On Wednesday 24 April 2019 18:06:47 Chris Albertson wrote: > On Wed, Apr 24, 2019 at 2:07 AM Gene Heskett <ghesk...@shentel.net> wrote: > > On Wednesday 24 April 2019 03:42:00 andy pugh wrote: > > > On Wed, 24 Apr 2019 at 01:30, Gene Heskett <ghesk...@shentel.net> > > > > wrote: > > > > Several, your personal config probably makes a far bigger > > > > difference than the actual version number of the kernel since > > > > you have to be pretty well up on YOUR hardware to re-configure > > > > the kernel you are about to build. > > > > > > And there is the problem. I have never yet managed to get that > > > step right and make a version that works on my PC. > > > (I tried and failed with the NTULinux repo last night) > > > > URL? > > > > > And the problem to be solved here is building a kernel that can > > > ship with LinuxCNC and work with _any_ kernel. > > > > > > Is it, in fact, currently impossible to provide an RTAI LinuxCNC > > > ISO with any kernel newer that the one we ship with Wheezy? > > > > IDK Andy. Has the current rtai kernel been tried on stretch? Nah, > > its 32 bit, claims PAE but is not. Stretch is 64 bit. > > > > > If that is the case then perhaps it is time to abandon RTAI > > > altogether? > > > > Might be, I have only 1 preempt-rt versions running here, not on > > software stepping though, both on mesa cards, a 7i90HD+3 ea 7i42TA's > > running the Sheldon, and the 5i25/7i76D running the G0704. But I've > > not changed the kernel on the 6040, that machine WAS formerly > > softstepping the old HF micro so its still running the live iso's > > kernel. But I've no clue if software stepping would run on it. If I > > software step it, it would be buckets slower, 10 ipm or so rapids, > > where its doing 200 now. And I'd have to share all my home switches > > for lack of i/o. > > > > I hope at the end of the day, that we don't have to abandon software > > stepping as thats the lowest cost entry point. > > SOftware stepping on PC hardware with Linux is "hard" but software > stepping otherwise is cheap and easy. > > I have right here on my desk a little machine that uses a Raspberry Pi > with "stock" Unbuntu 18.04 and it runs two servo DC motors or steppers > with nearly > perfect accuracy using a $3 "Real time processor". > > As I type, I'd using Fusion 360 to design an enclosure for the parts > and connectors. > I'm looking at a 180x120x40 mm box that house the Pi3 and a driver for > up to 15 amps > per motor and do A/B encoder interrups at about 100KHz. $60 total.
And where do you get the motors for such a low voltage, high current scenario? > > Size but not cost will be reduced what I can make a custom PCB > > My point is that "cost of entry" need not be a PC with a Mesa card. > It could be a Pi3 with one (or more) $3 STM32 board. Both the STM > and Mesa > can do steps and read encoders way-faster than is required. Until you want to reduce quantization noise to the vanishing point. Before I put a 1000 line encoder directly on the motor, and was trying to run it with a 68 slot optical encoder wheel, I even wrote a last 4 sample averager in hal for my G0704, but the noise still drove the pid and servo driver wall to wall if PGain was over 1.5. With the encoder on the motor shaft, I can run PGain up to about 40 with reasonable stability. I left the old optical in so I could use its index. I put switches on the gear shift knob, and change the rpm scale used in hal. And with PGain at 20 I can ask it for 10 rpm in either gear and cannot stop the spindle with my hand. And if neither switch is closed, I leave the motor running at about 20 revs. The gearshift knob, except for its half detent, moves effortlessly from gear to gear, nice since those plastic gears have flat faced teeth. It makes no diff how fast its running because the motor response is so fast its down to creep speed long before the gear being shifted out of has been disengaged. Can anyone else here make that claim? Yeah, its overkill, but it "Just Works". > USB work > well. It is 10X faster then needed for this purpose. > > > Thge BBB is very atractive because it in effect has the above Pi3 and > STM32 eqivalent (the PRUs+) all on one board. > > The Beagal Bard Blue would be PERFECT as it has low power H-brindges > and conecters for A/B encodrs on-board. A small mill or router > would not even need motor drivers. high-power cables could plug into > the BBB. (BB Blue) > But this is a $100 board. Double the cost > > So the entry cost should be under $100 including the PC and smaller > size motor drivers. > > If a person already owned a small notebook PC the entry level price > should be > closer to $20. > > I've been using these parts to drive motors of all kinds and they are > "bomb proof" > you can't kill them. they have reverse plolarity, short and > temperaure protection > It is a simple H-bridge with PWM control. > https://www.ebay.com/itm/30A-Mini-VNH2SP30 link is not found > <https://www.ebay.com/itm/30A-Mini-VNH2SP30-Stepper-Motor-Driver-Monst >er-Moto-Shield-module-For-Arduino/311579385639?epid=1675425525&hash=ite >m488b941f27:g:~D8AAOSwHsRYEv~K> I combine two of the above with one > these Cute but only 16 volts max? That will take a whole new family of very low inductance low voltage motors to be designed and made available if one wants higher speed rapids. > https://www.ebay.com/itm/STM32F103 > <https://www.ebay.com/itm/STM32F103C8T6-Minimum-System-Developmen-Boar >d-Module-Core-Learning-For-Arduino/323747797666?hash=item4b60df5aa2:g:i >lcAAOSw9m5cAggN> and have abuot a $10 total for a universal > driver/controller and the PID loop and > encoder inputs are on the $2 STM32.. The STM32 connects to the Pi3 > or notebook PC with USB. Documentation? Many of us, me at least, have steered clear of such because while the board might be $2, the programming kit to put it to work seems to be waaaaay outpriced, often over $150, and requires us to learn a whole new language. To say that's off putting would be a major understatement. And frankly, connecting to a pi3 by way of the usb to run machinery is scary enough that those of us dealing with it on a pi3 on a daily basis as I am, would never in a million years contemplate running machinery thru a traffic jam that makes NYC's times square look abandoned. Taint gonna happen again Chris. Using spi skips that hardware traffic jam AND makes it possible to control my Sheldon. Its an architecture mistake, driven by cost, but every single bit of i/o except the spi and wifi on the pi3b all has to wait its turn at getting thru an internal usb2 hub, and its too slow to get that job done in a timely manner by at least a decade speedwise. An uncontrolled keyboard repeat can ddns the whole system. I think I've found a way to control it, but I have to do it at every boot. > I'm slightly frustrated because I'd prefer to see MK run on the above > stack but it doesn't so I have DIY solution. and Mk?LinuxCNC is > confined to the garage. > > But for my uses, today I'd have to figure on at least a 5i25 as a > > driver card, and that unavoidably raises the entry bar for new users > > by what, 80 bucks? A sad state of affairs. > > Your call. > > 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> > > > > > > > > _______________________________________________ > > Emc-users mailing list > > Emc-users@lists.sourceforge.net > > https://lists.sourceforge.net/lists/listinfo/emc-users 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> _______________________________________________ Emc-users mailing list Emc-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/emc-users
