Messages

3661

General Mach Discussion / Re: Machine servos stop but mach3 keeps running movements as machine is moving.

« on: September 12, 2018, 06:23:22 PM »

Hi bbutcher,

then you may have posted on the wrong forum

To my knowledge OP has posted in exactly the right board, its you who keep banging on about Galill,
for which there is a separate board.

Craig

3662

Galil / Re: A Galil Mach 3 communication problem I need help with.

« on: September 12, 2018, 02:53:53 PM »

Hi,

Mach4 does not support the DMC 17xx or 18xx

It the other way around, those model Galill's do not have the plugin required to work with Mach4. Have you asked Gallil
if they intend on releasing a Mach4 plugin?

Your suggestion to just use Mach 4 is great, except for the expense involved

$200 for Mach4, how much is you machine worth?....or your existing Galill?

Have you thought about LinuxCNC....its free!

Craig

3663

General Mach Discussion / Re: Machine servos stop but mach3 keeps running movements as machine is moving.

« on: September 12, 2018, 02:38:22 PM »

Hi,

but sadly enough it seems that Mach3 is completely unaware when the motors stop . So my question is , is there a solution for this.

Mach3 and Mach4 are open loop controllers.  This is because Windows is not a realtime computing system and has never made any pretentions of that
sort, Both Mach3 and Mach4 can read an encoder for instance but the potential delay between the reading of the encoder and Mach issuing corrective
instructions mean the feedback loop fails.

As you have pointed out, servos and their drives are closed loop. Thus if Mach issues a movement instruction to a servo drive and the servo can't keep up or is
overloaded and therefore is a significant distance from its commanded position then it will fault and signal an error. There are stepper motor drives that
have closed loop drivers, they come at a premium and really offer too few advantages to be economically viable. They do provide some reassurance to
you, not that they prevent the situation from happening.

The real issue is that your stepper are loosing steps and they should not. Properly functioning steppers with appropriately specified drives DO NOT loose
steps. It means that you have the max velocity and/or acceleration in the motor tuning set too high.
Can you give us some details...stepper model, drive model, power supply voltage, microstepping regime, steps per unit value, max velocity and acceleration.

Once I got my steppers dialed in they don't loose steps and haven't done for five years unless I'm trying to do something stupid.

There are other ways, the previous post has an idea. There are motion controllers which can monitor encoders and enact feedback corrections. Gallil is one
very good but very VERY expensive example, expect to pay $2000-$3000 for a Galill. Vital Systems HiCon is or can be made capable at around $1200 with
the right activations and probably the CSMIO at around $600. All of these solutions rely on the realtime feedback computing be done on the control board
and therefore relieve Mach of the task that it is to slow to do.

LinuxCNC is a realtime computing system, I have done some experiments with it and on my spare laptop it has a jitter of 6us which allows a communication
loop bandwidth of acceptable quality of 25kHz. That's way less than you could get with a dedicated realtime motion controller but is none the less very
useful. Downside is you are going to have to be a whiz at Linux and able to program in C,C++ to take advantage of it.

The problem you describe (loosing steps) is well known, if you read the forum you will find hundreds of posts about it. You will also see that most people
have solved it as well and been using Mach happily ever since. May I suggest before you consider closed loop servo/steppers, a closed loop motion
controller or a closed loop computing system like LinuxCNC that you spend some time to understand what is happening with your steppers and
the potential solutions to it. As I say many have with good success.

Craig

3664

General Mach Discussion / Re: Pul and Dir pins short

« on: September 12, 2018, 08:20:32 AM »

Hi,
how do you mean 'short on the board'?

They shouldn't be like that. Without manufacturers support who knows.
I have recommended to you before to buy a recognized brand....I do so again.

Craig

3665

General Mach Discussion / Re: CS Labs CSMIO/P-A

« on: September 11, 2018, 09:03:57 PM »

Hi Mick,
go on......where's your sense of adventure!!! There's nothing more exciting than having 585V DC link
capacitors discharge in front of your face in a shower of sparks, smoke and molten metal.

It makes your heart pump really fast....and thats got to be good for your health.

In truth its not quite as difficult as it seems, there is a large body of published literature on the subject
of active PFC.
Depending on your drive you could instead of hooking three phase wires to it hook two DC wires of your home brew
boost PFC single phase DC power supply. I work for a company that repairs industrial welding equipment
I could walk around the workshop and pick up all the secondhand parts I need within the hour. A $50US
Texas Instruments Lauch XL microcontroller board and your in business, and I wouldn't be stopping at 3kW,
I'd be aiming at 5kW cont and 10kW 10 sec overload.

3666

General Mach Discussion / Re: What kind of bit/end mill should I be using?

« on: September 11, 2018, 08:49:16 PM »

Hi,
well that does make a difference.

The temptation is to take a lot of very shallow cuts but it is a mistake. That would mean that the only
part of the tool that sees any work is the very tip, once it gets blunt you throw it out despite the rest of the
flanks of the tool being good.

Take fairly deep cuts, maybe 1/8 inch or more. Then use your speeds and feeds calculator to work out how fast
to cut. You are hoping that you will load the tool up so that the spindle is producing some real power.....thats
when the tool is at its most cost effective. Sure you can make it last longer by pussy footing around with it
but when you bore in, approaching its strength limit, you will do a lot more cutting before it wears out.

Under those circumstances cooling, or more importantly, chip evacuation and lubrication become important.
I personally have had great success with flood cooling despite the extra complication and mess. rcaffin, a regular
on this forum uses M(inimum) Q(uantity) L(ubrication), a mixture of kerosene and olive oil with compressed air.
He uses vanishingly small quantities with great results and has all but convinced me to try it too.

We were discussing it on CNCZone but feel sure he'd be willing to post some pictures of his setup, its remarkably
simple.

Craig

3667

General Mach Discussion / Re: CS Labs CSMIO/P-A

« on: September 10, 2018, 05:43:45 PM »

Hi Hood,
master/slave works really well on a gantry say with two ballscrews but Mick is talking about
having two servos locked together on one shaft. I think thats a good way to generate a lot
of expensive smoke.

Craig

3668

General Mach Discussion / Re: CS Labs CSMIO/P-A

« on: September 10, 2018, 05:30:46 PM »

Hi,
if you want 3kW single phase then go Allen Bradley 2098-DSD-030 and matching servo and
be prepared to pay extra for the line reactor/filters.

It is the biggest single phase servo drive I know of.

The only other alternative is to strip the DC link circuitry off a three phase 230V drive and make your own DC llink,
you will almost certainly need buck/boost PFC correction beyond 3 kW but in theory you could make it whatever
 size you want

Craig

3669

General Mach Discussion / Re: CS Labs CSMIO/P-A

« on: September 10, 2018, 04:57:29 PM »

Hi,
imagine for instance that one servo is mounted 1 degree rotated in its mount relative to its
partner.

When trying to share load when executing an indexed move (position), the leading servo will try to be at 90⁰
at moment t=nnn.nnn and likewise servo 2 will try to be at 90⁰ at the same time. However the leading
servo will actually be at 91⁰ at time t=nnn.nnn. The trailing servo will treat this as the load advancing
on its ideal commanded position and therefore try to retard the leading servo.

This was an artificial construction to illustrate the idea, but any two position servos will inevitably be slightly different
to each other, and if that error differs by more than the 'zero error window' they will fight each other.

In torque they should share equally. In velocity mode imagine one servo is programmed to be a fraction faster than the other.
Which servo powers the load and would not the slower servo be dragging the faster one? There again I have made a
worst case to illustrate the idea but two velocity servos are never going to be identical. What happens to load
sharing then?

I'm not concerned about the power factor.

Yet!!!

Craig

3670

Mach4 General Discussion / Re: Old turret lathe

« on: September 10, 2018, 04:42:01 PM »

Hi,
when a line of  Gcode is read, interpreted and the PVT data is sent to the motion controller the
call returns to read the next line while the current line is executing.
If the next Gcode line is a jog move to an OB axis it processes that and dispatches that to the
controller. Unlike a coordinated move it can start imediately, thus you have Gcode executing
and you can still issue OB axis moves at the same time.

Craig