Questions and Answers about Combat Robotics
from Team Run Amok


Designing a Big Horizontal Bar Spinner
 

In January, 2016 'Ask Aaron' received a series of questions from a new builder in Texas planning to build a combat robot with a familliar layout in hopes of competing at BattleBots. The builder had a great many questions concerning the design and components, and the discussion went back and forth for a couple of weeks.

The resulting exchange is of such length and range that I thought it deserved its own webpage here at 'Ask Aaron'. Don't ask me to tell you the "Did you ever hear the story of the Texan who..." jokes the builder attempts to tell -- I'm keeping them to myself.

 


 Day One: A Texan Wants to Build a Spinner

Q: I am designing a two wheeled robot (plus front caster) with horizontal bar spinning weapon. I know, I know, it's been done and while a few spinners have done well, most do below average.

I used the torque, acceleration and battery calculator to which you refer. Inputs were 250 pound robot using two T74 motors with AmpFlow dual ESC (160 amps per channel) turning 10 inch wheels. I am planning on using five in series Turnigy 7.5 aH 90C 7.4V hard case LiPoly batteries to run the motors at 37V. Assuming a worst case 100% load (I know, unrealistic, but I am looking for worst case numbers to bound the problem) and coefficient of friction of .9 (again, given the caster unrealistic), I come out with 13.30 aH for 5 minutes. If I use a more realistic 70% load and CoF of 0.5 I get a bit more than half of that, 5.17 aH. So batteries I need are going to need to supply between 5 and 13 aH for 5 minutes. When I go to the battery chooser, I can't find my battery.

Question #1: Is there some way of putting the specs for the batteries I plan on using into the battery calculator?

Assuming question 1's answer is negative, I tried to get as close to the same impact as my battery. 90C (no burst values in battery specs) at 7.5 aH gives me 675 amps. I used a variety of Turnigy 7.4v batteries in the calculator to get an estimate. I adjusted the number of batteries in parallel to get the constant current or then the burst current amps equal to the 675. One Turnigy battery had a constant and burst value the same. What I estimate from that data is that one bank of 5 7.4V batteries in series should be enough and 2 banks would definitely be enough.

Question #2: Do you concur on the 1 vs 2 banks or have I piled to many guesstimates together?

On to the spinning weapon. Using the spinning weapon calculator and an Ampflow A40-300 motor running at 37V with 4" driver, 1.8" driven pulley belt drive, I come up with a ~20 aH battery requirement assuming 6 spin ups and ~10 aH assuming 3 spin ups. The blade spins up at what seems to be a respectable 3500 J/s for a 250 lb robot. The issue is the battery and controller for the spinning weapon. If I use the torque calculator values for the A40-300 motor at 37V, the stall current is 740 amps. So my options on controller are a contactor that handles 400 amps continuous, 700 amps for 3 minutes, 1000 amps for 70 seconds, 1400 amps for 30 seconds, 2000 amps for 10 seconds and 3000 amps for 1 second at 48V or the AmpFlow A160 controller with the two channels combined for 300 amps (per specs, not 320 amps with simple arithmetic of combining two 160 amp channels). The contactor with support logic is a little less weight so I would prefer to go that route. The ESC seems to me that during the first few revolutions the motor would want to draw more current than the ESC would allow.

Question #3: Which would you go for, the contactor or the ESC?

If I use the same five in series Turnigy 7.5 ah 7.4V 90C hard case batteries for 37V, then I would be supplying 675 amps to the A40-300 (90C x 7.5 aH), 70 amps below the stall torque, so all is good. But 7.5 aH won't last the match assuming even at a low 3 start up spins. If I put two banks in parallel, I have enough for 4 spin ups and three banks gives be enough for the 6 spin ups with a bit of margin. But then I am supplying up to 1350 amps or 2025 amps.

Assuming nothing is binding the weapon, then the motor should be fine drawing as much current as it needs to spin up and then settling down to a much lower current draw to keep the bar spinning. Three banks should give me enough aH to last the fight. I am also assuming that the spinner motor will be idled right before a hit and turned back on to spin up as soon after a hit as possible, so the motor shouldn't bind when hitting.

Question #4: Am I using the calculators correctly or am I missing a critical point?

The robot is designed to run up right or inverted. So the only "righting mechanism" would be if I was knocked on my side (e.g. against an arena wall). In that case, I would want to try and use the spinner bar to jar the robot enough to put it back on it's wheels. Even with three banks of batteries, I think a momentary turning of the robot spinner on and off should be okay. The contactor could handle the 740 stall amps for the momentary try easily. The motor will either be able or not be able to rock the robot up right and even assuming a total bind on the motor (robot somehow pinned) causing the full stall current to apply, a momentary surge shouldn't kill the motor. If it does, then I have already lost the match because I am on my side and will need to swap out a motor for next combat anyway.

Question #5: Am I thinking straight about using three banks of batteries in parallel to potentially supply 2050 amps to a motor with a stall current at 740 amps is reasonable to try and rock the robot onto it's wheels or is having that much current available going to immediately fry the motor?

I understand that I am running the drive and weapon motors in an over volt situation. From what I can tell that will lower the lifetime of the motors but they should last through several combat matches. On the other hand, I am also potentially providing the motors with all the current they want so burn out at the higher voltage is a real possibility.

Question #6: Is there anything I can do other than running at a lower voltage to help the battery and/or to refurbish the battery after a tournament?

Question #7: Is there some way to tell that the motor is about to give up the ghost so I swap out before a match?

Question #8: Could I reasonably provide even more voltage to the motors in question?

Question #9: How much of the 7.5 aH of the batteries can I count on drawing in real life, down to 0 or will the voltage and current drop off be more like 50% of the rated amp hours could realistically be pulled from the batteries.

Question #9: Given all the above, do I seem to be on the right path or do I need to go back to the drawing board from scratch?

Please feel free to amplify on any of the answers or add an answer to a question I should have asked. I appreciate your taking the time to help out noobs to the robot community like me. Thanks. [Beverly, Texas]

A: [Mark J.] You anticipated that there would be questions you should have asked, and I feel obligated to take you up on your invitation to answer one.

Q: Mark, should a self described 'noob' start their career in combat robotics by building a 250 pound spinner-weaponed robot?

A: No.

The only place you can compete with a 250 pound combat robot is BattleBots, and they have become very particular in who they accept. If you don't have a successful combat record you don't have a realistic chance of being selected as one of the very limited number of competitors -- particularly with an "it's been done" design. I see disappointment, great expense, and rejection in your future.

Once upon a time BattleBots was an open tournament and anyone who could build a robot and pass a technical inspection was welcome to enter. Those days are gone. Build a robot for a lighter weight class and gain some experience in regional tournaments before you drop big money to go play with the big dogs.

I'll assume that you'll choose to ignore that advice -- any robot builder worth their salt would do the same, but my conscience demands that I give you fair warning. Now, back to your questions.

Question #1: You don't need to enter your battery specs into the 'chooser'. For an assumed 0.5 effective coefficient of friction the main page of the Team Tentacle Torque / Amp-Hour Calculator provides you with the maximum current draw of the drive motors (Total Peak Amps: 88.6 amps) and the estimated battery capacity required for matches of 3 minutes (3.1 amp-hours) and 5 minutes (5.2 amp-hours). Your battery pack for the drive system needs only to meet those requirements while providing 37 volts.

Question #2: A single bank of your proposed LiPoly cells has an entirely adequate amp-hour capacity for your proposed drive train, and is massive overkill on current delivery. The wheels of your robot will break traction at a torque level where they are drawing less than 90 amps total; the motors will not stall.

Question #3: There are several advantages to using an Electronic Speed Controller to control a spinner weapon:

  • Reversing your spinner weapon to clear a jam or rock a stuck robot off a rail can be a lifesaver.
  • An ESC is usually both smaller and lighter than a contactor of sufficient capacity to control a given weapon.
  • It is MUCH easier on the weapon motor and battery to 'ramp up' the voltage over perhaps half a second during weapon start-up than to dump the full voltage and amperage to the weapon with a switch. This will avoid that big current surge that you're worried about -- the faster the motor spins, the lower its current consumption.
  • Tournaments have a maximum 'spin down' time for spinner weapons. Many ESCs have the ability to apply an electrical braking force to help meet this requirement, eliminating the need for a mechanical brake.
Note 1: 3500 joules of energy storage is NOT respectable for a heavyweight robot weapon. Are you sure you didn't leave off a zero? You didn't share your spinner weapon details, so I can't check your calculations.

Note 2: Forget about powering off to 'idle' the spinner before a hit. You're going to be way too busy to worry about switching the motor off in time for a hit. If the weapon has enough stored energy you won't need to worry about the weapon 'binding' -- your opponent is going to be flying away from you and the weapon will still be spinning.

Question #4: Something is very wrong in your calculations. There's no way that your weapon should be pulling that much current in so few spin-ups. Again, you haven't given me the details of your weapon, so i can't run the numbers for myself.

Question #5: A permanent magnet DC motor cannot pull more than its stall current at a given voltage, regardless of the current available from the battery. As the speed of the motor rises above stall, the maximum possible current consumption decreases linearly with the speed increase until the motor reaches maximum 'no-load' RPM, were only a small current is required. You need not worry about supplying too much current, but the current capacity you suggest is unnecessary and a waste of weight.

Question #6: LiPoly batteries are damaged by two things:

  1. Current drain above their rated capacity; and
  2. Discharging them below 3.3 volts per cell.
If you keep within these limits the battery should be fine. Once damaged, a LiPoly battery can no longer be safely used and must be replaced. There is no 'rehab' for a damaged LiPoly cell.

Question #7: I'm gonna take a 'seventh inning stretch' in the middle of this answer; I'm losing feeling in my fingers. Signs of a failing motor:

  • Discoloration of the motor windings, accompanied by an acrid smell. These are signs that the insulation on the motor windings has been overheated and is failing.
  • A blue discoloration of the copper on the motor commutator. This indicates high heat that may be weakening the structure of the commutator -- usually caused by prolonged high current flow.
  • Excessive motor brush wear. Know how long the brushes are on a new motor, and learn how much the brushes wear in a match. When they wear to about two matches worth left it's time to replace the brushes, or the motor if the commutator is showing wear as well.
These are all signs that the motor is ailing, but a good looking motor can fail abruptly and without warning. That's part of what makes robot combat so much fun.

Question #8: I wouldn't.

NPC T74 does not overvolt well, and I'm a bit worried about it even at 37 volts. The smaller T64 does pretty well at 36 volts, but the T74 has the same brushes and commutator and pulls considerably more current at a given voltage. I know that Robot Marketplace says the T74 can be run at 36 volts, but NPC says to keep it to 24 volts.

AmpFlow motors have a similar issue with overvolting. They are high performance motors to start with and offer very good horsepower for their weight. Pushing to high voltage can greatly reduce their effective life and result in abrupt failure. If you need more horsepower for your weapon, go to a larger motor.

Question #9 (the first one): LiPoly batteries are rated for the amp-hours of current they can provide before being drawn down below their recommended minimum voltage of 3.3 volts per cell. Very high rates of discharge can reduce the available amp-hours of current due to losses thru the internal resistance of the cells themselves. Performance will also vary with the manufacturer of the cells. It's best to test actual performance in the environment the cells will be used.

Question #9 (the second one): As previously noted, there is something awry with your weapon calculations. Take another shot at that, or send me the details and I'll see what I can do. Alternately, you can be sensible and go build yourself a nice hobbyweight robot and get some experience before attempting a heavyweight behemoth.

One additional concern: the NPC T74 has a gearbox cover made of cast aluminum that's brittle and has a nasty habit of shattering. There used to be a third-party machined billet aluminum cover available for these motors, but they are no longer made. The few builders who still use these motors have bought up the remaining examples of the billet covers. I wouldn't recommend using the motors without those stronger gearbox covers, and I think you'll have a hard time locating a pair. Best luck.

 


 Day Two: Let's Get This Straight...

Q: Thanks for the quick answer to my previous question.

Yes, the design is for BattleBots at 250 lb, but I have armor and configuration options that bring it down to the 220 (100 kg) weight class. As I see it with battlebots, you either have to have serious chops on the combat robot circuit (which I don't have in my wildest dreams) or a hook that appeals to the TV producers. I think I have the hook, time will tell. Even if not accepted by battlebots, I have options for getting the experience at the 100 Kg class.

As far as start small, I prefer to go big or go home. Yes, an ant weight would cost less and pose a different set of problems, but the big bot is just more interesting to me. I have been warned of the impending heartache, your conscious is clear.

Let me ask the battery question a different way. With a 0.5 coefficient of friction, I need 5.13 aH for 5 minutes, a 0.72 coefficient of friction, I need 7.45 aH for 5 minutes and a 0.9 coefficient of friction, I need 9.31 aH for 5 minutes. (I want to design for 5 minutes for other tournaments and/or a false start that requires a restart of the combat as per battlebots rules). The proposed battery bank (five in series Turnigy 7.5 aH 7.4V 90C hard case batteries for 37V) would provide either 7.5 aH (one bank) or 15 aH (two banks in parallel). The 90c x 7.5 aH =675 amps would be fine either way. The robot is 250 lb, the two wheels are NPC PT-444 tires which are listed as 4.10/3.5-4 (so in theory a 3.5 in wide tread, but the cad drawing on robotmarketplace has the width as 3.25 typical). Single caster up front is carbon steel.

One bank won't get me to the full 5 minutes if the coefficient is 0.73 or higher. So I can put in two banks at the accompanying weight and cost penalty and be totally safe under all considerations or find a way to better guess the coefficient of friction to know if one bank is enough. If the robot was built, I could just do empirical testing, running the bot at full speed back and forth for 5 minutes. Unfortunately, this is the design phase and while I can design in the space of two banks and fill with foam if I don't use the second one, I prefer if possible to get it right in the design phase.

Question #1: How do I drill down and determine the coefficient of friction?

A: [Mark J.] A Texan who wants to go big or go home. Well, if you can live with the stereotype so can I.

The Tentacle Calculator assumes that the full weight of the robot is supported by driven wheels. If you click on the 'Help' button in the calculator and scroll thru the input value descriptions you'll find an explanation and expansion of the correct setting for the coefficient of friction value to adjust for other conditions:

...The default value '0.9' represents a best case. You may enter lower values for known 'slick' conditions, or to compensate for a portion of the robot weight being supported by other than driven wheels. Example: for a robot with 60% of the weight on the driven wheels under good traction conditions, enter the product of 60% and 0.9, which is '0.54'.

Since your robot is two-wheel drive with a big weapon hanging off the front, you'll have well less than 100% of the robot weight on the driven wheels. Figure out the center of mass for your design and calculate the percentage of weight on the driven wheels. Multiply that by the assumed traction conditions (0.8 is pretty close for NPC rubber on slightly greasy steel) and enter that into the calculator.

Q: On the spinning weapon, the performance is ~3500 joules per second. So 3500 joules after one second of spin up, 7000 joules after two seconds of spin up, etc. So after 10 seconds, 35,000 joules and over 100Kj after 30 seconds. This is all assuming 37 volts on the A40-300 motor. If I drop down to 24 volts, the weapon is adding ~3000 joules per second. So after 10 seconds spin up, it is 5000 joules less. After 30 seconds, it is 15,000 joules less. Based on data gathered around the web (so who knows how accurate) competitors like Tombstone would come in between those two numbers. Obviously, I would rather have a bigger hit, but I would also rather have a working spinner. Option one is to play it safe, have a smaller hit and not burn out the motor. Option 2 is to over volt to 37 volts, risk motor failure and deliver a smaller hit. Of course, I have the option based on the match to alter the number of batteries in series to deliver 22.2v or 29.6v or 37v depending on the competition.

I looked at other bigger motors and as with all things associated with combat robots, there are trade offs. PMG132 and LE-200 motors have had reports of being unreliable in combat because a strong shock (being hit or hitting your opponent) knocks the armature out of whack and starts chewing up the permanent magnets. The ME0708 is reputedly more reliable and would run at a much higher voltage, but the peak torque listed on the Motenergy web site (38 Nm) and performance chart on RMP (3600 no load RPM at 48V) would deliver less energy per second for the same given blade. I could over volt that motor above 48 V to get up to the same hit energy, but then I'm over stressing a motor with Chinese ideograms on the diagram. The ME0708 is much larger (costing more frame weight), harder to mount because of the size (costing more frame weight) and almost twice as heavy. ME0709 gives more torque, but the size is to big for the 10" wheels. Going to larger wheels impacts drive motor performance and is a slick vs knobby tire tread. Also, the ME0709 is much larger (costing more frame weight), harder to mount because of the size (costing more frame weight) and almost three times as heavy. If you have a suggestion for another large spinner weapon motor to look at, I am very interested.

Absent another horse being entered in the motor race, my leaning is to vary the voltage based on competition, bring a spare motor (or two) and plan on rebuilding the motor after every tournament. Costs are up, but as you have said many times in different ways, if you plan to be competitive, plan to spend money. The weight savings on the motor/frame can be better used on a bigger/better blade.

A: Nope, nope, and nope. You're making several incorrect assumptions about electric motor performance, weapon energy storage, and the usability of spinning weapons at very high rotational speeds.

You're attempting to calculate the stored energy of the weapon as a linear function of the energy output of the weapon motor, and you're treating the motor output as a constant across the speed range. When 'bogged down' at low speeds the electrical energy input to a permanent magnet direct current motor is very high and the output torque is also very high -- but the speed is low. Since 'power' is the product of torque and speed, the power output is also low. As the speed of the motor reaches its maximum RPM the electromotive back force reduces the current consumption of the motor and available torque drops off to zero. Again, since power is the product of torque and speed the output power drops to zero. The top chart on the right shows the relationship of torque-current, and power output across the speed range of a permanent magnet direct current motor.

When you translate that power curve into a curve that shows the energy actually stored as kinetic energy into a rotating weapon you get something that looks like the lower chart on the right. The stored kinetic energy plateaus as the motor approaches maximum speed and additional time does not add to the stored energy level.

You've got very little time to build up a significant amount of energy stored in the weapon before your opponent 'box rushes' you and stops your weapon before it can become dangerous. I had assumed that you simply made an error when you mentioned in your last message that you had a larger pulley on your motor than on your weapon. A typical rotary weapon will have a small motor pulley and a larger weapon pulley, offering torque multiplication to spin the weapon up to speed more quickly. Going the other way will seriously bog down your weapon motor, causing it to pull huge amperage for way too long (meltdown) and slow the weapon spin-up far too much. You'll be a sitting duck with a melting weapon motor.

I strongly suggest that you read the Ask Aaron Spinner Weapon FAQ for general information on spinner design. The FAQ has a link to the Team Run Amok Spinner Weapon Excel Spreadsheet that can acurately calculate your true weapon energy storage and graph the spin-up performance of your weapon system. Very handy!

Q: Your arguments on the ESC vs contactor make sense, but there are a couple other considerations. The Ampflow website (which also warns against over volting) gives suggestions on what to do if you do over volt. One of those is to adjust the motor so it only spins in one direction. This knocks out reason #1 you give if I stay with the A40-300.

A: Advancing the commutator timing for running in one direction does not prevent the motor from running in the other direction, but it makes in less efficient. You'd still be able to spin the weapon in reverse long enough to clear a jam or rock it off an obstacle. Still highly useful.

Q: The contactor (shock tested to 25G) with battle switch to activate is just under 1.2 lb while the A160 ESC is 1.5 lb, which vitiates reason #2 you give absent finding another high current but lower weight ESC (and I have looked, but that doesn't mean there isn't one I haven't found yet). I'm planning on using bushings for the weapon (like Tombstone) and he finds the natural spin down do to friction easily meets the 60 seconds requirements. This is still a good reason to consider the ESC.

A: This contactor you found sounds very good -- a little too good. Its specs greatly exceed the performance of contactors with which I am familiar. I wonder why other builders have not found this contactor and put it to use? Those ratings wouldn't happen to be for AC current, would they? Controlling DC requires a very large de-rating of the AC specs. Alternating current tends to extinguish it's own arcing as it passes thru zero net potential on each cycle, where direct current does not self extinguish and is MUCH harder on contacts. Better check those specs.

Q: Your third reason for the ESC brings me to a larger question and your answer will hopefully guide me to a better understanding of motors.
You propose a softer ramp (which would presumably also mean a slightly slower spin up) to lower stress on the components. But if I am using an ESC that limits current to 300 amps and the stall current based on the torque calculator is 740 amps at 37 volts, the motor would be getting less than half the stall current. So in that crucial initial spin up phase, the motor would want more current, but the ESC would act as a double choke with a softer ramp and a lower max current. Less torque would be applied to those initial revolutions of the weapon because the weapon motor was current starved. Once the weapon starts spinning up, the torque requirements drop and the current demand drops and ESC vs contactor is a don't care because the motor required current will likely fall below 300 amps. If you want to use a standard blade for calculations, assume an aluminum 7075 1.5" x 4" x 48" blade and the A40-300 at 37 volts. Assuming a free spinning weapon, my understanding is that the motor will want the stall current initially and then fall off on current linearly as the weapon spins up. Tombstone's driver said that the one thing he wished he could change was to supply more current to his weapon (he emphasized he was looking for more current, not more voltage).

Question #2: Am I missing something here on what happens when the motor initially spins up a weapon and what am I missing?

A: You're not missing anything; amperage equals torque, and more torque means a faster spin-up. If your ESC has true amperage limiting (most don't, but the AmpFlow and Rage Bridge do) you don't need to ramp your throttle 'cause the current limiting will take care of it. If you're running an ESC without true current limiting (perhaps only thermal limiting) ramping will serve a similar purpose and protect both the ESC and the battery.

The current roll-off is linear, which in your case means that the weapon motor will still be attempting to pull better than 300 amps until the weapon is spinning up to nearly 60% of it's maximum speed -- which with your gearing could be quite a while. Let's take a look at the spin-up specs for your weapon system: 1.5 x 4.5 x 48 inch aluminum bar powered by A40-300 @ 37 volts via a 0.45:1 pulley drive [chart at right].

Great Horny Toads!!! Your proposed weapon takes seven minutes to spin-up, pulls more than 300 amps for the first two minutes, and sucks down better than 40 amp hours of battery capacity per spin-up! Half way into your first match your weapon motor will be a glowing lump of incandescent copper. Nope.

Change the pulley system to a 1.5" driver and a 6" driven and you'll have a weapon that spin up to 22,000 joules in five seconds and pulls 300+ amps for less than two seconds. A three-minute match with 5 spin-ups uses only 2.7 amp-hours of battery capacity. That's workable.

Q: The motor turn off was an idea from Tombstone's driver. We were planning on two transmitters with one person driving and the other handling the motor. I think part of his reasoning on turning off the motor was to lower stress on the motor vs having a sudden change in RPM's due to the hit while powered up. The weight penalty for a second receiver is pretty minimal.

A: It gets very hectic in a combat robot match, but if you have the luxury of a second pair of eyes and hands on a second transmitter you can certainly give this a try. It probably won't do any harm.

Q: On the motor failing, thanks for the tips. My plan to rebuild the motor after a tournament is in line with your brush wear tip. This does raise another question. I have seen other teams comment on the A40-300 that they lost a match because their wiring burned up. They replaced it with a heavier, high temperature wire. The A40-300 is listed as having 10 AWG wire. I can't find what kind, but I would estimate this is rated for 60 or 70 amps. This seems awfully low to me. Perhaps because the wire is rated at say 70 amps and 600 volts, if you only have 37 volts the wire can carry a higher current safely. I was already planning on using 2/0, 3/0 or 4/0 welding wire from batteries to switch to esc/contactor depending on potential current draw of the motor. It makes sense to me to change the wire to the battery if I can. If nothing else, it lowers the resistance and thus the voltage drop to the motor.

Question #3: What do you think about changing out the factory wire for as thick of wire as I can mate up to the motor?

A: Your gain from lessened resistance will be minimal, and that large gauge wire is surprisingly heavy. Maybe just a second run of 10 AWG in parallel to the existing wire and a little free air space around the power wires to give them a chance to dissipate a little heat. Many builders cram their wires into very tight spaces that invite heat build-up.

Q: Your concerns on the T74 are noted. The T64 at 37 V is slightly slower than a T74 at 37 V (3.69 vs 3.55 seconds side to side in 48 foot arena), but much faster than the T74 at 24V (5.28 seconds). So it makes perfect sense to save a bit of weight and space to go with the T64. I also see it in quite a few other teams bots which is a good sign.

But your comment on the gearing housing has me thinking. The housing on the T64 seems to be the exact same as that on the T74, so I presume the same issues would apply. Assuming that I don't commission a milling from a solid block of Al 7075 a new housing (it sounds like I might even break even selling the new frames to others), what should I do.

A: Yes, the gearbox housings for the T74 and T64 are the same. Let me ask around and see if I can find a couple of the billet housings for you. The motors are so little used anymore that I don't think you could find many buyers for a new run of milled units.

Q: I looked at team Whyachi and they have several gear reduction units that would work and presumably be as rugged as needed. But then I need to find a 3" motor to mate to it (e.g. P3 planetary gear box and commission team whyachi to make some custom 10" wheels to mate to the shaft). Midwest motion products has an interesting 36V motor (which I could of course consider over volting mwahhahaha), the MMP D33-655D-36V (4091 no load speed RPM, 1250 OZ-in stall torque) seems to mate to the P3 gear box. At 16 to 1 gear reduction, the motor and gear fit in size and seem about right on weight. But now I lose the benefit of the wonderful torque calculator to let me see if I am faster/slower than other alternatives.

Question #4: How serious do you consider the brittleness of the gear housing vs all the issues of changing to a rock solid gear reduction but a corresponding issue of finding a reliable battle proven motor?

A: The gearbox brittleness issue is pretty much what killed the NPC motors as viable combat robot power options. It's a very serious flaw.

Question #5: If I should change motors, any hints on good 3" motors to consider or on how to get the motor specs in to the torque calculator to compare the alternatives?

A: The 3" A28-150 AmpFlow motors will mate up nicely to the Whyachi gearboxes. That makes for a very strong and well proven drivetrain.

Thanks again for what you do in answering these questions. I know I've been long winded, but want to give you all the information you need to answer the question (though I probably left something out anyway). [Beverly, Texas]

 


 Day Three: I Think We Understand Each Other

Q: Thanks again for the quick answers. You probably have moved at least one loss to my future win column with your answers.

"A Texan who wants to go big or go home. Well, if you can live with the stereotype so can I."

Did you ever hear the story of the Texan who...

[Entertaining Texan joke redacted. I can't match it with an Oregonian joke and I can't allow a Texan to tell a better story here.]

I looked at the drive motor suggestion and I like what I see. An Ampflow A28-150 mated to a Whyachi P3 gearbox is shorter length, width and height, lower weight and the mounting is much better with 8 vs 4 bolts and larger bolts. Plus I can add mounts to the body of the motor for more support. Even with a two stage gear box, a 1" shaft x 3" long, the more and bigger bolts, two mounts with heat fins on the motor it is still 2.5 lb lighter and the upgrade in support and a 7075 aluminum gear box is unbelievable. Of course by the time all is said and done, my wallet won't thank you ($870 per gear box), but the end results look good.

But of course this brings me to another question.

I calculated my CoF based on your suggestions. Dividing up the weight of main compartment vs the blade/spindle and dividing up the extension from main frame to blade I got about 65% of the weight over the drive wheels and 35% over the front caster. Multiplying by your suggested 0.8 gives me ~0.518 for a CoF, close enough to start with.

I then did the gear tip math.

Step 1: 250 lb * 0.518 = 129.5 lb maximum tire force
Step 2: 129.5 * 0.4167 (10" tire) = 53.96 torque to get max tire force
Step 3: 53.96/216.9 (stall torque of 22.2v of A28-150 motor) and I get a 4 :1 gear ratio
Step 4: Multiply by 1.5 or 2 gives 6:1 or 8:1 ratio

The PM3 gearbox gives me five gear options: 2.91:1 -- 4:1 -- 8.46:1 -- 11.64:1 -- 16:1.

The 8.46 gives me 3.43 on side to side time for a 48 foot arena, but it is constant acceleration. You need 104 ft to get to top speed, so that's not happening. Even going up to a 11.64 gear box I have side to side of 3.42 and need 47.6 ft to get top speed so that's not happening either. Going all the way to 16:1 gear and I slow my side to side a bit (3.88 and 21.2 ft to top speed).

The amp hour story is also not good. The aH for 5 minutes (assuming 100% worst case) is 16:1 10.27, 11.64:1 14.12 and 8.46:1 19.43. So I am barely making it on two banks of batteries for 11.64 and higher ratios. I have to go down to 74% operation to get the 8.46 to squeak in enough battery life. Given that for all three of these voltages I will always be accelerating I'm betting higher than 75%.

If you just talk about going half way across the arena, they are all at 2.23 seconds because all are accelerating the whole way (limited by tire spin). If I up the voltage to 29.6, all three of these gear ratios are limited to 1.92 seconds to go 24 ft (half the arena), but I still have the battery life issue.

So it would seem to me that I am better of with the 16:1 ratio as my time to the center of the arena is the same and I get much better battery life. But your tip math (unless I've done it wrong) says I should go with 8.46 or 4:1 and thus add a third or fourth bank of batteries, but I don't see the advantage.

We'll put to the side the issues of 22.2 V vs 29.6 V over volting. I also ran the numbers at 24 V, which would move me away from those nice Turnigy batteries. Same issues.

Question #1: Given the above, it seems 16:1 is the right ratio to run. Do you agree or did I do the math wrong somewhere?

A: It's really difficult to achieve even 70% of 'Peak Current Drain' for a full match -- particularly given a very large weapon and poor pushing capacity. You'll either promptly end the match or be pressed back against the nearest wall. Since 100% of peak can only be maintained by pushing at full throttle with wheels spinning it's only a dedicated pushy-box that can burn thru that much current, and then only if it gets into a prolonged pushing battle. Watch some matches and get an estimate of how much time the 'bots actually spend pushing hard.

Your selection of the 16:1 gear reduction is correct. A few weeks ago I added an extra section to the end of Ask Aaron: Optimum Gearing for Combat Robots. That section is called "Adjusting Gearing for Special Conditions" that clarifies what 'optimum' is and how to adjust that starting point for considerations like arena size and current consumption. It begins:

The 'optimum' gearing allows the motors to produce their full output power without 'bogging' and consuming excessive current. While that is 'optimum' for the motors, the gearing may not provide adequate accelleration to achieve the best speed in small arenas or in cases where available current must be limited to the capacity of specific speed controllers. Adjustments to the 'optimum' gear reduction may be needed in these conditions.

Thereafter follows an example of an adjustment of gearing for a specific arena size and a limited maximum current flow. Your calculations mirror my own quite nicely, save only that the distance to the center of the arena needs be reduced by the length of your robot or the location of the starting box -- a minor adjustment.

Q: Of course changing motors means changing tires, weight calculations, etc. But I think the issue is the same.

On the spinner weapon storage, try this experiment. Assume 6216 RPM (Am40-300 at 37 volts) and 27.1 Nm with a 48" x 4" x 1.5" aluminum blade. Now divide the initial joules at 63% of RPM max by the seconds it took to get there. Vary the blade length and the drive ratios and do the same initial joules divided by the seconds to get there. I suspect under all the different conditions you will come out close to the 3500 J/S I talked about. I agree that the blade will eventually max out on stored energy, but for combat useful periods of time, it seems to be adding 3500 Joules per second.

A: Well, a joule is 1 watt-second. If you've got a motor pumping 3500 watts into the weapon system then you can expect an energy increase of 3500 joules per second. I don't like to get builders thinking in terms of a simple linear estimate of weapon power increase 'cause it's only sorta linear and only for so long. A lot of builders would take that estimate and run too far with it -- and I'd get the blame.

Q: I am using 2.22 in the drive ratio, I probably misstated driver vs driven pulleys.

I'm going to go through your spinner weapon archive again.

I get your peak motor performance, but I'm still not sure about the ESC vs the contactor. If the AmpFlow 160 ESCwith output paired to generate 300 amps is used and the stall torque is at 440 amps, then from when I start spinning up to the time the motor is spinning rapidly enough to drop the current requirements below 300, then I am current limiting the motor. Rough numbers and assuming the linear nature of the current/torque, that looks to be about a third of the spin up time( 300/440). So if I use the contactor, I am giving the motor all the current it requests all the time. With this contactor, I can give it 440 amps for 5 minutes (not that that would happen as the motor would elect a pope in that time).

Question #2: Isn't the 300 amp current limited ESC current starving the motor during at least the first part of the spin up, until the RPM's get to the point that the torque requirements drop to a point that is less than 300 amps?

A: Yep, it sure is. It's also keeping the motor brushes from vaporizing. The weak part of the AmpFlow motors are the somewhat small contact area of the brushes to the commutator. Limiting the current flow helps to keep them from becoming one with the ozone layer. I'd suggest watching brush wear very carefully, even at a 300 amp peak current.

Incidentally, AmpFlow has just announced a new motor you may want to look at for your weapon. The new A28-400-F48 pumps out 11.5 HP @ 48 volts. As yet unproven, but interesting...

Q: It sure looks to me like the contactor is going to give me faster spin up times.

The contactor is an MX34FD from Gigavac. Even a blind squirrel finds a nut occasionally, so I may have stumbled onto something. Great form factor (much smaller than SW200), better temp range, shock tested, any orientation, etc. Cost is $133 for 1 to 4 quantity, so even reasonably priced. Lots of options for auxiliary circuits. The only knock I have on them is they don't seem to have a SPDT version. Check them out. If they are as good as they seem to me, I can feel like I've given back even a little bit.

Thanks

A: I asked around about the Gigavac contactor. Seems like this is the worst kept secret in combat robotics. Several teams tell me they've been using various Gigavac products and they think they're awesome. Everybody thought they were the only ones who knew about them and nobody tipped me 'cause they knew I'd spill it. Well, you can consider it spilled.

Consider using the MX34CD model with the 24 volt (16 to 34 volt range) coil.

 


 Day Four: The Payoff

Q: If you like getting paid in funny stories, by all means let me pay you.

That wasn't the first time the Texas boy...

[Nope, still not gonna let a Texan joke sneak into 'Ask Aaron'. Who knows where that might lead?]

Your reply raised some interesting ideas. I think I can now get my spinner up to over 100 KJ in 20 seconds, deliver ~230 KJ in 60 seconds. 5 minutes and five spin-ups will need four banks of batteries, but well worth it. Yee haw. That I believe will be competitive.

Assuming the previous CoF and 70% peak current drain average, one bank of drive batteries will last me the match. I can also now get to 1.72 S for the 24 feet. I get what you mean about being a bit closer because of the length of the robot, but I kept the distance at 24 feet to compare to earlier set up.

Thanks for the advice.

A: Good luck, Tex. Here's hoping that your opponent gives you 60 seconds to spin up to that 230,000 joules. The hit could be epic.

 


 Day Five: Wait a Second...

Q: Ok here is another question. If I want to limit top end RPM, I can reduce voltage or put in a gearing system. The problem is that both of those also impact RPM when I first start spinning up. Is there a way to run at one voltage at start up and then switch to a lower voltage as I spin up?

My first thought is to take a voltage tap off my stack of 7.4V batteries at two points (top voltage off the top of the stack and mid voltage from the middle of the stack). I then have two contactors with the first contactor providing the top voltage at initial spin up, at the appropriate RPM (say 63% max RPM) the first contactor would be turned off and the second contactor would be activaqted to provide mid voltage.

I then thought about doing something with the ESC, but the ESC I have won't go to the required voltage and ESC that work at that voltage won't deliver the start up current requirements. So I'm starting from scratch on finding an ESC to control the spinner motor.

Any better way to do this and is there a "gotcha" somewhere in the idea? Accept for the sake of argument that there is a good reason to limit top voltage. I'll "pay" in advance for the answer. A young man was waiting at the bus stop...

[Lemme guess - the bus stop was in Texas. This is the third time, but we're still not THAT good friends.]

Thanks

A: I can think of a couple reasons for keeping the RPM under control. I'm a fan of keeping it simple -- consider this approach:

  • Increase your pulley reduction ratio.

  • Reduce your voltage.

  • Balance those two and you keep the same spin-up rate you initially had but it now levels off at a lower max RPM.
It's easier on the motor and battery, it allows use of a single contactor (or an ESC at the reduced voltage/amperage), and it reduces weight by one contactor and several LiPoly cells.

 


 Day Six: Wait Several Seconds More...

Q: I didn't ask my question well. I need to keep the RPM of the motor to a maximum value. I'm thinking of using a gearbox that has a max RPM value. The torque max is fine (in fact I have lots of room before max torque on the gearbox).

Assuming 5 minute match with 5 spin ups, a no load current of 4.4 amps, a stall torque of 70.58 Nm through a gear box and a 48" x 4" x 1.5" aluminum bar for the weapon for all below.

  • If I use 44.4V and 3463 max RPM (through the gearbox) with 1:1 pulleys, I get: 42784 Joules in 8.4 seconds and 97285 Joules in ~25 seconds using 6.45 aH. That is a respectable amount of energy to deliver in somewhat reasonable time frames.

  • If I use 29.6V and 2309 max RPM (through the gearbox) with 1:1 pulleys, I get: 19015 Joules in 5.6 seconds and 43238 Joules in ~17 seconds using 4.42 aH. Based on your suggestions from the spinner calculator instructions, this isn't bad but it isn't as nice as above.

  • If I use 29.6V and 2309 max RPM (through the gearbox) with 2:1 pulleys, I get: 4754 Joules in 1.4 seconds and 10809 Joules in ~4 seconds using 1.38 aH. This is pathetic energy.

  • If I use 29.6V and 2308 max RPM (through the gearbox) with 0.5:1 pulleys, I get: 76060 Joules in 22.4 seconds and 172951 Joules in ~67 seconds using 16.6 aH. This is pathetic time.

  • If I use 29.6V and 2308 max RPM (through the gearbox) with 0.67:1 pulleys, I get: 42359 Joules in 12.47 seconds and 96319 Joules in ~38 seconds using 9.41 aH. The energy is good, but the time is poor compared to the original 44.4V
So changing the pully ratio can get to the same joules of energy, but not in the same time. I need to spin up at 44.4V and then switch over to 29.6V to keep the RPM down.

I can do the two contactors to switch over, but that raises issues beyond weight and number of parts complexity.

  1. The timing of the switch over.

    • Could be done manually by the person controlling the weapon, but human elements are the most likely to fail.

    • Could be done with a shaft encoder that measures the RPM and sends a signal to switch to the lower voltage. A human reset would then switch back to 44.4v to spin up after a hit.

    • Could be done with an electronic time delay that signals number of seconds after spin up, but after a hit, the starting RPM will probably not be 0 (unfortunately).

    • Could be done by an ESC if it had a program input that limited maximum voltage if encoder value was above x RPM or based on a count of pulses from transmitter (same issue on count as with time delay). AmpFlow ESC has the encoder inputs and a max output drive value which should let me limit top velocity (aka top RPM), but the amplfow won't work at 44.4V

  2. The batteries are going to drain at different rates. All the batteries are being drained at the same rate. The first four batteries in series will always be in use and the last two batteries in series will only be in use during initial spin up. My gut tells me that uneven draining of the LiPoly batteries is not a good thing. I could somewhat compensate by rotating batteries through the stack, but at best it will be a crude balancing and at worse I'll mess up the rotation and make it worse.

  3. I think I figured out the master switch issue, I'll just run the "negative" wire through the switch. instead of the "positive" wire.
So I'm still stuck on the desire to run at two voltages. I agree with your philosophy of KISS to win. But I also want my big hit to win. I figure that most of my hits will be with 1 to 10 seconds spin up time. If circumstances allow the 60 second spin up before a hit, it'll be spectacular, but I can't count on it.

Would you please provide the equations from the spinner spread sheet so I can better understand what is happening in the time leading up to the 2/3 RPM. I'd like to calculate energy stored at 1, 2. 3. 4. 5 ... seconds. You are very much correct that ~2 seconds is all I can count on if the opponent box rushes me and I don't dodge.

Thanks

p.s. Since you heard the last joke, I won't tell you the one about the man who showed up for work Monday morning with two black eyes.

A: I'm severely disappointed in you, Tex. You've got no trouble at all ignoring warnings about overstressing your poor weapon motor, but you've decided to get all girlie about maximum speed ratings on your gearbox? Specs are for Yankee losers! Glare at the gearbox, spit in its eye, and tell it what you're gonna do to it if it fails. We ain't playin' Old Maid here.

More seriously: you explained what you were doing well enough -- I just didn't have time to run some example numbers to explain what I was going on about. I'm not sure what specs the motor or gearbox you're thinking about at the moment, so let's go back to your original weapon motor:

  • Ampflow A40-300 motor running at 37 volts: 6170 RPM, 41.8 Nm stall torque, 524 stall amps; and
  • a generic 2:1 gearbox.
With your 48 x 4 x 1.5 inch aluminum bar and a 1:1 pulley, that poor overloaded motor spins the weapon up to 1250 RPM in 5 seconds with 21K joules, and after about 30 seconds builds to a thoretical 120K joules at around 2950 RPM (you should live so long) using 9.4 amp hours to spin up 5 times in a 5 minute match.

Here's an alternative:

Drop the voltage to 33.3 volts and bump the pulley ratio to 1.8:1.

  • The motor now spins to 5550 RPM, produces 37.7 Nm torque at stall, and pulls 472 stall amps.

  • The weapon will still spin up to 21K joules in 5 seconds.

  • Better, the weapon spins up to 8K joules in 2 seconds, 57% more than the 2 second energy with the 1:1 pulley at 37 volts.

  • The gearbox speed is reduced by 11%, and battery power requirement drops to 2.9 amp-hours for the same 5 spin-ups in 5 minutes.

The weapon tops out near 30K joules in under 10 seconds, but I'd MUCH rather have the quicker spin-up to useable energy levels than the hypothetical mega-joules of energy that you're never gonna see. I also really hate the complexity of two contactors, control systems to switch between sets of batteries, and gearboxes that feed belt drives.

Every new builder has a file of complex designs that are gonna revolutionize the sport. I'd show you some of mine, but I burned the file after my first tournament. Our current designs have had hundreds of very clever builders designing, trying, and revising for more thn 20 years. Simple works, and reliability wins. Scrap the gearbox, pulley-drive the motor to spin to max RPM in 10 seconds, and win some matches.

I've uploaded my own unprotected copy of the Excel Spinner Spreadsheet. It has a tab named 'Data Table' that's hidden in the usual release. The table there is used to create the chart in the 'Calculations' tab. The calcs are speed-based, not time-based -- but you can tinker with the '% RPM' to tweek the 'Seconds' to whole numbers. Download your copy quickly 'cause I may decide to take it down before it confuses regular users:

Unprotected Spinner Spreadsheet

 


 Day Seven: That's Just About Right

Q: Thanks for the spreadsheet. This really helps put things in perspective. I see what you mean now.

Let me re-iterate the assumptions. 250 lb robot. 48" x 4" x 1.5" aluminum blade. 7.5 aH per bank of batteries with battery serial increments of 7.4V. I used as benchmarks your suggest 4 J per pound in 2 seconds, 16 J per pound storage so 4000 J, 2.25 seconds representing a really fast box rush in 48 foot arena with them starting at the edge of their battle box and me staying at the back for 36 ft distance, and then a 5 second and 10 second spin up time representing a chance to gain some momentum. Match assumption is 5 spin ups in 5 minutes.

Starting point was 44.4V system with 1:1 drive that would need to switch over to 29.6V to limit top RPM. So a complicated system.

  • In 2 seconds it generates 5600 J which is 22.4 J/lb.
  • It takes 1.6 seconds to get to 4000 J.
  • 6900 J in 2.25 S
  • 23,546 J in 5 S
  • 51,954 J in 10 S
  • 10.76 aH (so 2 battery banks of 6 batteries each)

Then I tried the same system at 29.6V

  • 2 S = 4297 J for 17.18 J/lb
  • 1.9 S to 4000 J
  • 2.25 S = 5278 J
  • 5 S = 16760 J
  • 10 S = 33287 J
  • 8.8 aH so 2 battery banks at 4 batteries each

So good numbers, just paying a price to run at lower voltage.

Then I tried a 1.25:1 drive at the 29.6V

  • 2 S = 5617 J for 22.468 J / LB
  • 1.6 S = 4000 J
  • 2.25 S = 6678 J
  • 5 S = 17352 J
  • 10 S = 26998 J
  • 5.9 aH so one bank of 4 batteries

So with a 1.25:1 drive I match well at 2 seconds, but start to fall behind at 2.25 S and beyond, but I save 4 batteries.

I then went to a 1.5:1 drive at 29.6V

  • 2 S = 6536 J for 26.144 J / lb
  • 1.4 S to 4000 J
  • 2.25 S = 7475 J
  • 5 S = 15968 J
  • 10 s = 20483 J
  • 4.32 aH so still one bank of batteries

So I can beat the 44.4V up to 2.25 S, but fall further behind at 5 and 10 seconds

I then tried a 1.1:1 drive at 29.6 V

  • 2 S = 4903 J for 19.612 J / LB
  • 1.76 S to 4000 J
  • 2.25 S = 5773 J
  • 5 S = 17318 J
  • 10 S = 30903 J

Then for fun I tried going up to a 60" blade, 29.6V, 1.1:1 drive, everything else the same.

  • 2 S = 3075 J for 12.13 J /LB
  • 2.5 S to 4000 J
  • 2.25 S = 3775
  • 5 S = 14049 J
  • 10 S = 34600 J
  • 13.76 aH

So the bigger blade will deliver a bigger hit if you can get to 10 S at the cost of every other measurement.

Going down to a 36" blade at 1.1:1 drive

  • 2 S = 6875 J for 27.5 J / LB
  • 1.3 S to 4000 J
  • 2.25 S = 7825
  • 5 S = 14242 J
  • 10 S = 16525 J
  • 3.06 aH

So much better 2.25 S and below numbers, but much worse 5 and 10 S numbers.

36" blade with 1:1 drive

  • 2 S = 6579 for 26.316 J / LB
  • 1.4 S to 4000 J
  • 2.25 S = 7625 J
  • 5 S = 15681 J
  • 10 S = 16700 J
  • 4.16 aH

So getting close to maxing out at 10 S. But the initial numbers up to 5 seconds look like a 48" blade with 1.5:1 drive.

You made some comment about Tombstone maybe using tarot cards to decide which blade to use. Perhaps he has analysis like this which in turn means he can decide does he want a big hit that will take time to build or a fast spin up.

[Some of it may be that, but several of his blades have effectively the same moment of inertia. I still think tarot.]

I had planned on having different blades and pulley ratios, now I have more science behind the selection. The shorter blades don't deliver the really big hits if you can spin up, but they do let you deliver good sized hits often.

A 36" blade at 1:1 ratio could deliver 20 hits on one bank of 29.6 V batteries in 5 minute match with about the same or more energy than 48" blade with 1.5:1 drive (assuming 5 second or less spin up).

Obviously the 36" blade won't have the reach of the 48" blade, but we are talking about being able to deliver a lot of hits twenty 4000 J hits in 28 seconds. Alternatively, twenty 15681 J hits in one minute 40 seconds. The 36" blade will also mean less frame, less weight in the blade (and thus more for armor), shorter extension supporting the blade.

So now my questions.

Question #1: Can I use the same torque value (based on stall torque) at 29.6 V that the manufacturer quoted for their 48 V part. Or to put it another way, is stall torque independent of voltage?

A. Speed, torque, and power are all voltage dependent:

  • Double the voltage = double the no-load speed
  • Double the voltage = double the stall torque
  • Double the voltage = four times the power (power is the product of speed and torque)

So, speed and torque each need to be multiplied by (29.6 / 48) = 0.617, and total power is multiplied by (29.6 / 48)2 = 0.380.

Question #2: If you had to pick between the 36" and 48" blade (with all the trade offs), which would you go with (or build to accept either)?

A. The longer blade can store a given amount of kinetic energy at a lower RPM. A lower spin rate gives better blade 'bite' into your opponent. I'm a big fan of 'bite'. Unfortunately, the longer blade is also less structurally sound. We're talking about very high energy impacts here. Don't hit the wall.

Question #3: I am now refining the blade based on the Riobotz tutorial. Anything else I should consider?

A. RioBotz has the general design elements correct. Given the high energy carried by the weapon I'd avoid a hub design that drill lots of holes thru the blade and weakens it. Maybe one central hole to locate and a hub that clamps around the blade?

You have helped a lot. I have a faster bot, better idea of battery needs, much better weapon and a lot more science behind weapon design. Of course the cost of the bot is up about $3000 over where I was before we started talking, but if you want cheap, play tiddlywinks.

Thanks.


 A Few Hours Pass...

You know I would have more questions.

Question #4: your suggestions of 8 joules per pound in 2 seconds and 16 joules per pound kinetic storage, are those minimum numbers that you shouldn't show up to the party without, middle of the pack competitive numbers or top of the pack numbers?

A. Those were entry-level numbers at the time the spreadsheet was written (2007). They'd still get you by at local barn fight, but they'd be non-competitive at a major event.

26 joules per pound in 2 seconds and 32 joules per pound storage in ~2.5 seconds and 135 joules per pound in 5 seconds would seem to be competitive whatever the benchmark.

A: I would agree.

 


 Day... I Think I've Lost Count

Q: I've thought this before, but now I think I have settled on a blade design. I should have known TANFL, so I don't get 48 V stall torque at 29.6 V.

I'm getting 27 J/lb at 2 seconds, 90 J/lb at 5 seconds and 153 J/lb at 10 S. It will require at least two banks of batteries and I'll design with space for a third bank. Even then, if I spin up for 10 seconds, hit with 38+Kj and repeat every 10 seconds for a 5 minute match, I'll get only 24 spin ups from three banks. Of course if I can deliver 24 38+Kj hits I hopefully won't need a 25th hit.

Now I start looking at different blade dimension and pulley combinations.

I almost hate to do this, but you are such a good source for pointing me in the right direction, so I'll ask another series of questions.

[Editor's note: 'almost' counts in horseshoes -- not in combat robotics. Ask away.]

Now that I have faster drive motor set up, more powerful weapon motor and the motors and gear boxes are running within spec without any extra complexity, I need to figure out how to mount the spinning bar weapon.

I like the idea of a dead axle design because I can use the dead axle itself to support the end of the extensions that are supporting the blade. Live axle would be easier to mount the blade and pulley but I don't see it handling the shock as well.

So I am thinking the spindle components would be:

  1. 1" 6061-T6 aluminum plate bolted to the frame with 5/16? grade 9 bolts to mount the shaft. The bolts let me remove the plate to attach a new shaft/blade.
  2. 1? bore x 1.562? O.D. x 0.03? thick thrust washer inside a 1.75? bore x 2.5? O.D. x 0.03? thick thrust washer (both 52100 steel)
  3. 1? bore x 1.562? O.D. x 0.78? thick needle thrust bearing (3000 lb dynamic load capacity) inside a 1.75? bore x 2.5? O.D. x 0.078 thick needle thrust bearing (5600 lb dynamic load capacity)
  4. 1? bore x 1.562? O.D. x 0.03? thick thrust washer inside a 1.75? bore x 2.5? O.D. x 0.03? thick thrust washer (both 52100 steel)
  5. 1" shaft x 6" stroke with 0.625" x 1.25" tapped ends (e.g. redi-threads heavy duty shaft 10600 RC 60 steel) 5/8? grade 9 bolt on each end
  6. 0.591? thick x 6? wide x 6? long 6061-T6 aluminum 2.0472? bore for RA100RR 1? bore spherical bearing (3550 lb dynamic load capacity) bottom clamp hub (extends 6? wide x 1? thick x 0.75? up blade side tapped for 4 x 5/16? grade 9 bolts on each side)
  7. 48? x 4? x 1.5? 7975-T651 aluminum blade with 1.5? bore hole to accept part of the spherical bearing at top and bottom
  8. 0.591? thick x 6? wide x 6? long 6061-T6 aluminum 2.0472? bore for RA100RR 1? bore spherical bearing (3550 lb dynamic load capacity) top clamp hub (extends 6? wide x 1? thick x 0.75? up blade side holes for 4 x 5/16? bolts on each side)
  9. 3? diameter with 1.25? deep inset for 2.0472? bore for RA100RR 1? bore spherical bearing (3550 lb dynamic load capacity) on top and bottom 6061-T6 aluminum hub with integrated Bx pulley extending up 1? shaft
  10. 1? bore x 1.562? O.D. x 0.03? thick thrust washer inside a 1.75? bore x 2.5? O.D. x 0.03? thick thrust washer (both 52100 steel)
  11. 1? bore x 1.562? O.D. x 0.78? thick needle thrust bearing (3000 lb dynamic load capacity) inside a 1.75? bore x 2.5? O.D. x 0.078 thick needle thrust bearing (5600 lb dynamic load capacity)
  12. 1? bore x 1.562? O.D. x 0.03? thick thrust washer inside a 1.75? bore x 2.5? O.D. x 0.03? thick thrust washer (both 52100 steel)
  13. 1.5" 6061- T6 aluminum plate welded to the frame to mount the shaft.
Total height is 8.5? with the blade 2.216? off the ground.

This gives me thrust bearings on each end and four spherical bearings going up the shaft (either side of the blade in the blade hubs and either side of the pulley hub).

My concern is that this is a potentially delicate shaft assembly with the 6 bearings as the most vulnerable parts.

I have a drawing to help understand how all this fits together, but I'm not sure how to attach it to the question.

Is there a better way of attaching the blade/pulley to the robot frame?

A. Tex, I'd sure hate to see your design for a baseball bat. Lord knows how many moving parts it'd have.

Team Run Amok Design Philosophy

A combat robot is a tool for defeating other robots. The best tools are simple, reliable, and easy to use.

As you previously noted, a certain very well known big bar spinner uses bushings to support their weapon. Nice simple bushings. Foolproof bushings. Bushings that don't come all apart when subjected to an unexpecedly huge off-axis shock load. Bushings that help spin-down your weapon when the match is over. The builder of that well known robot knows what he's doing.

Dead shaft - yes, for the structural reasons you cite.

If I properly understand the design of the removeable plate that locates one end of the weapon shaft, it should be fine. It wouldn't hurt to give me a look at the drawing. You can set it up in some image-sharing site and send me the link, or just email it to me: joerger@toast.net

 


 Another Day...

Q: What do you think of the 10" ampflow wheels on aluminum hubs vs the 10" NPC wheels?

The ampflow are lighter (2.4 vs 5 lb), thinner (1.5" vs 3.25/3.5" diagram vs spec), solid vs foam filled, mount directly to 0.75" drive shaft vs need another 0.5 lb of hub.

The NPC should give a little more cushioning effect when the robot is hit, tossed or otherwise abused, but for a two wheel robot, you are "paying" over 6 lb for that cushion effect.

Any reports from combat that would prefer one vs the other?

A. You don't see the large plastic-rimmed Colson style wheels in exposed applications on successful heavyweight robots. Recommended for use only with armor coverage in those situations.

I've used the NPCs, and when they say 'foam filled' it's a VERY dense foam that offers very little cushion. Traction is good, and they will absorb a good impact and stay round enough to finish the match.

Of the two I'd pay the weight penalty for the NPCs, but not for the cushion.

Q: Any 10" tires that will work with a 0.75" or 1.0" keyed drive shaft you would recommend beyond the NPC?

A: You might be interested in the BattleTreads. They're centrifugally cast one-piece all-foam tires that get denser toward the outter edge. Similar tires are available from various sources. I believe Team Plumb Crazy got their foam tires from Harbor Freight, and Northern Tool also has a variety of foam wheels and rims.

The foam tires may be buffed down to a smaller diameter as needed. These foam tires are much lighter than foam-filled pneumatic tires and will fit on the same style wheels, but hub solutions for these wheels may require custom machining.
 


 Thinking About the Bar

Q: I'mmmm baaack! Muwahahaha

Did I ever tell you the time a Texan... [EXPUNGED]

I think I have the blade worked out. I'm using teeth that can be inverse mounted, so one side is slightly longer (1") than the other (i.e. trying for a semi-one tooth design).

  • Option 1: A 36" x 4" x 1" 7075 blade with S7 teeth at the end. I can generate a ~51Kj hit in 2 seconds and max out at ~70K in 4 seconds. Batteries will support 14 spin ups in a 5 minute match (12 banks of 22.2V @ 7.5aH per bank = 90 aH). The pulley ratio is 0.411 (a 2.3 and a 5.6). The RPM is ~4500 at 2 seconds and ~5500 at 4 seconds.

  • Option 2: I can use a 48" x 4" x 1" 7075 blade with S7 teeth that delivers a ~21Kj hit in 2 seconds, ~46Kj in 4 seconds and max out in 9 seconds with ~72Kj. The RPM are ~1900 at 2 seconds, ~2800 at 4 seconds and ~3500 at 9 seconds. Pulleys are 2.3 and 3.7. I can get 13 nine second spin ups in 5 minute match.

  • Option 3: I can use a 36" x 4 x 0.75" Titanium grade 23 blade without teeth added so each end is the same length. 2 seconds is ~44Kj, 4 seconds is ~69Kj and I max out in 5 seconds at ~71 Kj. 13 spin ups. ~3800 RPM in 2 seconds, ~4700 RPM in 4 seconds and ~5000 RPM in 5 seconds. Pulleys 2.3 and 5.1.

  • Option 4: 48" x 4" x 0.75" Titanium grade 23 blade without teeth added so each end is the same length. 2 seconds is ~15Kj, 4 seconds is ~37 Kj, and I am almost to max in 10 seconds with ~70Kj (max is 400J higher). RPM 13 spin ups. Pulleys 2.3 and 3.3. RPM ~1450 in 2 seconds, ~2250 in 4 seconds and ~3000 in 10 seconds.
So all blades develop about 70 Kj, it's just a matter of time. Option 1 develops the energy fastest, but at a higher RPM. Option 4 develops the energy the slowest but at lower RPM.

Question #1: Am I doing any harm to your spinner spreadsheet equations if I just add the teeth to the mass and use the 1" longer side for the calculations?

A. Assuming that the impactors are not huge your calculated answer is not going to be far off. The teeth are denser (steel) than the bar (aluminum) and are sitting way out on the ends of the blade where they have the greatest MOI. The actual energy storage will be a bit larger and the real spin-up time correspondingly a little longer.

The 'correct' way to add in teeth/impactors is to manually calculate their mass and treat them as a 1" tall ring with a radius equal to the center point of their mounting location and a thickness of whatever is needed to equal their actual mass. I don't think I'd bother.

Question #2: 1" 7075 aluminum and 0.75" grade 23 titanium feels thin to me. I can go thicker at the cost of slowing down the energy storage. Any guidance on how thick to make a blade?

A: A real engineering analysis of the stress levels on this blade when it hits some unfortunate chunk of matter at 70 Kj requires high-end engineering software to which I do not have access, or a room full of 4th year engineering students, which I also lack. My guidance is to learn from the experience of other builders and mercilessly copy their successful design elements for a starting point.

Question #3: Any reason to make the titanium vs aluminum blades? Titanium cost is much higher though without teeth, slightly simpler to make.

A: See answer to question 2 (above).

Question #4: Assuming the bot can support all four blades and I can make all four blades, what situations should I use which blade? My thought was to use the 36" aluminum blade against most competitors and bring out the 48" aluminum blade when I need the reach.

A: I can't think of many circumstances that would benefit from an additional six inches of weapon 'reach'. I suppose a terrified opponent might bodge some special defensive structure in the pits to keep your blade at a distance...

A longer and slower spinning blade might be useful against a spinner-killer titanium scoop design for some added 'bite -- but see my answer to question 5 (below).

Question #5: I know that lower RPM is better for bite. What sort of trade off makes sense for lower RPM and lower joules at a given time vs higher RPM and higher joules sooner?

A: Lower RPM improves bite because the time between the passage of impactor teeth decreases, allowing more of the opponent's robot to enter the sweep of the weapon at a given closing rate. However, shortening one arm on the shorter aluminum bar allows it to function as an effective 'one-tooth' design that doubles the passage time of the impactor and doubles the bite. That will allow your aluminum bar to retain better 'bite' at higher RPM than your proposed titanium bar at lower revs.

Your 1" backset on the shorter bar side is just about right for a 5500 RPM weapon speed and a 10 MPH closing rate. If you 'just miss' with the long side of the bar you'll have a penetration of 0.96" when the short end passes, allowing it to squeek by and give the long bar an impressive 1.92" bite. That's money!

Question #6: What is competitive joules per second in a 250/220 lb weight class today? These numbers way out strip your guidance in the spreadsheet instructions but you said earlier that that guidance was based on competition several years ago.

A: Everybody has a 'brag number' for their weapon, but a whole bunch of energy storage does little good if the weapon just skitters across the opponent and doesn't dig in. I don't know of any active 'bot that can claim an honest 100 Kj, and the best I can verify comes in at around 80 Kj. Half that figure is still plenty to be very dangerous.

Please note that the Spinner Spreadsheet makes no effort to adjust for aerodynamic drag, which can be substantial on a big, blunt weapon at several thousand RPM. You may not ever climb to the RPM you're counting on for those really big numbers. Regardless -- you've got plenty of energy storage.
 


 Lions and Tigers and Bars (oh my!)

Q: The longer reach isn't for intimidation. I can think of two situations it can help. First, a longer blade covers more area. So a robot trying to get around to the sides or away is easier to thwart with a larger blade. Second, against another horizontal spinner if the blades don't inter mesh, the longer blade will hit the opponent before the shorter opponent's blade.

Ray Billings described his blade selection this way:

"I usually have 2 or 3 different bars available. For this season I had a long aluminum bar with steel teeth, and medium length solid S7 tool steel bar, and a short thicker S7 bar. The long one is for anything that hits back horizontally (like say Shrederator) or where I think I might need the extra reach (Counter Revolution). The medium length bar is for wedges, since the end is angled to cut into armor. The short one is for drums and smaller verticals to supposedly be less likely to break. Supposedly"

I'm also a bit confused by your bite calculations. 5500 RPM works out to ~575 radians per second. (2 x pi())/(Radians/second x 1 (# teeth)) works out to 0.011 [second between impactor passages].

Using the acceleration calculator, I adjusted the A28-150 for current, Kt and RPM/V because the stored values don't match the specs on the new motors. Stall current of 287, 8.7 OZ-IN/amp, 156 RPM/V. I moved the drive motors forward to put more of the weight over the drive motors (~75%). That results in the CoF from 0.62 to 0.69 for a range of what I can really get. I selected an 11.64 gear ratio. 250 lb, 10 in wheels.

I get accelerations MPH from 4.3 to 4.97 for 1 foot up to 14.2 to 14.3 at 24 feet. Calculating inches per second of 4 MPH to 14 MPH, I get 70.4 to 246.4 inches/second. Multiplying those inches per second times the 0.011 and I get 0.77 inches to 2.7 inches with your 0.96 inches coming in at 5 mph. That corresponds to accelerating from about 1 to 2 feet away.

I'm now thinking that I need a 3 or 4 inch offset (which is trickier to design) to allow a bigger bite at higher speeds. From 8 feet away, I accelerate to 10.1 MPH which calls for a 2 inch tooth. The 1" offset only buys me the bigger hit area if I am very close (under 2 feet).

If you look at the 4500 RPM (blade at 2 seconds), the 3" offset makes more sense with the 1" only working at 1 foot acceleration distances.

If I run the numbers for the bigger blade, I'm getting d max numbers from ~1 inch up to 10 inches depending on the MPH. In short, I'm finding the 1" only making sense if I stop one foot away and then accelerate into the hit (not good tactics).

If I look at the Titanium blades with n=2 for the teeth, the 1" makes sense for higher RPM and < 2 foot acceleration runs.

Have I screwed up my math some how?

A. All of my calculations are guaranteed accurate or double your money back. Your error is missing that the 'short' blade must only clear the distance traveled forward in 1/2 a weapon revolution:

  • 5500 RPM = 91.67 revs per second
  • 1 / 91.67 = 0.011 second per revolution
  • 10 MPH = 176 inches per second
  • 176 inches per second * 0.011 second = 1.92 inches distance per full weapon revolution
  • At 1/2 revolution, the 'short' blade end need be no more than (1.92 / 2 =) 0.96" shorter than the 'long' blade to clear.

Q: All of this is leading me towards a titanium blade with 4" offset but no extra teeth. So I would add in a counter weight to compensate for the weight. This would cover the 2 second spin up RPM and beyond. Any thoughts on this?

A. I like an offset blade. Remember, the speed need to calculate for is the closing speed. If you're charging toward your opponent at 10 MPH and your opponent is so unwise as to be charging toward you at 10 MPH, that's a 20 MPH closing speed. Some things are in your control and some are not. More offset is good offset.

Q: When Ray Billings was asked about the weight of his blades, he said "67 to 75 pounds", presumably for the range of his short, thick S7 blade up to his longer aluminum with S7 teeth. My blades are coming in at 17 to 23 pounds, so Ray has a lot more material. The S7 blade would be in the neighborhood of 36" x 4" x 1.75". I'm going to have to recalculate with heavier blades which is going to mean slower spin up times (or find that room full of engineering students). I'm also going to run his motor into my spreadsheet to see what sort of spin up times he is getting (and thus if I am competitive).

A. Given that S7 steel is nearly three times as dense as aircraft aluminum alloys, I wouldn't be so sure about the long blade being the heaviest.

A heavier blade carries more stored energy at a lower speed -- you may just need to bump the belt reduction a bit to get the same energy and spin-up time as the lighter blade.

I've seen a lot of bots with good energy and spin-up times that weren't 'competitive'. Remember, it isn't all about the weapon.
 


 There Comes a Time...

Q: Thanks for the answers, I see where I misunderstood your 1" number.

I've run spread sheets for different blade configurations. I've added a column of joules per inches ^2. This is the number of joules generated in a given time (e.g. 2 seconds) x inches per MPH x height of blade. The inches per MPH uses 17.6 inches per second per MPH. So your closing speed is linear factor to this figure of merit.

I varied the width of the blade (down to the hole for the shaft being no more than 1/3 the with of the blade), the length of the blade, the number of teeth (1 by offset or 2) and the height of the blade, all while holding the aH consumed for thirteen 10 second spin ups. The constant number of spin ups is accommodated by varying the pulley ratio.

Results are that the ultimate max joules is dependent on amp hours available, not blade characteristics. Based on the acceleration calculator showing that with a fairly wide variety of CoF values I stay under 7.5 amp hours, I converted one of the two drive motor 44.4 V bank of batteries into two 22.2 V banks of batteries for the weapon. The result was the pulley ratios changed to accomplish the 13 spin ups and the max joules went up to ~83 Kj.

Doubling the voltage for the weapon to 44.4 V actually hurt performance because the amp hours available was cut in half. So with thirteen spin ups, the number of joules created went down (because the pulley had to be adjusted to provide that number of spin ups). Obviously if you count on fewer spin ups, then you can adjust the pulleys to get a higher energy storage level.

Adjusting the blade length while holding the width and height constant gave ever increasing 2 second joules values as the length of the blade decreased (i.e. the blade could spin up faster). However, the joules per square inch peaked. The point of the peak depended on the width of the blade, but while the joules are going up as the blade gets shorter, the RPMs to get those higher joules varying the pulley ratio to keep to 13 spin ups changes the RPM faster than the joules. A 3" wide, 1" high titanium blade peaks at 37" while a 4" wide, 1" high blade peaks at 34".

Varying the length, you can get to ~76 Kj in 2 seconds with a 30" x 3" x 1" titanium blade, but the inches per MPH for two teeth drops all the way down to 0.08. So you have to have a closing speed of 12 MPH to get one square inch of dmax (12 MPH x 0.08 inches/MPH x 1" high blade). At 12 MPH, you are applying those ~76Kj on that one square inch. At 6 MPH, you are applying the same ~76 Kj on 0.5 square inches.

In contrast, a 36" x 3" x 1" titanium two tooth blade is generating ~52 Kj of energy in 2 seconds, but the inches per MPH is 0.13. So somewhere between 7 and 8 MPH closing speed with generate the same square inch. A 12 MPH closing speed will generate almost 1.5 square inches of dmax but still only ~52 Kj of energy in 2 seconds.

This is telling me that a longer blade makes more sense as a "blunt weapon" because the energy is going to be less but applied to a larger area (all else being equal) while a shorter blade may be better brought to a point since the lower impact area any way. A point or chisel would apply the larger energy of the shorter blade onto an even smaller area, penetrating armor more easily.

Varying the height means falling joules at 2 seconds, longer times to reach max joules and increasing dmax impact area. A 36" x 3" x 4" high titanium two took blade generated ~10.5 Kj in 2 seconds, but has a 0.57 inches per MPH. So at the same 12 MPH, your dmax becomes almost 7 square inches.

Of course you eventually run into limits of how long, high, etc you can make the blade, how many battery banks you can have, etc. and still make weight. But I'm not really finding a sweet spot for the blade per se.

The pulley variation seems easy enough to achieve. The variable pulleys easily cover from > 1:1 down to less than 0.333 to 1.

The amount of offset needed varies with the inches per MPH. With the 0.08 inches per MPH, we could have the two bots crashing into each other at the ~14 MPH top speed (assume for both) and you would still only generate a bit over 1 square inches dmax.

Any ideas on how to optimize the impact area vs the energy stored? The optimization probably has to include the opponent. A faster, short blade with a chisel tip would make the most sense against a wedge (to pierce the front armor if that is all you can get at) while a longer, slower blade for a vertical spinner to get a larger piece of his blade impacted.

A. I'm worried about you, Tex. I'm picturing you next to a wall covered by charts, photos and drawings connected with lengths of brightly colored yarn.

Get some fresh air, eat a big hunk of red meat, build the robot, and go hit something with it.

A couple thoughts on some issues you've raised:

  • Piercing is bad. Piercing weapons get stuck. You don't need 70 Kj to pierce. Do not pierce. Hit them. Hit them hard.
  • Do not confuse 'bite' with 'square inches of impact area'. A rotary impact 1" tall by 3" deep is MUCH better than an impact 3" tall by 1" deep because a rotary impact is almost invariably a 'glancing' blow. The deeper the arc of the blow passes into your opponent's structure, the closer it gets to being a 'direct' impact. Deeper is better.

 
 Epilogue

It's been about a week since I last heard from Tex. I have, however, heard from a number of applicants for the second season of BattleBots on ABC about their rejection calls. One note was from a BattleBots veteran who's design was rejected for not having enough 'bling':

ABC was picking robots based on appearance. We got thrown into the bar spinner category, and Trey [Roski] said they had a ton of entries that looked exactly like Tombstone...

Given that Tex was pitching a Tombstone clone I suspect that his 'hook' wasn't tempting enough to get a positive response from BattleBots. I hope he builds a 'bot anyhow.
 



To see the most recent questions or to ask a new question, go to the


Copyright 2016 by Mark Joerger -- all rights reserved.