Gecko Motors and Drives

[GHR (Unregistered Guest)]

I read a few posts regarding the retrofit of the PRT Alpha with Gecko motors and drives.

How much of an improvement did you see with overall accuracy and smoothness of curved cuts? Which specific components are required for a complete retrofit, and what is the estimated total cost? How much fabrication is required to make the new motors fit the existing mounts, and are new pinion gears needed?

Thanks.

[gerald_d]

To my knowledge, no PRT Alpha's have been retrofitted with anything else. Where are these posts that you are referring to?

There is no such thing as a Gecko motor. There are things called Gecko Drives, but they drive standard motors.

I have experimented with a Gecko setup on a standard PRT machine (not Alpha), and I was not blown away by the improvement. The hype was bigger than the bite. Converting an Alpha to Gecko's sounds like a downgrade.

[richards]

I agree with Gerald. On my test bench, I have four Gecko G212 drivers that drive some Oriental Motor stepper motors. With the testing that I have done, a Gecko driven system would probably give similar feed speeds to the Oriental Motor Alpha drives - if the motors selected were carefully matched to the Gecko drivers. Remember that the stepper drivers and the stepper motors are only one part of a CNC machine. There are many more parts that are required to make a reliable system, including software. Even though the Mach software, that I use on the test bench, is excellent software, it is not nearly as user friendly as the SB3 software supplied by Shopbot. And, to be productive, you would still have to buy a CAM package, such as PartWizard or Vector, to generate the tool paths - unless you really enjoy writing code line by tedious line.

[stevem]

Gerald, what do you consider as "blown away"?
Are you getting less than double the move speeds?
Did you not get a huge improvement in smoothness?

In order to take full advantage of the Gecko drives and Mach software, one would have to replace the original stepper motors with ones that have more torque. The original steppers are the bare minimum for reasonable cutting performance.

I have to admit that I'm still using the original steppers, but still getting 11 IPS move speeds with a huge increase in smoothness.

[gerald_d]

Steve, here (http://www.mechmate.com/scapenotes/notes/messages/2020/4760.html) are the results. The samples marked M2 are gecko/mach2 driven. I explained the test somewhere else (can't find it now) but it was a quick Saturday morning test and a lot more tuning could have been done. Anyway, I might change my mind when I get the bigger machine running with the new gantry and all.

[benchmark]

Hi Steve

What is you Shopbot and computer setup ???? my PRT seems to fall over itself when it gets above 4 IPM speed in either move or jog.

Regards

Paul

[richards]

Paul,

I believe that Gerald is working in metric. Forty to eighty millimeters per second translates to moderate speeds in inches per second.

[benchmark]

Opps !! typo....I should have been 4 IPS (101.92)

Mike, I knew that Gerald is using metric.. but Steve said he is getting 11 IPS move speeds
(297.4 mm)

Paul

[gerald_d]

Move speeds are very subjective. If it is very low-force cutting, then the move speed can be close to the jog speed. Going in straight lines is also very different to going around radii....etc. It's a bit like guys comparing the published top speeds of their cars.


Paul, if 12" = 305mm, then 11" must be 25mm less - okay it is nitpicking, but we have to show the Yanks that we understand their archaic inches.

[benchmark]

305 divided by 12 = 25.416666666666667

I like my woodwork accurate...LOL

[gerald_d]

....not bad when the inch is actually 25.4mm. Times 11 gives you 297.4mm???? (Should be 268.4mm)

Here's another useless one for you.....
Take pairs of the first odd numbers: 1 1 3 3 5 5
Divide them into 2 halves: 1 1 3 \ 3 5 5
Divide the second half by the first: 355/113
and you have a much more accurate value for pi (p)than 22/7. Useful for cheap calculators without a p button.

[benchmark]

I like your Trivia

My earlier post said 297.4 should be 279.4
I am not sure if it's my Dyslexia or the Alzheimers...
never mind pass the rubber tape measure

PS. where did the 268.4 come from

[gerald_d]

Ouch, I used 24.4 instead of 25.4 *blush furiously*

[GHR (Unregistered Guest)]

Oops, my original post had a typo.

I was referring to the PRT NON-alpha. Stupid late night mishap.

In any case, I'm getting the same scallopped cutting that Gerald showed in his pictures. To me, this is quite unacceptable and limiting to what I want to do with this machine (mill solid wood parts for furniture).

In this thread: http://www.talkshopbot.com/forum/messages/312/8846.html#POST26852
Brady Watson stated that his friend is getting perfect acrylic circles cut after the stepper and Gecko upgrade.

Gerald, in your test photos, are you using the same Vexta motors in all four tests? Would the Gecko Drives and Mach2 software be more effective with better steppers? If so, which steppers and where do I buy them? Thanks.

[Brady Watson]

GHR,
The setup that I spoke of is using Mach2 with a stiffened gantry and stock PRT motors. Before you take the plunge an go the Mach2/3 route, take the time to stiffen the PRT gantry and the table itself. It is amazing how much vibration and shaking that the old sloppy PRT gantries take up compared to after I welded it. You can clearly see the table shaking after I welded it up...so stiffening the table is a BIG factor in cut quality.

Another very big contributor that is not often mentioned is the pinion gears. These are considered a consumable. If you do a lot of repetative movement with 3D files like I do, then chances are you need to replace them...trust me, replacing them with new ones makes a HUGE difference in quality.

Then when you are done all of that stuff, look into the Mach2/3 stuff.

-Brady

[stevem]

I rarely cut wood, but will do some testing to see the quality of the cut on MDF. Will post pics.

BTW, the plastic I normally cut on the Bot has a very smooth finished edge.

[benchmark]

Hi Steve

What is you Shopbot and computer setup ???? my PRT seems to fall over itself when it gets above 4 IPS speed in either move or jog.

Regards

Paul

[stevem]

My Bot is stock, except for the following changes: welded steel tube X gantry, Gecko drives and Mach2 control software.

Just to clarify, I use “move” and “jog” interchangeably to describe rapid moves between cuts. The most efficient cut speed for the Kydex plastic that I cut is 2.5 IPS. The steppers seem to run out of torque at cut speeds above 4 IPS when cutting MDF .5” thick in a single pass. If I was cutting a lot of wood, I would replace the steppers with ones that have more torque.

[gerald_d]

"are you using the same Vexta motors in all four tests? Would the Gecko Drives and Mach2 software be more effective with better steppers? If so, which steppers and where do I buy them?"

Yes, same steppers in all the tests. The Vexta steppers are of the best you can get. You have a wrong understanding of the basics if you think that the "brand" of stepper makes a differerence.

I support Brady on the issue of worn-out pinion gears. The tips of the gear teeth are not supposed to contact the roots of the valleys - only the flanks of the gears must make contact. For a gear drive to be smooth, the "radius" of the gear must be smooth and constant. The smooth radius is through the flanks, and not via the tip at one point and then the valley root at the next. Grease also makes a big difference in cut quality, as does spring tension. In fact everything between the motor armature and the rack makes a difference, particularly if gearboxes and toothed belts enter the picture. Gearboxes/belts can improve the resolution of the steps but they introduce either more backlash or friction.

Have you guys noticed in those pics that the big scallops actually line up with each other - surely that tells us we are looking for something mechanical, not related to driver electronics and software?

Steve, I can't remember now, but do your steppers have gearboxes?

[stevem]

Gerald, yes, steppers with gearboxes.

As an aside, I have seen several Bots in operation, but have yet to see one that has grease on the racks. Am I the only one that greases the racks and pinions?

[gerald_d]

Steve, you already have more torque than the guys without gearboxes. If you do want to go for different motors, have you considered the servo route instead of steppers?

(A search on rack + grease is interesting)

[stevem]

Gerald, I believe the steppers without gearboxes are double stack and have the same torque at the pinion.

As for servos, I’m waiting for the G-rex external pulse generator and Mach4 to become available. This setup should give the same performance as the Alpha drives and motors while using regular steppers with encoders.

[gerald_d]

Steve, the guys with Alpha drives seem to be getting higher speeds, but they don't seem to be getting significantly higher quality. ie, they seem to have similar degrees of chatter/scallops/whatever. I somehow think that we already have as good as it can get with steppers and that a quantum leap to servos is necessary. I gave up on chasing the Mach numbers after burning my fingers with a couple of Mach3 versions. (Just had a peep and saw that Mach3 has had about 6 version changes in the last 4 days!) I wouldn't hold any breath for Mach4 yet.

[Brady Watson]

Gerald,
FYI Mach4 is the 'industrial version' that is supposed to cost upwards of $600 (last I read)...and they are not really 'there' yet with the GRex, but they are getting there...

-Brady

[richards]

Brady,
The GRex junior (G101/G102) model that I'm testing is an excellent module. The MachIV prototype software does an amazingly good job of spinning the steppers - at high speeds. The pulse rate is much smoother than the pulse rate produced via Windows and Mach3 - which should be expected, since the G1xx system produces its own interrupts and doesn't time slice the Windows XP operating system. Of course, I will be very surprised if the GRex and MachIV are released before summer of 2006.

Gerald,
I must be missing something in my understanding of servos. If I understand correctly, the resolution of a servo is determined by the encoder (and the ability of the servo driver to monitor the encoder). If the resolution of the encoder is the same as the step resolution of a stepper motor, and, if the stepper motor is not driven so hard that it misses steps, it seems to me that the movement generated would be identical between the two types of motors. I know that the 'big iron' machines use servos, but I've always assumed that they've choosen that type of motor because of the mass that they're moving. In a low mass machine, like the Shopbot, where a lot of structural weight is not being moved around, do you think that a servo would reduce chatter/scallops?

[gerald_d]

Mike, I don't know the answers, and I am throwing up things for discussion only.....

I know that you think the "scatters" are not related to speed, but my linked pictures reminded me why I believe that speed is a factor. Stepper torque reduces drastically with increased speed, while I believe that servo torque stays constant? Steppers are inherently rough with obvious magnetic "detents", servos are inherently smooth? Need to scratch around in here a bit......Steppers vs Servos (http://www.google.com/search?sourceid=navclient&ie=UTF-8&rls=GGLD%2CGGLD:2005-07%2CGGLD:en&q=stepper%2Bvs%2Bservo)

[Brady Watson]

Mike,
Thanks for the clarification. Great to hear about the smoothness of the Mach4/G101 setup. I wanted to use these for a project, but figured that I would wait until others wasted their money figuring it all out 1st


Gerald,
My understanding is that steppers and servos, from a torque perspective are inversely proportional. This is why you would be hard-pressed to find a servo system that does not implement some sort of gear reduction. Servos typically need to run 50-80% (from what I understand) of their max RPM in order to be efficient. So...a servo system still needs to be designed to have a 'sweet spot' in order to get the type of performance we are looking for. The gear reduction goes a long way for low-speed performance and cogging issues.

Selecting the right encoder can be tricky because the higher the cpr, the lower the possible top speed (but higher the resolution) depending on how much bandwidth your computer can handle when throwing out a pulse stream to the servo drivers. You may not need .00004" 'step' resolution with a 2000 cpr encoder that gives you a top speed of 8IPS(just as example)...but you may want to run a 500cpr encoder that gives you .00016 'step res' and 32IPS max speed...if you follow me. Servo performance is in many cases dependent upon your processor's ability to throw out pulses...this is where the G101/102 comes in...with serious pulse speed/'bandwidth'.

-Brady

[gerald_d]

My crude theory is that a stepper "chokes" itself to run out of breath (torque) at high speed because of the increasing impedance at the higher frequency - in other words, the current can't reach the motor because it is too busy going backwards and forwards in the cables!