Introduction: ☠Weedinator Part 1: Drive Mechanism☠

Picture of ☠Weedinator Part 1: Drive Mechanism☠

In the world of professional agriculture, a lot of focus has been put on large, incredibly expensive machines that work in huge open fields where just one crop is grown. Whilst this is incredibly efficient and produces very cheap food, it's not good for pretty much everything else!

There does exist a substantial backlash against this farming model where small farmers grow 'organic' vegetables on small farms with respect to the environment and indigenous wildlife. Some might call this the 'Permaculture' movement, but I'm really not sure if permaculture would embrace modern 'robot' technology or not as historically robots take away people's jobs and help destroy communities.

Anyway, here I introduce the Weedinator - an autonomous agricultural electric tractor that can be used on small farms to cultivate, till, plant and weed seed beds. It will be able to travel up and down 56 inch wide beds, several times a day if necessary, performing a multitude of quite delicate tasks which a big tractor might struggle with. Some of the key advantages are:

  • Less need for heavy cultivation such as ploughing which destroys soil structure.
  • Preservation of soil health by keeping essential natural micro-organisms and nutrients near the top of the seed bed rather than burying them.
  • No fumes or pollution especially in glasshouses or polytunnels.
  • Will help small farms compete financially with big industrial producers.

Will the Weedinator destroy jobs or will it encourage lots more people to set up small farms and help protect the environment? (Let me know what you think).

So if you want to build an autonomous tractor, where do you start? In part 1 of this project I'm just going to show how to build the drive mechanism - it's all about welding and fabrication and there's nothing much about the electronics ..... yet. The aim is to get a solid mechanical system working in 3 axes -

  1. Drive
  2. Steering and
  3. Suspension

Eventually, as the project progresses, the chassis will become populated with all kinds of weird and wonderful gadgetry including a digital compass, GPS positioning, GPRS , object recognition ..... The list goes on! Looking into the crystal ball, imagine a huge 200 kg drone that moves at about 1 mph along the ground fitted with an onboard CNC machine.

Step 1: Welding Safety

Picture of Welding Safety
  • Never look at the welding arc either directly or otherwise and protect other people and animals and anything else with eyes from the same. It's your responsibility to protect others from blindness.
  • Always wear a mask and never expose any skin to the powerful UV light emitted from the arc. This also means legs and arms and neck. Never do welding in shorts and T shirt. UV light can cause skin cancer.
  • Check very carefully for flammable materials in the work zone, particularly piles of sawdust. A blob of metal can easily fly off and land in a dark corner somewhere to gently smoulder - you go away for 1/2 an hour and come back to your shed on fire.
  • Be aware that it's surprisingly easy to set your own clothing on fire, particularly your underpants.
  • Long hair should be tied up to avoid catching fire.
  • Make sure there is adequate ventilation, especially if you are welding inside a tight space.
  • Keep a suitable fire extinguisher close to hand.
  • If you want to look cool and protect you scalp from hot blobs of molten metal, wear a baseball cap back to front. Traditionally, this should be as brightly coloured as possible, preferably with a floral pattern!

Step 2: Tools and Equipment

Picture of Tools and Equipment
  • MIG welder on CO2 or Argon mix
  • Plummer's wrench
  • Grinder
  • Drill
  • Clamps
  • Files
  • Leather / hide / copper head hammer
  • Set square
  • Vernier calipers
  • Emery paper
  • Spanners and socket set
  • Gauntlet gloves
  • Auto react welding helmet

The welder needs to be of at least 180 amps with a good quality feed mechanism for 0.8mm wire. Be aware of the cheaper Chinese models as they are generally useless. Keep the welding nozzle clean of splatter to ensure clean wire feed.

Step 3: Drawings

Picture of Drawings

Drawing files are attached!

Step 4: Components

Picture of Components

Raw Materials:

  • Round bar, bright, 20mm x 1000mm
  • Round bar, bright, 25mm x 320mm
  • Round bar, bright, 40mm x 1000mm
  • Round bar, bright, 50mm x 300mm


Special thanks to DMM Llanberis for allowing me to use their lathes and milling machine!

Step 5: Steering Sub-assembly

Picture of Steering Sub-assembly

The steering shaft ends on a 50mm cylinder and uses a low profile 45mm pillow bearing block to sustain the leverage forces created by the offset of the wheel to the centre line of the steering. The bearing is bolted into place and is used to ensure correct positioning of all the other components. Tack welds are made with 6 second blasts of the MIG welder at high power in strategic positions which allow both strength and the ability to remove the weld at a later stage. All tack welds should be accessible by either a grinder or a drill bit for removal in case of a positioning mistake (see one of the later stages).

Clamps are used to locate the bracing gussets.

Step 6: Drive Subassembly

Picture of Drive Subassembly

An 18" stilsons or plumber's wrench is the ideal tool to bend small pieces of metal in a vice. The part shown is deliberately designed with slots where the bends need to be - the bend will naturally occur along the slots, which are welded up afterwards.

The 25mm pillow block bearings ensure that the steel plates are in the correct position before the gusset is clamped on and welded tight.

Step 7: Joining Steering to Drive

Picture of Joining Steering to Drive

The 50mm free end of the steering shaft is tack welded to the top of the drive gearbox. The sheer weight of the steering sub assembly prevents distortion of the joint, as long as the tack welds are positioned in sequential opposite pairs and not all welded on one side and then the other. Distortion is the arch enemy of the welder and is often caused after a large, hot weld cools down and shrinks.

Always tack weld on opposing, opposite sides of a component to help prevent distortion.

Step 8: Suspension Wishbones

Picture of Suspension Wishbones

The suspension needs to be positioned such that none of the moving components crash into one another. This mechanism is quite complex, so it's easy to get the position slightly wrong, which is why it's important to use tack welds appropriately. The first time I welded in the suspension I got it slightly wrong and so had to grind out and drill out some of the welds. Fortunately all of the welds were accessible although I did break a couple of drill bits in the process.

Step 9: Suspension Motor Subassembly

Picture of Suspension Motor Subassembly

The suspension is driven by a stepper motor on a 14mm spindle and is most easily put together on the bench as shown before positioning on the main drive system.

Step 10: Wheel Disc

Picture of Wheel Disc

The wheel is attached by means of a steel disc, which itself is attached to the drive shaft with 4 small gussets. The gussets are cut so accurately that they can be used to align the disc itself so avoiding a wobbly wheel.

Note that the welder's earth lead is placed close to the welding zone and not to any other arbitrary part of the assembly - this is very important as otherwise current could pass through the bearings and cause a small weld within the bearing itself which would effectively ruin it.

Step 11: Motor Attachment

Picture of Motor Attachment

A small tube with a slot milled on it is used to allow the motor shaft to fit into the gearbox (see drawing). Small 4mm cone seated grub screws are used to secure the tube to the motor and a key is inserted in the slot to match up to the slot within the gearbox. A file is used to make the key itself from a length of 6mm square key steel.

Step 12: Test Bed Electronics

Picture of Test Bed Electronics

Since this is more about the mechanism, the electronics used is currently very basic - a simple stepper motor controller and an Arduino Nano - just enough to get the motors working with some acceleration.

The controller is a M860H which is supposed to give up to 7 amps at 50V, but I could only squeeze out 2.6 amps before the main drive motor started to miss steps. Nonetheless, it gave quite some considerable torque which seemed large enough to propel the vehicle forward. I'm going to try another system - a closed loop stepper motor setup which is a lot more sophisticated and reportedly gives better torque performance.

Step 13: Arduino Test Code

Picture of Arduino Test Code

# define CSpin 3
# define STEPpin 5 # define DIRpin 6 # define RESETpin 7 int del = 2000 ; long i=0; void setup () { pinMode (8, OUTPUT) ; //SLEEP digitalWrite (8, LOW) ; pinMode (STEPpin, OUTPUT) ; pinMode (DIRpin, OUTPUT) ; digitalWrite (DIRpin, HIGH) ; //High is clockwise Serial.begin (115200) ; delay (10000); } void loop () { while (i<500000) { i++; digitalWrite (STEPpin, HIGH) ; delayMicroseconds (800) ; digitalWrite (STEPpin, LOW) ; } }

Step 14: Testing the Mechanism

The driver assembly was bolted to a short piece of 200 x 100 x 5mm box section and held firmly in a large bench mounted vice. Firstly, the wheel rotation was tested, which worked well with the correct speed and reasonable torque. The torque was tested by crudely trying to slow the wheel down by hand and was deemed to be adequate. Torque is directly related to the current going through the motor windings and can almost certainly improved by upgrading the motor driver to a hybrid closed circuit type, which has good reviews within the online community.

The suspension also worked well and managed to lift a considerable weight - about 75 kg. In use, the suspension will be working in reverse, acting against the chassis. Something worth noting is that the suspension did not 'unwind' when no current was applied to the motor, which is very useful.

The steering had much better 'lock' than expected, with about 120 degrees of rotation which is very much better than most normal tractors. The only downside was that it had more backlash than expected.

Step 15: Improvements

Picture of Improvements
  • Steering Backlash - whilst the steering system could just be used 'as is', after all, it's only travelling at 1 mile an hour, I thought it should be addressed. One possible solution is to use a cable as in the diagram above to effectively pull the wheels together, holding the backlash on one side of the gear assembly. Phil, one of the Weedinator team members, suggested incorporating Ackermann geometry into the back lash design, which could be done by using an off centre or elliptical pulley wheel for the cable. Another possible solution would be to use a separate motor on a linear actuator, similar to the suspension motor, to act as an adaptive tensioner on the cable which reacted to the Ackermann geometry programmed into the steering. This will all be covered in Weedinator part 2.
  • Steering Backlash and Simplification - see video above - The steering shafts are joined together and the steering motors can be made to 'jam' against each other very slightly to take out the backlash. Simplification of the suspension could reduce possibility of rotational strains occurring in the chassis which could jam up the CNC mechanism.
  • Motor torque and control can be improved by using 'Closed loop' stepper drivers and rotary encoders. Fancy electronics built into the driver prevents any steps from being lost and raises alarm signals upon finding errors. Also, torque is increased on higher speeds.
  • Testing the machine at very low speed is absolutely fine but higher power and speed can be achieved by using servo motors - up to about 1KW - but these are currently way beyond my budget!


razor7177 (author)2017-11-16

Awesome job on the drive unit, but try brushless low rpm motors like a 130kv rc outrunner motor and do a servo-like mechanism. You could tandem the motors for higher torque, just a thought. Again awesome job and excited to see the outcome.

Tecwyn Twmffat (author)razor71772017-11-17

I'll research that - thanks!

razor7177 (author)Tecwyn Twmffat2017-11-17

I could send u an example for the brushless drive motor design if u want..

Tecwyn Twmffat (author)razor71772017-11-17

yes please - that would be really useful.

razor7177 (author)Tecwyn Twmffat2017-11-18

this is just a sample

Tecwyn Twmffat (author)razor71772017-11-19

Thank you!

jongscx (author)2017-11-16

I'm curious about the design choices made that led to this presumably large and heavy machine having its drive wheels powered by steppers. I get that a) it's only going 1mph and b)prototype, but it's also large, pneumatic tires, highly geared down, driving on loose earth.

I'm just thinking the need to control to fractions of a degree aren't that critical and could've been solved with simpler (read: cheaper) motor and velocity feedback with fine positioning of the platform handled some other way.

Was this just a case of a small initial prototype scaled up with larger components?

Tecwyn Twmffat (author)jongscx2017-11-16

The choice of suitable motors is an ongoing conundrum! I chose stepper motors, not specifically for their fine control, but more for their large torque characteristics at low speed. Imagine how much less torque a cheap servo motor would produce? The torque is required to navigate rough ground and drive over obstacles such as mole hills etc.

The stepper motor on the main drive is supposed to be able to deliver a massive 12Nm, which = 1200Nm on the output of the 100:1 gearbox. It's not doing anything near that yet so there's a bit of tweaking to be done with alternative electronic controllers and higher voltages. At the moment it's just about ok and should manage most mole hills at low speed.

skrubol (author)Tecwyn Twmffat2017-11-18

Why does it need that much torque? That's probably at least 10x what the tire could put to the ground in good conditions. Stepper motors are quite inefficient, you will likely run out of battery just getting to and from the field. Also as far as not needing much top speed, again the concern of getting to and from the field.. Also stepper motor torque drops drastically at higher RPM.

BLDC would be the most suitable option. A 2Nm NEMA 34 BLDC motor will have 2Nm all the way up to 3000 RPM, whereas that stepper probably will get down to 2Nm around 1200-1400 RPM. The bldc motor can also have short term overloads up to 3x the torqe, whereas a stepper will just lose steps. They also generally have a hall encoder built in, so you can do plenty precise positioning when running through a gear box.

Tecwyn Twmffat (author)skrubol2017-11-19

Thanks skrubol - that's a really good tip. I found some 8Nm motors here:

Not sure what the price would be yet.

jongscx (author)Tecwyn Twmffat2017-11-16

Oh, wow, that is quite a difference.

I was originally comparing the stepper setup a cobbled together PMDC motor/driver pairing, but next to a comparable complete off the shelf servo setup, it is significantly cheaper. Nice!

vishnumaiea (author)2017-11-16

This looks brilliant. My first project next year will be a small 6 x 6 rover. I'm researching about mechanisms for the shock absorbers for each wheels. I could use many of the info you have provided here for that. Thanks :)

Is that 6" x 6" ? We work in (mostly) metrick on my ship :)

Oh no, I meant 6 wheels driven rover (like 4 x 4) :)

Ah yes ..... so how big will it be and how fast will it travel?

NightFire (author)2017-11-16

Is there a need for suspension? None of my tractors have suspension and I've never seen a tractor with suspension, though I've never dealt with commercial agriculture equipment.

Tecwyn Twmffat (author)NightFire2017-11-17

Normally a tractor will have rigid back axles and a front axle that swivels in the XY plane. I could have copied the normal tractor design, but this solution saves a lot of weight and enable all four drive units to have the same design so easier to build and maintain.

Polyhistor (author)2017-11-16

One area I would be concerned with is the exposed stepper lead screw; exposed to dirt, mud and chemicals may adversely affect the operation of the suspension. Just a thought ...

Yes that's a good observation and it would need to be protected with a rubber gator.

dsbahr (author)2017-11-16

Wow - Just amazing work. I can't wait to see how this progresses. I would love to get something like this on a smaller scale (read cheaper - easier for someone without your tools and skills).

AND I love the idea that something like this could really really help the smaller farmers. I think this might actually be the bridge between the conflicting benefits of Organic versus Green/Eco-friendly (low carbon foot print) farming. Historically those focusing on Organic farming actually have a higher carbon foot print per pound of produce than do larger mechanized operations (this obviously ignores the benefit of less pesticides on my food - but the point is organic growers are much more labor intensive per lb of food produced). A machine - for all it's exhaust that you do see - still has far less of a carbon load per acre or lb of produce than the number of people it would take to maintain the same area.

That's been one of the main dilemma's - grow healthy, or grow economically (including eco economics). Check out the book "The Omnivore's Dilemma" - great read.

But - something like this looks like a great fit to bring both objectives into reach.

Great Job!

Tecwyn Twmffat (author)dsbahr2017-11-16

Thanks - hopefully it will indeed be of benefit.

MarkC7 (author)2017-11-16

Wow! Looking forward to the rest of the build!

bricobart (author)2017-11-14

Every time you do it again, blowing me off my chair! Just when I started to feel happy about our first home grown organic vegetables & steps in permaculture you're coming with this! Since I've not your skills I'd rather try to prevent those weeds to come instead of having to fight them, but this is only possible on small scale I know - not for someone like you who's definitely decided to erase famine out of this world! Very impressed friend! Still wondering how the weedinator will differ good from evil, and how you're planning the destroying itself?

Tecwyn Twmffat (author)bricobart2017-11-15

Most of the weedinator's work will be in preventing weeds like you do yourself, except it will use a rotary claw to tickle the soil's surface.

Obviously determining good from evil is a problem, which is where the object recognition comes into play. Also, when it starts to become self aware, there is the danger that it turns on it's master :(

NeilRG (author)2017-11-14

Beautiful work!! Excellent instructable. I especially appreciate your Hot Tips on welding properly. In step 10 you reference the importance of grounding so as not to ruin the bearings. As an electronic and mechanical engineer I appreciate the effort & thought that went into this project. Its not something I directly have a need for so I am writing mainly to offer encouragement.



Tecwyn Twmffat (author)NeilRG2017-11-14

Thanks Neil - I tried to make it a bit more relevant to the instructables readership - I'm sure nobody will actually build this machine (except Elon Musk of course!)

deluges (author)2017-11-14

Holy molly. I haven't read all of it yet but I will and this will be a very interesting project to follow. Very smart suspension system and beautiful implementation.

Truly great work man. I'm sure permaculture and robots will get along very well. Hell, make sure your designs don't get stolen by Elon Musk for his Mars colonization project!

PS watching those gear animate with that psychedelic music made my day

Tecwyn Twmffat (author)deluges2017-11-14

I don't mind if Elon Musk steals the design as long as he pays for it! Not suitable for Mars anyway as they have about 1/3 our gravity.

About This Instructable




Bio: Ugly pirate roaming the seas in search of Treasure.
More by Tecwyn Twmffat:☠Weedinator Part 2: Satellite Navigation☠☠Weedinator Part 1: Drive Mechanism☠Arduino Cell Phone 4G Signal Booster / Repeater Part 1
Add instructable to: