Mach3 lathe spindle pulleys problem

Background .
I converted a Boxford TCL lathe to Mach 3 but retained the DC Spindle motor, and Lenze controller with Iso Board.

Had a bit of trouble getting the spindle index function sorted but thanks to some help from folks on here (you know who you are) I now have a very accurate "true spindle " readout which corresponds well with a mechanical tachometer .
And with spindle speed set to 2000 the analog signal generated by the pwm output is 9.6 v and reduces more or less in proportion to selected spindle speed so that looks fine .

Now I know that this is actually the motor speed , and the true spindle speed will vary according to the pulleys in use at any time , and that I need to select a number for the pulley setting along with the ratio of the selected pulley combination.
But if I change the ratio in spindle pulleys (and I have tried varying ratios from 1 to 8) but this has no effect at all upon true spindle speed .
In addition whatever spindle pulley number I select the display continues to show S1

I have looked for other related settings but dont see any which are relevant. I have read the manual and although there is a section on pulleys and the need to specify the ratios I dont see any guidance on other settings
What might I be missing.
Suggestions anyone ?

IIRC the spindle pulley settings didn't work correctly in turn, especially if you were not using the parallel port. Certainly the closed loop spindle control just didn't work.

The Mach forum is probably your best bet at finding a definitive answer though.

Using UC100 for connection to BoB
Have left the same query on Mach support forum so will see what comes back.

Also mostly as others said later.
The spindle feed / speed like threading depends on the plugin and hw you have.

So my csmcio-ip-s results will not be the same as yours, or the centipede I had in the past, or the smoothstepper early version.

And inputs from analog sources may be accurate, as regards relative spindle speed, but equally may not be accurate.
I tested this really well with Sergey who developed the centipede.
We proved very clearly that most index signals, single pulse, are not too good and reliable and crisp.
He developed a specific board for that, to make a single crisp reliable pulses.
This was several years ago.

My lathe now uses industrial servos at 1:3 belt drive, HTD 8 MM x 30 mm, 1:3.
The controller csmcio-ip-s can read them, from servos with 10.000 counts/turn.
The readout fluctuates +/- 1 rpm, which is an error in csmcio plugin.
It´s still extremely accurate.

The spindle speed is still extremely accurate.
About 0.01 rpm accurate in fact.


This wont be of much use to You, as mach results depend on the controller and or hw you have.

Quote:

---

Originally Posted by John11668

Background .
I converted a Boxford TCL lathe to Mach 3 but retained the DC Spindle motor, and Lenze controller with Iso Board.

Had a bit of trouble getting the spindle index function sorted but thanks to some help from folks on here (you know who you are) I now have a very accurate "true spindle " readout which corresponds well with a mechanical tachometer .
And with spindle speed set to 2000 the analog signal generated by the pwm output is 9.6 v and reduces more or less in proportion to selected spindle speed so that looks fine .

Now I know that this is actually the motor speed , and the true spindle speed will vary according to the pulleys in use at any time , and that I need to select a number for the pulley setting along with the ratio of the selected pulley combination.
But if I change the ratio in spindle pulleys (and I have tried varying ratios from 1 to 8) but this has no effect at all upon true spindle speed .
In addition whatever spindle pulley number I select the display continues to show S1

I have looked for other related settings but dont see any which are relevant. I have read the manual and although there is a section on pulleys and the need to specify the ratios I dont see any guidance on other settings
What might I be missing.
Suggestions anyone ?

---

This lathe is a Boxford TCL acquired for £350 and I have used the existing structure , steppers and motor drive / control..
Effectively just added a pair of cheap drivers and a BoB and a bit of my own time (Which I dont count ) so total outlay on the machine is still around £400.
Cant really justify outlay on exotic motors so just something to play with until I get the confidence to invest in something better.

With regard to spindle indexing, the original Boxford spindle sensors did not stand up to the process, and with some unsuccessful experimentation and with some well considered advice from a friend, a different type was chosen which has turned out to be well up to the job. Correlates well with the measurement using a mechanical tacho so i am happy to use this and accept the reading which it gives me on the screen . Fluctuations seem to be only a digit or two in a thousand so good enough for my purposes.

I was always told that I would not be able to thread successfully with Mach but it looks like I will just need to discount pulley ratios too.
Will just play a while with what i have then look at changing to linux. I chose to keep Mach 3 cos i was familiar with it , but looks like it falls short on a 64bit operating system.

Thanks for inputs anyway.

Quote:

---

Originally Posted by John11668

I was always told that I would not be able to thread successfully with Mach but it looks like I will just need to discount pulley ratios too.
Will just play a while with what i have then look at changing to linux. I chose to keep Mach 3 cos i was familiar with it , but looks like it falls short on a 64bit operating system.

Thanks for inputs anyway.

---

Mach 3 can produce perfectly acceptable threads.
Only thing that won't work on 64bit, is the parallel port driver. Mach itself still works perfectly fine, you just need to use an external motion controller.

I use an external motion controller (UC 100) so am still exploring the options here .
Trying the wizard but it it comes up with an error message saying that no lead is specified in the G76 commend so am looking into that now .

My PC is 64 bit with windows 10 and connected via the UC100 controller to USB port .
Are you saying I should still be able to change spindle pulley ratios and cut reasonabe threads with this rig . If so what settings do I need .

Posted on the Mach Forum.
Admins must be away. Nearly 2 days and still awaiting approval :apologetic:

I just had a look, and your post is now showing up.

No replies yet though :subdued:

John, you know that I'm running LinuxCNC with my lathe - because I really wanted to move from Mach3, and I'll be honest, from what I've just read the Mach3-Turn is poorly supported - I guess you've done that research already. Just be aware that if/when you do pull the pin on jumping onto LinuxCNC (I know you mention it over several threads, and I wish there was an alternative!) you're (a) not going to be able to use your UC100 - so direct PP or expensive controller (my Mesa 7i76e with shipping and tax cost nearly what you paid for your lathe! - 302 Euro). (b) PP is going to need fairly fast task threads for thread-cutting - so you'll need something above the bottom-rung of the desktop ladder, and - (c) maybe the most obvious thing for me - the support for MOP wizards just isn't there - so you'll be hand crafting G-Code to operate the lathe. There is at least one attempt to integrate some basic turning add-ons but these are flakey, at best (e.g. you need to know that they just change some of the modal states - e.g radius/diameter mode internally with no recovery - not knowing that I've had some entertaining crashes). I do like LinuxCNC in some respect but be well aware that it comes with a health warning. Maybe best done when the workshop is nice and warm.

So
Is Mach 4 possibly an option?
How does it handle spindle pulleys and will it manage threading?

Found this video from 2017 in which the guy seems to like the features but mentions threading / spindle index issues but have to wonder if it has moved on yet https://www.youtube.com/watch?v=rqIqn3lg3wA

Any Mach 4 users out there to advise ?

And what is the function of "pokeys " and might this be able to overcome the problems I am finding

Also found this on Warp 9 site which suggests that Mach 4 later builds manage the threading function.

"""Lathe Threading
Both Mach4 and Mach3 support threading operations with G32 and G76 commands. The start of each threading pass is synced with the (required) Spindle Index Pulse.

Something broke G76 in Mach4 build 4385 through at least 4563. G76 and G32 work just fine in Mach build 4360 (available here) and lower, so we recommend using Mach4 4360 with ESS build 259.

UPDATE: Lathe threading with G32 and G76 are working properly with Mach4 4580 and ESS 268 or 269.
(This was not an issue in the ESS plugin, so any ESS plugin version should work.)



The speed of the Z and X cuts will be adjusted to compensate for the instantaneously measured spindle RPM, calculated by the spindle Index pulse or the (much more accurate due to multiple RPM updates per revolution) spindle Encoder A channel.

Even if you are using the Spindle Encoder to measure the RPMs more accurately, you must still have the Spindle Index signal hooked up as well. Without the Spindle Index signal, we can't determine the 0 degree location required for starting a threading pass.
You only need to hook up Spindle Encoder A for the RPM measurements to be made. However, without the Spindle Encoder B channel hooked up, the encoder will just dither back and forth between values. With the Spindle Encoder A and B channels hooked up, the encoder value will count up or down (depending on the rotational direction).


Threading works with either ESS PWM spindle control or Mach4 OB Step Dir Spindle control (a STEP DIR Spindle MUST USE MOTOR 5 for the OB motor). When the GCode starts a threading pass, we halt all motor motion on motors 0 through 4 so that we can wait for the index pulse to tell us to go.

If you are running a Step/Dir Spindle on motors 0, 1, 2, 3 or 4 it will be instantly stopped at the beginning of each threading pass, and you will get a clunk or a reduction in spindle speed for an instant, until the Spindle Index signal pulses.
On motor 5, we allow the Step/Dir spindle to maintain it's current velocity while waiting for the index pulse. This is why you want to be on motor 5 instead of motors 0 through 4 for your Step/Dir Spindle.
The PWM spindle is not affected by this.


If using a Step/Dir Spindle, you may wish to do these three steps to help prevent clunking:

1)Make sure your Step/Dir Spindle is configured as motor 5 (see the previous paragraph).

2) Go to Mach4 Menu -> Configure -> Plugins… -> ESS v... - Warp9 Tech Design, Inc -> Motors tab and check the box for "Enable AntiClunk Mode for Servo Motors".

3) Change your screen set's stop button so it will feed hold fold instead of just dead stopping your Step/Dir Spindle motor.""

Might there be a way forward here ?

Mach 4 is an option.

How external controllers handle threading, is entirely up to the controller. Mach simply requests a thread, and the controller does it.
You can always install Mach 4 and give it a try.

Pokey's originally was just an external button panel, but they now do various controllers- https://www.poscope.com/product-cate...ys-io-devices/

Installed Mach 4 but wont run without a controller.
So totally different from M3 too. Then I understand that it will only allow 6 mins of use unless you buy a license.
Cant imagine much of a trial under those constraints.

I installed Mach4 last week to test something, and it ran fine for me without a controller attached.
Without a license it limits how long code will run for, but I think it's a random time limit.

I can make things happen on the screen but it wont do anything with the machine ,(not even Jog)
My machines have both run in Mach 3 using the UC100 USB controller but nothing in M4 so I am guessing it needs ethernet controller .
I dont mind the expense as long as it works . Same applies to the license . I guess both of them could be moved forward to my next project .

Who has used M4 with such a controller and how did it go .

CNCDrive list a Mach 4 plugin for the UC100, so it should work.
You have installed the Mach 4 UC100 plugin - http://cncdrive.com/UC100.html ?

I hadn't but will give it a shot , in the morning perhaps .
Looks like I may also have to read the Mach 4 manual too

So downloaded the plugin via the auto install route http://cncdrive.com/MC/UC100%20full%...UCx00setup.exe
Seemed to give a bit more functionality and I was able to change the keyboard functions (all I wanted were x+ z+ x- z-. on up down left right arrows ) but I got in a bit of a mess and decided to start afresh so uninstalled Mach4 and downloaded again.

Now wont install
Keeps coming up with "cant find ******************* " which you can ignore but then immmediately comes up with another "cant find, and another and another

Grrrrrrrrrrrrr!

So this is now winding me up
I have installed the plugins and they now appear in my M4 plugins file

Attachment 29410

But I have restarted my computer (and M4) and I just get the message as follows.


Attachment 29411


I have just tested this on my office computer but of course no UC100 device is connected

On the workshop machine (where a UC100 is connected ) I do not get the error message but when I select to configure/select motion device then UC100 does not appear as an option .
I only see "simulation device" which i presume is part of the M4 software as supplied.

So where do I find the wisdom which will tell me what I am doing wrong :apologetic:

Quote:

---

Originally Posted by John11668

I can make things happen on the screen but it wont do anything with the machine ,(not even Jog)
My machines have both run in Mach 3 using the UC100 USB controller but nothing in M4 so I am guessing it needs ethernet controller .
I dont mind the expense as long as it works . Same applies to the license . I guess both of them could be moved forward to my next project .

Who has used M4 with such a controller and how did it go .

---

Have you successfully run a machine on Mach 4 m_c ? and if so with what motion controller
I have raised a question on the cncdrive forum but they are telling me that the problem is something missing in M4 . Typical when you have two products running hand in hand that if it fails to work each blames the other

No I've not.
I do have a license for it, as I bought it years ago when they done a pre-release offer, but I've never had any reason to run it.
The only reason I installed it last week, was to check how macros are stored, as somebody was having an intermittent issue with a machine hanging at a certain point, and to tell them where to find the files so we could try and workout what could possibly be causing the problem.

I have a response from CNCdrive forum which looks authoritative if maybe a bit complicated .
Will post it here if it works

" MAke sure the .net framework 3.5 is installed (enabled in the Windows features on Win 10.)
Run the UCx00 automatic installer application, select MAch4 and select the UC100 if you using that.
Make sure to plug in the UC100 to the computer's USB port. (You can also verify that the USB drivers were installed properly with checking the Windows device manager.)
Run Mach4
Enable the plugin in MAch4.
Select the UC100 as motion control device in MAch4.
Enjoy. "


Unfortunately it didnt work!

OK
so sent a query in to Artsoft . Mach support supplied a remote connection link and while I was having my dinner the problem was resolved in my absence . Came back after dinner and a notepad message told me UC100 was now connecting.
I will ask tomorrow if they can tell me what was wrong for the sake of others who may have similar probs .
I will now embark on the configuration process, but already I have the tool post whizzing about at silly speeds . Probably need a bit of motor tuning but that will be tomorrow.

Thanks to all that have tried to help . Will come back if I can get more info on what I might have done wrong.
Meantime I will continue with configuration and motor tuning then maybe see if I can do spindle pulleys and threading.

Watch this space :suspicion:

John

So it seems I did not enable the plugin. :orange:

Now I have to say that finding your way around Mach 4 is anything but intuitive so that is my excuse .
This is not helped by the simple fact that the manual applies to the 2014 version, and version 2 of M4 which has only been around since early 2018 has not yet had an updated manual produced.
So you try working through the old manual , and find the various configuration pages in the current version bear no resemblance at all.

Now while the functionality looks good in M4 with a lot of (maybe too many) options on the front page, and once you find your way around it may indeed be superb but I am still wandering in the jungle , and the guy who made the signposts, does not seem to have ever worked with M3 so they seem to be in a language much of which is foreign to me .
There is just too much on the front page . I cannot believe it would ever be used by the typical guy on a production machine

Obviously I have spent a lot of time on the forums in the meantime and there seem to be precious few who have mastered it, and I get the impression that most have stepper spindles , and those who have VFD control, or like me a DC motor with pwm are even fewer so constructive advice seems to be in short supply.

Doddy tells me that Linux CNC will involve a steep learning curve . I am wondering if Mach 4 might be just as steep.
Added to the fact that (it seems) threading will not be supported with UC100 so I will not be much further forward .

It seems that this device http://www.pmdx.com/PMDX-411 will cope with threading but is limited to 17 usable pins, so makes me wonder if I might be better saving myself a £100 on this and put it towards an ethernet device.

Meantime , back to playing with the wizards available in Mach 3 . :05.18-flustered:
I have managed to string a few together to make some Knobs and tapers and stuff, but I guess it is going to be a while before I get to make stainless steel bolts to suit my mates old Ariel red Hunter :disillusionment:

Go linuxcnc John
It works out of the tin
And there is support on the forums for it
I’m no expert but have got my orac fully sorted on linuxcnc and to coin a phrase it just works
It’s been faultless

The biggest hassle was getting things like a WiFi pendent and control panel buttons to work but I now have Hal and I in files that I will happily share with you so you are tweaking rather than writing from near nothing

Paul


Sent from my iPhone using Tapatalk

Fwiw..
My main use is lathe, and I am using mach3 mostly, and slowly migrating to mach4.

Have more than 2000 hours of use on mach3 lathe.

My 12x24 industrial lathe is a very deep refit, industrial quality (on version 4).
32 mm ballscrews (z), 0.01 degree indexing on C axis, 0.2 micron theoretical resolution, scratch built saddle on hiwin 25 linear rails, 32 mm ballscrew on x.

Now moving to linear rails on z, hiwin 25, to get sub-micron mechanical resolution (0.3 micron theoretical resolution) using ac servos of 750W, 220V.
1:2 belt drive, HTD 8 mm / 30 mm, taperlock pulleys from bearingboys uk.
This is the work of this week, about 40% done.

2 toolchangers, 4 tool revolver and 7 tool servo turret at back (not yet mounted), about 60 kg in mass for the turret.
Basic lathe was 350 kg originally, now around 500 kg.

I am very happy with mach3, machx.
Using a csmio-ip-s controller with lots of 24 V io.

Afaik, the spindle speed display is not related to the pulleys.
It just displays what the configured index signal tells the controller the speed is.
The display =/= what the actual accurate spindle speed and position is.

My spindle speed and position are accurate to about 0.04 mm on the circumference of a 200 mm disc, while cutting.
The actual spindle speed is accurate to much better than 0.1 rpm, perhaps 0.00x rpm.

It tracks to around 500 error counts, (when the spindle drive faults) but the error is never more than 10 counts in variation.
Typically, the error is around 1-2 counts, of 1 : 3x10.000 = 2 counts per 30.000 counts/rev.

In case of hitting a hard turning interrupted-cut edge the error count might be 10 counts for maybe 0.1 secs.
The led on the servo drive hardly has time to flicker.

The z and x axis drives fault faster than the lathe can break itself, in about 0.01 secs, as anything steel and cast iron can survive the minimal bend they get in that time, even when using upto 1000 kgf push force.
The servos fault in about 0.01 secs, 0.02 secs, on hitting something or overloading.

So the tools and toolholders and mounts bend before breaking, and the servos fault.
Typically within about 1 mm or less, sometimes much less.
X has about 1200 kgf push, z about 1800 kgf.

Using a *good* hw controller like the csmio-ip-s is the key.
It reads the servo encoders of 10.000 counts in real time for threading.
And outputs upto 4 Mhz of perfectly accurate pulses.
And has industrial 24 V io.
Endless io.
Unfortunately, it is quite expensive.

But so are all other options, when adding in 4 Mhz speeds, encoder tracking, 6 axis, 120+ IOs, 24V IOs, etc etc.
Differential drives.
Analog inputs and outputs of high quality with *good* plugins with *good* support.

The csmio-ip-s supports glass scales for extra accuracy.
I intend to install them at some point.
My goal is to make threads of sub-micron accuracy, industrially.
I have spent 17 years and 150k € on the journey.

My many points are:
1.Machx is quite upto the task of high end stuff.
2.Better stuff is not quite free, and 500$ will not buy you all the hw you need for a controller.
3.It is very very hard to actually do stuff really very well.
-- For wood routers and basic 3d woodwork most-anything will work.
4.Linuxcnc will also work very well for high end stuff.
Nothing against linux, I used it professionally and industrially in my xxM€ IT shop very successfully in great scale.

If your index signal is on the spindle, then the pulley should have nothing to do with it.
Whatever pulley you have in use, you get a given speed, and the index signal should show that speed.
No ?

In My Experienced Opinion:
No matter what, your index signal is likely flaky.
At varying speeds, depending on pulleys and/or variable drives, the index signal is going to be fuzzy and not crisp on some or all speeds.
Most likely all speeds.

So the signal triggering is going to vary by some amount in ms partly related to the speed of the spindle, and not constant at all.
Ideally, your index signal needs to indicate the exact angular position of the spindle, related to the exact angular position of the thread, to less than 0.02 mm error at the outer diameter of the workpiece.
On a typical sub-20 mm D thread you are looking for an angular resolution of about 0.05 degrees, +/-, better than 0.1 degrees.
Otherwise the sw will cut the thread undersize or oversize or wonky depending on where the sensor triggers.
Where the sw is performing perfectly correctly - but it just gets wrong info in from the index sensor.
The index signal wanders, because it is fuzzy, and not consistent.
It is not crisp enough.
Not repeatable enough.

I tried 4-6 different sensors and 2-3 different hw solutions and bobs, until we worked on this with Sergey from centipede many years ago.
Scanning at 500 MHz with special hw and sw, we proved conclusively that the index sensors are fuzzy and their triggers are not consistent.
The lathe speed with a very heavy chuck aka high inertia was extremely constant, but the index sensors were triggering all over the place.
This meant that the threading algorithm was chasing the sensor reading, and resulting in uneven threads, of varying reliability.
At some speeds, D, depths, temperature and tools it might work well.
Afternoon, in use, lathe runs at 550 rpm. 18 C.
Morning, 8C, the lathe runs at 480 -510 RPM because everything is cold.
One or the other might work well, sometimes, but would not be reliable.

After he built a new crisp hw board the results improved 1000x.
A cold lathe at 480 rpm in the morning would perfectly chase a thread made at 550 rpm the previous day, with less than 0.01 mm error.
I used dull copper and a scratch pass to prove this.
0.01 mm errors are perfectly visible.

Today, my opinion is that only optical encoder indexes are accurate.
But most cheap hw boards will not read optical encoder pulses on z, /z (differential), because they are so short in duration.
It´s possible to stretch the pulse with some electronics, I never tried it.
Should work, I think.
Likewise, encoders on belt drives should work perfectly fine, imo.

After 5 years and 1000+ hours and 1000€+ I just bought the industrial controller and the industrial servo drive and built an industrial-level belt drive for the spindle.

HTH.
;)

FFS. Do we really need to tolerate Hanermo's fantasy lathe ramblings again?

Please reply in a civil manner.
You appear as a dick.

If You have factual concerns on anything I wrote, they are quite welcome.

I try to help people who may not have had access to such a wide variety of stuff, so that they can avoid some of the many mistakes I made.
If You don´t like it, You are welcome to leave and be silent.

Quote:

---

Originally Posted by m_c

Do we really need to tolerate Hanermo's fantasy lathe ramblings again?

---

Quote:

---

Originally Posted by hanermo2

Please reply in a civil manner.
You appear as a dick.

If You have factual concerns on anything I wrote, they are quite welcome.

I try to help people who may not have had access to such a wide variety of stuff, so that they can avoid some of the many mistakes I made.
If You don´t like it, You are welcome to leave and be silent.

---

I've been called worse, but at least I actually produce evidence that I have machines, not keep posting the same specs of a wonder lathe nobody has ever seen a single piece of evidence off. How long have you been posting about this wonder lathe? 10 years?
NDA still your stock excuse for the lack of photos?

Anyway moving onto your post, it's maybe factually correct, but if you had read the thread, your commentary is off little or no use to the OP, who understands the limitations of his machine, and was looking for advice regarding pulley ratios, which turns out to be a limitation of the UC100.

You seem to have built a marvel there Hanermo and clearly had significant problems to resolve along the way.

I on the other hand set out with a cheap and nasty Boxford Lathe with a view to conversion, and to learn the principles as I go.

X and Z were no hardship as I had previously converted a mill, so Mach 3 settings for those followed intuitively. Spindle speed control was not too difficult once I sussed out the workings of the lenze controller and with a cheap ebay BoB I can happily vary the spindle speed using pwm control .

It quickly became clear that the spindle sensors were not giving a great quality output but this too has now been overcome with an "off the peg" piece of kit. and now the encoder and spindle true readout work well. I sincerely thank forum members here for holding my hand and getting me to the stage I have now reached

I can now string a few wizards together to produce simple components but I had not initially given much consideration to the spindle pulleys. This lathe motor is hardly powerful so needs gearing to give decent torque and the inability of Mach 3 ICW the UC100 controller to handle spindle pulleys and threading has turned out to be a stone wall . It seems that Mach 4 gives much more versatility but currently does little to overcome this particular problem. Lack of an up to date manual does not help. Machsupport feel they can assist but as things stand we are stuck.

So this is where I am . One day maybe I will graduate to a more grand piece of machinery like your own, but at the moment it is my "old bike" that is the basis of my problem . When I learn to ride this one I will maybe move on to an all singing and dancing version like yours