Monday, 29 March 2010

Pick Your Driver

The most widely available drivers on eBay are based on Toshiba's TB6560 chip. They are advertised at 36V and 3.5A, but you can try that only once - and only briefly. Checking the data sheet for the TB6560 shows that considering motor EMF and safety margins, you don't want to go beyond 30V, or better stay with 24V and 2.5A. I wrote an email to one of the eBay shops about the built-in current reference resistor and it turns out that they are limited to around 2.5A anyway, which is funny given it is advertised at 3.5A. There are similar vibes and concerns on cnczone. Bottom line, at around $73 for 3 axis delivered, they are the best bang for your buck at 24V and 2.5A, but if you want to go higher or want to squeeze better performance out of your steppers, then they're not the right option.

So:  TB6560: 24V, 2.5A, $73 for 3 axis.

One thing to learn from the TB6560 drivers is that you have to treat advertised specifications with a large grain (boulder?) of salt.

Next, there are a range of drivers of varying Voltage/current which all seem to be manufactured by JMC in China. eBay vendors verycnc, yuntat, cnccomponent, and wantaimotor (eBay, web) seem to be the main contenders, and comparing specs, it looks like the drivers offered by Keling Inc (ebay, web) and Zapp Automation (eBay, web) seem to come from JMC too. The Chinese vendors have "entertaining" manuals, whilst Keling and Zapp have the ones you can actually use.

Some offers don't come with a parallel port breakout board (marked "nc" below), which shouldn't really matter as I intend to use an Arduino as a controller, but it would be good to have a breakout board thrown in as a fallback.

Some of the "JMC" drivers don't have micro-stepping, some have too low or too high Voltage specifications, which reduces the possible range of 3-axis drivers to (prices in AUD inc shipping):
  • 75V, 5.2A: $250 nc
  • 60V, 5.6A: $296
  • 60V, 5A: $247, $220, $273
  • 50V, 4.2A: $237 (at auction), $256 nc, $236 nc, $205 nc
  • 50V, 5.6A: $281 nc
Finally, there are the highly regarded GeckoDrive drivers. The best choice for me seems to be the G251.The Gecko Driver has some nice extras, like mid-band resonance damping, recirculation mode, and microstep to full-step morphing. It's not clear if the other drivers offer that as well. Documentation is very professional. So:
  • 50V, 3.5A, $254 nc
I've sent a few questions to vendors and I got some specs to read, then we'll see...

Feel the Energy !

Energized by the Australian Grand Prix here in Melbourne, it's time to calculate motor requirements...

Roton provides some good formulas, as well as a nice introductory discussion on screw threads. Of particular interest is the list of friction coefficients (eg, steel on steel dry - 0.80; steel on steel lubricated 0.16), as well as the maximum allowed rotation speed of a screw thread resulting from the way it is mounted (Type of end fixity versus critical speed factor).

On mycncuk, irving2008 made a nice post including an Excel spreadsheet to calculate needed stepper motor size. The spreadsheet is pretty ok, though I had to adapt it to compare a list of motors in a single step, and I believe he got the calculation of the rotor torque wrong.


My gantry will have around 28kg and it will sit on linear ball bearing, so friction is low. Cutting primarily wood, and maybe aluminium, I would need a cutting speed of maybe 1,000mm/min to 1,500mm/min.

I'll be using M12 (12mm) screw threads, with an x-axis length of 1,000mm. I intend to mount the screw threads in fixed-supported mode. Screw pitch is 1.75mm. I assume screw efficiency is "lousy", ie, 25%, though with lubrication this may go up. It'll be an interesting test to see if nut-heating is going to be an issue. I also assume a cutting force of 20N.

Putting it all together, I would need to 0.55Nm at 2,400 steps/second for cutting, and 0.3Nm at 4,300 steps/second for rapid movement (if the motor and driver can handle that speed). Taking a safety margin of 3 means I need a motor of around 1.6Nm or better. I'd like to give it an additional buffer to allow for future growth (meaning, router version 2), so 1.6Nm to 3Nm would suit my needs.

Now, the ideal voltages for eBay-available stepper motors with 1.5Nm or more ranges from 50 to 80 Volts. So, in order to get good performance from these motors and considering that my 1.75mm screw thread requires high speed, I will be looking at a 48V power supply. 24V is too low, 80V makes it more expensive for the driver and more limited on stepper choice.

So, in summary:
  • Stepper motor: 1.6Nm to 3Nm, ideal voltage near 48V.
  • Power Supply: 48V 

Saturday, 27 March 2010

You spin me right round

I dropped in at CNC and Cupcake World and got myself six ball bearings to hold the threaded rods. I now proudly own:
  • 3 x 6201-2RS: dual-rubber-sealed, radial, deep-groove, self-lubricating ball bearings. ID=12mm (ie, M12), OD=32mm, W=10mm. Grease lubricated.


  • 3 x 5201-2RS: dual-rubber-sealed, double-row, self-lubricating, angular contact ball bearing. ID=12mm, OD=32mm, W=15.9mm.

...all I know is that to me
you look like you're lots of fun,
open up your lovin' arms,
watch out, here I come.
You spin me right round, baby, right round...
(Dope)

Originally, I though I'd use 2 x 6201 at one end of the threaded rod, but the recommendation was to use a single 5201 instead. We'll see how it all works out.

I'm now locked in to using M12 threaded rods and a 3-axis design. I'm getting close ! I need to decide on and order stepper motors and drivers, draft a general router table design, buy the building materials and get going. If only I wouldn't have to do various autumn chores around house and garden...

Wednesday, 24 March 2010

Bit Bender

Just some wild bendy ponderings...

How much does stuff bend ? How much support does the router table need to stay effectively true ? How thick does my MDF support beam have to be ? Is Aluminium any better ?

Calculating deformation (strain) depends on how a supporting beam is held at its ends (fixed, supported, or free). EngineersEdge list a number of useful formulas. For example, if the support beam is supported on both ends and has a single weight at its centre, the formula is y = F L^3 / ( 48 E I ), where y is the displacement height, F is the force (weight), L is the length of the beam, E is the Modulus of Elasticity, and I is the Area Moment of Inertia.

Bit Bender ©.
Use German beer for greater precision.
Das kommt vom Reinheitsgebot.

The subsequent calculation, of course, is obvious, but just in case you are exhausted from reading up on nuclear physics last night, I shall do the footwork for you:

Assume a weight of 20kg (a bit more than half the gantry weight as we use at least 2 support beams). Assume beam length L = 1m (approximate length of router table). Assume a support beam made from MDF, width 16mm and height 100mm.

As you undoubtedly know, MDF has a modulus of elasticity of E = 4 GPa = 4x10^9 N/m2 and Aluminum has E = 70GPa, ie, it bends 17.5 times less than MDF given the same volume and shape. Now the Area Moment of Inertia for a rectangular shape can be found on Wikipedia and is I = b h^3 / 12. So, for b=16mm and h=100mm, we get 1.33e-6 m4. Inserting all these values into the deflection formula, we easily find - naturally and without pen and paper: y = 0.77mm.

In layman's terms, an MDF support beam of 1 metre length, 16mm thick, and 100mm high, which is supported at both ends and holds a weight of 20kg at its centre will have a central displacement of 0.77mm.

A similar calculation can be carried out for a support beam made from a 50mm x 3mm aluminium beam. It will be displaced by 1.87mm. Things look better when both ends are fixed. MDF moves by 0.19mm and aluminium 0.47mm.

There is a difference in price: MDF = $2.92 (based on a 1200x900 board).  Aluminium = $21.98 (based on a 1000mm Aluminium angle). And a difference in weight: MDF = 1.1kg. Aluminium = 0.79 kg.

An MDF board of 76mm height and 16mm thick will bend less than 50mm x 3mm aluminium. Kind of interesting.

Based on this, it would be fine to use 100mm MDF support beams for the table.


...and I hope it's all true.

Monday, 22 March 2010

Putting X on the map

With general router dimensions defined by the set of shafts (x 1000mm, y 600mm, z 400mm), I gave some initial consideration to the design of the x-axis.

I don't want to move the target surface as this would effectively give me half the run-length, even though it would be a more precise design as the gantry would be fixed.

For the moving gantry approach, I'm thinking about 4 basic options:

1. Floating target with separate support table

The target cutting surface floats like a bridge along the entire x-axis length. The gantry rests on a separate, lower table. The threaded rod that pushes the gantry is centrally beneath the target surface.

Floating Target with separate support table.

Advantages: A single, centrally located driving threaded rod. No gantry weight on target table. Less dust and dirt goes on the rails and threaded rod. Everything is on a table, edge-to-edge (no damage from bumping against the table; easier to move around).

Disadvantages: Big and heavy. Taller design. Threaded rod harder to reach as it's sandwiched between target and table. Target support and table support must be relatively strong to prevent bending. Possible gantry bending error as threaded rod is not in line with shafts/drill point.

2. Floating target

Here, the the target table also supports the weight of the gantry and no separate support table is required.

Floating Target.

Advantages: Simple, light design. Used by various existing routers. Threaded rod kept clean and easy to reach.

Disadvantages: Dirt can reach rails. Target support must be strong to prevent bending. Somewhat complicated under-the-table cross-over. Possible bend error as threaded rod is not in line with shafts/drill point.

3. Dual Motor Design

Here, the target need not be floating and two threaded rods drive the gantry on each side.

Dual Motor Design.

Advantages: Simple, light design. Less bending as threaded rod is at level of shafts.

Disadvantages: Requires 2 motors (about $180 extra for motor, driver, etc) as I'm not keen at connecting the two threaded rods with chains or belts. Threaded rods are not perfect, so the two threaded rods can lose positional alignment which will require software calibration/inversion.

4. Single Motor Design 

Similar as above, but only one threaded rod and motor.

Single Motor Design.

Advantages: Simple, light design. Less bending as threaded rod is at level of shafts.

Disadvantages: Increased probability of positional errors when routing on the side that is away from the motor.

Natural Selection...

Well, I first considered the two-motor design as a good option, but then withdrew. Having to calibrate the two threaded rods, buy an extra stepper, driver, etc, all adds complexity and risk, which makes it comparable to option 1 and 2.

It also seems to me that option 2 wins over option 1. After all if I already have to make a fairly stable supported target table, then I might as well make it somewhat more stable and put the gantry weight right onto it. And the dust problem might be solvable with thin protective "fences".

So, I'm left with option 2 or option 4. I can't help but think that option 4 might create greater errors than option 2. That may require some analysis.

For now, I'll be going for option 2. Others have chosen this path, and in particular for a mathematician, there is safety in numbers....

Tuesday, 16 March 2010

The Grampians

I apologize for the interruption of the steady flow of highly creative cnc insights. I just returned from a "survey" of cnc hardwood resources in the Grampians and will return to my blogging duties soon.


Fortunately, we don't have grizzly bears
(Ursus arctos horribilis).