On 11/10/26 05:31, andy pugh wrote:
On Mon, 9 Nov 2026 at 18:20, Sam Sokolik <[email protected]> wrote:

In his configs - the base thread is 150000ns - which without reset
functionality (I don't know if the gpio in the rpi allows for that) is
3.3khz.

hal_gpio does export a reset function, so you can get 1 step per
thread cycle (rather than one every second step cycle)

There is a faster way that Bertho's spi driver (rpspi) uses, Andy. It uses 3 gpio pins to drive the 7i90, at 42+ megabaud transmit, and 25 megabaud for receive from the 7i90. That is the omly connection between the pi and the lathes interface. No ethernet involved...

A mesa 7i90 and 3 7i42TA's gives me 72 IO pins to drive my 11x56 Sheldon. This includes signals to some Crydom clone relays to control the lathes power, 2 stepper drivers (4 if needed), 3 encoder inputs and a PCM output to a mesa spinx1 to control the VFD's speed & direction.

One encoder in is from an AT667 triplet watching the spindles main gear and 2, each with mode pin s in place of the missing index on those dials so they are 3 pins each, reading a pair of $22 100 PPI dials I put on the new apron which effectively are the X&Z cranks I took off, drilling 8 1/4-28 boltholes in the front face of the carriage & bolted on a big 1/2" thick plate to anchor the Z ball nut, hold the X motor, the encoder dials and mode push button switches. The mode buttons change the encoder dials in how many steps per click are represented, in a 1-2-5-10 etc gain sequence from a fraction of a thou per click to as fast as the motors can drive it which is about 125 IPM for Z and around 60 for X. There is no disconnect as the dials can do it all.

The rj45 jack on the rpi4b isn't used with that setup other than furnishing a net connection to the world for updates of the software and OS. I can run firefox, prowling the planet, but firefox is an IRQ 700 lb hog, messing with the lathes smooth motions, so I don't run firefox.

The pi is mounted upside down which make that spi bus ribbon about an inch long, and is src terminated by 120 oh chip r's on a board Jon Elson designed. I don't think in the last decade plus it has ever made a mistake. And since I swapped the original motor/drivers out for closed loop 3 phase stepper servos the ONLY loss of home is when I check the auto-shutdowns these drivers can initiate if the motor miss-behaves. I can run a carbide chipped tool into a stationary chuck jaw, and it winds about a step, trips everything off in its tracks, the carriage and tool bounce back about 10 thou, at 60 IPM, no broken chip in the tool or mark on the chuck jaw. Lift the tool holder off the post, re-home the lathe, fix whatever it ran into, put the tool holder back on the post and resume the job. Just one problem, I've only tested it, never happened mid-job since I changed motors. A decade & change later it still just works.

Yeah, its $300's more costly than a SainSmart parport card but it just works. I feed it g-code and it does it with no arguments. I have kill-a-watts, the pi when driving the lathe, draws 13 watts + the monitor. Only power failures stop the pi and I did have a small UPS on it which has since died & been bypassed. Its running right now.

Obviously I'm proud of it but I had lots of help from the people on this list at the time I did it. You know who you are, and I thank you, a lot.

Cheers, Gene
--
There are four boxes to protect liberty
Soap, ballot, jury, ammo
Please use in that order!



_______________________________________________
Emc-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/emc-users

Reply via email to