guys, i want to use spectra line instead of belts on one of my i3. Has anyone did anything similar?, im using cerberus machined spools and idlers. Looks like i will have to put a spacer on the x stepper so the line doesnt rubs on the spool base and the xend wall… Will upload pics as soon as i get the spacer printed and in place.
I have to ask, why ?
Spectra lines were introduced to reduce cost, they are inferior to belts in every way. These days you can get really cheap GT2 belts ($2 per meter) so I really don’t see the point of using lines.
are those really cheap gt2 belts (that i have to buy overseas) good?; my experience tells me not. Is easier and cheaper to make a spool rather than a good concentric pulley. Also, sprectra, once tensioned, wont give any slack.
ALso, why not? 
@Shachar_Weis sorry but you’re wrong. If nothing else, they can improve print quality as there are no belt tooth/artefacts you often see with the cheap belts you mention. Go look at the surface quality of Tantillus or my own Ingentis prints if you have doubts. I’m not saying spectra is perfect but it has its place and certainly has its advantages over belts.
@Shachar_Weis the belts more expensive, quality can be an issue and the best belts for the job run to $5 a metre. Then add in ali pulleys and belts and spectra looks good value.
@Tim_Rastall I’m not convinced that that’s true (I’ve pushed x, y and z resolution much farther than most) but even if it is, I’d say that it’s better to switch from belts to synchromesh (which had teeth to prevent slipping and make steps/mm mathematically reliable, but with a helical profile that is smoother and quieter than gt2), rather than spectra. Toothless drive mechanisms like spectra and threadless balls screws create the potential for mechanical slippage, but also add uncertainty to steps/mm calculations so that they have to be calibrated empirically.
@Whosa_whatsis , what can you recomend to use as an idler for synchromesh?. I have a few meters plus a few pulleys, i was planning to use them too.
We used to use a bearing with a printed grooved ring until I was working on the Bukito and realized that the groove made by stacking two 625 bearings with a washer between them made the perfect idler. Solid (the printed rings occasionally split under high tension, super-consistent (imperfections in printed parts can result in a very slight eccentricity), and the effective diameter is the same as the motor pulley (to within about .5mm, anyway) which makes alignment easy.
@Whosa_whatsis actually, steps/mm is relatively easy to calculate for spectra as long as you know the diameter of the shaft you have the line wrapped around. I have 120lb stuff wrapped 10 times around a 10mm shaft on Ingentis and there is no sign of slippage once it’s tightened (and I have really tried to make it slip). I have run movement tests at 250mm/s with 9000mm//s accel to the point where the stepper motors overheat and everything returns to home positions without any measurable slipping.
I don’t disagree that synchro is a good option and if I had a limitless bucket of money, I’d use it but it’s expensive and hard to source in NZ. Spectra is cheap and easy to source.
I have to +1 for spectra too, my large delta runs on spectra and after calibrating makes really nice prints. I also just replaced the spectra drive on my corexy with GT2 belts, and to be honest I see absolutely no difference in quality or accuracy, and the belts are a lot harder to terminate and are far more fussy about going around corners and being absolutely the same height all the way around. I’m seriously considering switching back to spectra. One caveat though is you need high quality machined spools/capstans for spectra like grabbers ones. My printed spools never worked well.
Spectra / lines instead of toothed belts does not give a positive drive, which can introduce slipping.
It is ok for rostock because that is a different setup to an i3 and still manages to positively drive using spectra line… but a direct replacement of the toothed belt on an i3 would introduce the POSSIBILITY of slippage (possibility, not saying you can’t get lucky or reduce it to a minimum, but the possibility of slippage IS THERE)…
Also, (correct me if I’m wrong, I don’t own a rostock / tantillus etc) in a rostock with spectra line, the spool will wind the line on and start overlapping at a certain point… This means the circumference around the spool has changed (albeit slightly, it has changed) which will introduce positioning error.
Steps per mm are also not easily mathematically calculatable. With a 5mm pitch belt and a 10 tooth pulley (for example) you move one revoloution and you’ve moved 50mm… NO ifs or buts… Changing to a 20 tooth pulley? guess what - your e-steps have exactly halved…
With a 15mm diameter spool, how much spectra line per rev? what if it slips a tiny bit? what if over time there’s grime and buildup on your drive spool?
Belts are also reinforced… and they are designed and made specifically for the purpose we’re using them for…
I DO see how it can be of benefit to different bots like delta bots, but it is 100% inferior to belts. I’m a fitter and turner by trade and work on a lot of high precision and high speed machinery… I don’t ever see people opting for spectra line (or any industrial equivalent) It’s toothed belts or gears or racks etc (positive drive) for operations requiring positioning or precision…
My corexy is not a delta and it works perfectly with spectra. Slippage is virtually impossible with the correct number of turns around a decent spool. with 8 turns around the Grabber spool the steppers will drop steps before you can make it slip… actually I think the spectra will snap before it will slip! It is easy to do a one time calibration of steps/mm with a dial gauge so being able to mathematically calculate it is irrelevant. Unless you have actually used spectra it is easy to poo poo it 
But my practical experience with two large printers using it is very positive.
@Jarred_Baines so you know smooth belts are better than toothed belts then in this context.
As spectra is basically a smooth belt.
And I’m from a fitter engineer background and know its all about the pulley not the belt.
@Jarred_Baines it’s not 100% inferior to belts, go search the Tantillus forum on the Reprap forums and see what people’s experience is of going from spectra to belts. I’ve built belted bots and I’ve built spectra bots, both solutions have their merits. We’re not talking about industrial machines so your comparison doesn’t apply here. The holding friction of 10 wraps around a steel shaft is waaaay above the holding torque of a Nema, that means no slipping.
@Nigel_Dickinson Smooth belts allow smoother transmission of movement, this is true, and for an extremely SMOOTH machine (perhaps you prefer smooth parts to accurate parts) spectra / smooth belt systems WILL come out on top. And I was 100% incorrect saying that spectra is 100% inferior because there are definitely benefits to smooth, flat belts.
But a smooth belt is always driven by friction, even with your knurled spool (I’m guessing this ‘grabber spool’ is knurled or something for extra friction?) you cannot guarentee that the belt will return to the exact same POSITION even if it never ‘slips’. Consider a big print which moves the axes back and forth over 1000 times. I won’t go into the ‘how’, but a fitter / engineer would understand that since it’s a friction drive and not a positive drive (like a toothed belt or a chain and sprocket) there is the POSSIBILITY of error due to the fact that there is no ‘tooth’ just a braid ‘being gripped well enough to not slip’.
The only CNC machinery I’ve seen using smooth belts ALSO incorporate encoders, and so it doesn’t matter if the belt slips, or doesn’t return to the same position, because the encoder lets the CNC control know if it’s in the right spot or not. But with stepper motors and no encoder we are relying on the belt returning us to the EXACT same position every time.
A smooth belt/line/cable can also not return to the exact same position even if it doesn’t slip. Think about what happens if it moves very slowly with gentle acceleration in one direction, and jerks quickly in the other direction. The fast jerk will stretch the belt/line/cable momentarily as the tension fights inertia. This can effect where the belt/line/cable makes contact with the spool and starts gripping it, so the tension is not constant, and over many cycles, the belt/line/cable’s position will drift.
Then, of course, there’s the obvious problem of the nonlinear response to spool rotation and variable tension introduced by wrapping the line multiple times around the spool so that is “walks” across the spool as it turns. This means that the point where the line leaves the spool will move rather than staying in a straight line parallel to the carriage’s motion to the carriage attachment point. When this line is not parallel, its geometry changes as the carriage attachment point gets closer/further from the spool, in addition to the spool end moving, so the total length of the line will change and it will have to stretch (slightly) or go (slightly) slack as a result. This tension change plus the difference between the nominal length and actual length ( = sqrt(nominal length^2 * offset^2) ) of the run of spectra results in nonlinear conversion of rotational to linear motion.
^^ What he said 
Drift is the word, even without slippage there is drift on a friction drive… if I have 2 pulleys, one double the circumference of the other and I run them all day long you can bet your bottom dollar that if pulley 1 has done 1,000,000 revs, pulley 2 has NOT done exactly 2,000,000 revs.
This phenomena is MORE pronounced with changing directions / loads.
Yes, drift is a better word. Each time a tooth on a belt/pulley meshes with one on the pulley, it self-corrects for drift, but spectra drives have no mechanism for this unless you home every layer (and the switches most of us are using probably don’t have the repeatability to do any better). Even so, high resolution prints will show a visible line on the layer transition if you make a correction for drift (the correction will be a number of microsteps, while the actual difference will not be). The higher you push print quality, the more little issues like this stick out.
@Whosa_whatsis
A good point about the cable walking but you need to do the maths to see how trivial it is for thin spectra. For me the offset arrising from the cable ‘walking’ is about 15mm for 300mm of linear motion. So, the spectra’s length in this extreme case (hypotenuse of the right angle triangle) is 300.375mm. I’ve never printed a part big enough for this to be a measurable problem and I imagine some other error creating factor would mask it anyway.
It’s also worth remembering that the larger the shaft, the lower the number of rotations required for a given linear delta (and hence a lower ‘walking’ value): A 30mm (diam) spindle gives 94.25mm linear delta per rotation, so for an average I3 you’d only need 2-3 rotations and hence only see a few mm of walking.
The issue of positional repeatability is one that could be a problem in theory but I’ve yet to see it manifest in prints that have gone for 12+ hours. I also (out of interest) put marker lines on the spectra and on the shafts, I’ve yet to see them shift.
@Tim_Rastall that all makes sense, and I understand that the effects are not measurable in your case. I’ve seen a lot of problems with the printrbot simple posted that I’m fairly certain are due to these issues, as well as some that are obviously caused by slipping.
Even the 0.125% error you mentioned is enough to make me cringe because it is designed-in error. The platonic ideal version of every machine should have zero error, with any error in the actual machine being introduced by deviation from that ideal, at least that’s my opinion.