Team Kaizen Build Diary

[Ocracoke]

Thx. Yeah, it was. It was a pain to keep running at all and as much as I like the machine, it really does need a redesign to address a lot of the problems with it. I'll keep this version around for now but I am hoping to have a new shell and some various upgrades to what is there for EXR Cheltenham.

[Ocracoke]

Antweights
With BotFest 3 in September, I've got a bit of work to do in getting the Ants I have ready to go. Typically, with the Ants I have for BotFest, I loan these out to a different driver because I feel that me competing in my own event is a bit contentious plus, I am normally running the main bulk of the show on my own so I need to run the administrative stuff as well. However, since BotFest 2, my Ants haven't seen any action in an arena.

Amai is running fine. It needs the shell welding/gluing back together in places but otherwise is pretty much ready to do. I did have to swap the battery out with a different one as I'll explain later. I finally picked up a DSM2 controller so Ramu will finally see some action in an arena now! I'll need to order a new LiPo for it as well but it will run again. The famous Skittles robot Osu has been retired from combat but I'll still use it in games and such. I need to check the LiFe battery in it but I suspect it is fine.

Ikari, on the other hand, was dead by the end of BotFest 2 and it went into that event very ill with a faulty motor and/or speed controller so it needed rebuilding. After salvaging some wood for a frame from the BotFest arena, picking up a NanoTwo LT and some new motors, Ikari 3.0 is born. It still needs painting and a bit more reinforcement but it is all there, ready to go. Because of the space conditions, I had to use the battery from Amai but in reality, this is not a big deal. It is quite quick and features a lifting arm mounted to the bottom of the robot to jack it up to act as a lifter (kind of like how Shu! worked).

Apologies for the rubbish photo, the lighting in my office is not great at night.

20190713_003537.jpg

Next thing to do is get a new arena floor cut, painted and effects added.

[Ocracoke]

Azriel
I've decided to take the plunge to refit Azriel with brushless drive following a successful conversion I did a few months back. I was planning to refit the axe motor with a brushless motor as well but I think that would require a bit more work to make it work correctly so for the moment, drive only. I'll be using XK-3674 1650kV inrunner motors in this one, which is a little narrower and longer than I would like but that is what I have so it'll do. I've also picked up another 3S LiFe battery so it'll be running at 6S LiFe.

If it turns out to be too fast on the 100mm wheels, I'll put this one on 75mm wheels because one of the things I liked about Azriel at EXR Bolton was the controllability of the machine. It'll always be a case of finding a balance I guess. By my calculation, the new motor/gearbox setup should be spitting out ~2300RPM on the output shaft ((19.8v * 1650) / 13.7) at full power which I think is ~27mph on the 100mm wheels and ~20 on the 75mm wheels. Might need to configure some limiting as well :P

Ramu
This is the Shakey robot I bought last year from Greg which until recently, I've not been able to run because I didn't have a DSM2 controller (thanks Euan!). It is a simple rambot and it does need a little bit of attention but it should still functionally work. There are little things I want to do to it, like paint the thing and perhaps remake the front wedge on it but otherwise, it'll go again. A new battery has been ordered for it.

Jibril
The designs for Jibril 2.0 are nearly done, down to picking the materials now and will be ordering the materials in one go at the end of the month. I'll be reusing most of the internal components out of Jibril 1.0x but replacing things like the silver steel shafts with hardened ones in places and locking down the gears on the axe gearbox properly this time because the Scythe has fired, in the 15.75:1 gearbox configuration, a magnificent total of one time during the entire weekend. I may also end up rewiring the thing because right now, the wiring loom in there is a bit pick'n'mix and as EXR Bolton proved, the soldering is a bit rubbish in places.

[McMullet]

Is Ikari made of wood, or are my eyes messing with me? I'd really like to make an ant at some point.

I'll be interested to follow your brushless conversion on Azriel. I find the sheer range of BL motors a bit baffling.

[Ocracoke]

Ikari really is made of wood and polycarb. All of my ants feature using it in various levels. Azriel's conversion is going to be interesting, in theory. In practice, who knows? It should be fine.

[Ocracoke]

Azriel
So I thought I would go through the various stages of doing a brushless conversion. As a forewarning, I am not claiming in anyway that this is the best or ideal way to do this. This is simply the method I am using, your mileage may vary etc. Anyway, today I did the physical conversion and later on in the week/weekend, I'll do the brushless speed controller programming.

To give an overview of the things typically listed on a brushless motor, this motor is listed as a Turnigy XK3674 1650kv motor. That means that the motor diameter is 36mm and the length is 74mm and, for every volt applied to this motor when under no load, the motor will spin 1650rpm as I understand it. Originally, I chose this motor when I thought about doing a brushless conversion on Kaizen at the time. As a result, this motor is specced to run at 8S LiPo, meaning a no load speed of (3.7 * 8S) * 1650 = 48,840 rpm with a theoretical max current of 70A. For the budget I had at the time, it seemed to be the best balance of output speed and power I could find. Wherever I'll regret this later on remains to be seen.

Here are most of the parts I am using. From left to right, the Turnigy XK3674 1650kv motor, 2x M3x10mm bolts and a Gimson 13.7:1 motor with the 775 motor attached. The only thing not pictured is a 17T 0.6Mod pinion gear but as that is already attached to the brushless motor.
20190724_182401.jpg

First thing I needed to do was cut down the M3 bolts a bit as the depth of the holes in the motor is quite shallow. If you are doing this, I'd recommend the 6mm bolts, I think they'll fit nicely. The next thing to do is shave the shaft off the motor with the pinion gear attached. The easiest method I found to do his is to get the pinion gear on the shaft as far back as the D shape goes and then cut off the excess. Not doing this will cause the shaft to foul in the gearbox.

Before:
20190724_182747.jpg

After:
20190724_183117.jpg

The next thing to do is get the 775 motor off. To do this, lay the motor on the ground and with a suitable allen key, undo the four bolts on the gearbox. These run the length of the gearbox to connect to an adapter plate between the motor and the gearbox. Make sure you don't, whatever you do, allow the assembly to be vertical as it will all fall apart and will require greasing and cleaning before reassembly.

Once the 4 bolts are out, pull the motor away from the gearbox, which will reveal the plate. Undo the 2x M4 bolts with the allen key.
20190724_183339.jpg

Once the M4 bolts are undone, get the M3 bolts and place one in the adapter plate and then get the brushless motor and get that bolt in. Once done, swing the plate a little until the opposite bolt hole is lined up and then put the second bolt in. Note that not all brushless motors will fit this (under 50mm diameter I think should) and even then, not all motors have the right M3 or M4 spacing for this plate.

20190724_183624.jpg

Once the plate is on, get the motor back into the gearbox, which will require a bit of patience to get it seated properly, especially given that the gearbox will want to seperate apart. I try using a pair of fingers around the face of the gearbox to the shaft to hold it in place as I am reinserting the motor. Once in, screw in the 4x bolts when it is all lined up.

The net result is this.
20190724_182415.jpg

One thing to be careful is the gearbox, when opened, has a very thin metal plate to keep the gears in the gearbox. This plate often comes out with the adapter plate. Make sure that is placed back in the gearbox body correctly (i.e. flat) in the body before replacing the motor otherwise it'll prevent the motor from spinning in the gearbox and thus, seize it up.

Given the much narrower body of the Turnigy motor over the 775 motor, it might be worth supporting the rear end of the motor with a HDPE block. Next time, it'll be the ESC programming.

[adamclark]

this looks really useful, is there also a slight weight advantage as well as the increased power ? double whammy

[Ocracoke]

The 775 motor pictured weighs about 400g and the brushless motor that it replaced is about 310g according to the spec sheet, so about 180g saved so yeah, a little bit of a weight saving.

[Ocracoke]

Azriel
So following on from the physical conversion of the motor to brushless, next we need to get the brushless ESC to run the motor in two directions as smoothly as possible. I do intend on putting out a hex file which I'll be using in Azriel but I won't put it up just now as I'll explain later.

A reminder: Your mileage may vary if you decide to follow this.

Today's plethora of parts on the square paper are the following from left to right: A USBasp programming device (this doesn't seem to be in stock at HobbyKing at the moment), a 3x and a 2x terminal connection block, a old Featherweight grade wiring loom (I seem to recall this came out of The Honey Badger 2.x) and a Redbrick 100A v2.

20190726_135419.jpg

First thing first, I plugged in the motor into the 3x terminal block. I wouldn't do this in a fighting robot but for a workbench test, I found this to be sufficient.

20190726_135701.jpg

I then broke open the Redbrick and took off the heatshrink wrap, which I am fairly sure would void any warranty it was under but given this is a fighting robot, I am fairly sure no warranty would ever cover this.

20190726_140043.jpg

I then hooked up the ESC and the rest of the wiring loom together and picked up the 6S battery set from Azriel. As an aside, the below is how you wire up this ESC (I find the lack of colour coding annoying).

20190726_140630-1.jpg

Now, at this stage, I tested to see if the motor spins at all (I might post a short compliation video of this soon) and it did. Great, so now it is time to get programming the thing. The USBasp programming device is a great thing, I picked one up when I was flashing TZ85As and worth its weight in gold. For those who have never used one, you need to line up the red dot on the device with the circle on the Atmel processor as below when connecting the two together and, as a aside, don't have the ESC on when programming. When the device makes a connection, it'll supply its own power to the ESC during the programming.

20190726_141546.jpg20190726_141630.jpg

And here is my bench test setup. It is fairly rudimentry but because of the proximity to the computer where I do the programming on, I can flash a new hex file pretty quickly.

20190726_141941-1.jpg

Now originally, when I started this, I followed a few guideline settings from the Ask Aaron archives (http://runamok.tech/AskAaron/motors.html) but these do not work in this ESC/motor combination, just resulting in a lot of jittering. Now, in preparation for this, I watched a Youtube video on how AC motors work and figured that, in the SimonK code, the amount of power being delivered to the motor at startup was too much. To put it very simply, in the video below the hand is the amount of power being supplied to the motor and the fan is the rotor itself (2:01 in the video):

The standard SimonK code reads as the following:

.equ    MAX_POWER    = (POWER_RANGE-1)
.equ    PWR_COOL_START    = (POWER_RANGE/24) ; Power limit while starting to reduce heating
.equ    PWR_MIN_START    = (POWER_RANGE/6) ; Power limit while starting (to start)
.equ    PWR_MAX_START    = (POWER_RANGE/4) ; Power limit while starting (if still not running)
.equ    PWR_MAX_RPM1    = (POWER_RANGE/4) ; Power limit when running slower than TIMING_RANGE1
.equ    PWR_MAX_RPM2    = (POWER_RANGE/2) ; Power limit when running slower than TIMING_RANGE2

Even under what I assumed to be the COOL_START variable, the motor jitters so I changed it to read the following:

.equ    MAX_POWER    = (POWER_RANGE-1)
.equ    PWR_COOL_START    = (POWER_RANGE/48) ; Power limit while starting to reduce heating
.equ    PWR_MIN_START    = (POWER_RANGE/24) ; Power limit while starting (to start)
.equ    PWR_MAX_START    = (POWER_RANGE/10) ; Power limit while starting (if still not running)
.equ    PWR_MAX_RPM1    = (POWER_RANGE/8) ; Power limit when running slower than TIMING_RANGE1
.equ    PWR_MAX_RPM2    = (POWER_RANGE/4) ; Power limit when running slower than TIMING_RANGE2

Now when I start it, the motor spins up OK for a bit but about half way into full throw on the transmitter, it jitters a little bit before resuming cleanly all the way up to full power. Clearly, I still have some way to go in getting the spot on but the fact is, I could flash both ESCs and Azriel would move fairly happily on them. However, it isn't perfect as the motor and the ESC get quite warm after a minute so there is evidently some work to go on refining this. As a test, I also put a KEDA 5663 outrunner on this as well and it jitters a lot more than the inrunner but this is proving to be a good basis to work with - even that motor spins up partly on the code I have running now.

I'll try and update this over the weekend.

[McMullet]

Thanks for the detailed write ups! It's great to see the kind of steps even if, as you say, there may be some variation in the nitty-gritty.

[team death]

I'm pretty sure I could handle the motors conversion part but the programming of an ECU looks quite daunting for a novice , on a scale of one to ten , how hard is that programming ? .

[Ocracoke]

Thanks for the detailed write ups! It's great to see the kind of steps even if, as you say, there may be some variation in the nitty-gritty.

Many thanks. With brushless tech, there is always going to be some variation, likely down to the different builds of the motors themselves along with any differences in the ESC boards themselves.

I'm pretty sure I could handle the motors conversion part but the programming of an ECU looks quite daunting for a novice , on a scale of one to ten , how hard is that programming ? .

It does require patience and some knowledge of how AC motors work (as I noted in the last update post). To actually do the flashing procedure itself is easy enough with the tools I mentioned but knowing what values into put in there can be a lesson in patience. I'd say on a scale of 1-10, this is a 3 or a 4 though I should point out, my trade is software development so I probably have a skewed viewpoint on such.

I've got it a little smoother on 6S LiFe but with that said, I am working on something to make the whole programming brushless ESCs easier to work on.

I'll have a update later in the week, parts for Jibril arrived today and I want to get on with that as well.

[Ocracoke]

Jibril
So this robot... oh this robot. As I mentioned in my last post, the body parts arrived yesterday and last night, I did a mock build of the robot. This is the first robot I'll have where I did a little bit of CAD work on it.

mock_build.jpg

Obviously, that isn't the finished look, there is a lot to do. This was more just to see what the robot will roughly look like. The side arms are 5mm ARMOX panels that should add to the look of this thing when done. I am hoping to get the basic bodywork done by the end of next week to start transferring all of the electronics out of the old robot into the new one. I am aiming for a day at EXR Colchester with this so plenty to do.

[Ocracoke]

Jibril
I've started putting in the threaded inserts into the bulkheads. 20x M10 threaded inserts, it is a bit of a slog but at least progress is finally underway. At the rate of progress, I am hoping to transfer the drive components in by the end of this week. I may have to make a slight adjustment in the central bulkhead for the weapon as so not to compromise the integrity of the shell but it isn't a drastic change, should be fine.

Azriel
I've not had any time to work on the physical build but I've had some time to write up a program that makes alternations to the SimonK firmware so it should just be a case of plugging in some variables and it generate a file which can then be flashed to the ESC. I need to setup my test rig again to try out the new files being generated (and adjust if required) but I have some hope that it'll work (though I have just noticed a bug in it so... a way to go just yet :P)

[Ocracoke]

Jibril
The basic body of the robot is now complete and ready for component installation. Unfortunately, real life has reared its head and I've just not been able to get as much work done on it as I would like. At this rate, I am unlikely to make it with this version of Jibril to EXR Colchester so what I might do, barring a massive leap forwards, is to build a proper axe support and reinforce the gears in Jibril 1.1 (to make it 1.2) and run that at EXR Colchester for the day before running Jibril 2.0 at EXR Maidstone.

Currently weighing in at 13.9kg with the top panels and the bolts to hold the lids down. Now that the basic bodywork is done, I can start adding in the internal components to make it actually move. I will say though, the overall shape so far is rather inspiring. Guess that shows how far I've come in building these things. Would like a rematch with Earthquake one day.

basic_body.jpg

Azriel
I'll be putting Azriel back together over this week I think. I need to get some new wheels and wheel hubs but nothing that can't be sorted fairly easily. It'll be interesting to see how the brushless drive performs in this machine. Since Jibril is still going to be using its F30-150 motor, I can't swap out that into Azriel until the A28-150 motor goes into the new robot so for the moment, it'll be sticking with the CIM motor.