Need Help!- Am i limited to 100kHz (pps) when using Mach3?

[Cocacolalover]

This text is from Mach3 manual

What maximum rapid speed will we get? With 35,000 step pulses per second we get 5.83 revolutions [35000/(1200 * 5)] of the leadscrew per second. This is OK at about 9 seconds for 5" travel of the slide. Notice, however, that the speed is limited by the pulse rate from Mach3, not by the motor speed. This is only about 1750 rpm in the example. The limitation would be even worse if the encoder gave more pulses per revolution. It will often be necessary to use servo electronics with electronic gearing to overcome this limitation if you have high count encoders.

Is there any alternative to obtain a faster processing of pulses?
Using a dedicated controller board would solve this?
By my calculations, in my case, even at 100kHz, it would be far from max velocity the motor and rack allow.
Like it says in the text, using an encoder with fewer ppr makes it move faster, but increases basic lenght unit (CNC theoric minimal movement).

Am i going to run away from a pc based CNC....?

Thanks in advance

[H.O]

Mach3 supports up to 100kHz thru the LPT-port. You need a seriously fast computer to run it reliably at that speed though.

If you get a Smoothstepper you can get up to 4Mhz steprate, it connects to the PC thru USB and is pretty easy to setup. Some things it does better than the LPT-port, some things it does not so good - backlash compensation is still non existant but said to be "in the works".

There's also the KFlop motion controler from Dynomtion but I have very little info on how much is implemented in their plugin.

What kind of speed and resolution are you aiming for?

For example, with 10000 steps/inch resolution and 65kHz kernel speed you should be able to get 65000*60/10000=390IPM, if you have a computer that can run Mach3 reliably at 100kHz you'd get 100000*60/10000=600IPM. Same thing for metric, 1000 steps/mm, kernel speed 65kHz: 65000*60/1000=3900mm/min.

/Henrik.

[Cocacolalover]

Hello Henrik, thank you for answering me

I didn't gave too much detail because i was afraid to scare people away

Our project is a router table
This study concerns X axe only
I'm using:
Servomotors (2 in X axe)
5:1 gearbox (in study)
etc.

Motor revs per unit = 0,03125 revs/mm
Out pinion revs per unit = 0,00625 revs/mm

Motor nominal speed = 2000 rpm
Out RPM = 400 rpm

So max linear velocity is 1,07 m/s (this is velocity capacity due to motor conditions only)

Assuming that encoder is 2500 CPR (Cycles per revolution)
Motor ppr = 10000 ppr (4 times encoder CPR due to quadrature pulses)
Out ppr = 50000 ppr

So i need 312,5 pulses to move 1 mm (312,5 pulses/mm)
Then
BLU = 0,0032mm/pulse = 3,2 microns/pulse

(i'm actually accepting a resolution up to 25 microns as worst case)

So, assuming i can work MACH3 at 100kHz, axe X would be able to move at 320mm/s max, wich is pretty low for our pre-defined specifications (around 1000mm/s).

We are projecting a high accuracy router table (total error should be less than 100 micron), and high resolution (aluminium mold machining).

Everything else is already calculated (motors torque and rpm, rack, pinion, etc) but without being able to actually produce pulses at the required rate (around 300kHz ), speed won't be achievable.

*Only possible solution is taking an encoder with lower resolution, but it will affect the X axe resolution too (well, ok, maybe i'm being very demanding about resolution...)*

And, yes i am aware of Kmotion controller board for servo motors, as many other, but pulses are originated from Mach3 right?


Well, Merry Christmas

[Larken]

Cocacola
Is your router table built so accurate to need a 25 micron resolution ? Maybe on a super precision milling machine you need that resolution.

Can you use a step multiplier in the drive to reduce pulses/turn ?

With smooth-stepper, machs usb trajectory correction rate is only 1000 times a second, so at 4mhz, the machine axis can theoretically move 4000 pulses with out being corrected.
Really with a 4mhz step rate you should have a trajectory correction rate of at least 100khz. other wise your toolpath would be all over the place.

100khz step rate is really fast with a reasonable resolution.

Larry K

[H.O]

OK, you have 312,5 steps/mm and the top speed is 64000mm/min which means you'd need 333.3kHz step frequency. So, in short, with retained resolution you won't be able to reach your desired speed with the LPT-port driver - you need an external pulse engine/motion controler like the Smoothstepper or KFlop.

With that said, 60m/min is really fast! How big is that router?

The KFlop can drive analog drives if equiped with the KAnalog add-on card but if used by itself it can produce step- and direction signals, just as the LPT-port but at much higher frequencies. This is a quote from Dynomotions website:

KFLOP has 8 axes of Step and Direction. Output pulses up to 2.5 MHz can be generated.

See specifically the example section on Step and direction output

/Henrik.

[TOTALLYRC]

Originally Posted by Cocacolalover And, yes i am aware of Kmotion controller board for servo motors, as many other, but pulses are originated from Mach3 right? Well, Merry Christmas

Hi Cocoa lover,
When using an external motion controller such as the Smoothstepper, Kflop or DSPMC/ip, all Mach3 does is provide the trajectory information and the step pulses are generated externally on the motion control software. "Step" rate far in excess of what Mach3 is capable of through the p-port are now possible.
The smooth stepper will go 4mhz, I don't know what the kflop will do and the DSPMC is guaranteed to 2Mhz but in actuality the chips are capable of much more, I would talk to Rufi as to your specific needs.

I am with the other guys here in questioning your need for low Micron resolution on a router.

The resolution means nothing if the machine flex is orders of magnitude worse than the resolution.



Mike

[Cocacolalover]

Larken

Cocacola Is your router table built so accurate to need a 25 micron resolution ? Maybe on a super precision milling machine you need that resolution.

Yes, aluminium mold machining requires that kind of resolution

Can you use a step multiplier in the drive to reduce pulses/turn ?

I was forgetting one important point, 1000m/s is jog speed (no machining), machining speed will be much slower than 320mm/s, so probably, i won't even need the 300kHz pulse rating. (by my calculations, with 35kHz i could do 112mm/s wich is a good feed rate for aluminium)
During free movimentation, i don't need the machine to have an high resolution, so do you think electronic gearing (reduce pulses per turn) could apply during jog speed only? Will mach3 be able to do this?


H.O

With that said, 60m/min is really fast! How big is that router?

4x2- modular strucutre, able to increase in 2m increments along X axe (6x2, 8x2, etc)


TOTALLYRC

all Mach3 does is provide the trajectory information and the step pulses are generated externally on the motion control software. "Step" rate far in excess of what Mach3 is capable of through the p-port are now possible.

COOL, that's what i need.

The resolution means nothing if the machine flex is orders of magnitude worse than the resolution.

It's an high rigidity structure weighing an estimate of 700kg, and still being optimized, for lower deformations on critical zones. We are obtaining deformations around few microns (depending on the force direction) for forces that we will never experience during real machining (this using abaqus and solidworks of course)
Shure, the real thing will have an higher degree of innacuracy due to the process of assembly. But, we will try our best to attain client requisites.

[H.O]

During free movimentation, i don't need the machine to have an high resolution, so do you think electronic gearing (reduce pulses per turn) could apply during jog speed only? Will mach3 be able to do this?

Sorry, but there's no support for this built into Mach3.

Sounds like a cool machine, any chance we can see it?

Mike,
Has support for step and direction been added to the DSPMC now? I know there's been talk about it being added but I didn't know it was done. The web page doesn't say anything about it, as far as I can see.

/Henrik.

[TOTALLYRC]

Hi H.O.

If the step and direction has been added I haven't been told.
While Rufi and I talk often, he doesn't tell me everything.
I see that some of the ground work has been laid in the xml file but as far as actual implementation I don't know.

Even though you can't use step and direction yet the pps still applies to the ability to count the encoder pulses fast enough.

Mike

[Cocacolalover]

Originally Posted by H.O Sorry, but there's no support for this built into Mach3. Sounds like a cool machine, any chance we can see it? Mike, Has support for step and direction been added to the DSPMC now? I know there's been talk about it being added but I didn't know it was done. The web page doesn't say anything about it, as far as I can see. /Henrik.

Ok, electronic gearing is implemented in servo drives, not the software, sorry.

The machine is still in draft phase. There's nothing real to see, only virtual models and numbers (and if i show you pictures i might be fired , but we can discuss numbers freely)

I wasn't aware of that DSPMC/IP controller, seems pretty good
I am pointing to galil solutions though...


Regards

[mactec54]

Hi Cocacolalover

If cost is not a problem for your control, you then should look at Softservo Systems, your rapids can then go to Max 300m/m positioning speed, & high-speed/high-precision machining 60m/m, no problem with Encoder count from 12 bit to 17bit it may be more, I run at 16bit 16384 p/rev X4 Plus 3 dimensional dynamic look-ahead contour control (3DALCC) there is many more things your machine could get out of a control like this.

http://www.softservo.com/S100M.php?catID=3&subID=3

[yantra3d]

I can vouch for the kflop as I've been running a slant bed lathe w/ steppers and 2hp VFD for over 2 years now and more recently a closed-loop servo mill using an AC servo for spindle motor drive with the added snapamp board from dynomotion and they have been really outstanding in terms of reliability and performance. The onboard CPU does not require the pc or Mach to generate pulses and so there is no bottleneck to worry about in terms of speed - even though it does use a USB interconnect to pc.

For what it's worth I am very happy with Dynomotion's hardware and the software is pretty powerful allowing virtually any configuration you want via programmable interface. It also has sw to tune the motors, etc...

The boards are very high quality made as well and the support has been absolutely fantastic. I would definitely recommend their products for your application. If you have any questions I'd be happy to offer my experience on the kflop/snapamp and kmotion software. Best of luck!