Motion Controllers: Mixing Servos & Stepper Motors

[RCaffin]

I believe that the ESS supports Mach4, which is encouraging. However, I still run Mach3 .062 and an ESS, and so far see no reason to change that.

Tapered threads - stock standard feature in Mach3, been there for a long time. As for 'multi-axis' capability, Mach3 has had SIX-axis capability for ages. And that is FULL SIMULTANEOUS SYNCHRONISED 6-axis. See posting #50 for some 4-axis stuff which had X, Y, Z,& A all running simultaneously in full synch.

Cheers
Roger

[SCzEngrgGroup]

Tapered threads - stock standard feature in Mach3, been there for a long time.

Tapered threads are not a "feature" of Mach3. They are a "feature" of whatever CAM you use. ANY controller capable of 3-axis moves can do straight or tapered threads.

Regards,
Ray L.

[ger21]

Looking into the Galil/Mach option some more; it appears that several years ago there was a major evolutionary diversion point as far as machining technology when controllers went from linear-interpolation to high-order contouring guidance.

It looks like I was somewhat mistaken about this when I posted earlier. Looking through the Mach4 Galil plugin manual, I see that the "contour mode" is specific to the Accelera controllers, and it is supported by the Mach4 plugin.
There's more info on it here, but I don't really know what the difference is.

http://www.galilmc.com/download/datasheet/techref.pdf

But I still maintain that if you aren't using a Galil Accelera controller, then Mach4's motion isn't really any different than Mach3's.

[barnbwt]

But I still maintain that if you aren't using a Galil Accelera controller, then Mach4's motion isn't really any different than Mach3's.

That's my understanding also, but with the corollary that because the older Galil units don't communicate the same way as the newer contouring-capable ones, there's an insurmountable incompatibility with the Mach4 driver. But for the drivers that 'move' the same way as the old ones but communicate like the new ones --the 21x3 series-- Mach4 can still work with some level of reduced functionality as far as the advanced controls stuff. I'll pursue those controllers for the time being, and try to figure out exactly what kind of success and capability folks have been able to tease out of them using Mach4.

TCB

[ger21]

I wouldn't expect to see a noticeable difference.
People have been getting accurate parts from Mach3 and it's competitors for years.
The only difference may be in perceived smoothness, and maybe only at very high speeds? Just a guess, though.

[RCaffin]

Hi Ray

Tapered threads are not a "feature" of Mach3. They are a "feature" of whatever CAM you use. ANY controller capable of 3-axis moves can do straight or tapered threads.
Misunderstanding. I meant that 'tapered threads are supported by the Mach3 Threading Wizard'.

Cheers
Roger

[barnbwt]

I wouldn't expect to see a noticeable difference.
People have been getting accurate parts from Mach3 and it's competitors for years.
The only difference may be in perceived smoothness, and maybe only at very high speeds? Just a guess, though.

That's my assumption; stuff like second-order jerk really only matters when you're choreographing a high-speed ballet with heavy masses or trying to dynamically damp acoustic vibrations with death-grip control systems. My little benchtop has none of those things, lol, and I have all the time in the world to run prototype widgets.

Folks do seem to loudly sing the praises of the new controllers, but I notice they're mostly dealing with production machine retrofits or large routers; areas where this kind of control would see the most benefit.

[RCaffin]

And where the budgets might have an extra one or two zeroes, and the customer might not have the skills to MYOG and might have to rely on the handhold from a major vendor.
Never forget: for big commercial companies Capex of that size is trivial, but Downtime is the ultimate hell.

Cheers
Roger

[barnbwt]

Okie-dokey, I'm finally closing in on a solution I think I'll be happy enough with to get started.

Mach3 & Galil driver (for the time being)
Used Galil 21x3 series controller
Used Galil 206x0 series stepper controller
5A 24VDC and 7A 48VDC power supplies (should be enough for the motion controller/servo drive/stepper drive logic circuits, as well as the two stepper motors. Servo drive power is 9A 110VAC --I'm pushing it on the 15A available current, but I think it's viable unless I bog down the servo badly)

Now I'm just trying to justify the use of the Galil stepper board module, since I'm not quite sure it does anything that a new external Chinese unit(s) can't for a lot less. It's really slick how it snaps right onto the controller, and how both can be powered by the same 24V source with only fan cooling. But I am most interested in how its specs stack up against hardware that isn't a decade old.

From the rather sparse spec sheet in the Galil catalog:
DC Supply Voltage: 12-60 VDC
Max Drive Current (per axis) 3.0 Amps (selectable with AG command)
Max Step Frequency 3 MHz (microsteps down to 1/64 I believe)
Motor Type Bipolar two-phase
Switching Frequency 60 kHz
Minimum Load Inductance: 0.5 mH

Most of it seems self-explanatory, but I think there's some sort of jargon disconnect as far as the step & switching frequencies. Max Step would seem to indicate the number of pulses per second that may be generated across one (or all?) axis (2 or 4 for this series), so what is "switching frequency" referring to? Is that the max number of pulses per second available at each axis, or the speed at which the drive processes command pulse signals from the controller? On a Leadshine spec sheet for instance, input pulse frequency is the only measure in the "kHz" range, and it's 4X higher than the Galil.

Assuming the Galil is as fast as the current drives, I really like the ability to run integrated 50V/3A steppers in a (potentially dual) closed loop, and still easily upgrade to servos at a later date, all from the same 24V/1A brain-card that takes up very little space. The fact it's "idle" current for steppers is 25% the maximum is also a bonus (most drivers seem to do 50% or more, but also aren't set up to run closed-loop). I don't think the Galil has the ability to automatically dial back on the microstepping at higher speeds for stronger operation like the newest cards, but that's the only downside I can see.

Anyone ever mess around with the Galil stepper drives? It seems like practically everyone just runs servos only if they bother to deal with a proper motion controller, so there isn't much out there. I myself am tempted by the NEMA23 servo motor options out there, but I just don't have the room for belts/gearing required to get sufficient torque at sane RPMs (I can't imagine ANY lathe operation requiring hundreds of inches per second of tool travel, lol)

TCB

[RCaffin]

<Caution: some bias evident>

Galil vs China: well, at least you know the Galil stuff will work. They might even support it.

Power supplies: you will rarely draw even half the full-rated power on the axes.

Switching frequency: basically, you can ignore this. It is internal to the driver.

Cheers
Roger

[barnbwt]

What about Galil controllers from China?

As far as the current draw by the motors, that's kind of what I was hoping; I'd have to be hogging through taper threads (or crashing the machine) to get it to trip a breaker. And while that limitation may keep me from doing much milling since I only have 3.6NM at the spindle, it will hopefully keep me from breaking stuff right away.

I believe I'm overthinking the frequency, too, I just don't like not understanding this stuff, lol. I ran some numbers, and to move the Z-axis (the only one that needs any kind of speed) from one end to the other in five seconds, my 1605 screws only need a little over 400rpm; I think that should be easily doable barring some massive impedance mismatch at the motor selection on my part. I'd only ever be moving that fast under load if I'm broaching something. To be honest, the power & space savings in addition to the closed-loop capacity make that stepper drive something of a no-brainer, unless I'm missing something (for closed loop, the new-made drivers aren't even that much cheaper than this used card, to be honest)

So I've now got a new-to-me DMC-2183 and SDM-20640 on their way to me from Thailand & Texas, respectively. Here's to hoping everything still works when it gets here. Thanks for weighing in!

I also found a rather useful if brief tutorial for setting up this exact combination of hardware, albeit in XP. Off to find out if "DMC Smart Terminal" or an equivalent can run in 10. (EDIT: It's called GDK now, and is Windows 10 compatible. Now I just need a Galil to connect to via ethernet)
Galil Motion Controller

TCB

[Jim Dawson]

The software you want is Galil Tools. GalilTools | Galil

I have never used the Galil stepper drives, I have used their on-board servo drives and have had good results. The on-board stepper drives are generally a bit underpowered for the motors I normally use. For precise motion control my prefered stepper drives are the Automation Direct STP-DRV-80100, not cheap, but they are very flexible and fully programmable via the computer. For more utility purposes I normally use the cheap import stuff from EBay. Both play nice with Galil controllers.

[barnbwt]

Does GDK not play nice with this controller, or something? It looks like Galil has it as their preferred/modern setup & tuning software on the main download page (it seems they change the name of this tuner/terminal interface every couple years)

This is kind of cool, I guess; if I understand correctly, Galil has made their GDK configurator C-based, so it's quasi-open source now;
https://en.compotech.se/blog/2015/06...mming-library/

"I have never used the Galil stepper drives, I have used their on-board servo drives and have had good results. "
Because why would you, right? Like I said, it seems if a person is willing to go to the trouble of using a Galil, they pretty much always are willing to shell out for proper servos. Makes me wonder if this stepper board might actually be a more popular item among the used/hobby set than industry.

Agree on the limited stepper-food these all in one boards offer; to be honest, my thought was if I outgrow this initial configuration I'd simply install servos for the axes (and use the 4 stepper ports for what-have-you, or get rid of the daughter board entirely). The used Galil board actually seems identical in a lot of ways to the mulit-axis unit I see Leadshine & some others offer (only rated to like 3.5A), but cost about the same. 10 amp drives, huh? I guess that's for size 34 or even 42 motors, right?

Last question for tonight; what's the deal with this board only requiring fan-cooling when operated at a full 3A (per Galil), yet the cheap import drivers requiring big heat sinks? I guess they're switching the pulses on/off via a different circuitry process that's less efficient? I'd kind of figured on the heat being more of an issue than this to be honest, but it looks like the only cooling I'll be needing is a small fan for the servo amplifier heatsink.

TCB

[Jim Dawson]

The reason I haven't used GDK is because it's $195, Galil Tools is free, works fine, and I'm cheap. Admittedly, GDK does have some nice features like being able to talk to both the GDK program and the host program simultaneously. Saves switching back & forth, but I build all the debugging/diagnostics into my software anyway.

I use steppers and/or servos depending on the project and the customer's budget. Basically, if you come to me to have a machine built or modified, you are going to get a Galil controller. There is no learning curve on my part and I already have all of the software modules written. I have a combination of servos and steppers on my mill.

EDIT: I do remember using a Galil onboard stepper drive on a dispensing machine I built. Used it for positioning the rotating drum with a NEMA23 stepper. In that case the drum had 10 stations, so I used full stepping thus 20 steps per station. Lined up the drum with one of the natural ''cog'' positions. Not really a very demanding application.

Yes, 10 amp drives will run most NEMA 42 as well as NEMA 34. That's what is nice about those drives, they will run the full range of common stepper motors from NEMA 17 and up, and are tunable programmatically to run correctly with any of them.

The heat sinking vs. fan cooling may have something to do with the quality of the electronics and the efficiency of the circuitry. The Automation Direct 10 amp drives are little tiny things, fit in the palm of your hand and don't have any major heat sinking or fan. The standard 7 amp import drive is twice that size and has a huge heatsink.

[barnbwt]

The reason I haven't used GDK is because it's $195, Galil Tools is free, works fine, and I'm cheap. Admittedly, GDK does have some nice features like being able to talk to both the GDK program and the host program simultaneously. Saves switching back & forth, but I build all the debugging/diagnostics into my software anyway.

I think GDK 'lite' is free to download from Galil's site; I've got it running on my laptop right now & everything. Granted, there's no controller to talk with & test yet, but I only see a request for the "pro" license on the o-scope simulator and feedback-loop tuner screens (and I'll be tuning my one servo loop via the Baldor MintWorkbench software for the time being). The important controller configuration and terminal interface screens seem to be fully enabled.

TCB

[Jim Dawson]

I'll have to check it out

[barnbwt]

Well, now that I have all the logic components in my possession, I can start putting my pin-out wiring diagram together & get the necessary harnesses assembled. So hopefully I'll be able to get my computer talking to the controllers in the near future.

Just need to do some minor surgery on the Galil 2183 card; the one I got a good deal on has a right-angle connection for the big 96 pin Eurocard/DIN41612 pin header (hopefully the combination of all these names for the thing will show up in search results for other poor saps trying to figure out what connector that is, lol), but the daughter board requires a vertical header so when plugged in, secondary pin headers will interface with the main board at several other locations. I figured this out before the purchase, but decided to chance it anyway (it's not like the board is under warranty or anything as it is) after learning that desoldering isn't that dangerous a process so long as you have the proper tools at hand. Hopefully others considering the Galil 21x3 cards will find this & can decide whether to wait on a card with vertical slots if they wish to use the various daughterboard options out there (I have to say, it puzzles me that Galil even offers this as an option, considering how heavily they seem to market these expansion cards, since it makes them all unusable without modification)

TCB

[Jim Dawson]

Normally any daughter boards are factory installed. I think the one you have is designed to be used with a cable.

[barnbwt]

So, I found out that there are plenty of cables/connectors to join the boards together at a right angle (or even parallel but at a short distance), but they're the tape variety. Now, I'm not very knowledgeable when it comes to circuitry, but I assume that noise is a likely problem even with short (1") cable lengths? There would be logic, sensory, and 48V(DC, I think) connections between the boards. Obviously Galil wouldn't weigh in on this configuration when I asked, lol. The upside to the cable solution is it isn't permanent, doesn't void warranty, shouldn't break anything, and costs about 60$ by my rough estimation. Tooling up and replacing the connector myself also runs about 60$ (I could find a buddy to do it for a sixpack or two, plus 3$ for the connector).

I did find one other fool who had this same idea; 2123 and 2143 | Galil
Of note, the Galil tech says that theoretically the connector could be replaced to reorient the connection. Also, this bit;

"The dmc-2143 is a four-layer board, the internal layers are 5V and GND. There are vias that pass these signals from the 96-pin connector to the internal layers. So the connector is not strictly through-plated. Keep in mind that if there is damage to the PCB the card will no longer be repairable."

Any ideas what they could mean by "not strictly through plated?" I was under the impression that all soldered via holes are plated entirely through the board thickness (though they may or may not connect to traces at each level though a copper pad around the hole). Though 4-layer, this board is still the 'standard' ~.07" thickness you usually see in PCB's, so I don't think getting solder to flow fully through a cleaned/fluxed via is the issue. So long as the existing solder/pins are removed carefully such that the via plating is not destroyed, it seems like installing the new header should be no different than the original installation of the one that was removed. Am I missing something?

TCB

[RCaffin]

I think your interpretation re 'through plated' is correct.

The problem with desoldering is that sometimes the pins can have an interference fit into the plated hole, and removing the pin can damage the plating. After all, the PCB material itself can NOT withstand the soldering temp for very long. Equally, the high temp can lift a copper track off the surface of the PCB - which can then break. The photo below shows what can happen - see pin 12.


Cheers
Roger