I've already posted a few times regarding drivers and so on for this machine, but thought it might be worth just doing an over-view of the whole project.
This video is worth very many words:
https://youtu.be/0vZgQbFCj2Q
Printable View
I've already posted a few times regarding drivers and so on for this machine, but thought it might be worth just doing an over-view of the whole project.
This video is worth very many words:
https://youtu.be/0vZgQbFCj2Q
Welcome and nice to meet you David,
Thanks for sharing and the introduction, I've also sub'ed to your YT channel and look forward to following along. Will go and have a look now to see if i can help regarding drivers.
David welcome to the forum. It looks like its going to be a nice project. Are you running this from the parallel port or with a motion controller? Have you also thought about Linucnc if you want in the future to use ATC ect and as many inputs and output as you want.
Thanks for the welcome(s). I've already done a great deal of the groundwork to get this thing up and running, so it shouldn't be too difficult to get it actually doing some work.
I spent this afternoon looking at the ways and ball-screws of the X-Y table and after a little adjustment it's working much better. This thing is, like most Swiss stuff, beautifully made and really nice quality components have been used including dual-ball-nut screws allowing the micro-adjustment of the pre-load. I'm going to stick to the 2000mm/min rapid as this is reliable.
I'm running from the parallel port of a 10 year old Dell PC - two parallel ports actually (in anticipation of the many inputs required for the extra limit switches). I really like your suggestion of using LinuxCNC. I've only been using Mach3 up to now as I have a friend who's been building small CNC machines for the last 15 years and is very knowledgeable about Mach3. However, I've always used a Mac and am not overly keen on the MS operating systems so switching to another Unix based OS appeals to me.
Would I be better off using a motion controller rather than the parallel ports?
Using a motion controller is night and day difference. I don't want to get into the Mach3 V Lcnc I use both. But If you say do a feed hold in Mach3 and it is in a cycle it won't stop until the cycle is finished with Lcnc it will stop immediately.
I you decide on the Lcnc route with steppers you would need a Mesa 5i25 or 6i25 (pci or picx) and a 7i76 bob that would give you at least 34 input and output, 5 axis 2 encoders. spindle control, 24V inputs differential signaling etc.
edit: You could also just use a 7i76e if you want an ethernet version without the 5i25 etc
Lcnc is a steep learning curve but it is a rock solid system. It will do what ever you want.
I think you said you had 4mm screw if so I don't think you can go much faster rapids with a nema34 I think the max motor speed would be 500-800 rpm.
Thanks once more Clive. Many things to look into and think about. I'm happy with the 2000mm/min rapids, I'm still somewhat incredulous that those old Superior steppers could run any faster.Quote:
Originally Posted by Clive S
Sorry about the cock-up with the video editing btw. Don't know how I managed to get two copies in one video.
Here's the latest on the Fehlmann table. Sorry that it's a bit long - will try and be brief in future.
https://youtu.be/nwekbZpsq6Q
Making good progress on other stuff and will post more soon - watch this space!
Coming along nicely I enjoyed the vid. Re the Nema34 they might run better with a higher voltage that 70v I think you can get 230v drives for them.
Hi Clive, just come in from the workshop after more fun. More vids tomorrow...Quote:
Originally Posted by Clive S
Do you think it sounds like they're missing steps or is it a step-generation/processing issue? Difficult to answer without test I suppose. The quill has always made the same noise from time to time. I've was driving it with one of those crappy HY-DIV268 things and when running a repeating cycle of ten up an down movements it always returned to the same place with better than 0.01mm variation - which I was very impressed with.
I think after going to the effort of buying the highly recommended AM882 drivers, in the short term at least, I'd rather replace the motors - they only cost £27 for the pair including delivery so I can afford to replace them!
In the longer term I still favour going all-out for servos. Those Teknic Clearpath things look very impressive. If confidence in Sterling would just improve a little I'd be tempted to order.
Motor operation and control issues:
https://youtu.be/EafSH0RX1Lk
One problem using the PP is that until the pc is powered up the pins can be in any rambled state and they can also chatter as the pc is powering up causing all sorts of unexpected happenings. That is the beauty of a motion controller this behaviour does not happen with Lcnc and a Mesa card.
edit Having said that you might be able to use the charge pump in Mach3 so that the bobs are not enabled until the charge pump signal is seen.
Mmm.. yes, I understand that, but it's more of an issue with the BOB itself in that they've used the NC rather than the NO side of the relay - can't see the reasoning behind this or a way around it.Quote:
Originally Posted by Clive S
BTW I'd love to move over to LCNC, but my impression is that it's for people with an interest in programming and computing rather than just wanting to get on with using a machine. I've looked hard, but I can't find the equivalent of the Mach3 User's Guide. I find that I can't get to a position where I can understand what I need to do to use Mach3 - and I regard myself as being moderately savvy with PCs.
Without seeing the videos (my broadband is currently not working, so am limping along on mobile broadband), if you're using a parallel port, you really need to use a charge pump for safety, regardless of what software you're using. As Clive has already said, parallel port pins can act randomly during loading. If the BOB you are using has no charge pump, the easiest option is just not to power up the machine until the computer is fully loaded.
The Clear tecknic servos do seem good, however you have to remember as they need the step/dir signal directly, it adds in a lot of extra potential for interference and losing/missing steps. If I was to use them, I'd be looking at adding differential signal drivers to avoid potential noise issues.
Plus I think they're more closed loop stepper, than what's more commonly referred to as a servo. Good in the fact they're better for direct driving ballscrews, but does mean top speed is limited.
Performance wise, you're more likely to be better of with some good Nema23 motors, and adding a 2:1 drive ratio. Nema 34 really need high voltage drives to get the best performance from them.
Thanks for the reply. What's a "charge pump"?Quote:
Originally Posted by m_c
Re Clearpath - they are true servos rather than hybrid servo/steppers. I take your point about interference. I might pose the question to Teknic and see what they say - they've been really helpful so far.
There seem to be as many opinions about drives and motors as there are drives and motors. I did canvass this forum and others quite extensively before buying drives and the consensus seemed to be that Leadshine AM882s at 70v was the way to go. Still, the Z axis is fine, and the X&Y drives can be redeployed. If I do make some change to the X and Y drives it will be to servos - not poncing about with these poxy stepper things any more:wink:
if the bob does not see a continuous signal from the controlling PC it shuts down.Quote:
Originally Posted by Agathon
Also called a 'watchdog timer'
From my bob manual,
The charge pump uses the 12 kHz signal from the parallel port generated by the
CNC software to operate a logic circuit that gives an active low output. Any
piece of machinery that uses powerful motors can be dangerous if controlled
by a computer that can be in an unknown state while being powered up or in
a software crash condition. Using the charge pump circuit to disable power to
motors is a safety device in that it only operates when the software is running
correctly and under user control. The charge pump circuit is also used to
disable the output signals so even if your stepper boards do not have an
enable pin they will be disabled automatically when the charge pump signal
is not present.
Mach3 generates a 12.5Kc/s pulse onto one of the pins of the PP This is used to enable the bob. When Mach3 starts up. It is used in case there is a software problem or e stop situation the pulse disappears and basically switches the bob off.Quote:
Thanks for the reply. What's a "charge pump"?
Re the drives is is usual for us to have a two to one reduction with a belt drive. I use these AM882 on my mill.
Your mill though is in a league of its own.
I'll do some research on charge pumps.Quote:
Originally Posted by Clive S
I have been puzzling about using Nema 23s with a reduction drive, but can't see how this will help in my situation. If I wanted to achieve 1000 rpm at the feedscrew the motors would have to run at twice that speed with the inevitable massive drop-off in torque. Even if the torque itself is multiplied by 2:1 (which it won't due to losses in the drive belts etc) the torque of the motor is likely to have dropped way below any mechanical gain. I can see how it helps with gaining torque at lower speeds, but looking at the torque curves of even the lowest induction and rotor inertia steppers it looks like less than a zero-sum game to me since their torque drops down to less than a third of their maximum. By the time any motor gets to having nearly 7000 steps a second they're going to have no torque at all. These topsy-turvy stepper things do my head in!
Attachment 22119
How do you get rid of attachments? Realised that the axes on the blue graph don't make any sense.
I'd like to get my hands on one to see, but the torque curves look more akin to those of a stepper system, than a servo system. Stepper and brushless motors are very similar, it's just stepper motors are synchronous (which is what gives them the detents), while servos are a/non-synchronous, which means you don't lose power/torque overcoming the detents.Quote:
Originally Posted by Agathon
Regarding the Nema 23 v 34 argument. Compare them running similar voltages. You'll generally find Nema 34 graphs are using a high voltage driver, while Nema 23 are done using a relatively low voltage driver.
I've just had a quick look to see if I could find a couple graphs to do a comparison, but the Nema 34 graphs I found were mostly using a 110VAC supply (about 155VDC), while the Nema 23 ones were using 30-40VDC supplies. At those low voltage, torque drop of is very noticeable, and crippling Nema 34s with only 70VDC also makes for a very similar torque drop-off. You need voltage to over come the back EMF at speed. Without that voltage, torque at speed is very limited.
I have to say all this stepper and servo stuff is new to me. I'm very familiar with squirrel cage motors and vfds, but steppers, and now servos are a very steep learning curve. I know what you mean about the Clearpath graphs, but Teknic have stated that they are servos and not steppers. Their torque curves are certainly much more healthy at speed than any stepper graph I've seen (which admittedly isn't that many). Certainly when you compare the torque/speed graph of the Clearpath 2.04 Nm (rms) (CPM-SDSK-3421S-RLN) servo I was thinking of buying to that of the Astrosyn 4.8 Nm (holding torque), there's no competition. Where the Clearpath has no problem producing its nominal rms torque at 1000rpm the Astrosyn has fallen to something in the region of 0.9 Nm at around 3000 steps/s (900rpm).Quote:
Originally Posted by m_c
Attachment 22120Attachment 22121
Just as an aside, if the Astrosyns were geared 2:1 they'd only be producing 0.4Nm
Having looked at the torque curve more carefully I see now that with my table axes running at 2000mm/min that the motors should be producing something in the region of 1.7Nm - which should be fine and indeed seems to be so. I can also see why increasing the speed by 50% causes them to lose steps as they've lost 0.5Nm in torque:grumpy:
Moving off steppers and on to the spindle motor:
I spent this afternoon setting up the inverter and its control in Mach. I found that by fiddling with the PWM base frequency that I could get 100% on Mach spindle control to correspond to 100hz on the inverter (which previously it would not).
Having sorted this out I turned to pulleys and defined the four motor speeds as pulleys. I then calibrated the spindle speed to a range that I thought would work without me having to move the mechanical variator (Reeves type drive) or even replace it with a single fixed ratio poly-v-belt drive. I reasoned I needed speeds between 150rpm and 4000rpm. However, having run the slowest speed of the motor at the lowest reasonable frequency I quickly concluded that the motor has insufficient torque at this speed.
So I've decided to use be able to use the variator in two positions giving tops speeds of 2000rpm and 4000rpm with plenty of low-down torque in the lower range setting. I'll be mostly machining cast iron and steel so the lower range will be deployed most of the time.
As I understand it Mach3 while it's possible to tell Mach3 the spindle speed it cannot change the pulley setting automatically. It would have been nice if one of the four motor speed control relays could have been linked and operated by this feature of the software, but I guess I'm hoping for too much! In any case I'll have to set the variator manually.
[pedant mode] The motor should have the same torque at any speed provided the VFD is not limiting current. It's the lack of gearing that means there is insufficient torque at the spindle[/pedant mode]Quote:
Originally Posted by Agathon
Mach 3 can via Macros. I've never done it, however you can use a Macro that takes the requested spindle speed, and changes gears accordingly. Probably worth having a search for gear change over on the Mach forum.Quote:
As I understand it Mach3 while it's possible to tell Mach3 the spindle speed it cannot change the pulley setting automatically. It would have been nice if one of the four motor speed control relays could have been linked and operated by this feature of the software, but I guess I'm hoping for too much! In any case I'll have to set the variator manually.
I hadn't understood the four contactors for speed were operating a mechanical device, I thought you were using the inverter digital inputs to select four preset speeds.
I feel your pain. I got the notion that everyone was getting better performance out of stepper motors than me and I now have 3.4's which I drive at 220 Volts. Problem is they run hot and I don't trust them at full tilt which is strangely self defeating. You can't win. I keep thinking servo motors are the way to go but I have always been just one rebuild away from wonderful so I am not convinced. Do you really need mega speeds? What is the max delay, end to end on the table? Is it really a problem?
I suspect the reason we can't have nice stepper motors is that someone has decided they must all be 200 full steps per rev. I got a Roland mill which had 400 full step/rev motors and it is freakin' amazing.
You are blessed to have that quick tool release, but I am not so sure about driving the Z through the quill rack. I put a ball screw to drive the Z, a ball screw that has to be released every time I hammer out taper tooling. It's only one M8 cap head but I've already had to helicoil the thread.
You should have a look at Mach3 Brains, watch this video to get an overview; https://www.youtube.com/watch?v=O8V7dZy02og
[QUOTE=m_c;92790][pedant mode] The motor should have the same torque at any speed provided the VFD is not limiting current. It's the lack of gearing that means there is insufficient torque at the spindle[/pedant mode]
While I don't claim any expertise on inverter drives, I've been using them for the last 20 years and my understanding and experience is that torque falls off either side of the nameplate frequency. Modern motors designed for inverters and "vector control" improve the flatness of the torque curve but the reason machine tool manufacturers using this type of drive specify such huge motors is due to the drop-off in torque. As an example, one of the Swiss firms I represent in the UK make a plain lathe of 70mm centre height designed for instrument making and horological work. The spindle is belt driven at 1:1 by a inverter controlled 1.1kW motor. In the past this machine was made with multi-pulley drive with a fixed speed motor of 300W!!!
You may have provoked me into boring you all with another video! The contactors just control the 4 motor speeds nothing mechanical. The variator (Reedes drive) gives the mechanical variation.Quote:
Originally Posted by EddyCurrent
Is that Reedes or Reeves, the one where the pulley opens and closes so the belt rides up and down ?
Sorry, yes Reeves.Quote:
Originally Posted by EddyCurrent
Thanks for the empathy - much needed! I think the thing is that most people are building routers with very low friction linear rails and low mass tables/gantries. What I've got here is a beast compared to a CNC router. In fact it's probably one of the lightest and smallest pro CNCs that was built. Since owning this machine I've been looking at other high quality, but old, CNCs with a view to doing the same thing, but these are all massive for the same envelope that the Fehlmann has. A couple of good Deckel CNCs have sold on eBay recently for just over a grand but you're talking about 2-4 tonnes of iron, which I don't have the space or the heart to be involved with. Maybe I'll eventually replace my manual Aciera F4 with a bigger CNC, but for now I'll sit tight.Quote:
Originally Posted by Robin Hewitt
You are quite right about the traverse speed. The afore-mentioned Aciera F4 has a rapid of 1800mm/min which has never made me feel like I need to pop the kettle on while it does its stuff. So 2000mm/min is fine on the Fehlmann - if it can do it reliably.
Having been involved with very high-quality manual machine-tools for a very long time I tend to want a professional solution to a problem. It's been a steep learning curve, but as I see it stepper motors for machine tools are now really an amateur thing. I want this machine to perform as an industrial machine should and I will be using it to produce high precision stuff for my own amusement and for my business (I say business, it's Halcyon-days are behind it and with things as they are at the moment it's a paying hobby, which isn't such a problem as I have another job too). Hence, I am drawn to servo motors. The only problem is I know sweet FA about them!
I'll see how driving the rack goes. As I said in the vid, this thing is a very high precision thing and I can detect no backlash in the rack. There must be some of course, but it's certainly less than 0.01mm. Fehlmann developed this machine into a 3-axis a year or two later and drove the rack with a servo motor - they seemed to have sold quite a few of these machines and they are still sought after in Switzerland.
Hammering-out taper tooling is always bad news. Have you thought about making a captivated draw-bar so that it pushes the tooling out as you unscrew it?
At the end of the day I would be looking to convert an S value in gcode to a combination of inverter speed and contactor selection, does that sound correct ?
So what is the relationship regarding gcode S value, inverter frequency, contactor selection, spindle output speed equal to gcode S value ?
If that can be determined then it looks like a Mach3 Brain can be used. If you watch the video linked to above you will see that their example has similarities to your own requirement
You've lost me a bit there, but I'll take a look at the video.Quote:
Originally Posted by EddyCurrent
Took a look at the Brains video - that looks very promising. I'll do some research.Quote:
Originally Posted by Agathon
I did make a little video on the spindle speeds - probably redundant now, but here it is:
https://youtu.be/hncxZvUoy1g
For induction motors, torque remains pretty constant over the rated speed range up to the rated speed, and is a result of the amount of current flowing through the windings. Although frequency controls the speed, as speed drops, the voltage required to drive that current through the windings also drops proportionally I.e. for a 3000rpm 240V motor, peak voltage will be around 120V when running at half speed.Quote:
Originally Posted by Agathon
Now because voltage is reduced, so is the power output, so if the above motor was rated at 1KW, at 1500RPM it would only be producing 500W. You could increase current (which some inverters can do to give a low speed torque boost), however you risk overheating the windings due to the extra current.
Above the rated speed/power, current becomes limited. So taking the above example motor, and trying to double it's speed to 6000RPM, you can do so by doubling the frequency, however unless you double the supply voltage, 240V will only be able to force half the current needed to get full torque through the windings, with the result you will only get half the rated torque.
The result is above the rating, the motor becomes power limited. You can have double the speed, but only half the torque, so even though the motor is spinning faster, you still only have 1KW of power.
This is why when fixed speed motors with gearing get replaced, they're often replaced with far bigger motors.
With gearing, say you take a 2Nm 3000RPM (probably about 600W), you gear it down 4:1 and get 8Nm at 750RPM.
To get that same torque at the same speed using direct drive, and still retain 3000RPM, you now need a motor capable of 8Nm and 3000RPM, so you need a motor with 4 times the power. (realistically you'd compromise with some gearing, a bit less torque, and running the motor so higher speeds are above the motor rated speed and into the derated/reduced current area).
That's the reason why most modern CNC machines come with such big spindles. For most end users, it not because they're going to be managing to use upwards of 10KW hogging metal with endmills, it's so they still have enough torque to drive a big facemill at a couple hundred RPM while still taking a decent depth of cut.
Very nicely and clearly explained. I was aware of the frequency voltage relationship, but hadn't really twigged the issue of running higher than the name-plate frequency and the consequent lack of volts. As the video shows, I don't really think that torque is going to be so much of an issue as far as the spindle goes, so I'm fairly hopeful about my plan to use the higher mechanical speed setting most of the time. We shall see.Quote:
Originally Posted by m_c
Having looked into Brains more I understand your previous post and think that it all looks doable even for a novice like myself.
I've done more fiddling with the table and found that at 2500mm/min the table is repeating position over 200mm for the x and 150mm for the y better than 0.01mm (in fact pretty consistently within 0.003mm) which I am extremely pleased with. I tested fairly extensively with 10 or 20 repeat cycles.
I am still getting the grunting noises from time to time, so I suppose this is down to the latency of the processor generating the pulses?
3 microns is an interesting number BUT what about the backlash?Quote:
Originally Posted by Agathon
What backlash? This thing is Swiss! I would think that the 3 microns is simply down to vibration in the mechanical loop between the digital dti and the mount. If you're not careful you might provoke another mind numbing video!Quote:
Originally Posted by Robin Hewitt
Try the attached spreadsheet (I use LibreOffice if you don't have excel)
Just enter the desired spindle speed in cell A4 and it will calculate the inverter frequency and contactor to energise.
That logic needs put into a Brain using Mach3 DRO 169-Cmd SpindleRPM as the desired frequency.
It uses an inverter frequency up to 100 hz and assumes the manual adjuster is set to 1.5 * motor speed, so for the 1400 speed it's set to 2100 RPM then the inverter will take it to 4200 RPM
Thanks, that's really kind of you. I do have Excel so I'll take a look.Quote:
Originally Posted by EddyCurrent
Can anyone tell me what G0 code will make the x axis move +100 back to zero and then +100 repeatedly - say 10 times? I'm looking around for a simple answer but can't seem to find one.
Ignore that last request I worked out how to do it - probably in a very labourious way but none-the-less it worked.