TTB Swerve Discussion

Don’t want to clog up the main thread here: The Thrifty Bot - Product Releases & Updates but mods, feel free to move it if you think this product discussion should be merged in.

I did a quick price comparison for the WCP and TTB swerves. The WCP Falcon Swerve, with 2 NEOS, 2 Spark MAXs, and the CANcoder, costs ~$680. The Thrifty Swerve, with a NEO, Spark MAXs, Neo550, VP (plastic and metal ring gears), and CANcoder, costs ~$650.

For that $30 reduction, you’re looking at

  • No bearings on the encoder gear
  • No bearing on the main module rotation
  • Plastic forks
  • Plastic ring gear for rotation
  • Neo550 for rotation instead of a Falcon

You also have to use the REV ecosystem in some way unless you use a different 550 motor and a Talon SRX for the rotation (which actually lets you save $10 on the encoder, but I digress). I’m trying to see what I’m missing about the product that makes it a better deal than what I can get from WCP or SDS modules. Maybe I’d pick it for the form factor, but taking that at the cost of reliability is something I would much rather not do.

I’m also worried about the swerve the interaction between the lower module and the steel tube around the Rendevous Point. For those unaware, swerve drives that hit the barrier at speed found themselves breaking aluminum wheels. With those kinds of forces, I’m doubtful that plastic forks, especially when split in two, won’t be a weak point for the system. Something like a single piece fork with extra reinforcements and a dead axle would make me feel a lot safer about the module.

Has there been destructive testing performed on these modules to compare them to metal modules? Maybe the plastic lends some flexibility and shock absorption?

23 Likes

The TTB swerve ends up being about 20 dollars cheaper then the WCP swerve, but with (imo) a lot of compromises, some of which you’ve already listed. Others being a lot of increased failure points, (such as the VP, and the (imo) weak plastic ring gear compared to a “traditional” belt/gear module).

A lot of these compromises could probably have been overlooked if it was significantly cheaper then the competition, but at a very negligible cost savings (80 dollars in the grand scheme of a robot is basically nothing), I just can’t see it.

I am really excited that TTB is going to be selling individual swerve parts, but for the actual swerve itself I really can’t see any reason to use it over the WCP swerve. If I’m missing something, someone please inform me.

12 Likes

I am the last person that should weigh in on swerve. But opinion statements on a brand new part that few have seen in person and fewer still have run on a robot enough to understand are not something to encourage. It’s how you end up with a robot cart that is better than your drivetrain.

14 Likes

Are you trying to compare the KOP chassis being rejected by some high schoolers fifteen years ago to two experienced mentors pointing out flaws in the TTB design when other, proven alternatives to the product are available?
Using a WCP Swerve does not a Weak Robot Cart make.

25 Likes

The KOP chassis in JVN’s story was (colorfully) rejected as weak without much to back it up.

The Thrifty Swerve here–still unreleased to the public–has had key components called “weak” and you’ve alleged “flaws”. While there’s certainly more understanding here now than in that pit then, it feels incredibly premature to use those terms to describe TTB’s design choices.

If someone were to call it “unproven” as a COTS product, I wouldn’t have my hackles raised this way. But let’s not pull the shovels out in the first 24 hours.

8 Likes

I agree with this sentiment. I really wouldn’t care as much about supposed design flaws if the pricing were significantly reduced. But in this case, I fail to see how many of the changes could be better than their slightly more expensive counterparts. Which is really what it comes down to, if the price is roughly the same. Why test something new for the same price as the old reliable solution?

8 Likes

I’d love input from somebody on 2767 but the saddle of the TTB module that other have said may be prone to breaking is almost identical to the saddle that 2767 used in 2019 and presumably also in 2020. I haven’t seen any evidence of them having issues with the saddle breaking so I’m inclined to trust the strength of the TTB saddle. It’s also worth noting that your team may already have the motors and versaplanetary gearbox. So while the motors and gearbox should be factored into a direct cost comparison, the barrier for entry into swerve is lowered by around $100 from the SDS standard kit.

9 Likes

We’ve ran printed saddles with live axles for 2 years now. That’s 4 robots (2 practice, 2 comp). The practice bots get driven a lot, I mean A LOT with no failures caused by the saddles.

If we were to go back in time and do a 24 hr review on an SDS module, some things that I would call flaws are:

Too heavy.
The Neo on the azimuth is too big.
Not enough tooth contact on the azimuth belt.
I need to run REV motor package.
Single flange mount is weak.
Too many gears.
Only has an absolute encoder.
Another swerve with that big azimuth bearing.
1 wheel = traction problems.

Now zoom back to the present and make judgment calls on my critiques. Half of them are based on opinion or a desire and the other half proved themselves not to be an issue.

One of the advantages the of the TTB swerve is weight, possibly 4 lbs. (1lb. per corner) The price you pay for this weight advantage might be strength. You’ll always be able to say the all metal, SDS module is stronger. But how much strength do you need? We’re going on 5 years of running aluminum on plastic azimuth bushing and 3 years on a plastic saddle and we are happy with the outcome.

27 Likes

Print it at home and you’re saving $120 / module (minus the cost of Onyx, maybe $20 worth / module?). So let’s say $400 cheaper per robot. That’s a significant savings. If you don’t have a MF printer maybe you could probably try this out of NylonX on a Prusa (would love to hear if someone tries this). If you don’t have a Prusa it would literally pay for itself in 2 robots.

8 Likes

Testing with a Prusa is on our roadmap and the filament is purchased. Hopefully will have data by the time parts ship. Onyx works great.

6 Likes

Some fair points are being made so I wanted to address the intent behind Thrifty Swerve first. When we set out to develop this module, these were our goals:

  • Bring something different to the market that leverages the years of swerve knowledge and development from 2767. A version of this module has been used by their team and iterated on for several years. I want to thank their team for their partnership in this process, if it were any other team this module wouldn’t exist right now.

  • Continue to make FRC more accessible for teams by being the least expensive module on the market. The 3D print at home kit especially was geared towards this goal. Hopefully now cost is less of a barrier for getting teams to adopt swerve.

  • Provide a module that used existing parts commonly found on FRC robots to simplify the process of getting started with these modules - in this case those ended up being the versaplanetary gearbox and mag encoder.

  • Create a very efficient, lightweight module by leveraging Markforged printing where it makes sense and by not using a large brushless motor on the azimuth rotation. The module in the pictures on the product posting including motors weighs a total 4.1 pounds.

The pictures show the NEO 550 because that’s what I had on hand here at home. Teams can use other options such Banebots 550 motor paired with an SRX controller. I know my team personally has spare mag encoders, 550’s, & versaplanetary parts that we would likely draw on for something like this, especially during COVID times to build this drive base up.

In regards to testing - more will be released on this in the coming weeks but rest assured these have been put through the ringer. Be on the look out for increased documentation and video of these modules on the field and in game conditions.

I get that people have reservations on using 3D printed components on FRC robots and seeing field interactions would be helpful. Just as a note, teams have used 3D printed swerve drive components before and we see this as the next step to making them more accessible for all teams.

27 Likes

I had always had the impression you were running continuous fiber through portions of your modules. Can you confirm that isn’t the case on the saddles your referring to?

Thank you for this - makes a huge difference.

1 Like

SF saddles do have fiber. SF has ready access to fiber printers and use them pretty much wherever they can.

In our testing so far, it has not been necessary. We will continue to prove this, and we vow to be transparent with our testing as we understand just how large a financial commitment this is for teams, and the last thing we want to do is have failures on the field by teams trusting us.

9 Likes

As Sam has stated, our saddles have CF in them. Also as Sam has stated, since most of our printers are CF printers, we are quick to use it, maybe a bit too quick.

5 Likes

I would still consider the AM serve and steer to be the lest expensive module on the market. The “sticker price” might not be as low as your module, but it comes with both motors and all the gearboxes that you need.

16 Likes

Anand, I just can’t understand why you’d be hesitent to run a 3D printed swerve :stuck_out_tongue_winking_eye:

I agree that these concerns are definitely valid, and could be better eased with more examples of seeing the modules used (ideally through a competition or long practice day). With any COTS product on the market, the more visual examples of running and testing, the better. I remember VEX made new mecanum wheels in 2015 and in order to show their durability, made a video showing a testbed robot driving around the 2015 field.

I like the idea of giving teams the option to remove cost by using common parts that teams may already have from previous seasons. Aside from the printed parts (which in all honesty, I can’t say I’d trust most teams to have access to a printer that could make decent enough load-bearing parts), I’d argue a lot of teams could have spare VPs, bearings, motors, and gears laying around, so the less they need to buy, the better when it comes to a “thrifty” product.

When it comes to comparing to other COTS swerve modules on the market, I’d say there isn’t really an exact comparison to be made. Each swerve on the market has its own strengths, and different modules work better for different teams in different scenarios. For example, as mentioned, without seeing too much testing myself, I would definitely be heistent to use a mostly plastic module on a field with any bumps or berms out of fear that the plastic would break when we’re seeing stronger aluminum have issues sometimes in these same scenarios. On a completely flat field, I may not be as worried about that, but also, that’s just me. I can see teams going with the plastic swerve because the accessibility of swerve is a higher priorty to them for their team in their circumstance, and I can see teams going with the metal swerves because the higher load capacities and reliability are a higher priority for them and their circumstances.

Overall, I’d say I am excited that TTB is adding another swerve option to the market because I love more options for teams, and I welcome any efforts to make the experience of such a complex system more accessible to students. While I can’t say I would personally jump onboard to using TTB swerve in 2021 with my team just yet, I am very interested in seeing how those who do use it like it, and would love to see more videos of tests and matches played with these new modules.

3 Likes

I’ve been holding off from giving my 24 hr review on TTB swerve, but I feel like it is fair for me to respond to these comments about the SDS module.

Would you really have said the MK2 was too heavy? At under 5 pounds it was the lightest COTS module available at the time. I still believe any module ≤5 pounds is doing pretty good.

If using the NEO for steering can result in a nicely packaged module, is this really a flaw? NEOs are small when compared to a motor with a planetary gearbox attached. Yeah, a CIM class motor can output more power than you would even need for steering, but it isn’t an issue to not use full power potential of a motor.

Time has shown this to be a nonissue.

This is simply not true. The MK2 is compatible with all CIM class motors as well as many gearboxes with CIM mounting.

I’m not sure I’m understanding exactly what you are referring to here. The MK2 has two flanges that mount to the adjacent sides of the robot frame.

While the MK2 does use one more stage of gear reduction than necessary to achieve it’s drive reduction, it is important for it’s packaging concept.

I would be curious to hear your opinion as to why this is a flaw. An absolute encoder is all you should need for swerve steering.

How is this a design flaw?

On 2910 I don’t think we have ever had traction problems with our modules. We did realize an increase in traction from 2019 to 2020 when we went to nitrile treaded wheels.

32 Likes

I don’t speak for anyone else but I thought the purpose of his post was to point out that opinions of flaws are exactly that, opinions. I think you may have missed the point being made, or I read it less like someone being combative.

14 Likes

In my post, I went back in time, passed knee jerk critique on your swerve as though the points are factual. You were able to come to the present and tell me how I was wrong.

I believe you have proved the point of my post.

I don’t think I need to dialog on the “flaws” of the SDS swerve because the points I put forth are opinions and not flaws at all or have proven not to be an issue, which I stated in my post. Time has proven that the SDS swerve is just fine, some would say great.

For all those who read my post and think I am dinging the SDS module, you are wrong. The point is we shouldn’t be so quick to judge.

Did I mention the SDS module is good? Ahh I just did.

17 Likes

Yup, thanks for clarifying, this is how I read your post, and I thought you said it pretty clearly to make the point:

2 Likes