Messages

6861

Mach4 General Discussion / Re: MACH 4 for 5-Axis Micro-milling Application

« on: July 21, 2017, 10:21:33 PM »

Hi,
the post by Vital Systems Support is spot on. As it stands with your intention to use Aerotech servos and drives you don't need the closed loop
feature. As the post points out it would be ideal for something like a linear encoder. With a little jiggery pokerey you could get an LVDT or even
a resistive sensor to be read as an encoder and so close the loop on that axis/variable. Great feature but not one you require.

I note that in your list you've allowed for a differential step/dir board. If your expected step pulse streams are under 100-150 kHz then differential signalling
is not required. If over that speed and/or in an electrically noisy environment then differential signalling is a great idea.
If you use 2500 line encoders (10,000 count per rev) and you want your servos to achieve 1000 rpm (16.67 rev per sec) then your step frequency without
electronic gearing, ie max resolution, is 10000X16.67=167kHz. You can see that the max signalling rate will determine the max speed of the servo without
the use of electronic gearing. Given the cost of the differential board relative to the cost of the linear stages I'd say go for it.


What arrangements had you in mind for your rotary axes? I'm gobsmacked at the cost of linear stages....let alone rotaries!!! Are you sure you don't want to offer
someone a job, I have a CV ready to go...LOL

Craig

6862

Mach4 General Discussion / Re: MACH 4 for 5-Axis Micro-milling Application

« on: July 21, 2017, 03:12:01 AM »

Hi,
reread your post...no you don't need to purchase the closed loop option, that saves $595 right there! I'm not sure about how they have phrased their advertising
material, it rather suggests that to achieve accuracy you should BUY the addon. Certainly you seem to have gained that impression...its wrong. Any decent AC servo
and drive has a following error allowance, programmable and usually default of 20 encoder counts. If the servo gets out of whack from the position you specify
it throws a fault which you can decide how to handle. I assure you any decent servo and matching drive will easily achieve accuracy of  a few encoder counts, it
has less to do with the motion controller and more to do with the smarts the manufacturer builds into the servo/encoder/drive.

I suspect you will have more use for the 'extended IO package for $250 and if you wish to read analogue values at all the Analogue package for $200. Your spindle
is grossly unsuitable for rigid tapping so you wont need that. Until you've got some good Mach4 miles under your belt I wouldn't bother with the API and/or Basic/C
macro features either.

Craig

6863

Mach4 General Discussion / Re: MACH 4 for 5-Axis Micro-milling Application

« on: July 20, 2017, 01:28:16 PM »

Hi,
that's serious bucks!

I can understand that you want to do it right.

The stages would work fine with steppers. Steppers don't usually have encoders and are therefore open loop. I guess the majority of hobby machines are
arranged this way. Steppers, even good ones, are relatively cheap. Used within their limits they work great, no loss of accuracy. If however you push them
too far they can loose steps and without encoders you might not know and unless you have a self correcting drive you cant do anything about it.
The single worst limitation of a stepper is that they lose torque authority at speed, anything over 300-500 rpm and the torque drops off rapidly. Under
that speed they go well and at least in terms of torque do better than servos.
My mill uses 5 phase steppers matched to a low backlash planetary gearbox. They give me 1um resolution, 500 plus kg of thrust at the expense of being
slow, rapids of 1200mm/min. I haven't missed a step in several years. So steppers are quite capable with care.

Servos and matching drives are closed loop. Should a servo lag behind the drive will increase its output to have the servo catch up. Servos are fast, most are
rated to several thousand rpm and capable of high speed rapids suitable in a production environment. I note that Aerotech offer servos with encoders of
either 1000 line or 2500 lines. The servo with a 1000 line encoder has a resolution of 0.09 degree (5.4 arc min), not bad...The servo with a 2500line encoder has
a resolution of 0.036 degree (2.2 arc min), even better... Also I see that they offer a brake, you should consider a brake for the Z axis. You don't want the weight
of the spindle to cause the axis to slump under gravity when the machine powers down.

Clearly you want an accurate machine and are prepared to pay big bucks to get it. Accuracy is largely determined by the stages, the motors and matching drives
and how carefully you assemble/adjust them. Provided your motion controller works it doesn't affect accuracy. Any of the current crop of Mach4 capable controllers
will work fine, HICON is one of the more expensive ones but enjoys class leading reputation as well. If your happy to pay....stick with it.

Craig

6864

Mach4 General Discussion / Re: MACH 4 for 5-Axis Micro-milling Application

« on: July 20, 2017, 05:08:57 AM »

Hi,
I've just been reviewing Vital Systems site about your closed loop idea. I think your barking up the wrong tree.

The HICON controller can, when you pay to have it activated, read an encoder, do the numerical calculation to produce an error signal. The error
signal must be amplified and applied to a servo. Vital Systems do make such an amplifier but it suits their own Maxsine servos only. If you buy the
Areotech stages without servos and buy an fit Maxsine, yes it would work.

If you use Aerotech supplied servos then you must use a suitable driver, usually from the same manufacturer. Servo drives accept encoder inputs and
step/dir inputs (among other protocols) and drive the servos directly. The servo drive closes the loop, not the motion controller. Why bother paying an extra
$500 for a feature you don't need?

The primary activity of a motion controller is to take position, velocity over time numerical data from Machs trajectory planner and to do the calculations
to produce pulse streams to motor drives. They do not as a rule drive the motors directly, that's the job of a servo drive. The HICON controller is a very capable
controller and it can 'close the loop' but this is an added feature not the most vital one.

I note also on the Aerotech site that you can order stages with or without a motor and various choices from steppers to 2500 line AC servos. If you don't mind
me asking what sort of money do they want for the stages you have in mind?

Craig

6865

Mach4 General Discussion / Re: MACH 4 for 5-Axis Micro-milling Application

« on: July 19, 2017, 03:20:02 PM »

Hi,
I think that the closed loop package will probably work, remember you have to purchase the features required and so add to the cost of the controller
about $400-$600.

On the other hand Aerotech offer drives for their servos, the drive closes the loop. All that is required to control the servo is step/dir and maybe enable/fault
signals. Usually servo drives of this description have a software setup program which allows you to set speed/current limits, PID parameters, electronic gearing,
index homing and so much more. You can probably achieve the same thing manually but with a GREAT deal of work/research and hand bit/byte programming.
I recently used a AC servo for a spindle motor, a 1.8kW Allen Bradley servo and I used a matching Allen Bradley servo drive which is programmed with Ultraware,
proprietary setup software. I didn't much like having to pay for the software but setting up the servo parameters with a GUI is a doddle and makes the software
worth every cent.

Motion controllers issue pulse streams to steppers/servos and while some controllers can read encoders and 'close the loop' the servo manufacturer knows exactly
what encoder he is going to use and the electrical characteristics of the servo and usually do a better job.

Craig

6866

Mach4 General Discussion / Re: MACH 4 for 5-Axis Micro-milling Application

« on: July 19, 2017, 03:18:27 AM »

Hi.
looks like you're going to spend SERIOUS money using those puppies....are you sure?...really sure?

My experience building a mill, and with nothing as classy as the stages your talking about, is that as you learn and uncover the tricks and traps
that CNC presents it will blow your budget, maybe so badly that the project never gets completed.

My recommendation would be to use the Aerotech  drivers for the motors you choose, no guessing, you'll know they're right at probably similar cost
to the 'extras' associated with closed loop capable HICON board.

What drive does the spindle use? Most motion controllers do a pretty good job with VFDs open loop. Unless you need really accurate speed control or
indexing then 'no feedback' is the norm.

My experience was that I focused on travel/ rigidity/ accuracy/ resolution/ ease of construction and so on of the axes and the frame of the machine.
Towards the end of construction I started thinking about the spindle. This was a mistake, the spindle speed, power and torque, the materials being milled
and the tools to be used should be  at the TOP of the list. Realistic appreciations of these factors will determine to a VERY large extent how the rest of the
machine looks. It will also uncover any 'budget blowout points'.

I am guilty, not badly so, but guilty none the less, of being a perfectionist. The level of perfection I seek often exceeds my skill and certainly my budget. This
can be a real disaster in this hobby (obsession), I have at times talked my self out of enhancement projects because I couldn't achieve the level of performance
or perfection. I have become better at saying 'to hell with it and box on, make the thing and see how good it is'. While never perfect the results are usually worthwhile.
I need the practice of overcoming 'perfectionisum'....

Let us know how your going, thinking and dreaming about it is part of the fun!

Craig

6867

Mach4 General Discussion / Re: Mach4 controlling a spindle with MODBUS

« on: July 18, 2017, 01:43:21 PM »

Hi,
a while ago I helped a guy write some LUA code to run a VFD over MODBUS

http://www.machsupport.com/forum/index.php/topic,34023.0.html

maybe it'll help you too.

Craig

6868

General Mach Discussion / Re: usb to printer port

« on: July 18, 2017, 01:33:25 PM »

Hi,

http://www.mini-itx.com/store/

have plenty of boards with built in parallel port and for under $100.You'll still need 32 bit Windows and all the usual limitations of parallel
port motion controller apply. I bought a dual core Atom board about 3 years ago and loaded it with Windows 7 Embedded Standard 32 bit, WES7.
WES7 is a free download and if you register with MS free for six months. Buying a licence is not straight forward, you need to sign an agreement with MS.
Cost me $200 NZD, about $180 US at the time. Used the boards built-in port and another PCI port with Mach3 for 2-1/2 years, no problems.

More recently shifted to Mach4 and an Ethernet SmoothStepper with the same 'little baby Atom board'. The low computing power of the Atom runs
Mach4 fine but I would buy something with a bit more go if I were to buy again, i3 with 4 gig would be good enuf....

I would suggest that you need one PC to run your machine and another separate machine for CAD/CAM. Your machine controller PC should do nothing
else but control your machine. Anything else loaded on it like CAD/CAM, internet, networking etc just increases the chance that your machine will crash.
If you don't believe me just carry on and when bits of carbide are whizzing around your face after having exploded I promise you'll be convinced!

Craig

6869

General Mach Discussion / Re: usb to printer port

« on: July 17, 2017, 10:04:43 PM »

Hi,
no...a USB to parallel converter may be adequate for a printer but it will not work for Mach.

The original Mach uses a 'parallel port' but really its much more than that. The PCs own CPU is pressed into service to create accurately timed
pulse streams to drive steppers/servos. For various reasons a PC CPU is not very good at producing accurately timed pulses, it a real credit to
Art Fennerty that he managed to get it to work as well as it does. In any event the CPU is our 'motion controller' and now it has to communicate
those pulses to each driver simultaneously, so it switches the output pins of a DB25 parallel port on and off, and can do so simultaneously.

USB is a serial communication protocol and so it appends messages front-to-back and sends off chunks. The messages cant arrive simultaneously,
they have to be reassembled. Worse USB send packet every few milliseconds so any sense of accurate timing is absolutely shot. If we use the PC
CPU to create pulse streams we need a 'realtime' communication protocol to signal the drivers. USB wont do, in fact neither will Ethernet despite being
way way faster. There are such realtime communication protocols unfortunately such solutions that exist for X86 Windows PCs are very expensive.

What can and does work is to send 'trajectory' data from Machs Gcode  interpreter thru either USB or Ethernet to an external motion controller.
In this case the controller is most likely to be an FPGA, DSP or industrial microcontroller. All these ICs are optimised for generating pulse streams.
Note that the trajectory data can be sent serially, it gets assembled into a buffer table on the controller board. The pulse streams generated by the
controller chip turn output pins on and of simultaneously as is required for the motors to be co-ordinated.

As for your situation you have a couple of choices.
1) Install a PCI or PCIe card with a parallel port and use the PCs CPU as your motion controller.
2) Use an external motion controller, the cheapest and simplest are a UC100, a PMDX-411 or a PoKeys 57cncD25

A parallel port card is likely to be around $20. You may have to do a bit of research...not all parallel port cards are equal. Also because X86 Windows CPUs
aren't the best as motion controllers you will have to accept some limitations on speed, timing accuracy and especially sensitivity to other software running
on your PC.

The external controllers I've mentioned are USB and the output connection is a DB25, you might call it a USB to parallel converter, but they are in fact
way way too clever for that. They are around $100. Speed and timing accuracy is good. You can use either 32 bit or 64 bit OS. These devices all have Mach4
plugins if you want to see what all the fuss is about. There are even better and more capable controllers as well but at greater cost. If you budget allows
it I would recommend an external controller, I have used parallel ports but I wont be going back now!

Craig

6870

Mach4 General Discussion / Re: Mach4 V3390 and HiCON

« on: July 12, 2017, 08:33:43 PM »

Hi Mauri,
kool, I've looked and looked at the HICON and have heard they are good. I use an ESS and it works fine
and of course less than $200. My mill doesn't require anything special, what servos I run have their own closed
loop drive, neither do I require analogue input although it would be nice, could probably use the plain
HICON board as you have with succsess.

I've got a couple of customers thru work that I could supply Mach4 to run plasma tables. Warp9 is working on THC
but its not ready yet, and even if the ESS can handle it there is still the need to write a Mach4 panel...

HICON has an added feature for THC and is looking pretty good....$1355 US including Mach4 Hobby.

Craig