On Tuesday 30 April 2019 22:22:14 E. Liddell wrote: [...] > Ah, well there we will have to disagree. I've always bought AMD CPUs, > and I'm quite pleased with my current Threadripper. Mind you, I'm not > using it for CNC or other realtime-sensitive applications. > > E. Liddell > Actually, I've been generally pleased with this now elderly 2.1GHz phenom. But its not well equiped to run machinery. latency-test shows 70,000,000 ns for the 1 millisecond servo-thread. Thats 70 milliseconds. Take away the base-thread which runs at 25 microsecond intervals and it is still around 50 milliseconds. And it would be 10x worse than that if I was using an nvidia driver. So I'm using nouveau. I'm not much of a gamer ubless you count solitaire. :) For GP computing, I've always favored amd just to keep competition alive. To demo the diff between software stepping, one of those 6 u-s machines has run a small mill for years, at a maximum table speed of about 12 ipm. A damaged ball screw put that mill out of commission, so that computer is now driving a 600x390mm gantry style mill, but it now has Mesa fpga based cards for the i/o because one parport doesn't give near enough i/o. But with those cards to offload the high frequency base thread from the cpu, that machine is moving at 210mm/second top speed, almost 20x faster. It gets homed at about 100mm/sec, and overshoots the switch less that half the overtravel it has. One switch is a button, and is mounted on a spring to protect the switch from being smashed. A 2" piece of .0325" thick sheet alu is the spring. The max flex might allow a 24lb piece of paper to be inserted as thats how fast the machine can be stopped. The other 3 similar function switches are std mini-style microswitches with roller tipped levers. Neither allows a hard crash, stopping the machine about 10 thou from it. But thats not the "home" point because once the switch has tripped, the machine moves away from the switch at 1/20th the search speed because the software only samples the switch at millisecond intervals. So when the switch releases, is the actual home position reference, and that gives me .0001" accuracy. The switches aren't that accurate of course but the individual switch in a bag of 10 for 4 dollars is very very repeatable. Thats about how far the motors turn when being microstepped at 1/8 step, a full step being 1.8 degrees of the motor shaft. Top speed is limited by two things, 1st being the speed of the opto isolators in the motor drivers limits us to a step rate of 200 kilohertz, and how fast the motors inductance can be overcome to establish the magnetic position. To do that we use power supply's of 10x or more than the nameplate label, its the current regulation that counts. Your trivia factoid for the day :) Cheers, Gene Heskett -- "There are four boxes to be used in defense of liberty: soap, ballot, jury, and ammo. Please use in that order." -Ed Howdershelt (Author) Genes Web page <http://geneslinuxbox.net:6309/gene>