Belt vs rack rails


This must have come up before but I haven’t found it so what is the reason for using belts instead of racks? I am interested in building my own setup and when reading about expandability it seems to me that the belts (when the x direction) are the biggest problem.
I would like to build a bot in a greenhouse (cold climate here in Sweden) and I thought that having racks mounted upside down up on high tracks would eliminate much of the risk of getting dirt into the traction system while also allowing me to make a (much) longer bed. It would also be fairly simple to extend the length later on.

What am I thinking wrong? Should mention that I am not a mechanical engineer so I lack a lot of experience in this field.



I would assume that any reliable and reasonably accurate system of location would work. You are going to have a little more slop in a rack and pinion system, but due to the lower accuracy requirements, I can’t see how that would make a difference. This was a path I am researching so if you have a good source for racking that would survive in UV then please share it.
The belting is $2.50/ft, so that is the upper limit. So far I found McMaster has some for $1.50/ft All the other stuff I found was only hobby lengths ~10". The one thing you will have to worry about is issue if it ever slips a gear. The belts have 270 degrees of wrap so are unlikely to slip, but a rack may only have a tooth or two in contact at any time.
The path I am presently researching is to use the belt as a rack and then roll over it with a tank tread style system of more belt. This would provide even more teeth engagement than a wrap but requires much more complexity. The advantage is you could more off the end of a rail to a different rail system.


Depending on size thermal expand and contraction does come a factor in play as well as you scale up but there are other things about rack and pinion.

Note the Lubrication is key section on rack and pinion to avoid excessive wear. PatrikL the main reason rack and pinion systems get dirty is the lubrication material glues what ever floating dirt to it yes upside down mounting helps a bit but does not solve it.

Belt systems avoid the Lubrication issue but they normally suffer from more thermal expand and contract issues as they get longer.

RichardJHauser single belt based systems as size expands can have more slop than rack and pinion due to thermal expansion and thermal effects on how much belts stretch. Duel belt systems can run into issues with teeth not meshing if fixed belt and the drive belt end up at different temperatures it is why the patented method has dual belts running complete length. Thermal expand and contraction becomes a bigger and bigger nightmare as you scale up. Tank tread style might cause your more problem than you can dream running on a fix belt that at times going to be different temperature than the track tread belt . Tank tread on a smooth track would have less thermal expand nightmare to worry about question is can you get enough traction to get exact movement…
There is another question with how accurate a farmbot needs to be does it in fact need belts in the small size or could farmbot use a spool drive and heavy duty fishing line. Stuff like fishing line is way cheaper to make and acquire than belts or racks. To be stupid some farm equipment used hoses on spools as lines to pull quite large machines around.

Basically every bit of toothing be it belt or rack equals a increased cost compare to using toothless. Question is will toothless be good enough. All the prototypes of farmbot I cannot see any tests using toothless.

With racks there is a lot of steel racking out there that will survive UV but they are not rust proof. Even the stainless steel ones will slowly rust out in the environment. My problem is I don’t know of any rack that if you look closely is rated for where the farmbot operates. I can think of many toothless options like heavy fishing line that are UV rated and would be fairly cheep to replace if there is a issue.


I wasnt exactly considering the cost involved, more the function, but it does add a sizeable part of the total I guess. My thought was that assuming it is mounted so it cannot slip but rather stalls the motor there would be less risk or losing exact position. As you say it is probably unlikely that the belt slip either. What happens if a small rock or something jams the gantry, would the belt slip or the motor stall?

I had not considered the lubrication glue effect and that might be a reason it itself to not consider it. If not then the rust issue probably is.

I think in general the wish to be able to extend the bot a lot in the x direction might not actually be such a good idea. From what I have read in other threads, not only does a genesis bot actually run for an hour or so (can’t find it right now so might be wrong) each day but it might also be quite important that it can do its work at the right time of the day. Scaling it up x2 in x and y could be the limit of what is a good idea and then any issues with belt expansion is probably not an issue anyway. What I am saying is that maybe more bots is better than too big a bot anyway.


PatrikL scale up lot larger is more than possible even allowing for a hour per day to perform tasks. Bash in weeding only need to be done once ever about 7 to 14 days in most areas. Depending other tasks even items like watering don’t have to be performed every day.

Making wider you run into gantry strength issues.

Issue with the idea of more bots is getting the per unit price down.

What happens if a small rock or something jams the gantry, would the belt slip or the motor stall?

Worst case with a toothed belt if motor does not stall it strips. Strips meaning it comes a toothless belt that is no longer able to move anything. Toothed belts basically don’t slip they just ruin self. Now something like spool drives they slip. So with toothed belts your motor size is limited to less than the strength of belt so it stall motor instead of stripping the belt.


@oiaohm As for jamming, depends on the object. My guess is if it is thick it would jam, but if it is large and thin, like a leaf, then it could wrap the gear and cause it to slip.

As for bash in weeding, I think that is an interim step. Once you can realistically track the size of the plants you are growing through their lifecycle, I think you could adjust plant spacing and succession cycles to squeeze out most weeds. A weed can only grow if there is space left for it and will die if it looses the race to get to light. If you mix your plants at the right spacing to fill all the altitude niches then weeds will be out competed. Think of it as a miniature version of the three sisters companion planting system of corn, beans and squash. The corn goes high, the beans in the middle and the squash at the bottom. The squash leaves cover the ground and shade out weeds.

So I expect the later evolutions of the systems with good soil built up, the whole process will be around precise planting, watering and harvesting,so tasks per day should be low allowing much larger areas to be farmed.


@oiaohm As for thermal expansion and lubrication, I can see how thermal expansion could be an issue with accuracy for long runs but since they have a fiberglass core, I think the issues would be minimal. The link I found suggested about 1.5" per hundred feet expansion between 0 and 100 deg.F. This wouldn’t be noticeable for my proposed belt to belt interface. As for lubrication, please remember that we don’t need gears with tight interfaces and I wouldn’t suggest using metal racks for cost and for the plastic’s better lubrication. The nylon gears I found could either go unlubricated, or could be lubricated/washed with water to keep them clean.

I like the idea of getting rid of the belt and going with a toothless option, but I’d suggest something like kevlar as then you nearly eliminate thermal expansion.


Rack based systems you do need lubricated of some form. Yes if the material you are using is something like nylon you get away water. The issue what happens on racking systems without lubrication is starting and stopping ends up jarring against the rack either chipping the teeth off the gear or off the rack over time. Basically at first rack based systems appear to work without lubrication then one day fail quite badly yet a lubricated one could go for centuries. This was found when doing the history on some steel laths and the number of repairs. Rack part of the system in those laths was one of the first sections to show if someone had not been keeping lubrication up .

My guess is if it is thick it would jam, but if it is large and thin, like a leaf, then it could wrap the gear and cause it to slip.
Slip should only happen with a toothed belt if the belts not tight enough. Other wise the teeth of gear and belt should basically cut through the thin material. I did not think the belt system on farmbot would have enough slack to-do this. Horrible part here is if the tooth belt is too slack it is wearing faster that is what it should. I did not consider this possibility because if it slack enough to allow the leaf todo that the belt was not at correct tension. So yes leaf wrapping around gear is possible but if that is happening you had something else wrong. So tooth belts don’t slip unless something else is wrong they strip or stall. If you want slip as possible response you use a non toothed solution. Slip is something that all toothed based solutions are not designed todo and should not be possible in correct configuration be it belts, racks gears… strip or stall are the two outcomes a correctly constructed system will do. Slipping in toothed systems is something else is wrong.


I can see how thermal expansion could be an issue with accuracy for long runs but since they have a fiberglass core, I think the issues would be minimal. The link I found suggested about 1.5" per hundred feet expansion between 0 and 100 deg.F.
Belts are never that simple the fiberglass core changes at one thermal expand speed and the rubber does another. Belts can do all kinda of strange things with thermal expand like appearing to shrink because the belt is curving.

You have to use the spec for the constructed belt to know what it is in fact going to-do to you. Single material items are way simpler to calculate for than anything composite. The rubber expanding on one belt and not the other might get you in a location where the teeth refuse to interlock even that the length has not changed very much at all.

Toothless would make the maths scaling up simpler. But toothless normally ends up having to use some other form of tracking due to the fact it can slip and that is part of the design.

Everything is a trade off there is no such thing as a perfect machine.


I think you are right


What about using rubber tank tracks

Then every 50cm attach a small magnet to the rail.

The bot moves along the frame with a little bit of inaccuracy, but every 50cm recalibrates itself using a cheap hall effect sensor