Gear reduction understanding

Tommy, thats not what you said to me on this thread!
forum.linuxcnc.org/30-cnc-machines/31509...pinion-drive?start=0

This is a very useful thread I think notwithstanding the extra work I was given from it.

On page 2, the  discussion on gearing starts.

Note that on the feedback given I did not try the direct drive because I knew the ideal travel was about 1" per rev with a range of 15-30mm
We've since done a lot of engineering design which confirmed this.

rodw wrote:

Tommy, thats not what you said to me on this thread!
forum.linuxcnc.org/30-cnc-machines/31509...pinion-drive?start=0

I am fully aware of that, but he insists it works for him, so why bother?

Interesting topic.
It would be interesting to see some data that shows the optimal distance per revolution and why, rather than anecdotal evidence. When I built my table I saw the inch per revolution figure tossed round but due to constraints at the time (and laziness) I ended up using GT3 belts at 48mm per revolution. I thought I may add gearing at a later date but the cut quality to me appears good and I haven't yet seen a reason to change. 

phillc54 wrote:

Interesting topic.
It would be interesting to see some data that shows the optimal distance per revolution and why, rather than anecdotal evidence. When I built my table I saw the inch per revolution figure tossed round but due to constraints at the time (and laziness) I ended up using GT3 belts at 48mm per revolution. I thought I may add gearing at a later date but the cut quality to me appears good and I haven't yet seen a reason to change. 
 

I don't have charts or torque curves to post, but on the two big tables I built I had two variables to play with: pinion diameter and reducer ratio. 

To answer your 'why?' question, my objective was to have have the motor torque start to fall off around 500ipm, ensuring that at any reasonable cut speed they would have full torque, plus quite a bit for rapids as it's painful to watch a gantry creep along for 12 feet at 200ipm.

Servos had a top rpm of 5k, and the torque started dropping off at about 3500rpm (acording to datasheet)

In addition, one of the tables has an interchangable router spindle and the owner wanted an electronic & mechanical resolution of <0.001".

Drive system for both is a home-brew version of the Bell-Everman Loop Track arrangement.  Wound up with a 15T pinion (0.947" pitch dia) & 10:1 planetary reducers on all four drive motors (for commonality of spare parts).  This gave an effective distance per motor rev of 0.2975", and a theoretical top speed of ~1000ipm.  Servo drives were programmed to be just under the 200kHz input frequency at top speed.  The Bell-Everman system can be treated like a rack and pinion for speed/rpm calculations.

The gantries can hit around 1100ipm and the heads can go a bit faster.  Long-term testing at full speed revealed the need for braking resistors on the gantry motors if I left the speeds way up and both owners declined the added expense.  They're both happy with the tables throttled down to 800ipm with fairly aggressive acceleration left.

As has been noted multiple times: if cut quality is acceptable, don't change anything.

In my case, however, both owners told me "I want a Hot Rod" and they got what they asked for.  The gantries are fairly light & stiff, heads are light, and the drive system is remarkably stiff & backlash-free.  When programmed aggressively, they're pretty frightening to watch - like oversized pick-and-place machines.

phillc54 wrote:

Interesting topic.
It would be interesting to see some data that shows the optimal distance per revolution and why, rather than anecdotal evidence. When I built my table I saw the inch per revolution figure tossed round but due to constraints at the time (and laziness) I ended up using GT3 belts at 48mm per revolution. I thought I may add gearing at a later date but the cut quality to me appears good and I haven't yet seen a reason to change. 

There is not really a definitive rule but if you do the engineering, as spumco says both gear ratio and pinion diameter are both different tuning parameters. If you want to get to the moon in your spaceship or build a hot rod, you need to maximise performance. I worked with Marshall, an engineer in the US while he developed a stepper motor design model over 3 months. The model was then tested on my table using LAM Technologies drivers. The main benchmarks we reviewed was the motor voltage required (some motors only needed 30 volts), the amps required to achieve the desired performance (some motors only needed half their rated amperage) and the temperature rise to make sure we did not melt the insulation. But then you also needed to review the torque curve to see if the selected motor had enough headroom for your rapids. This is where higher voltage than what is required helps.

So changing the pinion diameter and the gear reduction makes a difference. We tried various combinations from 60mm pinions and gearing ratios from 10:1 and down to get the best results so we could achieve 30 m/min rapids, 10.24 m/min max cut speed and 5 m/sec/sec acceleration (approx 0.5G). This means we accelerate from a stand still to 10.24 m/min in 30 milliseconds.

From all of this, we found that my seat of the pants design was close to optimal when benchmarked with the model. Whilst there was a case to change the reducers from the existing 5:1 and 3:1, it would have involved pretty major mods to the gantry ends so we just went with it as it met the model design. We also found that every motor was different so it really is not possible to generalise  design parameters.

I was able to get my X axis up to 60 m/min rapids and 0.8G accelleration (8000 mm/sec/sec) but this was too much for the NEMA34's on the heavier gantry so we settled on 5m/sec/sec and 36 m/min  rapids. At this level of performance from a stepper, Lam's boost technology is mandatory.

So whilst Phil's 48mm pinion circumference is probably OK, but it is likely suboptimal A 1:1 gear ratio would fall far short of our design parameters as it would not be able to harness the massive low down torque steppers produce to achieve the desired acceleration profile.

This video attempts to outline what we were looking at 
www.loom.com/share/64ef8fb135ad47bcaf006761e222b076
Remember Acceleration is king with plasma.
 

rodw wrote:

So whilst Phil's 48mm pinion circumference is probably OK, but it is likely suboptimal A 1:1 gear ratio would fall far short of our design parameters as it would not be able to harness the massive low down torque steppers produce to achieve the desired acceleration profile.

I understand the relationship between voltage, accelleration an velocity, I also undserstand that changing pinions and gear ratios are tuning parameters.

What I don't understand is the 1" per revolution "rule". I guess it is just a rule of thumb because I haven't seen any supporting data.

I assume my table being "high geared" does utilise the high torque range of a stepper which is at low rpm. I can get 0.8G accelleration and 80M/Min velocity so it can't be that bad. I only use 0.3G and 30M/Min because I don't see any benefit from higher values.