How I Get Industrial Robots for so Cheap

Robotics and Automation
4 followers
#machine #Robotics #robot
Timestamps:
00:00 Getting more robots
03:44 What to look for
10:28 What else you'll need
14:14 Getting a robot running
36:58 Possible issues
AI Generated·May contain errors
The biggest question I get on my videos is people asking where they can get their own robots for
cheap. Today you're going to find out what I look for when buying used machines, how to get them
running, and what can potentially go wrong. We're on our way to pick up more robots from a place in
Ohio, about a two hour drive. They had an online auction a few weeks ago that I was able to win
some robots from, a few more than I originally had planned on getting, so my friend is going
We were at Six Axis Solutions, an industrial robotics integrator. They set up and repair
robots for all sorts of companies. Any company can have an industrial auction that includes robots.
For example, an automotive plant may clear out all of their old machinery when switching to make a
newer vehicle. However, in this case, the company's entire thing is robots, so there was some pretty
cool stuff here. Most of the items from the auction are already gone. This room is full of
used robots that are still in good condition and will likely be put back into production. It's
crazy how many there are all in one place and how clean they are. Most of the robots I see for cheap
are completely coated in grease and dust. This is also my first time seeing the giant FANUC M2000
robots, which absolutely tower over everything else. Some of them are big enough to mount one
these little R2000s on the end of it. I did find it kind of funny that the motors on them aren't
much bigger than the motors on the R2000s. They just have two per joint instead of only one,
and two control cabinets to fit all of the servo drives. This place has a lobby that matches the
robot theme. Lots of robots on display, along with one of the truest sayings I've ever read,
and a flamethrower behind the desk. There's also an outdoor area full of older robots.
This is the kind of condition I'm more familiar with. They're the ones you can often find for
cheap. If you're looking for a robot or need some integration work done, it might be worth
reaching out to them. There's a link to their website in the description. Now the fun part
begins. I need to get all the robots I won from the auction back home. My dad and I are in the
truck with the bigger trailer taking the majority of the stuff. My friend has the rest in his car
and U-Haul trailer, which is pretty hilarious. Running out of space was the only reason I ended
up not getting even more. Most of the robots I got were sold as parts robots, so they're missing
pieces and probably had something wrong with them to start with. That was the only way I was able to
afford so many. Usually these small robots are very expensive since everyone wants them. They're
the easiest to use and move. It's kind of ironic that the big robots, weighing almost 3,000 pounds,
are almost always cheaper to get second hand than the tabletop robots, even if they include
controllers. No one wants to deal with moving the big robots around. As I take a look through these
robots, I think now is a good time to talk about what I look for when buying used industrial items.
I'm largely going to be talking about industrial robots, but most of the information and things to
watch out for are the same for other industrial machines as well. Industrial auctions are a good
way to get cheap machines, but aren't always super easy to deal with. They're normally run by an
online auction house that you need to sign up and verify with before the auction starts. Then there's
a couple days where you're allowed to look at the items in person to see their condition. Auctions
charge a premium on top of the final item price, usually around 15 or 20 percent, and then an
additional loading fee or even require you to pay a certified rigging company to come and load your
items, which will almost certainly cost more than the robot if you're looking for a good deal. There
are also industrial surplus stores that buy and resell industrial machinery, which are often easier
to deal with than auctions. I've put links in the description to some of the places I've got machines
from. If you get anything, be sure to tell them I sent you. It helps increase the chance of me
being able to work with them in the future. Also, places like eBay or Facebook Marketplace will
sometimes have good deals. However, by far the most powerful source of used industrial goodies
is knowing someone who happens to be getting rid of stuff. More often than you'd think, robots get
sent straight to the scrapyard, so asking around and knowing the right people can get you some
robots for cheap or completely free if you're lucky. So how do you know what's a good deal?
Well, a lot of it comes with experience. I'll put how much I spent on all the different robots and
machines I've bought up on the screen to give you a better idea as they appear. First up is determining
if it seems like it's in good enough condition or not, and includes all the required parts. For a
fully operational robot, the biggest things are the robot itself, the controller, the teach pendant,
and the cables to go between. If you only want the robot and plan on replacing the motors with your
own or using something like the open source robot controller I'm building, then you only need the
robot and maybe cables if you can't find the mating connectors for sale somewhere. Robots without
controllers will often be substantially cheaper than ones with, so they're what I usually get.
Only one of the robots I got from the auction came with its original controller. Make sure
everything looks intact. No missing motors, encoders, smashed connectors, or cut cables.
I've seen a lot of damaged connectors from something bumping into the robot, or cut cables
from people removing the robots from wherever they were installed. Apparently the five seconds
to unplug the cable is too long compared to the three seconds to cut it off. Tags or writing on
the machine can be super useful. Many companies will tag faulty equipment and write the reason.
Might be as simple as NFG, or maybe a bit more colorful and tell you what the actual issue is.
Open or missing access covers are often not a good sign. Chances are someone was in there trying to
I always check extra close when I see something like that, and keep it in mind if I'm trying to
troubleshoot an issue later on. If you're looking for a robot and controller pair, check that the
numbers on the controller and robot match. Controllers are usually designed to run multiple
robots using the same hardware with different software settings, but these settings aren't
always easy to change or in the standard documentation. I didn't check the numbers
matched on my Fanuc R2000 I bought years ago and ended up getting a controller from a different
model robot. I was unable to get it running with that controller, which was what started the whole
open source robot controller project. Supposedly you can change settings deep in the controller to
get it working, but I only had people tell me years after I had already sold the original
controller and started making my own. I'm still making progress on the new version of my servo
drives that are capable of running three motors from one board. Here the motors are running in
open loop mode to test the drive itself and help me spot issues before ordering the next batch.
Be sure to subscribe so you don't miss the full power testing. Thanks to PCBWay for helping
sponsor these projects. They provide manufacturing and assembly services for custom PCBs, as well as
custom laser cut and bent sheet metal or CNC machined parts. They made all of the circuit
boards and sheet metal enclosures for my robot controller parts, saving me a ton of time. The
200 people who use the code PCBWay-ExcessiveOverkill10 can get $10 off orders over $30.
You can apply the code at checkout, or add it to your account to save for later.
The discount code and link to sign up are in the description below. While testing my drives,
I found I managed to flip some of the connections on all of the CurrentSense amplifiers and had to
do a lot of tedious bodges. I already have four more of these boards which are now non-functional
in their current state, so they're going to be on my store as leftovers if anyone wants one.
Not sure what you use them for, but they aren't doing me any good. My store is linked in the
description if you'd like to check it out. The pre-order for the development edition
light-up badges is also on there, and it's still open for another few days before I start
manufacturing. If you want to get your hands on one, now's your chance. I'm only doing a
limited run for this edition, so once they're gone, they're gone. Anyway, back to the video.
Since my full controller isn't ready for sale yet, the best way to get a robot up and running is with
You should check inside of the controller to see if there are loose hanging connectors or empty
slots, indicating a board might be missing. It's common for old equipment to have parts stolen to
keep other machines up and running. Sometimes there are intentionally spare connectors or slots
though, so it's really best to compare to the manual. The manual is probably the most important
thing to find before you make a purchase. Maintenance manuals are incredibly valuable
when it comes to getting old stuff like this up and running. Certain machines will have the manual
or at least electrical schematics stored inside a control panel. Unfortunately, that's normally not
the case with robots, and many brands like Fanuc don't make manuals available to just anyone.
You need to be a company with money to throw at them, so you might have to search online pretty
hard to find any or rely on forms to read solutions to common problems. There are usually manuals for
controller programming, controller maintenance, and robot maintenance. Finding all three is ideal,
but the controller manuals can get you almost everything needed. I don't have the robot
maintenance manuals for any of my robots. Now that you know what I watch out for, there's a
couple things that aren't as big of an issue as you might think, such as if a robot is absolutely
covered in grease and dirt. Sure, it might mean that the previous owner's maintenance program was
a bit lacking, but there are some applications where robots will be absolutely covered in dirt
after just one day, and they're built to handle it. Most robots are pretty well sealed up and
be perfectly fine with light water like from rain. Sitting outside for a while probably isn't
going to hurt anything as long as all the parts are still attached and sealed up properly.
If you see a machine that looks pretty bad, keep in mind other people are probably thinking the
same, so it could be a good chance to get it for way cheaper than usual, if you are willing to
maybe fix some problems or spend more time cleaning it up that is. If the machine itself
looks good, or good enough for the price, there are still a few other things to consider on your
end. Industrial machines are big and heavy, so they aren't easy to move. However, I've yet to
find anything that can't be moved with enough friends armed with crowbars and steel pins.
Make sure you have a plan for how to safely get it where it needs to be. I highly recommend robots
with bases already attached. They make it much more stable when sitting there and give you better
mounting holes for bolting it down, which you definitely should do. Robots are very dynamic.
The majority of the loads their mounting bolts need to handle are from the robot moving,
not just the static load of it holding something still. Even if a robot seems stable when moving
slowly, it may jump all over the place when running at full speed. You'll want to make it
sturdier than you think and carefully test it up to full speed to see how it handles.
You'll also need to figure out how to power the robot. Most industrial machines want three-phase
input power, which isn't commonly run to homes in the US. Some countries do have three-phase,
which is convenient, but you still might need a transformer to get the correct voltage.
I live in the US and have standard 220 volts run to the house, so I built what is called a rotary
phase converter. At its core, it's a normal three-phase motor that is spun up and powered
off of only single phase. Once it's running, it generates the remaining phases, although typically
not super well, so some capacitors are often added to help balance out the voltages. These systems
are common and pretty easy to build. You can get the parts for quite cheap if you're already looking
at industrial surplus. However, that only gets me three-phase at 220 volts. Some machines want
480 volts, or even 560 volts. To get those voltages, I have a couple three-phase step-up
transformers, also bought as industrial surplus. The actual total power these robots need might
be less than you expect. The supply requirements will usually be stamped on the controller or in
the manual, but the average power is often significantly less. Robots are very peaky loads,
so while moving slowly or stationary, they barely draw any power. But when doing fast moves, they can
easily draw 10 or 20 times more for brief periods. My big robots are rated to only draw around two
and a half kilowatts on average. Smaller robots will use even less, as low as only a few hundred
watts on average. You'll also need to plan out some safety setup. Unless you happen to find one
of the fancy collaborative robots meant to work around people, any industrial robot you find will
probably be capable of causing great injury to say the least. It might vary from breaking your
hand to throwing you through a wall, but either way, you should stay out of reach and always be
aware of what the robot is doing. Robots usually have teach pendants with special safety rated
switches you must hold down to enable motion. If you let go, the robot will stop immediately.
They also have an e-stop button on the pendant, and terminals in the controller for you to attach
your own additional e-stop buttons, or safety sensors to tell the robot to stop if someone
gets too close. Should you set up e-stop buttons and a whole safety enclosure around your robot?
Yes. Do I? No, at least not usually. I almost exclusively run my robots in teach mode, which
requires me to be holding the pendant for the robot to move. These robots can also run in full
auto mode as well, which doesn't require you to hold the pendant. This is how robots run in
production. Teach mode means I am at all times able to stop the robot in under a second by letting go
of the safety switches or pressing the e-stop button, and I am paying attention since I know
the robot is going to move. Then when I'm working around the robot and not holding the pendant,
I have the e-stop button pressed or the entire robot switched off for extra insurance.
If you choose to run in full auto to automate something, I highly recommend at the minimum
painting the kill zone around the robot and setting up some barriers with e-stops around the room,
especially if the area can be accessed by anyone who might not know the dangers of the robot.
Ideally, the robot would be totally enclosed with safety scanners or light curtains to detect if
someone was in the danger area, which would then prevent the robot from running. If you're going
to buy e-stop buttons or safety devices, buy them from an actual automation supplier like
AutomationDirect. And not just any old button, get the ones meant to be e-stop buttons, and for sure
not Amazon or AliExpress grade ones. I've seen way too many of those fail for me to bet my life on
Once your safety setup is complete, test it thoroughly and test it regularly. Wires can
get damaged or shorted together, so it's a good idea to make sure everything is still in working
order periodically. With that out of the way, I'm going to try to get the one robot that came with
a controller up and running. This next little portion is unscripted, so cut me some slack if
I misspeak. First things first is getting this thing mounted so if it moves it's not going to
fall off the table. Even though these are relatively small and relatively lightweight,
I believe this weighs around 70 pounds, it is still well within its capability to break your
finger or worse. I threw together this little wooden platform that I can just screw the robot
base to and then bolt down to the metal table underneath so nothing's going to slide around or
move once this is running. Now don't be deceived by the 220 volt sticker on the front. This is in
fact a three-phase controller unfortunately. Our main 220 volt three-phase comes into the top of
this disconnect which is accessible from the front of the controller and basically turns on and off
the main power to everything. After that it runs down to this relay here which is controlled by
the e-stop board. This is the safety system for the entire controller and robot which makes it so
if you press the e-stop button on the front of the controller or on the teach pendant it shuts down
power to everything so there's no chance of it continuing to run. It also has terminals that you
can hook up to external e-stop buttons such as whatever the robot ends up actually being built
into. You typically want to have e-stops anywhere someone could potentially be within reach.
It also has fence inputs which are so there can be doors around the robot and when those doors
are open they will break a contact and then trigger the robot to stop as well. Now in our case
someone has already conveniently jumped over all of those safety inputs so we should be good to go
right from the start and not need to mess with any of that stuff. After this relay the power goes into
the bottom of this module on the left. This converts the 220 volts three-phase into roughly 300 volts
dc up at the top. After that the dc power gets jumped to these two servo drives here which are
responsible for actually running the motors in the robot. The motors in there I believe are 200
watt servos which are very small when it comes to industrial robots. This larger drive runs
three servos and this smaller drive handles the remaining two. The module here is the main cpu
which actually handles all of the motion control, running the programs, connects up to the teach
basically any of the actual processing happens on here. The servo drives themselves are extremely
dumb. They don't do much of anything at all other than exactly what the cpu is telling the motors
directly. The large connector here is the encoder feedback from the robot. Each of the encoders has
I believe two twisted pairs that run back for data. One of them is the request signal sent from
the controller to the encoder and then the other pair is the data being sent back from the encoder
to the controller. I believe this also connects to some of the additional kind of auxiliary
controls on the robot like the aerosolenoids built into it and the connector near the wrist on the
end of the robot. This cable here is the teach pendant. This controller did not come with the
teach pendant unfortunately but luckily this is an RJ2 controller which is the same variant as
the robot I have in the back there so I can just take it from my other controller and bring it here
to test with. The second cable that comes back from the robot is actually hidden underneath and
runs to the bottom of these two servo drives which is where all the motor connections actually go.
So after having a good look over everything I'm not seeing any loose connections or any of these
modules that have gotten popped out of the rack they're supposed to be seated in so I think we
should be good to go. We can get power applied, get these cables hooked up to the robot, and see if
this is going to move at all. The main reason I actually bought this robot was because I wanted
these cables. I actually already have I think three other of this exact same robot model. One is
actually attached to this sliding rail up here on this table. So I was never able to find the
cables for these robots but now I have them which means I can figure out what the numbers are on
them and get some more ordered. Hopefully if they are even still manufactured or there's new old
stock of them somewhere. Very satisfying. I drilled holes in the corner of this piece of plywood so I
should be able to bolt this down to this actual table so the robot isn't going to slide all over
the place if we are actually able to get it running at full speed. All right, nice and snug.
Get the teach pendant, get three-phase connected to that, and we should be good to try it out. Here's
our teach pendant from my other controller. Hopefully there's nothing like horrifically
wrong with this that ends up damaging this teach pendant because I don't think I have any extras.
All right, I guess nothing to it but to turn it on and hope for the best. All right, it's on.
What do we got on the pendant? Diesel alarms which means the encoder batteries in the robot
itself are dead which is not surprising at all. It's not really a major error. Pretty common on
a robot once it sits for long enough. So the controller runs. So far everything seems like
it's actually fully functional. The issue I'm having now is all these errors that are still
set on the controller. The vast majority of them, or I believe actually all of them, are related to
the batteries in the robot being dead. These robots actually have batteries inside of them
that power the encoders on the back of all the motors. So whenever the controller is disconnected
and normal power is removed from the robot, the encoders will actually stay powered and retain
the position information to show the controller whenever it's reconnected and powered back on
what position the arm is in. That way you don't need to recalibrate the whole robot every time
you turn it on and off which is super convenient. The downside is those batteries obviously do not
last forever and whenever they die you get a whole bunch of errors and the controller basically
refuses to do anything. There are switches on the back of the pendant that need to be held in
for the robot to be able to move. You don't need to hold in both switches but you need to hold in
at least one and it's meant so that you really need two hands to be holding this thing if you
want to move the robot because you need to hold in one of the switches on the back and press shift
and press those jog buttons down there which are now out of reach. So you might be able to do this
with my other hand. Do some trickery here. I'm sure this is how they intended it to work.
Anyway you can see the error messages we get whenever I try to jog. Enable input is off.
Now I'm not sure where that enable input is. Okay according to the internet that
MCTL-013 enable input is off is related to some sort of software configuration in the
controller itself. So I'm seeing a few different options here. One of them is this system variable.
These buttons are not super functional. It's called op work. UOP disable.
Pretty close. Up work. UOP disable. One. All right. We'll hit jog. Error status. Reset. Try again.
No. Still won't work. I think we need to replace those encoder batteries.
But at least we fixed the enable issue. That is progress. So I'm looking for the batteries
here in the robot and I kind of assumed they'd be there in the base since it's the easiest to get to
but they are in fact up here conveniently covered by the top part of the arm. I have
no idea how you're supposed to get at those if the arm happened to be in an even worse position
where it was leaned back and there was almost no room above this cover. But there are two batteries
in there. They are these weird little industrial grade lithium cells or something. I'm definitely
not going to have any of these laying around. So I guess I should probably get the correct
replacements but in the meantime I think I'm going to find some other batteries I can plug in there
and get this thing moving. Trying to test which of these pins are connected to see if these
batteries are in series to get six volt or not. Yes they are. Okay that makes this easy.
Then I can just put six volts and jump together these two inner terminals.
Well I guess they're already jumped together. I just need to put six volts on this pin to that
pin and since the connectors are polarized I know which is positive and negative already.
All right we now have our replacement batteries. We need four of those since they're one and a
half volts each. These original ones were three volts. They're just wired in series in the robot
and to hold them I unfortunately did not have a single four cell battery holder. So I have a
3-cell and a 1-cell soldered together with the connectors. I cut off these old batteries where
they should be so we should be able to plug all this in and get it to fit. And then it goes here
goes here. I'll leave that cover open for now. Whoops. Well in my haste to plug everything in
and try this out, I accidentally swapped the connectors whenever I plugged them in.
Which instead of the six volts being on the outer pins going to the encoders, the six volts was on
the inner pins which that little circuit board shorts together. So these batteries were just
dead short and it actually ended up getting hot enough that it melted one of the contacts out of
this little battery holder. I could smell it and immediately knew I messed up so I unplugged it
right away. But that battery holder is already toast. All right so after that I went for the
much simpler and a bit more reliable solution. You just have a power supply hooked up there now.
Now if I shut that off the robot's going to forget where it is and lose all the encoder
position information. But just for testing that is perfectly adequate for now. Yep we only have
the pulse not established errors which are in our case acceptable because I haven't moved it yet.
Now comes the moment where I can try moving this thing. It is not calibrated yet so
none of the soft limits are going to work. I'm going to need to be careful.
Oh it's on. I hear the servo drives whistling away. Now I can try to move each joint very slowly.
All right so would you look at that they all move. Disturbance excess that's not a good sign.
Which joint is that? Five? Does it come back? Could have been if it just sat for quite a long
time. It was a bit stiff to start. But it very well could be that one or more of the harmonic
drives that run each joint could be in not super great condition. All right next up is going to be
getting all the joints recalibrated so it is mastered as Fanuc calls it. And to do that all
need to do is line up all these little witness marks. There's one on each joint they all need
lined up to the zero position and then we can calibrate it in the teach pendant and we should
be able to get into actually running programs at full speed and also doing xyz jogging. All right
so I just finished mastering all the joints and everything is saved and updated. Should be good
to go. So previously we could only jog in joint mode which is where we were controlling each
joint individually. What it being all calibrated now will let us do is jog in xyz mode. There's
actually several different modes but they all follow the same sort of general idea where we're
controlling the actual end point of the wrist instead of individual joints. So we'll be able
to jog the end point of the wrist in straight lines relative to the overall world rather than
just the joints on the robot. Basically all of the joints are going to move together to maintain a
straight path for the end point of the wrist to move in. Now we're going to start this off nice
and easy at 10% jog speed and we're going to try moving in z first. Yeah z down. Seems right.
Yeah it looks straight to me. All right then we have x which is forward and backward
and then y which is side to side and then we can try going a bit quicker. We'll go up to 50%.
It's actually pretty quick. Now this speed is of course nowhere near the actual max speed of the
robot or the joints. In any jog mode you're in the joint speed or at least the linear move speed
is typically limited quite a bit below the actual maximum speed of the joint and the robot. That way
whenever you're jogging around you kind of just you kind of need it to be slower to have the
proper control. Once a program is written and the points are all predefined you can crank up the
speed to the maximum and we'll be able to see how quick the robot can actually move then.
All right now that the robot is basically completely up and running we have it calibrated
moving in xyz position as well as joint positions we should be ready to finally write some programs
and then see this thing run at full speed. Programming these robots is incredibly simple.
The main way is by jogging the robot to whatever positions you want it to be in and then it'll
basically play them back in whatever order you specify in the teach pendant and that'll let us
run it at full speed compared to the limited speed we have now when we're just jogging.
So I'm just going to write a simple test program exercise the joints a little bit move around
basically not do anything but just see how fast this thing can actually go. So I wrote a quick
little test program it should basically just exercise each joint and then do a couple linear
moves. We are going to start it at 25% which I think is pretty reasonable. The joint moves are
all set to 100% speed and the linear moves are set to 500 millimeters per second which I think
isn't quite full speed for this robot. I think full speed might be like a meter a second but the joints
are going to move full speed regardless for the normal joint moves. So we can try this out and
see how fast it can go. All right well that all seems good. Should we do 50? Nah we'll just go
right to 100. Hopefully nothing has any problems. Maximum speed. Pretty quick. Pretty quick.
I'm excited to see how quick the newer versions of this one will be able to go. I assume they're
quicker than this. This is one of Fanuc's very first LR-Mates I believe. From what year is this?
Oh wow this is actually a pretty modern one. This is from the year 2000. I think I have some that
are from like 1997. This one up here is 1998 so pretty old. The newer ones down there and they
have some really new ones that are still current. They'll be six axis and everything so very excited
to try to get those working eventually. I need the open source robot controller for them though.
They did not come with any of their own original controllers so that'll be a project for another
day but eventually I'll get some of those running up at full speed just like this one is. The dead
batteries are basically the standard issue with Fanuc robots after they sit for a while. Not at
all a big issue to fix. There are a few other issues that are also usually easy fixes but some
can be huge problems. If you have issues with the controller the first thing is to check what the
errors mean in the controller manual. It usually has great instructions for how to go through and
Often there might be fuses blown due to improper or damaged wiring. Most controllers have a fuse
scattered about for control power and logic. That's one of the first places to look if you
are getting really bad sounding errors or it won't power on at all. Of course you should
have a look to see why the fuse blew in the first place. Could be bad wiring, could be a bad part,
could be something that used to be connected had failed and is now gone but the fuse was never
replaced. Just make sure you aren't blindly replacing fuses without knowing the original
issue. You could potentially cause even more damage. There's also the signature culprit in
all electrical systems. Bad connections. Always check connections first if there's an issue.
The contacts themselves could be dirty, the connector might not be fully seated,
maybe a solder joint has cracked. I've seen cables and airlines in robots completely worn
through after thousands of movement cycles or small nicks in cables that cause intermittent
connections or shorts. The list is endless. What I'm getting at is verify your connections first
before you start tearing everything apart. The other common issues I've seen that are less easy
to fix are the mechanics of the arm itself. Chances are any arm you find will have AC
servo motors in it, unless it's very old in which case it may have DC brushed servo motors. Basic
stepper motors are uncommon in industrial grade machines since they have much lower power density
than those other motors. AC servos are typically reliable but can still fail as you'll see in a
I have also seen the brakes attached to the motors fail. Most robot servos have spring-loaded
brakes attached to their motors that require power to unlock. Robot joints are usually back
drivable so if the motor isn't being powered to hold its position the arm would flop down.
The brakes hold the motor in place by default with no power so the arm doesn't fall. The brakes are
sized to provide a bit above the continuous torque output of the motor so if the robot
loses power while moving and the brakes slam on the stopping torque is still within the acceptable
range. The brakes shouldn't damage anything if they're needed to stop the robot from full speed
however during normal use they only ever hold the motor stationary and aren't actually handling
slowing down the motor. The normally locked brakes introduce a couple important considerations.
If you manually send power to the brakes to release them or remove a motor to work on it
the arm may fall or in some cases if the arm has a counterweight or big counterbalance spring
it might instead rise. I've heard stories about people removing a motor on one of these big robots
and sending the end up into the ceiling from the counterweight falling. I generally short the motor
phases together if I want to release the brake on a joint. It makes the motor stiff to turn which
prevents the joint from moving fast and keeps everything nice and slow. I got a box of assorted
robot parts from the auction including a couple motors that say they have bad brakes. On a good
motor and brake the shaft will normally be locked in place. A small amount of backlash is fine. Then
when power is applied to the brake it'll be free to rotate. On the bad motor the brake still draws
power and even makes the click sound like it's releasing but the shaft stays locked in place.
There is also no backlash at all unlike the good motor. I decided to take the motor apart to try
to find what the actual failure reason was. It turns out the brake itself is fine. It releases
and moves freely like it should. The real issue is the rotor itself has come apart. The magnetic
dust covering it caused it to jam up against the inside of the stator and prevent it from turning.
This is unfortunately unrepairable. Even if I cleaned off all the dust and reassembled the motor
the magnetic properties of the rotor have changed and the motor won't perform as well as it should.
Not to mention I also managed to scratch the encoder code wheel while I was trying to get it
off the shaft it was glued to. These small motors are not at all easily serviceable. If only the
brake failed it could possibly be fixed or just totally removed. Then the motor could still be
used for something else as long as it being allowed to rotate when powered off is okay.
One of the worst problems to have with these robots is the gear reducers themselves failing.
They're what connect each joint to the next. They contain a set of bearings keeping the joints
together and also some sort of gear reduction that lets the relatively fast servo motors drive
the relatively slow joint with very high precision. Although as you've seen relatively slow can still
be pretty fast. After years of use they can end up getting sort of crunchy and stiff to turn or
start to have more backlash than is acceptable. Not only are the replacements expensive and hard
to find, they're a huge mess to replace since they're full of grease. Even the grease is super
expensive and specialized for each robot and reducer type. On the robot I just got running
you may have noticed how loud and squeaky the joints sounded when they moved at full speed.
You might think the gear reducers or motors are shot but that high pitch sound is mainly from
the servo drives running the motors. Older robots typically have servo drives with a lower switching
frequency which makes them sound louder. So don't be immediately alarmed by the high pitched squeal
the joints make when moving. There are two common types of gear reducers used in industrial robots.
Small robots use harmonic drives, also called strain wave gear reducers. The one I have here
is a brand new frameless harmonic drive. The ones in robots are usually a bigger integrated unit
including bearings and seals to keep the grease in. This one is only the actual reducer. It's meant
to be put in a machined housing with bearings and seals added separately. They have a thin
flexible inner spline and rigid outer spline that has two more teeth than the inner one.
Then a bearing fits snugly inside the flexible spline. Crucially this bearing isn't a perfect
circle, it's actually a slight ellipse, bending the flex spline so it meshes with the external
spline at two sides. The input to the reducer is the center of that bearing. When it rotates it moves
where the splines are meshed together. The output is between the inner and outer splines, which would
be connected to each side of the robot joint in this case. Since the outer spline has two more
teeth than the inner spline, each time the input rotates one full rotation, the inner and outer
spline move by two teeth as well, resulting in a total reduction between 50 to 200 to 1 typically.
These reducers don't always have a graceful failure where they slowly wear out however.
This robot has been through some sort of substantial crash. The J2 reducer is completely
destroyed and the joint is free to move. The J3 reducer is still at least partially intact,
but it's stiff to move and not at all smooth like it should be. After pulling it apart,
it's clear this joint has also suffered a crash. The flex spline has started to twist and buckle
in on itself. Those bumps in the flex spline aren't supposed to be there. It should be perfectly round
like the new one. This kind of failure requires the reducer to be replaced. There's no repairing
the flex spline. These were listed as parts robots like I mentioned before, so I will eventually tear
apart all of them and see what parts are good and what needs to be replaced, then build as many
complete robots from the good parts as I can. The remaining good parts will either be used for other
projects, kept as spares, or put on eBay to help fund future projects. The bigger robots don't just
use giant harmonic drives. Instead, they use something called radial vector reducers, or RV
reducers for short. I'm not going to take apart one of my big robots to show you a reducer, but
luckily I got some brand new RV reducers as industrial surplus years ago and haven't used
them for anything yet. This is a relatively big one. All of the joints on robots this size typically
use RV reducers. My UP6 has small RV reducers on joints 1, 2, and 3, with the rest being small
harmonic drives. I think harmonic drives are easier to manufacture once the joints need to
be so small. I even have some insanely small ones about the size of a quarter. RV reducers have
cyclotal disks surrounded by pins they mesh into, with the count being off by one between them.
Since the lobes are so much bigger than the teeth on the splines and harmonic drives,
there aren't as many and the reduction ratio is lower. To help with this, there's an additional
reduction between the motor and the cyclotal disk through standard spur gears that brings the
reduction back to a similar range as harmonic drives. Now don't be fooled into thinking this
is a planetary gear reduction. There is no outer ring gear that the larger spur gears are running
against. Instead, they're attached to offset crank shafts that drive the cyclotal disks.
This method of driving the disks from multiple points with much larger bearings is what separates
RV reducers from plain cyclotal reducers, which are usually only driven from a single offset
crank in the center. The pins on the outside are also able to rotate, which changes the bearing
or wear surface from the small point where the cyclotal disk contacts the pin to the large
cylindrical surface the pin is seated against, reducing friction and greatly improving the life
of the reducer. The bearings supporting those crankshafts might seem oversized for handling
the loads from those relatively small input gears, and that's because they are. Those crankshafts and
bearings are what carry the full output torque from the cyclotal disks to the joint the reducer
is driving. There are two cyclotal disks in this reducer. If you look past the top disk, you can see
a second disk moving around underneath. Technically, only one disk would still work. The reason there
are two disks is for balance and additional rigidity. A single disk would create quite a
bit of vibration as it oscillates since it's off-center. The second disk is designed to move
in the opposite direction as the first, so the vibrations cancel out. This also means the loads
on the crankshafts from the cyclotal disks are balanced. Each shaft is pushing the disks in
opposite directions, and the disks mesh on opposite sides of the outer housing. That way, the
oscillating load from the disks mostly stays contained to the crankshafts, disks, and housing,
rather than needing to go through all of the other bearings to keep the disks meshed with the outer
pins. I'll eventually work on getting some of these other small robots up and running, maybe make
another video out of it. But first, I need to clear out space in the garage to work on the mobile
robot arm, so into storage they go. If you found this information helpful or want to keep up with
my projects, be sure to like and subscribe. Thanks for watching!
