Team Astro Build Log (Cheese bots)

All things antweight

Moderators: BeligerAnt, petec, administrator

User avatar
MarkR
Posts: 375
Joined: Mon Dec 18, 2017 12:46 pm
Location: Reading Hackspace
Contact:

Re: Team Astro Build Log (Cheese bots)

Post by MarkR »

Sounds like overkill cnc for such a simple part, you can probably make it with mostly hand tools out of the right thickness of polycarb.

As a quick tip: materials which cannot be laser cut, either stick some masking tape, then laser etch the design, or just print out the design on paper and stick it on the part.

I've usually just used a scroll saw with suitable marking. I also like to use a centre punch tool to make the holes be in the right places then drill with a drill press ideally - because if I try it with a hand drill they always come out wonky. NB: I don't have a laser cutter, drill press or scroll saw, I just joined the local hackspace. If you don't want to use a scroll saw, a coping saw (hand tool) works well too.

I've found that "good" plastics for robots are mostly un-laserable- at least they can't be laser cut, you can still use the laser to mark the surface.
Robots: Betsie - RaspberryPi controlled flipper bot with gyro stablisation - too clever for her own good?
Stacie - tidy flipper; 4wd driven by hair bands
JoeHLloyd
Posts: 25
Joined: Fri Jan 10, 2020 8:18 pm

Re: Team Astro Build Log (Cheese bots)

Post by JoeHLloyd »

I am quite fond of overkill, and like the mass-producableness of machine made parts. But you do make a good point. Good tip on the laser etching. For now I think I'm going to develop the 3D printed strategy.

Have been having another stab at fusion and I think I have reach critical mass and it's beginning to make sense now. Not the most intuitive program really. Imported the original chassis but gave up on the conversion from STL. It's a nightmare. Triangles for days. Anyway, because I wanted to use some of my original Harmesan wedge armours I have to keep the exact dimensions for the holes. Painstakingly measured those manually from the STL and re-built it in a new file.

Below is WIP. New bolt-on motor mount now comes with motor shield. I'm going to complete this chassis with idential dimensions for now, and then I'll look at lowering the disc and moving it forward a bit, which should make it a little more dangerous and less dependent on successful wedging, but will have to be balanced against manouverability loss and potentially reduced ability to self-right.

Image
JoeHLloyd
Posts: 25
Joined: Fri Jan 10, 2020 8:18 pm

Re: Team Astro Build Log (Cheese bots)

Post by JoeHLloyd »

I think I'm ready for a test print.

Key features of v2:
  • Separate motor mount
    • printable in different shear plane
    • bolted to chassis with M3 bolts - should be non-shearable
    • allows easier electronics install/removal without soldering
    • options for modular mounts that can move weapon position around (e.g. probably lower and further forward
    • incorporated a motor shield
  • filleted to high heaven. looks cooler and will hopefully be stronger
  • more extensive weight reduction to hopefully buy enough weight for the heftier weapon mount
  • hole for switch this time
  • weapon floor slot cut out for moving weapon around
  • backwards compatible with old armour and panels
All in all my experience with Fusion 360 has been...OK. I don't think it's a very intuitive program and although the mobile timeline is cool I have found some seriously weird dependencies that I've made by accident and it's all a bit of a clusterfuck. Don't look at my Fusion timeline, it will give you cancer.

That all said, I'm sure it would look a damn sight neater if I were to do it again now, I think I kind of get the program a bit better now. Should have probably started with Fusion rather than sketchup - anyone reading this I implore you to do the same.

Image
Image
Image
Image
JoeHLloyd
Posts: 25
Joined: Fri Jan 10, 2020 8:18 pm

Re: Team Astro Build Log (Cheese bots)

Post by JoeHLloyd »

Am trying to get v2 together for Robonerd.

Haven't been able to be very picky about how it's printed, have a few irons in the fire. The first to come through looks like this PLA+ print. Not my first choice, but at least it's not PLA, and a chassis is better than no chassis. Not sure how the weapon support is going to hold up. If the printer (a mate of a brother of a mate) got my instructions it should be solid with no infill. Will have to check. Either way, something is better than nothing, and I'd rather fight with a brittle chassis and at least get some arena time than no chassis.

I should also have a TPU one on the way, and an carbon-nylon SLS from a professional outfit. Will be interesting to compare the weights. Hope I've not made any mistakes because the SLS one cost real money. Hope it doesn't weigh a ton either. Still, live and learn, and this hobby is a lot easier if you're willing to throw a little money around.

Print looks good. Will hopefully come in the mail this week and I'll get to re-fitting the electronics. Should hopefully be a lot quicker this time around.

News just in: both pieces are 100% infilled. Weight is probably going to be an issue now. Still, will just have to test it.

Image
JoeHLloyd
Posts: 25
Joined: Fri Jan 10, 2020 8:18 pm

Re: Team Astro Build Log (Cheese bots)

Post by JoeHLloyd »

Prep I've been very busy the last few weeks and discovered to my horror I only had 1 night to finish Harmesan v2 and get it battle ready. Everything I was waiting for came in time and I was able to throw it together.

The PLA+ v2 chassis came in at 33g with the mount, 3g heavier than the v1 TPU, but with 100% infill. I think that's pretty good given the significantly beefed up weapon support, my attempts at weight reduction were more extensive than before. Will have to have a think about where any more weight can be saved. I was still worried at this point because v1 came in at 149g, and I still had 2x20mm M3 bolts to add. I ordered them in aluminium to save weight.

I couldn't clean up my old RX well enough so just cracked out a new one, great move, saved loads of time. I focused this time on reducing the wiring down significantly, because I probably had 50% too much last time, and it's surprisingly heavy. It made the final fitting very tight, but I got away with it. Also simplified the Y-cabling, and everything just looks so much better than it did last time. And it can be lifted in and out of the chassis pretty easily (apart from the switch, which wasn't very satisfactory and in the end I just left it "on" because of the play I couldn't reliably use it).

Final weigh-in was 150g on the nose, which means I saved at least 3g in wiring, probably more. As my final adjustments I finally added a ton of "expo" to the sensitivity curve for steering. Off to battle!

Fights
Many fights: vs Smashy Flipper: vs a Team Azur bot No Fox Given:
Image

Too many to address individually so I shall make some general comments.

Driving
  • It's SO MUCH EASIER with the expo settings. My god. I pitted myself a few times early on, usually through overzealous pushing of opponents, but I got the hang of it quite quickly. Got very good at feeding people into the big flipper, and generally evading trouble and out-manouvering. Actually was able to execute some fairly slick manouvers that I would have wanted to last time but didn't have the skill to do so. My brain still doesn't like reversing and turning, I think it's expecting it to behave like a car, so I decided to just always go forwards and will adjust my brain to the reversing at a later date.
  • I need to make sure I never take fights facing a pit if I'm on the line between the pits. Yes, I might pop my opponents into the pit, but I might also pop myself into the one behind me. This lost me a fight in the last competition and I think it's finally registered with me not to do this.
Weaponry
  • I did many fights with the weapon off, and Harmesan is a surprisingly effective wedgebot/pusher it turns out, even without acetate. This was pleasing because that's one of the reasons I designed it with a big wedge in the first place - as a solid backup weapon.
  • The death angle did make an appearance from time to time, but surprisingly rarely even with the weapon off. Guess it's pretty low CofG.
  • The weapon is pretty potent even at half throttle (ran at half power to conserve battery power for more fights). Had a couple of very good connections, one of which against a horizontal, which ended very badly for the horizontal (pictures above). Had much more luck getting under opponents this time, even without acetate. I think I just got unlucky last time. Although there were also not many very low bots to compete against this time.
Durability
I had about 15-20 bouts I would estimate. The critical flaw of the shearing weapon mount did not appear again despite heavy abuse. I think I can count that problem as solved.
Damage:
  • a crack in the rear of the chassis. Not sure if it was a rogue spinner hit from someone or the result of ricocheting. Nothing some glue wouldn't sort out.
  • Broke a ziptie.
  • Acetate blade destroyed by spinner.
  • Disc began to look a bit lop-sided at one point, so I tightened the motor mount bolts. Think the threads may have stripped unfortunately, but due to the design, it wasn't a problem and the weapon still span freely. It can't get too off-angle because the weapon shaft is supported on both sides.
  • Polycarb held up great and has a few new gouges in it.
  • Basically, even after 20 fights, Harmesan was still fully functional and dangerous. EXTREMELY happy about this. I didn't face any really big hitters (wanted to try my luck against Wu Ben, not present) but repeated abuse was no problem. It feels so good to walk away from so many fights without needing to rebuild the entire bot.
  • The disc tooth looks used now, but is still sharp. Just has a couple of dents in it. Should get some good mileage out of these. When this tooth looks too battered I'll simply flip the disc and use the other side.
Design Revisions/Future
  • I'm going to disassemble Harmesan fully and check everything. In particular I want to test the threads for the weapon mount and see if the aluminium bolts have bent. I might actually have the weight for steel ones if it's only a 1g difference or so. Will probably need to print a new mount
  • The weapon support has never really sat fully true, and the bearing tends to work its way out. I need to adjust this better and glue that bearing in place, should make the weapon even more solid with zero room for flex.
  • Need to secure that switch
  • Also need to secure the wiring better, the main lead got sucked into the weapon at one point, although luckily didn't break.
  • Need to re-print/re-design the left cover, as it was too tight a fit and messed up the weapon alignment, and after I clipped it down it tended to flap towards the disc. Also need to find a better way to secure this as the single screw allows a fair bit of movement, although it hasn't really proved to be an issue yet.
  • I will also design a weapon mount that lowers the disc and moves it forward by about 2/3mm, which should make it bite sooner, with less wedging action needed. This may be a trade-off with self-rightability, so will have to test. But that's the beauty of the replaceable motor mount - it's easy to test.
  • And I'm going to make the cable hole in the weapon mount bigger, getting it down without pinching a wire was fiddly.
  • I want to print this chassis in TPU as I still think it's the superior choice.
Summary
All in all, I am extremely happy with how v2 has performed. I feel that the major design flaws have been addressed. Weapon potency is there, driving is there, durability is there. Bring on the competitions!

Battle-scarred but functional:
Image
JoeHLloyd
Posts: 25
Joined: Fri Jan 10, 2020 8:18 pm

Re: Team Astro Build Log (Cheese bots)

Post by JoeHLloyd »

New
Image
Old
Image
New
Image
Old
Image

The new motor mount (which may not really be a replacement, it's just a test to see what happens) is 2.5mm lower and 3.5mm further forward than the original. This means:
  • Less ramp to climb (4.5mm less) before an impact
  • lower CofG
  • Death angle less likely
  • Hindered ability to self-right when flipped backwards onto rear wheels...but by how much remains to be seen
  • Has to be another 3.5mm further forward before it will hit things in front of the wedge
As usual, had some seriously weird dependencies in Fusion and in the end was easier for me to start this mount from scratch than it was to modify the original. Love filleting things. Got to leave them until last though. I would really like to do a more organic looking bot at some point, like something that is futuristic and looks like it was grown, with asymmetry and no right angles. Anyway, just an idea stored here for another time.

Should be getting a new chassis printed in TPU this week, just going to let this version percolate for a day and see if there's anything else I want to add before committing.
haz
Posts: 1855
Joined: Thu Apr 15, 2010 11:20 am
Location: Underwater Lair
Contact:

Re: Team Astro Build Log (Cheese bots)

Post by haz »

Looking like a great set of improvements - experience of mine is you cannot be too aggressive with your weapon mounting, the further out you stick it the more you're going to hit!

Honestly the only thing I could really say about the whole bot is I'm not too much of a fan of the blunt tooth profile and the sharp internal corners on your disc. I think you'd find it would catch and rip much better with a sharper tooth but, personal preference! I wouldn't like to be hit with the disc as it stands!

It looks like a really tough little bruiser of a robot, really simple to maintain and repair!
JoeHLloyd
Posts: 25
Joined: Fri Jan 10, 2020 8:18 pm

Re: Team Astro Build Log (Cheese bots)

Post by JoeHLloyd »

Thanks Harry. Yes, I do wonder whether I should make the tooth sharper. I might see if I can do that with a file, but I'll probably end up unbalancing it and may have to faff about filing on drilling on the other side of the disc. Although I should add the disc in the fusion file is not actually the real disc I use, although it's similar, and the tooth is the same.

I went to the Halloween Bodge bots last weekend and competed with Harmesan. Overall, I was pretty disappointed with the performance.

Fights at 20s, 2m20s, rumble at the end (fight 2 not recorded)
https://www.youtube.com/watch?v=QGQv693Et_4&t=1090s

Image

Battle Report

Won 1, lost 2.
Win 1: Opponent self-pitted after minor contact
Loss1: my switch apparently wasn’t fully in the “on” state and turned itself off after the first weapon-on-weapon impact against a horizontal. Left dead in the water. Very disappointing way to lose.
Loss2: Against a vicious overhead hammer-saw. This was going fine, I was getting under him and chewing him up a bit, until the hammersaw went straight through the weapon guard and thence weapon motor, crippling the weapon. Rolled onto the back side, and because of the new motor mount position the wheels were just in the air. I tested this, and it recovers fine from this position as long as the weapon works. Which it didn’t. Otherwise I would have had a good chance of winning by being the superior wedge. Received the most damage in this fight of any fight, and probably equivalent to all previous damage ever received. It was brutal. Actually, on reviewing the footage, I lost the weapon when the support was destroyed, and the motor can was destroyed after that.

That was me out of the tournament, but I still wanted some blood, so I set to repairing the damage. The damage:
  • Weapon support destroyed
  • Polycarb wedge severely bent (the hammersaw hooked in behind it and yanked it over the screws holding it on. Interesting to see the screws held in and the polycarb ripped over their heads. Wasn’t expecting that at all. I had not prepared for that angle of attack at all. The wedge is potentially salvageable, but probably not, I think it’s done for.Image
  • Weapon shield. The weapon mount was fine apart from the shield - this one was PETG as the TPU one rubbed the motor. The PETG just shattered and didn’t protect the motor at all. Will try again in TPU or carbon fibre sls nylon. I can’t afford to make it much thicker due to weight, but PETG is clearly not the answer. That little saw was so vicious it might not have mattered what it was made of actually.
  • Weapon motor. Stoved in, seized completely.
I had spares for everything except the weapon support. (Well, I technically did have a spare for this, but I hadn’t drilled a hole for it on the replacement polycarb armour so couldn’t use it. I had everything replaced (including soldering in a new weapon motor) in about 25 minutes, go go battlemechanic/techpriest. I wasn't thrilled about running the weapon unsupported, but it was that or nothing, and I thought it would be interesting to see what broke first - the motor or the weapon mount.

I threw it into the royal rumble at the end to try to get some redemption. I ran the weapon at full power for about 90 seconds I would estimate, and delivered some satisfyingly big hits, although against a very tough bot that survived with an only mildly bent titanium scoop. I sent it into the roof with one hit which was satisfying. I also decapitated a horizontal spinner, and ripped the back off a fleaweight cluster. Then I started getting some magic smoke, and the drive stopped, and I was pitted. The weapon still worked though, which is interesting. I haven’t taken the bot apart yet to find out exactly which part died. It’s a bit of a mystery - the rx and weapon esc were still working flawlessly, but both drive channels appeared dead. Can I have burnt out 2 channels on the rx but not the 3rd? I guess it was drawing a lot of current running the weapon at full speed for so long. But at least the weapon was still going - that was well-specced in the first place and I’d have been very irritated if that gave up, as the ESC is rated to double the expected maximum current draw. The bot was definitely hot when I picked it up.

What went well:
The big hits at the end were with an unsupported weapon, and the motor is completely unbent. This is pretty surprising, given I managed to bend the last motor whilst it was supported on both sides. Maybe the weapon support isn’t as important as I thought for a robust weapon. I think I’ll still keep it, as it also keeps the motor can on in the event of a big horizontal hit. Although maybe it wouldn’t be enough to save it if that happened anyway.

The big hits were also satisfying, weapon potency when it connects is good.

The new weapon position worked OK. I can’t say I noticed a big increase in weapon contact, but perhaps a little. It can only have helped. Maybe I should consider moving it further forward still.

What did not go well:
That bloody switch. The “on” state doesn’t seem very secure. I think I’m going to buy a different style of switch for future use. It also got bumped off in some of the casual fights. I either need to put it somewhere else or modify the chassis to shield it a little. It’s tricky to have something that can be operated with a finger than can’t be operated by an opposing robot.

The death angle. If the weapon is on, it’s not a problem. But, first rule of robot combat, if your robot has a death angle, sooner or later you’ll lose a fight because of it, and usually sooner. And it’s not the case that I have to spin to win. Harmesan is a pretty effective wedgebot/pusher, so losing the weapon is by no means the end of the fight for me. I have some ideas to solve this one. I think if I rotate the gearmotors forward by 30 degrees or so then the ESCs won’t protrude so much, and I can have flatter covers that won’t contact the ground when tipped over backwards. This wouldn’t have been possible with the old weapon mount, but the new one should allow it.

The overheard hammersaw. I can’t believe how brutal such a small weapon and small motor were. This is pretty much the perfect counter to Harmesan - it’s a rare design, such that I never really expected to face a competitive one, and have not thought much about how to defend against one - it came in at angles I had minimal armour. I should probably have just tried to pit him. I think my general strategy should be to pit my opponents, not destroy them. The longer you fight the more chance you have to lose or take damage, and the pit is the quickest way to end them. Other than better driving, what can I do to avoid such a brutalisation in future? The PETG shield was utterly useless. The hammersaw went through that and into the motor like butter, and punctured the motor can. ImageAt a minimum, that needs to be replaced with TPU or carbon-sls nylon. I could even beef it up by a mm as I have a spare gram to use. Alternatives to this could be a titanium shield. This would be quite a bit more involved to do - I’d need to bend the sheets myself, and possibly laser-cut the sheet beforehand. Then there’s the question - would 1-2mm titanium sheet actually repel the hammersaw? Yes, I think it would, as long as it didn’t catch an edge. A flat/curved sheet of metal is a good defence against a spinning weapon. See Beta vs Tombstone. As long as there’s no edge to catch, it should work OK. The problem with plastic, although it is tough, is that it’s soft, and a sharp disc is going to sink into it and bite, and transfer energy into it. Rip chunks out. Hmm. I think I’d like to try a Ti shield, but given the additional hassle in making one, for now I’m going to develop the 3D printed version, with the Ti shield on the backburner as a possibility. Something I wanted to do originally but couldn’t was have a single flat polycarb panel as each cover. That would provide much better armour than the 2mm printed covers. They aren’t serious armour really. They are just there to prevent a glancing blow from hitting something important, they are never going to repel a solid hit. Polycarb panels would repel a solid hit. I doubt I’ll be able to get the gearmotor ESCs flat enough to get away with this, but I’ll bear it in mind when I return to Fusion.

Magic smoke. Currently, I don’t even know what is actually broken. I assume running at a high current for extended periods of time melted some wires together, or burnt out something in the rx. If the former, I can use thicker wires potentially, if the latter...I don’t know. I could just not run the weapon at full speed for as long. But I don’t like that as a solution. It's tantamount to heresy, and merely a plaster over the problem. Could I change the wiring so that the ESC is completely isolated from the rx? I just don't know enough about electronics. Can anyone help me out here? I'll re-post my wiring diagram for scrutiny. For that matter, does anyone have any theories what burnt out and why?

AWS is not far away. At a minimum, I probably need to replace the rx, and I can at least field something. I think given the fairly extensive chassis changes/cover changes, I’m going to struggle to achieve that in the next 4 weeks and also manage to get it printed. Still, I have 2 weeks of staycation coming up, so I should have unusually large amounts of time to work on it. Might be able to do it.

Future
  • Rotate gearmotors so that the covers don't contact the floor when inverted
  • Print motor shield in Carbon-SLS-nylon or TPU (after a re-design to accommodate the less accurate TPU print
  • See if each of the complex covers can be replaced with single flat sheet of polycarb
JoeHLloyd
Posts: 25
Joined: Fri Jan 10, 2020 8:18 pm

Re: Team Astro Build Log (Cheese bots)

Post by JoeHLloyd »

Image

Yes, it is flawed. The weapon is armed at all times, with the option of turning the rx off when you think the whole thing is off. Really more of a danger feature as it currently stands than anything else :shock:

Does this mean I have a high current going through the RX when the weapon is running at full power?
T_Y
Posts: 153
Joined: Thu Mar 15, 2018 12:51 am
Location: York

Re: Team Astro Build Log (Cheese bots)

Post by T_Y »

There has been a recent emergence of powered overhead weapons, I think we'll start to see more people developing armour packages for them - I made 0.5mm Ti top plates after seeing Shakey's "Calm Down, It's Only a Drill!" eviscerate a LiPo at Ant Freeze last year.

I've found Titanium workable with hand tools as long as you keep the thickness down. I've cut 0.5mm Ti (both Grade 2 and Grade 5) with a pair of cheap Silverline tin snips and you can drill holes in it with good quality HSS/cobalt tipped drill bits at a slow speed.

One other option you could try is a lightweight but thick material - maybe foamboard? Or a 3D printed shield that has a very low infill density but is much thicker overall. That way, although the blade might cut through easily, the material next to the cut might stop the blade from going deep enough as the rest of the weapon assembly (motor, weapon arm etc.) makes contact with it. In the same vein, you could go for the "air is armour" strategy and just move the shield further out so there's a larger gap between it and the weapon motor.
Thomas - Approximation Robotics AKA "the one that does the mechanums"

Ants:
PLA-ST Is a Perfectly Legitimate Choice of Material
Percussive Maintenance 4: A New Spinner
Stack Overflow

Beetles:
Déjà Vu/Two
Post Reply