Mach3 user thinking about switching over....

PetefromTn wrote:

I have a sensorless vector drive on my mill and I am wanting to get spindle indexing done on it and I have a pretty nice optical sensor here that I thought I might use. Can a reasonably accurate spindle encoder be setup with say four white line tapes on the spindle pulley and this one sensor using EMC2

Rigid tapping needs an index as well as position information. That tends to mean that you need three sensors. It is just about possible to spoof it with 2, and for a lathe you can use one. The reason you need two with a mill is that you need an encoder that counts correctly in both directions, which means you need two sensors in quadrature
en.wikipedia.org/wiki/Rotary_encoder#Incremental_rotary_encoder
EMC waits for an index before it starts a coordinated move so you need that channel too. This doesn't make a whole lot of sense, really, unless you plan on doing multi-pass rigid tapping. It ought to be possible to cheat with HAL (Hardware Abstraction Layer) components (a bit of a lash-up of dividers and subtraction on the encoder counts would be possible) to save an IO pin.
You can see an example of a home-made encoder in this video of my vaguely similar machine:

That uses three "gap" type sensors. However the lathe spindle worked fairly well with an encoder made by wrapping a laser-printed pattern of stripes round the shaft and using reflective optical sensors.
You might find that you can use a drive pulley if it is a toothed-belt and the sensors have a short focal length.

I have my mill setup with a CNC4PC C11g breakout board and three Gecko G320 servo drives and I am adding a fourth axis soon.

With 4 axes and a 3-channel spindle encoder I think you would start to run out of parallel port pins. It is possible to use more than one parallel port. (I would be tempted to skip the breakout with a PCI parallel port, though, as they are so cheap) . (The Mesa 7i43, which plugs into the p-port and gives you 48 IO pins and hardware step generation is also worth considering if you do run out of pins)

I understand that you can download the software and load it onto a CDrom or DVDrom and temporarily load it onto your machines control computer and kinda make a test run is this accurate or is there more to it than that?

Booting from the LiveCD should do exactly what you describe. You can run various simulated configs and see how you like it. I think you can also create a config with the wizard to run your machine, but can't say I have tried that when LiveCD booted.

Another question is how does EMC2 work with tool tables and my chosen TTS tooling for repetitive Z height datum?

You can either measure tool length and type the numbers into the tool table editor, or you can touch-off the tools to the work and have the offset written to the tool table. With a TTS I would be tempted to go for the first option, but there is no reason you can't use both as the mood suits you.

Here is a small video of my machine running I just made...

Unfortunately that seems to be blocked from the country I am currently in, due to content from UMG.

PetefromTn wrote:

Specifically the rigid tapping sounds very nice. I have a sensorless vector drive on my mill and I am wanting to get spindle indexing done on it and I have a pretty nice optical sensor here that I thought I might use. Can a reasonably accurate spindle encoder be setup with say four white line tapes on the spindle pulley and this one sensor using EMC2 and get the positioning accuray necessary for this rigid tapping as long as I only go like say 150-300 rpm or so? I have my mill setup with a CNC4PC C11g breakout board and three Gecko G320 servo drives and I am adding a fourth axis soon.

Hi Pete,

As Andy mentioned you need an index pulse and a quadrature encoder. Looking at your video I would mount an encoder to your spindle and not fiddle with home made ways unless you just like to tinker. Encoders are cheap (I get mine from automationdirect.com) and easy to mount when you have a setup like yours.

I understand that you can download the software and load it onto a CDrom or DVDrom and temporarily load it onto your machines control computer and kinda make a test run is this accurate or is there more to it than that? I do not wish to lose my Mach3 settings until I am confortable with EMC2. Is there somewhere I can see the entire interface and perhaps play with it offline somehow?

You can boot from the LiveCD and play with all the sims but as your running from the CD you can't save anything or create a custom configuration. One easy way is to just put a different hard drive in your computer and install Ubuntu/EMC on that to test with. Also I think there is a way to make a bootable USB drive but I'm not sure how to do that. While a servo system it is not as important to have low latency as software generated parallel port step and direction pulses you should check it after you boot from the LiveCD to make sure your PC hardware doesn't have some gremlins hiding in there.

I also have never used anything but windows so I am not really good with linux I am sure.... Would just love to see my machine perform a rigid tap and have full closed loop control. That would really be something... Am I dreaming here or is this possible?

Ubuntu is actually easier to use than windoze and I've used windows since DOS 6.22 went out of fashion. The Gecko site is down so I can't check but as I recall the G320 drives take step and direction? To have a closed loop system you need to connect the encoder to EMC and EMC tells the drive how fast and what way to run.

Another question is how does EMC2 work with tool tables and my chosen TTS tooling for repetitive Z height datum? I would like to get setup with rigid tapping and be able to use multiple taps in either a quick change tap holder of some sort or individual holders I guess in some sort of 3/4 shank for the TTS system. Any ideas and information on this would be most helpful. Thanks and nice to be here with all of you... peace

Pete

I assume the TTS system allows you to get repeatable Z height for tools. What I do on my BP with a QC200 spindle (no tool changer but repeatable Z for tools) is to load the tool then place a 0.500" diameter dowel on the Z0 location (usually my material top but sometimes my vice top) and lower the tool to a point where the dowel won't pass under and as I raise up the Z until the dowel just passes under the cutting edge I stop and enter a 0.500 for the tool table and the tool number. I do this for each tool (using a different tool number) and it takes only seconds to do. Now when you do a manual tool change in your g code with a T1M6 for tool #1 you can have the machine configured to go to a tool change location and a popup prompts you to insert the correct tool. You swap tools and tell EMC to proceed and off it goes.

John

Hi Pete,

I missed the fact that you are using a drawbar, that does complicate putting an encoder on the spindle a bit. Can you add a small timing pulley to the v belt pulley on your spindle with a through hole big enough to handle your draw bar and put the encoder on the side out of the way?

You can combine all the home and limit switches on one input with EMC to save inputs. The only thing you loose when doing this is when you set up your homing sequence you have to home each axis separately. You still press the home all button once to home. The other thing you loose is the knowledge of which limit switch is tripped but doesn't seem to be much of a problem on a machine like yours where you can see the axis. You do need to do the math and figure out what your maximum pulse rate is so you can pick the proper encoder count. The encoder needs three inputs A phase B phase and Z (index).

The Geckos are fine you just don't have closed loop back to EMC with them. You just set it up like a stepper drive in EMC. The spindle will be closed loop back to EMC with the encoder feedback.

John

With a belt drive you would want to get your encoder feedback from the spindle unless you had a timing belt drive then the motor would be ok I think. Looking at your video (and it's not really clear to me) it looks like you could take a timing belt pulley and bore out the inside and fix it to the top of your v-belt pulley. Do you have a photo of that area?

Looking on the user map there is another user located at Lenoir City and quite a few in Nashville. I'm located in Swamp East Missouri...

John

PetefromTn wrote:

Yeah the Geckos are step and direction based servo drives, does that mean you cannot use them?

No, that's no problem, EMC2 is perfectly happy with step/direction drives.
You possibly have the option of using the drives in velocity mode, and closing the position loop in EMC2, but that would need a whole lot more IO and encoders on each axis. Probably not worth the trouble and expense. If what you have now is working, then it will carry on working with EMC2.

I do not have toothed (teethed) pulleys I have vee belt and pulleys so that too might be a problem?

No, absolutely no problem, but it does mean that the encoder absolutely has to be on the spindle and not on the motor as the belt ratio is indeterminate with a V-belt.

It does not sound like this is gonna be possible if I need three sensor inputs just for the spindle.... Is that correct?

I think that you could possibly manage with just 2. Either two sensors and a once-per-rev pulse using one of them as the index too (ideally you would have half the spindle black and half white, and the sensors at 90 degrees to get the quadrature). Better would be multiple pulses per rev with the sensors offset by half a pulse to get the quadrature. One simple solution which might work would be an encoder and 1:1 belt drive, as used here (This is another EMC2 installation, with a power drawbar):

If you look at his other videos you will see a 1" tap rigid-tapping.

What would you fellows say is the easiest simplest way to arrive at a rigid tapping setup for my belt driven milling machine and use the least amount of hardware and complexity of wiring?

Not wanting to sound too much like a Mesa salesman, I think that freeing yourself from the parallel port pin limit would be useful. For $80 you can buy a 7i43, plug it into your parallel port and then you have 48 IO pins, megahertz encoder counters, super-smooth hardware step generation etc. (There is no reason you can't wire your existing BoB to a subset of the pins)
That would be rather a commitment to the "EMC2 side" though. As a first step I would be tempted to experiment with a laser-printed encoder disc or strip and a second optical sensor to use just two pins. The index could possibly be a software button, I think, but creating a fake index in HAL ought to be possible. The p-port has 5 input pins. It is possible to have all the home switches and limits share one pin, so it is probably still doable with the full three channels.

If you remove the front cover on the spindle casting, can you see the spindle shaft? It might be possible to clamp a 2-part encoder ring round it (or even use a paper encoder wrapped round the spindle, like I did with my lathe)

I assume you have seen the video of the Tormach power-drawbar? If you need to do belville-spring calculations you might find this spreadsheet useful:
www.bodgesoc.org/DIN2093DiscSprings.xls

I guess you are thinking in terms of a rack-style toolchanger? Currently the support for those in EMC2 is not especially good. There is a project underway to introduce G-code-as-toolchange for toolchange variants which require coordinated axis movement (rather than simple quill-up). Conversely, for more complex toolchange systems the availability of Classic Ladder in EMC2 may be an advantage over Mach3.

I am a bit concerned about the phrase "Parallel Port emulator". What do you mean by that?