcmaster ones- while they work, the threads are super tight
1 Like
Kind of surprised that I havenât seen anybody mention it yet but if standoffs are a major issue for you Iâd recommend using 1/2in thunderhex as your standoff material. You can buy it in long lengths and cut it to size on a band saw, then either tap both ends to make a standoff or use it as a spacer with a #10 bolt. We make almost all of our standoffs/spacers out of thunderhex and we can make spacers in a few minutes and standoffs in about ten. Another similar option would be to use AndyMark Churro stock, it has a larger ID so its really easy to tap for a 1/4in bolt. Churro also comes in a variety of colors that can add some bling to your robot.
2 Likes
If you really want to make standoffs for #10 instead of spacers 3/8" Thunderhex would work for your purposes.
2 Likes
So this in and of itself probably is the key - rapid iteration implies youâre using processes and techniques that prioritize speed over (almost) anything else. But, sounds like youâre using âproduction - levelâ techniques. The overall mindset shift is where Iâd start.
Rapid Iteration = do things quickly to learn. Once youâve learned, go back and do it ârightâ.
Maybe Iâm over simplifying⌠but yea, 3d printed parts are a great way to do multiple things at once (ie, design it and print it, and go do something else useful while the print is going). Itâs not super precise, but even a length of PVC sliced off with a hacksaw works fine.
Speaking of PVC, Spectrum has a set of 3dprinted parts to hold lengths of PVC together for quick prototypes. Obviously, this wonât work great on a final robot, but thatâs not the goal. The purpose is learn what sorts of geometries work well and which done, and get close-to-final dimensions you can use to do it once, well, on a lathe or mill.
Other general suggestions:
Treat prototyping like a scientific experiment. Observe, identify issues, form a theory, design an experiment, perform the experiment, observe the results, change one variable at a time, repeatâŚ
Engineering is just the real-time application of the scientific method.
Get second opinions. Have at least one person responsible for being the expert on each thing youâre prototyping, but have âconsultantâ experts drop by to hear about what was learned and provide comments and ask good questions. A fresh viewpoint is often the best way to make breakthroughs.
Keep a detailed notebook, ideally with lots of photos. Take pictures of measuring tapes near key items.
Push hard early on, and know when to quit and go to production. The goal is never âperfectionâ - itâs always âworks good enough to score points on the fieldâ. Donât let your prototyping squeeze the production schedule to where you canât spend the time fabricating it right. This is why having someone be âproject managerâ can be a useful thing.
As part of the prototype, note which areas need frequent adjustment. Build these into your final design - setscrews, sliding-adjustable shooter hood release points, screws and slots to adjust mechanism travel hard-stops, all just examples. But, try to minimize these, as each one is an additional failure point that could come loose at the wrong time.
If your mechanism is requiring anything more complex than a simple thriftythrottle to control properly, be sure you get your software and electrical teams involved early on.
4 Likes
Other bits:
Take bits and pieces of advice here and there. Accumulate a lot of little tricks. Make sure youâre passing down a culture of them, as well.
A lot of people love the âbuilt not boughtâ mindset, but thereâs a certain point at which you need to âbuy the best, invent the restâ.
One key to rapid iteration is admittedly, to already have everything on-hand, and organized. Fully stock your fasteners. Fully stock your spacers. Fully stock your gussets. Think you might need it? Buy it ahead of time. Any time you spend making parts or waiting on them to arrive is time youâre spending not assembling, testing, and tuning.
Will a 1" spacer work where you really wanted to use 7/8"? Well, 1" it is.
Shift your mindset. Think about how to parallel path and pipeline. Can you have someone build a frame rail while someone else makes a spacer while a 3D printer prints a pulley? Do it in parallel, not series.
1 Like