CNC software highest step rate

What do you mean by step rate? Instructions/sec?

snipped-for-privacy@hotmail.com wrote:

Mach3 will go as high as 45 kHz, although I can't imagine why. Mechanical resonance limits practical speeds to below 25 kHz.

That's low IMO, you mean there isn't anything out there approaching

100kHz?

Typically, the same software is used by step compatible servo drives. Example, a 4X only servo drive (Gecko 320 for example) driving a 2600 RPM DC motor with a 500 CPR encoder mounted on the motor shaft needs a step rate of 88kHz in order for the motor to reach maximum speed. Resonance is a non-issue.

To clarify, I am not aware of faster speeds. Available step rate is so not the limiting factor on my setup. Unless you're talking about something else entirely, of course.

Sure, if you can get 2600 rpm into your system. Mechanical resonance overwhelms steppers at about 1600 rpm. What are you looking at that it becomes a "non-issue"? I had heard, I think, about the 320's adaptive micro-stepping. If it's that effective, I might give it a shot. 300 ipm seems plenty, but faster is always gooderer.

As for step rate on a PC, EMC runs on RTLinux. Dunno why it thinks it needs realtime to run a few simple steppers, but maybe worth looking at for your application.

[If you really want a higher step pulse rate, you could get a G-101 "G-rex" board from Geckodrive. They claim to put out over 4 million step pulses per second: ] [It's a non-issue because it's driving a servo motor; they don't have the same resonance problems as steppers. While the control can send the same step/direction signals that control steppers, the 320 drives will accept those and convert them to something servos can run with. 2600 rpm motors usually require some belt reduction to be useful in a CNC system, though. ] [Realtime is necessary to avoid having your milling job disrupted when the OS decides to reshuffle its memory, or do other "housekeeping" functions. Mach2 gets around this by infecting Windows like a virus, to grab first priority for step pulsing. Most other Windows-based systems, like Flashcut's, use a dedicated pulse generator of some sort, instead of forcing the computer to do it.]

Andrew Werby

>

This is not a totally software solution, but I have a device that allows step rates up to 300,000 steps/second on each of 4 axes. The boards can be daisy-chained for more axes, if needed. It is supported by drivers in the EMC software. See

more info.

Jon

How about inches / sec Martin Eastburn @ home at Lions' Lair with our computer lionslair at consolidated dot net NRA LOH, NRA Life NRA Second Amendment Task Force Charter Founder

Tim Killian wrote:

If you want to look at a complete system with high performance, check out

They have controllers which can handle

40,000 steps/sec per channel (max 4 channels), with high current capability, and G-code interpreter all included. Regards, Ian Kirby.

According to Mike Young :

[ ... ]

Apparently -- a *non*-stepper fed from something which thinks that it is driving a stepper. Feed the steps that fast and the servo will be moving smoothly, instead of jumping and stopping to excite resonances.

EMC runs on a real-time kernel because it can *also* talk to analog servo motors, with encoders giving feedback as to the actual position, which I think is more interrupt intensive than driving a stepper. You've got three axes all reporting change of position at the same time (sometimes), and updates to the D/A converters to reset the servo speed to compensate for load or whatever, and possible interrupts from the limit switches as well. You want to give those limit switches very high priority so it can stop things before you hit a hard stop and damage the ballscrews and nuts or something else.

Enjoy, DoN.

I'd been curious about that. Why are the limit switches not typically hardwired to disable the drive, but rather simply reports impending crisis to the software? It seems something already has gone wrong by that time.

So, EMC can run closed loop directly without additional hardware support? (Beyond power drivers for the stepper, that is. ?) That's something, given that old PC's are essentially freebies. It would seem a perfect waste of computational resources otherwise. Replacing Gecko 320's, for example, with simple drivers would be a pretty big savings for the hobbyist.

In general, are servos considerably different from a stepper with an encoder?

Not really. Just wanted to know the highest step frequency one can expect from popular CNC software these days. Looks as though it's DeskCNC at 125 kHz because it uses the serial instead of parallel port. You'll need a compatible controller in order to get the needed step and direction outputs so it's really a software/hardware combo like other pulse generation schemes mentioned except probably more costly with even lower step rate.

Let's make certain we're on the same page first.....The Gecko 320 is a DC servo drive. Besides cost, its 4X only support is another limitation for the hobbyist who is likely to be looking at surplus DC motors. I bought one for evaluation and concluded it was not the drive for me. Instead, I went with simple microcontroller based drives with both 1X and 4X support at about half the price. So if we 're talking about the same Gecko drive then simpler and cheaper drives already exist.

Greetings Oparr, What do you mean by only 4X support. What are 1X and 4X? And why would

4X only be a problem with surplus motors? I have bought 3 of these drivers and the specs seem to me like they will work with the servo motors & encoders I already have and the step and direction software I have. Eric

On Thu, 22 Sep 2005 23:46:13 GMT, "oparr" wrote: Oparr, I forgot to ask what the cheaper drives are that you have found. I will be building a positioning table for plasma cutting this winter if time permits. Less expensive drives would be nice. Eric

According to Mike Young :

I agree in general -- the axis should be disabled in hardware -- at least in the direction towards that stop -- but it is also possible that other moves are happening at the same time, and you want the controller told at the earliest opportunity to *stop* right now!. :-)

As for having the hardware limit switches disable *all* motion at once, this is a bit of a problem with a machine designed from scratch as CNC. I used an Anilam retrofit on a Bridgeport clone mill at work some years ago, and whenever someone managed to hit a limit, the whole controller locked up until you grabbed the appropriate handwheel (and there was not even an indication of which one hit the stop, so if more than one axis was close to the stops you had to tweak them all), and then you hit the reset button to regain control.

However, replace that with a machine designed from the ground up for CNC (e.g. my old Bridgeport BOSS-3 machine -- and I believe all past that at least up to the BOSS-6, and probably to the BOSS-8), and you have *no* handwheels. The X, Y, and Z axis steppers (or servos for past the BOSS-6) are the *only* thing which controls the axes. The Z-axis you could grip the end of the motor spindle and back it off, but everything else was buried under belt guards, and the X-axis was also deep under the table, as instead of turning the leadscrew (which can have whip at high speeds), it mounted the leadscrew rigidly, and turned the ball nut in opposed bearings.

So -- with all electronics switched off by hitting a stop, you wound up with needing to partially disassemble the mill to gain access to something which could allow you to back off of the stop in question.

Now -- servo amplifiers come with sets of contacts to inhibit motion in a single direction, so you can lock out the drive in the direction which created the problem, but still allow backing away from the stop using jog controls.

Indeed so. It is the only way to work when using real servos as they are designed.

Stepper? This is with real servo motors, which need an amplifier which sums the speed command voltage from the controller (a Servo-to-go card in the PC in the case of EMC) with the tach generator feedback from the motor to get a precise speed from the motor (and thus from the axis), and the encoder is used to tell the controller exactly where you are at the moment. The Servo-to-go card also handles that, with a counter for each axis to deal with storing encoder pulses which arrive when the controller software can't catch them, so it can update when it can get back to things.

Well ... it depends on what you call "simple drivers". You need an amplifier capable of producing say +/- 40 VDC at at +/- 7A to a typical servo motor. And you need one per axis. I've gotten such servo amplifiers at hamfests and from eBay auctions for quite reasonable prices. But I consider it a pity that Gecko does not make a real servo amplifier, with command voltage input for speed, and with inputs from a tach generator for verification of the speed. They are more energy efficient, since they are running as switching mode regulators, instead of the analog pass transistors used in the servo amps which I have gotten. Each amp has its own Rotron muffin fan and big heatsink to keep the output transistors cool. (And its own transformer and power supply.) There are switching mode servo amps available which run from a common power supply, but they are still quite a bit more expensive than the servo amps which I have.

The fact that you can set a speed command voltage to the servo amp and get a steady speed from the servo motor is why resonance is not a problem at high speeds with servo motors used as intended.

If the Gecko 320 is the one which drives servo motors (but pretends that they are steppers), then it does nothing with the tach feedback from the servo motor, and uses the encoder to tell which "stepper motor step" it is at. Feeding it several steps very quickly will increment a counter which is compared to another counter run from the encoder which produces a voltage to drive the servo motor in the intended direction until the count from the encoder catches up. At certain speeds, you have the servo motor moving at pulsing speeds, just like the stepper that it is pretending to be, and thus have the resonance problem again.

If the Gecko were to honor the tach feedback, it could run at a smooth speed even with step pulses as input. One pluse behind gets a command voltage of 1mV. Two pulses gets 4mV, three pulses gets 16mV and so on, so the motor would speed up as needed to catch up, instead of having to try to run at full speed for a single pulse of offset.

Considerably so. I think that I have covered most of it above, in answering the other points. Note that I've taken a DC servo motor and amplifier, and set it up so 10VDC input is full speed from the motor, and then set the voltage down to 0.0001 V (the smallest that I could repeatably produce from the power supply in question), and had to stick some tape to the output shaft of the motor (sort of like a flag), and spend some time watching before I could even be sure that it was moving. So -- with the speed command, you can get a much smoother surface when cutting a shallow angle than you can with steppers. Well

-- micro-stepping might come close, but it takes a lot more work from the controller -- each micro-step has to be generated, instead of simply outputting a voltage proportional to the desired feed in that axis, and checking every so often that it was where it should be. If not , tweak the command voltage a bit until it is running at precisely the speed you want. The other axes will be running at their own commanded speeds to give you a very smooth cut. (Yes, encoders will limit your resolution for precise positions, but the slow motor motion will carry you through the intervening spaces to eliminate major steps.) (In contrast, when I tried to turn a Morse taper with my Compact-5/CNC lathe, I could see the steps (which were 0.002" diameter). So -- someday, I intend to retrofit the Compact-5 with servos and an EMC controller.

Note that the Gecko drive, with a servo motor, will still be producing the steps, because it does not try for a steady-state speed.

I hope that this helps, DoN.

Typically, the incremental quadrature encoder has A and B channels with pulses 90 degrees out of phase when rotating. Not only does this allow for the decoding of direction information but also allows for a pulse rate four times that of either channel. Hence the 4X.

If a servo drive only supports 4X then the step frequency required from software to support any encoder RPM is;

(RPM/60)*CPR*4

If the encoder is mounted on the motor shaft as opposed to a driven shaft geared down from the motor shaft then you have a worst case scenario in terms of required software step rate. If the highest step rate from software is less than the above when RPM is the maximum RPM of the motor then the motor will never be able to reach top speed.

1X support will reduce the above requirement by a factor of four.

Follow the link below. Note the voltage and current limitations when compared to the Gecko 320, they may not meet your requirements;

This is nonsense. At no time is the DC motor pretending to be anything. It is being itself driven by a PWM voltage always. Anything resembling resonance issues in steppers is due to either improper tuning or exceeding the specifications of the system.

The guys at cncteknix use DeskCNC as their control software, because it (DeskCNC) uses its own embedded chip for the interpreter function. Their hardware controller is designed to optimise the DeskCNC chip. Regards, Ian Kirby.