[electrotwelve]

I'm trying to get some basic theory right about the maximum step frequency that my software should output. I got some formulas from here and lets say I want my machine to move at the rate of 60mm/s and needs 80 steps to travel 1mm. This means my step rate is 60 * 80 = 4800 steps per second. Does this mean that the step frequency is 4.8Khz? Each step would have a pulse on and off time and how would that be accounted?

[Jim Dawson]

Yes, the frequency would be 4.8kHz. Normally the output from the computer is a 50% duty cycle square wave, but could be a PWM looking signal with a variable duty cycle.

[grg12]

Yes, the frequency would be 4.8kHz. Normally the output from the computer is a 50% duty cycle square wave, but could be a PWM looking signal with a variable duty cycle.

I don't know how today's software behaves but 10years ago mach and linuxcnc outputed series of short spikes - not a 50% square. I designed and built breakout board with optoisolation only to discover that transoptors selected for step frequency (20kHz) were way to slow for step pulses.

[109jb]

I don't know how today's software behaves but 10years ago mach and linuxcnc outputed series of short spikes - not a 50% square. I designed and built breakout board with optoisolation only to discover that transoptors selected for step frequency (20kHz) were way to slow for step pulses.

LinuxCNC has used square wave forat least 10 years. I know this for a fact because i used an oscilloscope to view the wave firm when i started my first mil and that was over 10 years ago. I cannot imagine it being anything but a square wave even far before this because it would actually be harder to make the output a spike than a square wave.

Sent from my RS988 using Tapatalk

[grg12]

LinuxCNC has used square wave forat least 10 years. I know this for a fact because i used an oscilloscope to view the wave firm when i started my first mil and that was over 10 years ago. I cannot imagine it being anything but a square wave even far before this because it would actually be harder to make the output a spike than a square wave.

Sent from my RS988 using Tapatalk

I must admit i didn't measure linuxcnc - just mach.
Generating spikes is easier that square wave - all you need to create spike is to set and immediately clear port. For square wave you need to calculate how long the port should stay on, and shedule port clear on next kernel clock cycle - at maximum speed you need double kernel frequency.

[ger21]

Did you try adjusting the pulse width in Mach3 when you tested?

[grg12]

Did you try adjusting the pulse width in Mach3 when you tested?

Yes - but even with max value pulses were too short for my hardware - eventually i scrapped the board and build new one with faster optocouplers. "Slow" board used cheap LTV827 optocouplers, "fast" HCPL2630. Since then i dont have any problems with signal quality ☺

[ger21]

Many people mistakenly believe that the max is 5us (Mach3 displays 0-5), but you can actually set it to 15 or 20us, maybe even higher. You can also use Sherline mode, which gives a 40us pulse width.

Also, the quality of pulses can vary greatly from PC to PC. Some PC's will be much better than others.