[Smertrios]

I have a gecko servo setup using kelinginc size 23 servos (the smaller ones) and a 35v 15amp power supply. I've had the gecko velocity/acceleration set at...

X = 90,6
Y = 90,6
Z = 60,6

If I try I can get the speeds up over 120 but there is a very good chance of faulting the servo drives. Today I was experimenting and set the acceleration to high values compared to 6 using 90 acceleration and it seems to be working. *But* is it?

IMO: When using modes like constant velocity its very important to have the velocity set as high as possible to keep the rounding off of corners to a minimum when using high feed rates (for internal corners and 3d milling).

EDIT: changed the word in bold from velocity to acceleration

[SignTorch]

for any given velocity, you must maximize acceleration to minimize CV corner rounding

need more info on the whole drive system, size and gearing, wiring, loading, and comprehensive problem description,

if you're running the motor free spinning, or direct driving a rusty screw, or sliding on aluminum tubing, or moving a 100 pound gantry, or something like that, then that might be the problem, or it could be encoder electrical issues, or servo tuning, or supply issues, or any combination thereof

[Smertrios]

I have a TAIG with 20 TPI screws. The machine is about 6 years old but in near mint condition.

[SignTorch]

20 tpi ???

I have a 1/2-10 tpi acme screw on the larger 23 frame servo with a 60 volt 20 amp supply, 2000 ppr encoders, 1:4 belt gearing, that barely achieves 150ipm (1500 screw rpm)

it can go faster, but noise, vibration, and friction makes it impractical and unreliable, with geckos I can't tell if it faults due to position error or current limiting, but at 1500 rpm it will fault, and I have to run 45khz clock to go that fast (with 1:1 step multiplier on gecko (I think))

servo tuning sure affects whether it can even maintain 130 ipm reliably, but it doesn't feel or sound good at that speed so I run it at 100 ipm, and tune carefully, to prevent sporadic faulting

my X and Y axii are on rack and pinion and they don't have any problem, the screw drive is just way more problematic

without knowing your gearing, encoder count, step multiplier, clock speed, tuning quality, etc...

to get 90 ipm at 20 tpi would require 1800 rpm on the screw

I don't suppose you have 4:1 gearing or else the motor would have to turn 7200 rpm

90 ipm sounds too fast for your motors to me, screws are so inefficient, the friction alone overloads the motor, 35v is not a lot of muscle, especially without 4:1 gearing

why not try limiting velocity to 50 (or lower) and raise acceleration (as much as possible), that might give better overall performance than having really high velocity with really low acceleration

I run my Z screw at 100 ipm + 20 accelleration

and my X/Y racks at 200 ipm + 30 acceleration

but keep in mind I have 60 volts and bigger motors to work with

I mostly cut small detail with small bits well below 50 ipm, so good acceleration is way more helpful than high velocity for me....

[Smertrios]

With no ballscrews to backdrive IMO no additional gearing is necessary. The 20 TPI screws give the TAIG all the mechanical advantage it could possibly need. The movement is smooth!

[SignTorch]

at higher speed the mechanical advantage gives way to screw friction, only a small percentage of motor torque goes toward moving the load, every little bit the servo tries to speed up the screw builds up a lot more friction, the screw itself becomes more load than the motor can turn

if you try to go 90ipm (1800 screw rpm) with no gearing the motor is probably running about half speed (1800 rpm)

if you have 2:1 gearing, the motor can go to 3600 rpm, more motor inertia helps keep the screw turning, there's less load on the motor, higher speed and acceleration is possible, 3:1 might be more ideal

but if you can run 90 ipm with decent acceleration without gearing, that sounds pretty good to me, better than I would have expected, I'm not advocating that you need gearing per se...

[Smertrios]

I didn't state my question well. I was thinking about how to see the error during acceleration so I can find the right numbers to minimize toolpath errors small as they may be.

For cutting aluminum I dont worry about acceleration errors with the relatively slow speeds. I also cut paraffin wax to make molds from and if the toolpath has no errors from overshooting or undercutting the programmed position I can use maximum speeds. Thing is I dont have a way to know the amount of error if any...

[SignTorch]

if the motor lags more than 1 encoder count off position then the green light goes off on the gecko, if it doesn't blink during acceleration then that means the motor position should be good, if it flickers or turns yellow that would mean the motor is lagging

I don't know how to detemine actual error at the tool, except the little lighted 30x microscopes from radio shack or harbor freight work good for inspecting the cutter and the cuts

to measure backlash I just pocket a circle and measure the height and width

[Jeff-Birt]

to measure backlash I just pocket a circle and measure the height and width

That is not a very accurate method to measure backlash. To measure backlash you need a test indicator and a piece of stock to indicate against (mounted to table.) You put the test indicator in the spindle (stuff a rag under the pulley to keep spindle from rotating) and jog the machine so the needle just touches the side of the block on the table. Now jog forward about 0.010" to preload the indicator against the block. Now jog the opposite direction 0.001" at a time until the indicator needle moves. For example if you jog back three times and the indicator moves back only 0.0005" then you know you have 0.0025" of backlash.

[Smertrios]

The green LED must be for newer gecko drives. Mine is a 203 and just has a red led that lights up when it faults.

If this math is correct the error would have to be HUGE to matter much (maybe). I am assuming that even by just tuning by ear listening for oscillations the acceleration can be set quite high and never amount to a huge error but its just an assumption.

20 TPI screws
300 CPR encoders (1200 graycodes or positions)

1/20 = .05 (travel per revolution)

.05 / 1200 = 0.000041666 (travel per encoder position)

assuming the geckos maximum error before faulting is -127 to +128 then the most the position can vary from the programmed position is approximately

0.005333 = (travel per 128 encoder positions)

[SignTorch]

oh, yes the green light is on the 320x, your figures look good, that is the motor position, there would be additional position error at the tool tip

on backlash, same difference if you measure a hole or use a dial indicator, neither method is "very accurate", the spindle axis and tool flexes when cutting at speed more than when jogging .001 against a dial indicator, the former is the total working backlash (including any belt stretch, machine flex, overshoot, etc.), it's possible to measure near zero backlash on an indicator and still not be able to cut a round hole accurately (due to backlash).

[Jeff-Birt]

the spindle axis and tool flexes when cutting at speed more than when jogging .001 against a dial indicator,

Neither of these things are backlash though. The way I described measuring backlash is very accurate at measuring backlash. Tool flexure and lack of machine rigidity are completely different issues. (That both affect the quality of output.)

It is sort of like tire pressure and body roll will both affect the handling of your car, you can measure tire pressure with a good tire gauge and fill your tires accordingly. While you can maintain correct tire pressure it has nothing to do with body roll. (For body roll you would look at the cars suspension and/or sway bar.) Your premise applied to this analogy would be that correct tire pressure is meaningless as it does not also correct body roll.

[BAMCNC.COM]

Look up Taig x axis backlash on YouTube.

[SignTorch]

Neither of these things are backlash though. The way I described measuring backlash is very accurate at measuring backlash. Tool flexure and lack of machine rigidity are completely different issues. (That both affect the quality of output.)

It is sort of like tire pressure and body roll will both affect the handling of your car, you can measure tire pressure with a good tire gauge and fill your tires accordingly. While you can maintain correct tire pressure it has nothing to do with body roll. (For body roll you would look at the cars suspension and/or sway bar.) Your premise applied to this analogy would be that correct tire pressure is meaningless as it does not also correct body roll.

what? you're the one saying that measuring screw backlash in .001 steps with an indicator is "very accurate" and that machine play is not backlash

I said to measure [total] backlash I just cut a circle and measure it, nowhere did I imply that screw backlash is meaningless

backlash [ˈbækˌlæʃ]
n
1. (Engineering / Mechanical Engineering) a reaction or recoil between interacting worn or badly fitting parts in a mechanism
2. (Engineering / Mechanical Engineering) the play between parts

[Smertrios]

TAIG CNC has those youtube videos linked

I check my axis for backlash

X = .001
Y = .0025

It has not always been this bad especially for the Y but it has been a long time since I checked for backlash. After tightning the bearing preload and the nuts I get .0002 for the X and .0005ish for the Y. The Y has always been hard for me to adjust for some reason.

Most of the lash for the X was in the bearing preload or rather the lack of it and for the Y it was in the nut.

Back to topic...

Rapid to 90 with acceleration set to 70 is maximum now but its pushing a limit going that fast. With acceleration set to 40 the machine can rapidly reverse direction without faulting like what would happen with peck drilling but that is also pushing a speed limit. Rapidly changing directions at rapid speed doesn't happen often for the X and Y axis so I am ok with it.

When 3D millling instant reversals of motions are rare. IMO 40 is going to be ok and will give excellent cycle times without sacrificing accuracy.

EDIT: while on the subject of cycle times another thing that *might* increase milling speeds is to change the "general config..." setting for "Look ahead". The max is 1000 lines and if you have *alot* of small moves increasing this number from the default of 20 can increase milling speed. Thats only *if* 20 was not enough.

I tested this using some very small .001 to .00004 moves and the increase in milling speed is significant.

[SignTorch]

sounds like good numbers to me, I'll have to give those motors another look, the bigger ones are more than I needed

[Jeff-Birt]

@signtorch - in machining terminology backlash is the unwanted movement between things like a leadscrew/nut or two gears (in a gear reduction), in other words between the driven parts of an axis. Unwanted movement from other things like loose gibs, etc is not backlash. The benefit of everyone knowing the terminology makes it easier to talk about it and share ideas.

Also if you'll re-read my method of measuring backlash you'll notice I said to jog 0.001" at a time until the needle moves. If you jog three times, that is 0.003" and the needle moves 0.0005" the difference is 0.0025", this 0.0025" is backlash. (This is using a test indicator with 0.0005" graduations.) By adjusting the lead-nuts you can get the backlash down to less than 0.001".

@Smertrios - good point about checking the bearing preload, the coupler is another place to check. Too large a gap on the Taig stepper coupler and you can introduce 0.0005" of backlash easily.

I have a customer who runs a lot of lithopanes on his Taig, lithos use lots of short linear moves of all three axis. He sent me a test file that was about 300,000 lines long and took about 2.5 hours to run. I tested various acceleration and velocity settings to find out how it effected the total machining time of the litho without loosing steps on an axis. With the velocity set to 80IPM and the accelerations at 12(Z), and 14(X, Y) it cut 15~20 minutes off the cycle time and did not miss a step. With the acceleration set higher (say about 20) the cycle time of the part was only reduced about 2 minutes (in a 2.5 hour long run) and the Z axis would miss a few steps. The point here is that crazy high acceleration numbers do not really increase performance but they do greatly increase the likelihood that you miss a step from over acceleration.

[SignTorch]

@signtorch - in machining terminology backlash is the unwanted movement between things like a leadscrew/nut or two gears (in a gear reduction), in other words between the driven parts of an axis. Unwanted movement from other things like loose gibs, etc is not backlash. The benefit of everyone knowing the terminology makes it easier to talk about it and share ideas.

to suggest there is some difference between wanted and unwanted motion and some moving parts are not driven parts and then preach about knowing terminology and all that

just because I said I check backlash by cutting and measuring a hole

is not making this conversation any easier

please address the topic, not me, I'm not here to argue...,

[awerby]

You can figure out if you've got a problem by attempting to do something, like cutting a circle, and failing at it. But if you're trying to do something about it, then it really helps to narrow down the issue. If it's really backlash, then there are certain things you can do to minimize it. But those won't help if what's causing the problem is something else, like loose bearings, misalignment or bad leadscrews. If you're looking for help in fixing your problem, it's good to state it as clearly as possible. That's why it's important that we all use the same words to describe the same problems; otherwise we're working at cross-purposes, wasting everybody's time without getting you any closer to a solution.

Andrew Werby
www.computersculpture.com

[SignTorch]

I wasn't wasting time or being unhelpful

Backlash is backlash, it is the difference between the target position and the actual position

If you need more clarity then you can qualify it as screw backlash, gear backlash, or some specific type of backlash

Here's the first few definitions I find on google, it matches the dictionary definition precisely, argue away...

• Backlash is any non-movement that occurs during axis reversals
• According to the CNC Dictionary, backlash is any kind of unexpected play in an axis due to clearance or looseness of mechanical parts
• CNC Backlash is any kind of play which is found in cnc machine axis.
• Backlash is measured at different locations for different reasons
• I only have about .004 - .006 backlash but it's enough to make circular pockets eggshaped.
• backlash is the striking back of connected wheels in a piece of mechanism when pressure is applied. Another source defines it as the maximum distance through which one part of something can be moved without moving a connected part.[1] In the context of gears backlash, sometimes called lash or play, is clearance between mating components, or the amount of lost motion due to clearance or slackness when movement is reversed and contact is re-established. For example, in a pair of gears backlash is the amount of clearance between mated gear teeth.
• Backlash is play in the drive gears
• Backlash is defined as the amount of freeplay between the motor shaft and the load. It is determined by measuring the amount of angular movement on the shaft ...
• Is there a backlash against feminism? Who declares that there is a backlash against feminism and what do they mean by this phrase?
• A "backlash" is a popular negative reaction to something which has gained popularity, prominence, or influence.
• The term "Backlash" originated from gear engineering discipline. In JIS terminology, this word is described as: Play that exists between a pair of engaged gears' ...