[Robin Hewitt]

Robin, I'd love to hear some more updates/progress on the system you're working on

Thanks for resurrecting this thread I thought I'd lost it

Having figured out that the PC needs to do a lot more of the work if it is to tweak what's happenning on the fly, I reworked the circuit boards to remove a few annoyances, dump some unnecessaries and added a port to drive a digital inverter.

Then I reworked the milling machine with SSR's for the main motor and suds pump so I can switch them over to software control. I put everything inside the door of it's cabinet for easy access.

Next I have to mount the scales and plug it all in.

I'm going to have to drive it through a USB adaptor if it is to have any hope of working under XP. Bill Gates can't even keep up at 9600 baud for long even with FIFO's enabled.

[Robin Hewitt]

Is that some type of universal proto board ? Seems like a LOT of connectors on there!

You need a lot of plugs for the malarky

3 stepper drivers, 3 digital scales, 1 port to drive motor/suds/etc, 1 PC interface.

Then I have my 3 auxilliary ports for pligging in pendants, stops, tool locators, whatever, and now an extra RS232 for the motor inverter.

Did I forget anything?

[eightball]

i dont think the usb adapter is gonna help you achive higher transfer speeds if thats what you're looking for. Im pretty sure you're always gonna be limited to the speed of the serial controller on the device side. might be worth it go go full on usb and grab a usb2 mc

[turmite]

Not really sure I know what I mean either

Here's the idea I had...

I've been positioning the milling table by turning the leadscrews very precisely and trying to minimise backlash, it works okay.

I had a notion that if the milling machine could read the table and quill positions independantly of the screws, it could position it a lot better.

So I'm making a computer board that accepts g-code, reads the scales, turns the handles, plus a few other things that I think might be fun such as a pendant, tool height and X,Y absolute position gauges.

It's okay to tell me that it's already been done and is available for 50 bucks from Geekware Inc. I won't mind

It's already been done and available for 50 bucks from Geekware Inc.

I'm sorry, but I just couldn't pass that one up. I can tell you that not only what you have planned has already been done, but done using dual encoder feedback on 6 axis with position correction on the fly! It is not 50 bucks though! I will tell you this though, I have seen a video where the developer places his hand on his mill and pushes on an axis and you can hear the steppers moving to correct the position, then when he takes his hand off, you can hear them correct again.......or I should say, you can hear them end correction!

BTW nice work!

Mike

[cozymkiv]

It's already been done and available for 50 bucks from Geekware Inc.


Mike

Do you have a link to the website you mention above?

hennie

[Robin Hewitt]

i dont think the usb adapter is gonna help you achive higher transfer speeds if thats what you're looking for. Im pretty sure you're always gonna be limited to the speed of the serial controller on the device side. might be worth it go go full on usb and grab a usb2 mc

I don't need faster than 9600, I could probably work on less. What I want is a buffer so it doesn't lose information when Windows doesn't service the interrupt fast enough to receive my status block without holes in it.

IIRC USB works with a shared memory block so a buffer is implied. Hopefully big enough. I don't need much.

Downloading is fine. When I have room for more pre-digested G-code I point the serial input interrupt where I want the data to go and ask for 8k blocks until my buffer is full. I don't care if it arrives in one great lump, I can keep up

[digits]

You might have trouble with USB's latency if you are trying to get real-time position information into the PC. A serial port may not have much bandwidth, but it does have a small and predictable latency, assuming your PC's not busy doing something else real-time.

[Robin Hewitt]

You might have trouble with USB's latency if you are trying to get real-time position information into the PC. A serial port may not have much bandwidth, but it does have a small and predictable latency, assuming your PC's not busy doing something else real-time.

Fortunately the replies to the PC are only for drawing a picture on the screen and setting options. This only works (hopefully) because the scales and stepper drivers are both on the same dedicated cpu. Even so the steppers will get ahead of the scales which only report 50 times a second, dismally slow when you use lots of steps/inch.

I'm off to my workshop to continue wiring up the mill with it's new switch panel. Last time I did it I didn't know what controls I would actually want/use, with hindsight I want 5...

Main motor on/off
Suds pump on/off
Suds flow rate
auto/isolate/manual
and one to kill the steppers so I can do it the old fashioned way

[digits]

Fortunately the replies to the PC are only for drawing a picture on the screen and setting options. This only works (hopefully) because the scales and stepper drivers are both on the same dedicated cpu. Even so the steppers will get ahead of the scales which only report 50 times a second, dismally slow when you use lots of steps/inch.

I'm off to my workshop to continue wiring up the mill with it's new switch panel. Last time I did it I didn't know what controls I would actually want/use, with hindsight I want 5...

Main motor on/off
Suds pump on/off
Suds flow rate
auto/isolate/manual
and one to kill the steppers so I can do it the old fashioned way

IMHO, that's definately the way to do it - you want dedicated real-time H/W on your machine and just a PC to monitor and command it.

Happy tinkering!

[eightball]

I'm off to my workshop to continue wiring up the mill with it's new switch panel. Last time I did it I didn't know what controls I would actually want/use, with hindsight I want 5...

Main motor on/off
Suds pump on/off
Suds flow rate
auto/isolate/manual
and one to kill the steppers so I can do it the old fashioned way

I also planned for a couple additional switches...
lights - main halogens, spindle led ring light
lasers (for centering/aligning using 2 line lasers, mount on either side of the spindle and where the 2 lines intersect is where the center of the bit/endmill will make contact)
mist coolant -(3 way rocker switch) always on(opens air and coolant solenoids), off, manual control (turns the air solenoid on and activates foot trigger for activating the coolant solenoid)
Flood coolant - on/off


this is sort of a wish list, i only have about 1/2 these things ready to be plugged in to the control panel

[Robin Hewitt]

It's coming on well, looks a lot tidier, have a before and after pic

[sansbury]

Thinking about the challenges of the 50x second update cycle... So, just out of curiosity, what is the goal of the system? From where I sit I see four possible gains versus a typical open-loop system:

1. Remove backlash from less-expensive linear positioning systems
2. Reliably run motors very close to their torque limits
3. Compensate for lead screw inaccuracies
4. Replace home and limit switches

There's also the "it's cool!" reason which is usually sufficient

Anyway, from where I sit it seems like #1 would do the most good followed a distant second by #3. As I understand it, steppers are 100% reliable assuming that you stay within their torque capabilities, and "proper" limit switches are designed to work outside the control loop for safety, so those two are somewhat moot.

Point being, if all you wanted to do was to reduce backlash, then couldn't you essentially ignore the scale position except on the way in and out of an axis reversal? As you approach one you could slow down to within the realtime sensing resolution and then accelerate back out of it.

Lead screw accuracy is another issue but can also be compensated for by measuiring the screw and building a deviation table. This does create a lot of additional things for the software to do, but if the sensing interval were to prove problematic, it might be a way out.

[Robin Hewitt]

So, just out of curiosity, what is the goal of the system?

The goal is to wring every last drop of precision to be had from a cheap Chinese mill/drill combo fitted with ball screws. I'm hoping the scales will tell me what needs to be done

[turmite]

Do you have a link to the website you mention above?

hennie

I do, but I don't have permission to post it since it will be announced here hopefully early this week. I will try to get permission to post a video but that will strictly be the developer's call.

Mike

[rokag3]

I do, but I don't have permission to post it since it will be announced here hopefully early this week. I will try to get permission to post a video but that will strictly be the developer's call.

Mike

well 2 month later we are very interested because a dual encoder interpolator for 50 us$ and even if it does only that it's great

[cyclestart]

well 2 month later we are very interested because a dual encoder interpolator for 50 us$ and even if it does only that it's great

turmite posted "it will not be $50 though". Fairly certain the discussion can be found here
http://www.cnczone.com/forums/showthread.php?t=59763

A couple of members here (possibly more?) have the device. No cnczone member except the developer making chips with it so far to the best of my knowledge. It's early days.

[Robin Hewitt]

One intersting feature of the Chinese scales is that the metalwork connects to the positive side of it's battery. Makes for some interesting accidental circuits if you don't isolate it from the machine

[The Blight]

Just wanting to check if theres anything happening to this project.

[Robin Hewitt]

Just wanting to check if theres anything happening to this project.

Well, I ripped it up and started again twice, then I hit a bug which stopped me for a few weeks... I was turning on and applying the accelerations BEFORE data actually reached the step interrupt. The timer reload setting overflowed and everything went to hell in a handbasket.

While finding the bug I found several other potential nasties so the code is currently running sweetly. I am sure I will soon change all that.

Then came Christmas and I had to do other things, still busy and I will need to see a few weeks clear for programming before I put this program back in my head and get going again.

[The Blight]

I'm glad to hear that you are working on it! I have been looking into those cheap scales myself, but as a way to check that if the position is correct. What you are doing is a lot more advanced. Have you been thinking about making your own controller for the caliper? I found the signal it sends out to be junk. It can be used, but it's messy.

And you are damn stubborn! Thats a good quality!