Trying to optimize base period, etc.

I built a JGRO inspired machine a couple years ago and I'm now trying to optimize the setup in hopes of slightly increasing travel speeds. I have Longs Motors 280Oz-in steppers with DM542A drivers and a CNC4PC C50 breakout board. The setup I have been using uses the default "Other" stepper timing of 5000, 5000, 20000, 20000 with a latency of 16000. I have since upgraded to Ubuntu 12.04 and Linuxcnc 2.7.2. I have run the latency histogram with the new configuration and found the same latency.

Looking at the ini file I see the base period set to 100000 not the 31000 the stepconf wizzard suggested. The StepTimingCalculatorPlus spreadsheet suggests a minimum base period of 22usec with this setup. Using the 2500, 2500, 10000, 8000 stepper timing from the wiki chart the spreadsheet suggests a minimum of 19.5usec. Adjusting the base-periods in the ini file and step configuration in the hal file per the spreadsheet I get an error pointing me to check dmesg with a base period of 55usec or less. I get no error with 60usec or greater.

Why does stepconf set the base period to 100usec? What am I missing that my minimum base period seems to be 3x the spreadsheet's prediction?

All of this may be academic as I seem to be limited to about 3200Hz with this stepper/ driver combination regardless of configuration.

John

What do you have the micro-stepping set for? Full or half stepping will usually give the best results for top speeds.
What voltage are you supplying your drivers with? Increasing the voltage is often the only way to increase the top speed of a stepper. (if step generation isn't your limiting factor)

All axis are set for half stepping. Full stepping is not an option with these drivers.
Nominal 36V power supply, measures 38.3V at the input to the drivers. Drivers rated for 18-50V. The stepper motor data is attached.

I don't really understand the voltage limits of the motors. If the motors can take the higher voltage it sounds like a 48V power supply might be a worthwhile upgrade.

John

I don't really understand the voltage limits of the motors. If the motors can take the higher voltage it sounds like a 48V power supply might be a worthwhile upgrade.

You dont need to worry about stepper motor voltages and amperages, the driver takes care of that.

So long as your PSU is regulated, just go to the highest voltage your drivers will take.

going to 48v , will increase your speed , and a bonus you'll get more torq to
just don't go over the rated voltage of the stepper drivers

Tom,
The Y and Z axis are just 3/8-16 threaded rod. These are limited to about 30ips. The X axis is a ball screw with a 5mm lead (5.08 rev/in). The X axis moves at 96ips with a little lower resolution but adequate for my needs. In each case above ~3200Hz the motors stop turning and emit a louder vibration/tone.

You wrote:
"On the other hand you have the pulses and with 3200hz you mentioned divided by 400 for half step you get 8 revolutions per second or 480 rpm from the motors so no voltage or current will help there."

Are you saying these drives/motors are limited to 3200Hz and increasing the voltage will not change this? Stepconf says the max step rate is 32258Hz but it also says the minimum base period id 31000ns even though it sets it to 100000. As you suggested even with a 100000ns base period the system should be capable of ~10000Hz.

I am also looking at rebuilding the machine. Now that I have used it awhile I see many opportunities for improvement. Replacing the threaded rod with better lead screws with more lead is certainly one improvement. I always figured I would use this machine to build the next one. I'm just trying to better understand my driver/motor capabilities so I keep any future mechanical changes within their limits.

Thanks.

John

In each case above ~3200Hz the motors stop turning and emit a louder vibration/tone.

In this case it is probably not the 3200Hz that is stalling the motors but the motors are unable to provide the torque to move at that speed.
Keep in mine with stepper motors after a certain rpm their torque drops off very quickly, as noted previously in this thread higher voltage may help. But it does have it's limits.
For example...
www.automationtechnologiesinc.com/wp-con...4/11/Torquecurve.pdf
Balls screws are more efficient and available in ratios that may work better for you.

Rick G

A quick follow-up...

I changed to a 48V power supply and now each axis is able to move at >4000Hz. About a 25% improvement is great for such a simple change. As part of a general mechanical rebuilding of the machine I purchased a 250mm C-Beam actuator bundle from Openbuilds during their cyber Monday sale. It uses a 8mm leadscrew with a 8mm lead. The reduced friction of the leadscrew vs. 3/8-16 threaded rod allows much faster travel. Changing to 1/4 stepping I get about 120 IPS at around 5000Hz. It will probably go faster but my Y axis isn't stiff enough to handle the forces generated by these speeds.

Y-axis upgrades are next...

Thanks for the help.

FYI - The C-Beam internal wheels allow too much deflection. If you are interested in the Openbuilds stuff go with a wider plate and external wheels.

John

I'm new to linuxCNC and I haven't quite grasped the limits and relationships between things. I'm currently running the nema42 stepper motors w/ no load and am getting some errors. If I run @ 200 s/r or 400 s/r (2 micros) I have no errors. But if I go to 4 or more microsteps I either get the realtime error OR if I manually increase the base period in the BOB.ini file from 62,000 to something greater I get "following errors" in running the axis test file.

Is there a theoretical (empirical) limit to the number of micro-steps an parport can do? I've never seen my jitter higher than 5.1us. Setting higher jitter doesn't seem to solve the "unexpected realtime delay on task 1". It also doesn't seem to matter w/ the step timings (for example 2500/2500/5000/5000). Not really sure what the MA2722M should be but I'm probably in the ball park.

Just trying to understand the best practices. Thanks!