Friday 30 December 2022

Change of plan - bummer - Kitagawa chuck

Welcome back.

Much of the Xmas break has been taken up with endless shopping excursions, overconsumption, sitting on our arses etc etc. The actual number of workshop hours has been minimal, with most of that taken up with the tool touch-off issues I found within the Centroid software. Time to change that! 

My cunning scheme for using the DIYO touch probe on the Tree sort of worked out but it's a bit fraught when it comes to turrets spinning, boring bars and LH tools all moving in different directions. One of the main reasons for taking this unconventional approach was the difficulty mounting the probe in the spindle, mainly due to the fact that my the chuck that currently fits is the Kitagawa hydraulic 3-jaw, which in turn has a very limited (~2.4mm) movement on each jaw. 

I'm not planning on moving the jaws every time I want to touch off tools, hence a system that doesn't mount on the chuck. 

There are some useful videos on the Haas website and of those, the lathe "Tip Of The Day" (TOD) videos contain some useful content for the likes of me, learning the CNC lathe ropes. Here's one that took my fancy - a stepped conical mount for a chuck-mounted lathe tool probe. This accounts for the fact that many hydraulic chucks have very limited jaw movement by incorporating many small steps in the diameter. No matter what the current jaw position, this should fit.

My chuck seems to manage about 2.4mm movement between open and closed, so steps of 2mm might work. Furthermore, I have some 70mm diameter 6082T6 in the box. 

So let's model something up for my machine. The DIYO probe has a 6mm dia shaft and I need some meat for a grub screw as well as needing to avoid any deep grooving - typically the MGT tools I am using have a reach of around 18mm. This revolved section looks workable without getting overambitious. I used the Fusion 360 parameter table to drive the step length and radius increments:

I should be able to machine the stepped conical section using a VNMG insert and groove the necked down section using a KGMN200 (2mm) grooving / parting tool. I have the polished, uncoated (H01) flavours for use with loominum.

Stock prepared, tools ready for action:

This should work:

Will the chuck be up to the job? With the acetal I used for my pawn test piece, the runout is pretty reasonable - about 20um.

I need about 70mm from the front of the stock. This leaves some safety clearance.

But hold on. The runout is more than just visible. The fucker is a major imbalance issue, or would be if the Tree didn't weigh several tonnes. Try as I might, there is no consistent way to clamp down on this stock to give an acceptable level of runout. I can tap it with a mallet so that the runout is brought to an acceptable level. But alarmingly, this doesn't require much of an impact. Put it another way, the stock is far from secure in the chuck.

I've got a centre drill mounted in the turret, so I can find the centre of the stock when it is rotating and leave a small centre hole. I can then come out in the X+ direction and leave another witness mark. But hold on - the stock has moved in the meantime. Clearly the stock is held so loosely that creating that witness has dislodged it. That doesn't feel like a sensible basis for attempting the sort of machining I had in mind. Tits.

Yes, I have to conclude that all the time and money I've expended on the Kitagawa chuck has been a complete waste. The runout is appalling (basically it's buggered and irrecoverable) and it doesn't have enough grip to safely hold the stock. Apart from that it's been a dream. The latter has always been a concern at the back of my mind but I've not managed to come up with a means of measuring the gripping force. In the meantime I've been hoping against hope that it would be fine. Well it isn't, idiot.

What now?
Well it's back to Plan A - machine up a spindle nose adaptor so that I can use my various D1-3 chucks on this A2-5 nose. Back in March, I modelled up an adaptor in Fusion. I guess it's time to resurrect that. 

The machining will have to be done on the Bantam, as clearly I have no sensible (safe!) means of holding a large (140mm dia) lump of 4140 (aka EN19T) steel in the Tree. This means coming to grips with LinuxCNC again. 

Last time round, I started machining the main body but I'd clearly got a bit carried away with the feeds and speeds, as it stalled part way through. I don't seem to have posted anything about that aborted attempt but the stock is still mounted in the 4-jaw, so is almost ready to go again once I change the CAM and/or machine setup to have another go at it.

It's rather too large a part for the Bantam to run in top ratio (2000rpm range). I'll probably have to reduce the feed rates and/or the spindle speed. In the latter case, I will probably have to change the gear ratio and rescale the VFD control voltage accordingly ie the rpm per volt it expects to see. If I understand correctly, it operates open loop speed control, so if I simply change the gear ratio I suspect it will either object to a big speed mismatch (it can see the spindle speed through the encoder if it choses to) or apply the specified feedrate, assuming the spindle speed is correct. Neither would be ideal.

Next steps:
Let's dig out the CAD, CAM, machine setup etc and see about machining up that spindle nose adaptor. No point crying over spilt milk etc.

Saturday 3 December 2022

Centroid CNC12 lathe tool touch off for boring bars - WTF?

Been struggling to understand how to get boring bar tool length offsets correctly measured in CNC12 lathe. I've been finding this all very frustrating but I think I'm getting there now. I was planning to update from 4.80 to 4.82 but TBH I can't see any suggestion from the release notes that anything has been done that would affect lathe touch off. Besides, I now have a custom PLC on account of my ATC turret, which would further complicate matters.

Instead, I've done some experimentation / learning / calculations to better understand what is going on and I seem to have made some progress.

I started out by setting Xref to zero at the point of touch off of Tool #1, which I'm considering to be my reference tool. I'm not sure why I'd do this from what I've read in the manual but it seemed sort of in line with what Uwe said just now. Subsequently, the offsets are back around the -70mm for the external tools. But I still couldn't understand the boring bar offset.

On my machine, the boring bar tip is ~26.5mm closer to the turret axis (ie in radius units) than the external tools. That's equivalent to a difference of ~53mm in diameter. My probing "tip" is actually a piece of 3/8" square stock (so that I can probe it from all sides unlike a ball tip) ie 9.53mm across flats. The section in the CNC12 lathe manual dealing with boring bar offset measurements says "multiply the [boring bar] manufacturer’s offset by negative two (-2), and type the number into the X Offset field. The value you type should appear as being added to the measured X offset already measured." This seemed to be pretty relevant and indeed my "offset problem" distance is twice the width of the probe tip.

Quite simply, if I subtract twice the width of the probe "tip"(ie a total of 19.06mm) from the offset, I end up with the "correct" offset. By way of a check, when I change tools and approach a fixed DTI, the typical final "error" in the DRO position between my external tools and the the boring bar is 19mm before this correction is implemented.

So (and yes, I'm still new to CNC12 lathe):

  • Offsets are measured as diameter units, not radius.
  • As suggested by Uwe Mattern on the forum, the Xref should be entered only once. I can't say I'm quite clear how this works yet but it's a start.
  • The auto touchoff process doesn't seem to allow for the width of the probe tip when measuring boring bar / internal tools.
  • That's likely because the process outlined in the CNC12 lathe manual tells us to manually enter the boring bar offset. This is essentially the same process used for drills and taps but with the offset entered manually.
  • The manual covering lathe tool auto touch off is "commendably brief".
  • When I auto touch off boring bar tools with my probe, I will need to subtract twice the width of my probe from the measured tool offset.

Phew. Probably quite a simple problem but it took me some head scratching to figure out!

Shall I update Centroid CNC12 from 4.80 to 4.82?

What up?

I thought I'd got the tool touchoff process sussed out last weekend but now I come back to it this weekend, the offsets being reported during tool touchoff don't make sense. I seem to be getting -140mm instead of -70mm which is suspiciously exactly doubled. I'm also flummoxed by the boring bar touchoff procedure which has simply "compounded my confoundance".

Given this strange behaviour, it sounded as if it may be sensible to update to the latest version. I'm currently running 4.80, yet there is a 4.82 out now. There could be merit in the update. Or persisting uncertainty if I don't. What to do?

What's involved?

The link to the 4.82 update package is here and the update instructions are here.

On the face of it, the process is fairly simple. Rename the existing folder from c:\cnct to something else, such as c:\cnct_4.80. Then do a clean new install (which will be automatically named 4.82 in this case) and finally, "just" copy the macros, homing, library etc files and licence file across from the old folder. 

Finally, finally, go into the setup wizard and change the settings to reflect what you had before in terms of input and output settings etc. Ideally you'd have saved screenshots showing you what those were beforehand, otherwise you will have to rename the old folder back to c:\cnct again and run it up so you can check what they need to be.

And finally, finally, finally, you need to make any changes you made to the original PLC program to the new source (SRC) file and recompile it. Otherwise, the turret controls (in my case) will have been lost. This is a job for the Visual Studio Code PLC plug-in, to highlight differences between the new and old, particularly the changes I made for the turret, in my case.


Not for now, Fatty. The hassle factor is too great given it will likely make little or no difference to the offset measurement issue.

Sunday 27 November 2022

Touch-off probe setup and testing!

Here's what I used last time, to investigate how the Centroid touch off features worked. Perhaps there is some intermediate scheme that I can use to mount the probe. Removing the upright from this mag base looks like a good starting point:

Mounting it in a tailstock chuck, then adjusting the probe position looks like a promising approach. Something like this:

Firstly, I need to remove the ball end from the probe tip, as the lump of Bluetack isn't entirely appropriate for the job. Presumably it's a 2mm ball bearing brazed(?) to the end of the probe shaft.

Indeed. It held up well against the angle grinder - until it didn't. The ball itself was as hard as witches' tits until it flew off into the dust and dirt on the floor. Job done either way. This leaves a simple 1.5mm shaft:

I can't find the key (square) stock I know I have somewhere, so finally I gave up looking and went with this 3/8" brass stock. It's a bit larger than I'd planned but I can always replace it later. Meanwhile, I was too impatient to machine it down and instead settled for squaring it off...

...and drilling a 1.5mm hole in the centre.

Yes, I only possess one drill in this size. Don't mess up, Fatty - you only get one chance.

Fits nicely, drill didn't break and a bit of Loctite is ideal for bonding it to the shaft. If I want to replace it, I can simply heat it up with the blowlamp.

And here we are. Yes, the large brass lump looks a bit silly but so what - it will do the business.

Good, so let's give it a whirl. With parameters #281 & #282 set to 9.525mm (aka 3/8") and the "safety clearance" distance reduced to 5mm, we should be good to go.

Here's what I get when I run the Z and X touchoff 3 times (telling it the tool number has changed between events). The offsets seem to vary by ~6um or so. This may be a best case result of course but at least it isn't instantly disappointing.

Let's mount a few different tools in the turret. Then I can touch them off and see what happens when I change tools with the turret and tell the machine to go to the same position in X and Z. 

I will need to issue G43 after each tool change presumably, as indexing the tool doesn't in itself activate the appropriate offset. That's something else to familiarise myself with.

Here we go:
It's early days yet but it seems you don't need a G43 for the lathe, just T0202 etc, to load tool #2 and apply offset #2.

I've now touched off in Z & X (it does them in that order) for the first 4 tools:
  • VCGT160404 LH turning tool
  • MGMN200 2mm turning / grooving tool
  • CNMG120402 LH turning tool
  • WNMG080404 LH turning tool
I then set X0Z0 with the DTI reading 2.00mm. Moving to X30 (to clear the moving tools!), then indexing the turret and setting T0202 etc then asking for X0Z0 resulted in the X and Z offsets being correctly applied and the tool tip moving to the intended X0Z0 position. 

Using the Mitutoyo DTI (10um per div), I see tools #2-#4 achieving the same X position, with tool #1 consistently ~50um off. I'm guessing I didn't touch off as accurately for this tool. However, I seem to have sussed out the process and we appear to have a reasonably consistent result.

In practice, the tool change will need to be made with a G28 (tool retract) move, to avoid liquidising the probe, workpiece etc. But for now, I'm testing it out with MDI commands.

No video, it didn't happen of course.

Saturday 26 November 2022

Hard times - machining toolholders down to size.

Now that I've bought a dozen toolholders from AliExpress, I have the task of machining them down to the required height for the Tree lathe. These are all 20mm x 20mm x 125mm sized, yet the Tree turret is designed for 3/4" (19.05mm) tooling, as it originated in the US in the 1980s. Not much demand for imperial tooling these days but the difference is only ~1mm, so I will skim them down on The Shiz.

Quick test with the hardness tester suggests they are generally hardened to something like 40-45HRC. So carbide inserts it is. 

CAM setup:
It would be entirely possible to use Centroid conversational programming to set the machine up for this but it's simple enough to do it in Fusion and I'm more familiar with this process currently.

There's only one tool (BAP300 50mm face mill) involved, so I simply need to probe the corner of the stock and check the tool length offset (Tool #25) is still correct, then off I go. 2500rpm and 0.06mm per tooth, with step down of 0.2mm. So that requires 5 passes and just under 2 minutes. This isn't good for the inserts (hardened steel, interrupted cuts etc), so no need to push it any further and I will need to deburr the parts, which will keep me busy while the next part is being machined.

Machine setup:
Firstly, before I start, how flat is the vise / parallel setup? The Renishaw probe isn't much use here, as it only indicates a threshold. The Mahr 3D DTI gives a visual (analogue?) indication.

Answer - pretty good.

How close to the alleged 20mm are these Chinesium things? Pretty reasonable.

So, off we go:

Here's the first example. Not a bad finish but the inserts are all new of course.

Measuring pretty close to the required 3/4":

So off we go again. I have 12 of these AliExpress specials plus another couple of existing tools to do....

Getting there. I deburred as we went. The inserts were clearly suffering, as I found they became hotter as I progressed through them. I was machining them dry to keep the mess down and not thermally shock the inserts so I ended up wearing gloves.

This Teknik trigon holder from Cutwel has through coolant, which complicates matters. As I have a couple of trigon holders already machined, I'll leave this one for now. Apart from the various plugs that would get in the way, the coolant coupling would foul the turret on the Tree, so it's questionable I'd be able to use it there anyway.

How was it, Fatty?
Bottoms up:

Right way up:

Good. That wasn't too bad and I didn't start any fires in the workshop.

Any damage?
Sure enough, 4 out of 5 of the inserts are chipped. Seems a reasonable price to pay for the work, not least as these are genuine Mitsubishi parts from AliExpress that didn't cost an arm and 10 legs. Here's one of them:

Good. Now let's think about setting up the tool touchoff system on the Tree. For now I think I'll simply mount the DIYO probe in the tailstock rather than dick about making up a pukka assembly. I've not been able to make a lot of progress recently for various reasons, so I'll cut a few corners for the time being in order to see some swarf sooner than later.

Sunday 20 November 2022

Understanding the tool geometry offset library for Acorn Lathe - and testing it out with the DIYO probe

Wahaay. Relieved but not surprised to see the Centroid team have thought this through and implemented the various variants such as:

  • Internal / external tools (determines which X direction to probe in)
  • Left handed / right handed tools (determines direction of cut ie toward / away from headstock etc)
  • Drills, taps etc
  • Direction of spindle rotation (in my case, CW for external tools and CCW for internal tools, which is the reverse of "normal" (ie manual) practice.
How is it all set up?

See Chapter 5 "Tool Setup" of the Centroid Lathe Operator Manual:

Setting up auto tool touch-off parameters in CNC12 Lathe:
With the "Pro" lathe licence, the tool probing is included although Swissi hasn't created a special "probe app" for the lathe (yet?). The setup is pretty simple and is easiest taken care of by the wizard. There aren't many things to set up. Once you have told CNC12 which input is the Probe Detect signal, the only parameters you can / need to set up are the stylus widths in X and Z and the "safety clearance" distance. That's only 3 parameters:

In my case, I plan to have a 6mm x 6mm square probe tip, so #281 ans #282 will be 6.0. I set the clearance to 10mm but that seems rather girly, so I may reduce that in the near future.

Running auto touch-off:

Fairly simple when you know how:

For touching off in X:

For touching off in Z:

For touch-off in both Z and X:

Uwe Mattern posted a YT video recently showing his setup with the DIYO probe and Acorn (Acorn 6 in his case, FWIW):

Mine pretty much does the same although I can't mount my probe on the spindle axis due to the constraints of the axis movements.

I have a spare input on Acorn, so got this set up and did some playing, using a bit of Bluetack on the probe tip ('cos it isn't designed for tool touchoff).

No video? It didn't happen!

Here's the most commonplace touchoff move ie a RH external tool. I know, I know - in my case, technically it's actually going to be a LH tool but let's keep life simple.

By flipping the "nose vector", I can change the probe direction in Z. This is what I'd do to change from a LH to a RH tool:

There. This gives me the confidence to know that my slightly unconventional machine setup should work fine with the Acorn software.

Tool setter assembly for the Tree lathe

What up?

I need to be able to probe both internal and external tools, both left hand and right hand. Most of the probes use the "trilobe" concept, where the probe tip sits on the end of a sort of tripod scheme, with the 3 legs held against electrical contacts by a spring. It's what I have on the Renishaw MP1S and TSR27 probes used on The Shiz to good effect. This principle works fine for mill probing which only occur in 5 axis (+/- Z, +/-Y and -Z directions) but if you try to "pull" the tip away from the probe body in the Z direction, bad things would happen.

In an idle moment (couple of hours), I modelled the DIYO probe in Fusion 360 and uploaded it onto Grabcad. Looks good but the purpose was to be able to use it to design an assembly, taking into account the rather limited movement available on the key components that make up the cross slide, turret, tailstock etc.

In simple terms, the external toolholder and the internal (boring bar etc) toolholders don't present their tips at quite the same radial position. True, most of the external toolholders have a well defined length (125mm from the back the tool holder to the tip) and height (0.75" from the base of the toolholder to the cutting tip) but the radial position of the boring bar tip will depend on the diameter of the boring bar and the length of the tip. 

The centre line of the boring bar holder is already defined by the boring bar holder block and is about 25mm further from the turret axis than the external toolholder tip - and the larger boring bars may have a tip-to-centreline distance of as much as 17mm or so. That gives a range of tip positions of getting on for 50mm and by the time you add some distances for actual probing moves plus the thickness of the probe head itself, I'm seeing that I will need to lose a lot of the available cross slide (X axis) movement. 

Putting it another way, the position of the probe tip will have to be calculated fairly carefully to ensure it can probe the variety of tools I expect to use.

NB: On this machine, the probing move needs to be in different directions for the internal and external tools. The cross slide can only move away from the centre line (away from the operator), which means that the boring bars will be presented "upside down" with the spindle rotating clockwise. In contrast, the external tooling has to be "left handed", presented "right way up", with the spindle rotating counter clockwise.

This probe assembly is going to be hazardously close to the hydraulically powered turret, so a turret indexing move could easily destroy it. So my cunning plan is to mount the probe on a swing arm, with the tip pointing down. That way, if the turret indexes when the probe is in position, it stands a chance of pushing the probe out of the way rather than trashing it.

Obvs the first thing to do was to model up the key elements. The cross slide sliding limits are set to restrain the turret movement to its actual values - 245mm maximum centre distance (at X axis home position). The tailstock can also move along the Z axis and the upper swing arm is able to pivot down and out of the way. The construction details for the swing arm are to be determined but the dimensions and operating principle should be reasonably clear

Here's my solution. 

With the boring bar fitted, there is still a modest clearance between the boring bar block and the tailstock. I suspect I may want to fit a limit switch to prevent them clashing, as the cross slide is able to move the tool tip slightly beyond the headstock / tailstock axis:

With the external tool fitted, the cross slide is slightly further away from the tailstock but still has some movement left between this position and the home position. The probing move is in the opposite direction, clearly. I believe the Centroid CNC12 is clever enough to cater for this, using the info in the tool table - assuming you have correctly populated it that is.....

This looks workable, although I may be advised to test out the probing moves before getting too far into the build. The "build" itself is actually fairly simple, comprising little more than the swing arm assembly, now that I have done the calculations and figured out a scheme that appears to actually work.

Sunday 30 October 2022

More commissioning tasks

First, let's finish connecting up the monitor. The bracket seems to be just about up to the job now that I've fitted an additional brace to make it more rigidly attached to the enclosure. And now I have 3m long cables for the USB, HDMI and mains, they actually reach to the controls.

The mains needed a gland to bring it into the electrical cabinet. Not much point wiring all this stuff up and requiring the monitor to have its own socket.

Now to connect up the 110V isolation transformer. It doesn't need to be isolated really, nor does it need to be rated to 3300kVA but this is the only one I have and I'm not about to buy another, smaller one. I need to rewire the 240V side first. Note that these devices feature a "centre tap earth connection", so the output is actually 55VAC either side of protective earth on each output connection, giving 110VAC total.

There's no simple means of fixing this to the chassis, as it's intended to be a portable site transformer. However, I have no pride, so one of these strap things will do the business. Last thing I want is somebody (most likely The Stupid Fat Guy) catching it in passing and it ripping the cables out of the cabinet.

God, it's becoming a bit messy now but the end is in sight from an electrical installation POV.

That's most of yesterday's task list done now - or at least the ones that actually need to be done. The remainder will follow on when I finally have the thing running again.
  • Wiring up and plumbing up the coolant pump and its associated reservoir tank.
  • Mounting and connecting up the monitor and mouse/kbd.
  • Installing the 110V transformer (required for the hydraulic solenoids).
  • Connecting up the internal (IP66) lights.
  • Making up some form of tool touch-off probe for tool length / offset etc measurement.
  • Sealing the headstock / bed covers to the enclosure.
  • Cleaning up (and polishing?) the front and tailstock windows.
  • Checking out / replacing the Meanwell PSUs that seem to be tripping the RCD. Or wiring up a 240V isolation transformer on the main incomer.
  • Machining several toolholders down from 20mm to 3/4" height so that the tool tip is at centre height (part of this machine's imperial legacy).
I don't have an isolation transformer in the incoming mains feed, so I'm expecting to have to either replace the Meanwell PSUs or fit a transformer if I'm to avoid tripping the RCD every time I try to turn it on. Unless the Meanwells have magically fixed themselves.

Let's have a go and see what happens.....

D1-3 adaptor - dimensional cockup!

Trial fitting of the adaptor body to the Tree spindle nose reveals a problemette - the adaptor bottoms out on the taper nose before it seats...