Messages

2581

General Mach Discussion / Re: Slaving closed loop stepper motors

« on: April 19, 2019, 09:20:01 PM »

Hi,
you are right that certainly sounds weird and I don't have any bright ideas.

If you were to consider buying something new then GET AC SERVOS. Those DMM's that I linked to earlier will blow
your mind. Steppers are Ford Model t's, servos are  2019 model sports cars.

Craig

2582

General Mach Discussion / Re: Slaving closed loop stepper motors

« on: April 19, 2019, 08:32:52 PM »

Hi,
I am intrigued that you should replace open loop steppers that work with closed loop steppers that don't.

As you have discovered close loop steppers can lose steps just like open loop steppers and for exactly the same reasons.
While it is true that if unloaded a closed loop stepper will try to correct itself but is loaded it will try and probably fail,
I mean if it lost a step because of overload at speed then why should it respond any more favourably to another pulse
inserted in the pulse stream to correct it?

If you are going to replace steppers you should go for AC servos....no shagging around with steppers. Good AC servos
have come down in price and will wipe the floor with any stepper open or closed loop.

http://www.dmm-tech.com/

Does the stepper driver have a programmatic 'following error window'?. That is if the required position deviates from the
actual position it triggers an error. Some latitude between the two ( actual vs commanded) is required. If the demand is within
a few steps then the drive is highly likely to fault 'following error' whereas if the window of latitude is 50 to 100 steps then
it probably wont fault but at the expense of accuracy.

It may be that the window is set too wide and therefore accumulated error is causing the drift in position but still be less than
the error window.

Craig

2583

Mach4 General Discussion / Re: Step pulse duty cycle

« on: April 19, 2019, 06:50:02 PM »

Hi,
well it certainly sounds like you have experimented enough and have the calculations well in hand.

As I stated in my previous post that it is unrealistic to expect better that 400 step/rev resolution from an open loop
two phase stepper. You can set a microstepping regime with apparently far better resolution but that theoretical
resolution is not actually obtained. There is a sound explanation for that....but I would not, unless you require it,
regurgitate the reason for it.

A closed loop stepper is still a stepper with all the limitations of a stepper, reducing/vanishing torque at high speed.
A closed loop control system does not prevent that. For this reason I do not recommend that the premium paid
for closed loop steppers as opposed to well engineered open loop steppers and drivers.

One area where closed loop steppers do offer an advantage is resolution.
An open loop stepper can reliably be at either a full step position or a half step position. A half step position occurs when
both phases are energized and the resulting magnetic field 'drags' the rotor to a position midway between the two
full step magnetic poles on either side.

A closed loop stepper enjoys a distinct advantage in this regard. Should your required motor position required  to be
14% CW of the last full step position (therefore 86% away from the next full step position). As a guess you would reduce the
phase current in the 'leaving' phase  to 86% of max and increase the current in the 'arriving' phase to 14% of max.
It is almost inevitable that the rotor will not achieve its 14% position as required. However it actual position is measured by
the encoder and therefore signaled to the driver. Thus the driver will adjust the current balance between its two phases
such that the rotor indeed achieves its desired position. It may require a current distribution of 84% and 16% respectively.

This constitutes the sole major benefit of a closed loop stepper. The question obtains 'does the increase in resolution
justify the expense'

Lets compare open loop resolution vs closed loop.

Open loop at half step, resolution = 400 step/rev, with a 5mm leadscrew resolution = 0.0125mm or 12.5um
Closed loop (Leadshine encoder 6000ppr encoder) =6000 step/rev, with a 5mm leadscrew =0.833um

Clearly close loop has an advantage....but the minorest amount of backlash or flexure and that resolution is wasted.
Unless you have the need, and are prepared to build a sufficiently rigid machine with backlash free components, I would
recommend open loop steppers and drivers.

If you go that way you absolutely MUST buy steppers with low inductance, its as important as holding torque. With
23 size steppers you should aim to get steppers of 1mH or less inductance. Pair those with a good highvoltage
stepper driver like Gecko or AM882 drivers from Leadshine and you in business.

You should, excepting very heavy loadings, achieve 1000 rpm or with a 5mm leadscrew 5000mm/min.

Craig

2584

Mach4 General Discussion / Re: Step pulse duty cycle

« on: April 19, 2019, 04:02:37 PM »

Hi,
the duty cycle of the ESS step outputs is fixed by the ESS and is not able to be programmatically controlled.
I have observed the step outputs of my ESS/BoB combo on occasion with an oscilloscope and they are already nearly
bang-on 50% duty cycle as is.

The ESS can produce pulse streams up to 4MHz, so 125kHz is scarcely a crawl for the ESS.

The limitation here is the ability of the driven device, in your case the stepper driver. The manufacturer specifies a max
of 125Hz, but you would be advised to stay well below that.

What microstepping regime have you in place?. Its not uncommon for inexperienced CNCers to specify a way to high
microstepping regime on the misunderstanding that it is better and gives better resolution. It does not.
With a two phase stepper (ie normal) resolution beyond half step microstepping is fallacious. The real advantage
of microstepping is smooth motion thereby avoiding mid-band resonance.

I would recommend a microstepping regime of about 8 micrsteps per full step or 1600 pulse per rev.
If you wanted your stepper to spin at 1200 rpm (ambitious for a stepper) then the pulse rate will be:
pulse rate=1200 / 60 X 1600
   =32kHz

That is comfortably within the spec of your drive.

If however you choose an unrealistic microstepping regime of 64 microsteps per full step or 12800 steps/rev
the calculation becomes:
pulse rate = 1200 / 60 x 12800
   =256kHz

Which is well outside of the spec.

What voltage power supply have you provided the drivers? The general rule is that it should be as high as the drivers
will stand because the steppers will run fastest.

Craig

2585

Mach4 General Discussion / Re: Mach4 with Pokeys57CNC and FRO

« on: April 19, 2019, 03:39:57 PM »

Hi Tony,
yes the PLC can be a trap like that. Here is the procedure:

https://www.machsupport.com/forum/index.php?topic=39965.msg267488#msg267488

The reason that the script is failing is because it cant find the register handle and that is because we hav not
correctly specified the path.  The pics you have attached however detail it exactly.

It should be:

Code: [Select]

local analogRegHandle=mc.mcRegGetHandle(inst,'PoKeysCNC_41744/Analog input 41')
Note this has to be exact, including uppercase/lowercase and whitespace.

Because of the issues that you can get into with the PLC I like to write and debug scripts by storing
and working from within the macros folder. Once its right THEN copy it to the PLC.
None the less having the wxMessageBox statement in the PLC script is a mistake. Remove it.

Code: [Select]

local inst=mc.mcGetInstance()
local analogRegHandle=mc.mcRegGetHandle(inst,'PoKeysCNC_41744/Analog input 41')
if analogRegHandle==0 then
        mc.mcCntlGetLastError(inst,'Failed to find register handle')
else
local analogString=mc.mcRegGetValueString(analogRegHandle)
local percent= tonumber(analogString)*250
mc.mcCntlSetFRO(inst, percent)
end

Craig

2586

General Mach Discussion / Re: Help! How best to configure Mitsubishi MR-E-20A (and 40A)-QZ

« on: April 18, 2019, 09:20:16 PM »

Hi,
according to the manual there is an Estop circuit within each servo drive.

If you connect with a wire CN1 EMG input (pin8) to SG input (pin13) then the drive side Etsop is permanently deactivated.

Craig

2587

General Mach Discussion / Re: Help! How best to configure Mitsubishi MR-E-20A (and 40A)-QZ

« on: April 18, 2019, 09:03:35 PM »

Hi,
sorry missed the earlier link....please disregard my post.

Craig

2588

General Mach Discussion / Re: Help! How best to configure Mitsubishi MR-E-20A (and 40A)-QZ

« on: April 18, 2019, 09:02:13 PM »

Hi,
this post should probably be on the Mac4 General Discussion board as would get more replies....possibly.

My machine is outfitted with the same Mitsubishi drives listed,

Where is that listing? Would you post either the drive manual or at least a link to it.

I can't figure out how to configure the e-stop circuit, and clear my AL.E6 alarm

AL.E6 sounds like a drive alarm.... not really Mach at all?

Craig

2589

Mach4 General Discussion / Re: Mach4 with Pokeys57CNC and FRO

« on: April 18, 2019, 07:45:53 PM »

Hi,

Code: [Select]

local inst=mc.mcGetInstance()
local analogRegHandle=mc.mcRegGetHandle(inst,'Analog input/PokeysCNC_41744/41')
if analogRegHandle==0 then
wx.wxMessageBox('Failed to find Register Handle')
else
local analogString=mc.mcRegGetValueString(analogRegHandle)
local percent= tonumber(analogString)*250
mc.mcCntlSetFRO(inst, percent)
end
Would this work?
Note I am not sure how the Pokeys registers are named, I am familiar with the ESS. I put a test in there so that if
a valid register handle is not found then the remainder of the code will not execute and crash Mach with an invalid handle.

Craig

2590

Mach4 General Discussion / Re: Mach4 with Pokeys57CNC and FRO

« on: April 18, 2019, 07:27:22 PM »

Hi,
where is this code to be placed....the PLC script....where?

I understand that you are following an example but I think you have seriously over complicated what should be very
simple code. You have a 'main' and two supporting functions with input arguments for both functions and a return code
for one function returning a critical value. Why the complexity?

All that is required is the a register be read, converted to a numerical value and the FRO be updated.

Craig