The Sequel - Oh Blimey I bought a CNC Lathe (Beaver TC 20)

[awemawson]

Overnight the paint had hardened sufficiently to flat down the blemishes and when it was delivered spray the inside and undersides with the RAL9001 creamy colour, which seems to be quite a good match to the original - just looks cleaner !

Then despite the replacement proximity switches having arrived from Germany I earned some brownie points taking the wife to Rye for a coffee and look at a water sports place she wanted to investigate for grand child entertainment.

By the time we returned the paint had dried sufficiently for me to turn the bits over and spray the tops and outsides (Who said I'd calculated that    ) and before supper I was able to re-assemble the replacement proximity switch into the four way block and test all four ways.

I'll give the tinwork one more coat on the outsides tomorrow as this paint is going on at only 1.5 thou per coat according to my film thickness meter.

This version of the switch has a nifty tell tale yellow LED inbuilt - if the other had they don't work any more ! One feature that I hadn't appreciated was that the mounting screw goes co-axially through a jacking screw, used to align the front faces of the individual elements to the housing.

Then it was a case re-fitting the four way block (I set the gap to 0.5 mm as the proximity sensors range is 0.0 to 1.1 mm) and of re-threading the cables up the ducting, re-making the connections and giving it a soak test changing tools and moving about. At the time of typing it's been running 20 minutes, so early days - fingers crossed.

[JerryNotts]

Andrew,
What film thickness meter are you using. I presume you are  using wet paint, not powder coat.
Jerry

[awemawson]

Jerry, it's a rather nice device made by Sheen that I must have bought yonks ago. I'm ashamed to say that when I dug it out of the cupboard the 9 volt battery was dated 2004 - I always scratch a date on batteries in equipment when I install them - fortunately this one hadn't burst and actually was still measuring 8.5 volts, but obviously I changed it.

Yes it's a wet paint system. I gave the items their final outside top coat this morning out in the sunshine as it was a relatively still day.

Last night I ran that soak test spinning the spindle, moving the axis's and tool turret revolving for over an hour and it ran faultlessly. However my Infra Red thermometer told me the the heatsink on the Field Coil Drive card Thyristor was at 125 deg Centigrade  Now that's the maximum permitted JUNCTION temperature for the device which must have been far higher in fact.

So this morning I put my trusty AVO 8 in series with the field windings and they were drawing 4.5 amps - the card is only rated up to 3.5 amps apparently so I tweaked it down to a conservative 2 amps during testing, and it now is running at a slightly more civilised 60 degrees - still really too hot as, as it is mounted it is directly below an electrolytic capacitor which is being slowly cooked. Motor still seemed to work happily but of course I'm putting no load on the spindle.

Then it all went wrong 

I'd been doing manual turret changes, foolishly stopped it part way through one, pressed the reset button, and it went into eStop with an error saying that the PLC program was not running. Somehow I'd corrupted the program.

So I then cleared down and reloaded the controller - it takes about an hour at 9600 baud, and all was right again. However when I went to initialise the Tool Turret it wasn't reading the 4 way proximity bit for Tool 8 where is happened to be. Squinting in at the gap it looked bigger than the 0.5 mm I'd set it to yesterday on Tool 1 - sliding in a feeler gauge brought up the reading, so I reckon that the drum with the projecting pegs is very slightly eccentric, Tool 8 being almost directly opposite Tool 1 where I'd set it to 0.5 mm. So I re-set the gap at Tool 8 to 0.25 mm and hope it doesn't scrape the face off the sensors when it turns round ! (it doesn't)

Now I need to re-write my diddy as it was wiped out in the re-load !

Got to worm, fly treat, ear tag and foot trim the sheep first though !

[vtsteam]

Now I need to re-write my diddy as it was wiped out in the re-load !

Wha? 


Must be English or some other foreign language.

[awemawson]

A diddy program Steve is just an off the cuff small bit of code created usually at the console to test some function or feature - in this case the Tool Turret tool change reliability.

Written in Sinumerik's dialect of G Code, which in most cases is pretty standard. This one was only a dozen lines or so. But even so I should have saved it for future use by porting it out to a PC on my network . . . . but I didn't !

[vtsteam]

ah, a test loop 

[mc]

I'm trying to understand what bits of the turret actually slide/turn. Looking at your second encoder wheel picture in reply/post 240, I'm confused.

I'm assuming it's the hirth coupling showing, but what I can't figure out is what's attached to what, and what moves.
It looks like the encoder wheel is attached to the nearest side of the hirth coupling, but assuming that's the tail end of the main turret/tool disc shaft, that means it's going the wrong way to lock...

[awemawson]

MC there are two Hirth couplings. One with half on the body of the turret, and half fixed to the rear of the tool disk. The second has half connected to the central shaft and half connected to the belt drive system. They are arranged so the when one is engaged the other is disengaged. So the hydraulic cylinder pushes the tool disk and shaft to the left freeing the disk to rotate and coupling it to the servo motor drive. A combination of the SMCC card, the 820T controller and the Simodrive servo cards rotate the turret to the demanded tool and the hydraulic cylinder then locks the front Hirth and disengages the rear one, leaving the servo motor free to drive the powered tooling.

What differs from the patent application is that the encoder has been removed from the rear of the main shaft and put on the rear of the servo motor, and the addition of the four proximity switches that count the tool position in binary. In the first version the encoder could both count the disk and provide servo feedback as it was fixed effectively to the main shaft and hence tool disk. Once it was moved to the rear of the servo motor it could no longer keep track of tool positions, as at times it would be disconnected and not keep a fixed relationship.

[mc]

It makes sense now.

I've just had a read through of the patent, and hadn't realised the first diagram was for a basic turret without live tooling.

One final question, is the turret is also held locked by hydraulic pressure?
The patent mentions springs, but it would need a fair bit of force to keep things locked under load..

[awemawson]

Yes Moray, item #17 is a co-axial hydraulic cylinder that is double acting, operated by a two way spool valve.

I've come to the conclusion that there must be some setting up proceedure to get the front half of the rear Curvic Coupling synchronized - it carries the notch that the Index proximity switch detects. So that the rear Curvic mates smoothly when the front one disengages their angular relationship must be controlled, or when they come together either in the worst case it could be 'tip to tip' or if not so much out of place then a tendency to rotate a bit to engage.

[JerryNotts]

Its important to not become too hung-up on what you might see in patents. For many years I worked in a (British) company that patented almost everything that was thought of in its extensive laboratories. Part of the philosophy in that company and other R&D based companies was to issue so many patents for similar things that competitors found it difficult to work out which patents might apply to the item in front of them. Of course every patent could be backed-up by records showing how, if callenged, the patent was arrived at.
Jerry 

[awemawson]

I agree Jerry, but in this case nowhere else are there any drawings or descriptions of how the thing worked, and I've talked to three people who worked for Beaver maintaining these turrets. Even then drawings weren't available and they just had to muddle through and do what they could.

I think the patent description and drawings in this case are a fair representation of a slightly earlier, but VERY similar turret to mine, from which mine evolved. They have clarified quite a few details in my mind.

After a very short production run, as I understand it, they decided to buy in Baruffaldi turrets rather than make their own. This ties in with a history of Beaver's demise that I've read, where about this time they started outsourcing parts due to 'bean counter' pressure, as the Banks were calling in the overdraft.

[awemawson]

I spent this morning altering a Post Processor for FeatureCAM to generate code suitable for the lathe. Not there yet but distinctly 'on the way'. I also spent some time investigating the commands for the Powered Tooling discovering little quirks like:

M24 S100 <LF>  (M24 means start tool spindle, S gives it the speed to run at, <LF> is 'line feed')

Totally fails to spin the tool drive at 1000 rpm, and it creeps round VERY slowly

However:

M24 <LF>
S1000 <LF>

Works a treat  No documentation on this, and these little things take AGES to find !

Now in retrospect the very slow rotation is because it still holds the speed command value from the last tool change, where the servo turns VERY slowly . . . but if you don't know . . . .

Then I went to more physical things for a bit of sanity. The door on the tail stock end that gives access to the tails stock handle has obviously had an 'issue' at some time, in that the tin work around it's opening was a bit bent, and the original door is missing. However someone has started to make a replacement, bent up from galvanised plate, with welded corners.

It's a bit of a rattly fit, the opening for the catch has been cut rather roughly, and it's been left somewhere damp, and not been painted. So the welds were heavily rusted. I attacked it with a rotary wire brush and sanding disk and it's slightly better now. The galvanising had gone all white and furry! I can't paint it until the piano hinge that I've ordered arrives - it'll need packing out side to side involving drilling holes, and that's best done before painting.

The catch - a Push and Turn from 'Southco, Lester, PA' must have suffered in the original accident, as it's body was badly squashed and a bit broken off one corner. They are still available, but it looks like this size (35 mm x 78 mm) has been discontinued. 21 x 45, 40 x 85, and 57 x 141 no problem - but not this one 

A bit of judicious bending got most of the squash out, but this is thin Mazac so is just waiting to break - it'll have to do !

[awemawson]

I decided to have another go at the cut out for the catch on the Tail stock Door. Remembering how well lead free solder works with galvanising I soldered up the 'over cuts' where he'd been a bit reckless with his cutting disk, then flatted it down with a sanding disk.

Now at least the remaining blemishes are hidden by the rim of the catch

[awemawson]

Well I'm CHUFFED - though I say it myself 

This evening I've managed to draw a simple cone in FeatureCAM, post process it with my tweaked and adjusted post processor, upload it to the Sinumerik 820T controller using DNC4U, AND have the courage to run it full wack no restraints 

Now to be honest first time I'd nobbled the feed rates and spindle speed to very low values and had my hand on the eStop button (as you do !), but then once proved I let it rip, and both the machine and I survived

Being 'constant surface speed' turning and a steep sided cone the chuck didn't half spin at a rate of knots at the smaller end of the cone.

The post processor code still needs breathing on and tidying up, but it's definitely on course.

I still need to master work offsets and tool offsets before I can actually make chips, but before I do that I need to invest in some 25 mm shank index-able tooling as most of my existing stuff is 20 mm.

Now having a 'rear' tool turret and some VDI40 holders that jack the shank to the top of the slot and others that press it down, I need to standardise and make a few decisions. It affects the handedness of the tools and the direction that you spin the chuck , CW or CCW. I need to find a quiet hour to get my head around the variables without the usual interruptions of life.

[PK]

AND have the courage to run it full wack no restraints 

We have a saying at work; There's no avoiding the "Push the red button and trust in Jesus" moment.

[Pete.]

Well I'm CHUFFED - though I say it myself 

This evening I've managed to draw a simple cone in FeatureCAM, post process it with my tweaked and adjusted post processor, upload it to the Sinumerik 820T controller using DNC4U, AND have the courage to run it full wack no restraints 

Now to be honest first time I'd nobbled the feed rates and spindle speed to very low values and had my hand on the eStop button (as you do !), but then once proved I let it rip, and both the machine and I survived

Being 'constant surface speed' turning and a steep sided cone the chuck didn't half spin at a rate of knots at the smaller end of the cone.

The post processor code still needs breathing on and tidying up, but it's definitely on course.

I still need to master work offsets and tool offsets before I can actually make chips, but before I do that I need to invest in some 25 mm shank index-able tooling as most of my existing stuff is 20 mm.

Now having a 'rear' tool turret and some VDI40 holders that jack the shank to the top of the slot and others that press it down, I need to standardise and make a few decisions. It affects the handedness of the tools and the direction that you spin the chuck , CW or CCW. I need to find a quiet hour to get my head around the variables without the usual interruptions of life.

Now that's taking the bull by the horns!

So, is the turret fully operational now then Andrew?

[awemawson]

ish !

I don't fully trust it due to previous problems but at the moment it's doing what it's supposed to.

Just by co-incidence (not really !) I have a former ex Beaver service engineer check into one of our holiday cottages this evening with his wife and dog. I really don't want to make this a bus mans holiday for him but no doubt I will be making a full down load of his brain over the next few days ! (any one got any truth drugs !)

This is a nice chap I met first when I bought my Beaver Partsmaster, and the carriers mangled the ball screw by not using the shipping clamps. I'd lost touch with him, but by co-incidence he needed accommodation for a job he was doing locally this week.

[WeldingRod]

Ok, I've gotta ask... my sole reference for chuffed-ness come from the British Baking show.  My kids are asking; what is chuffed, and why should someone else need to say.it about you?
I'm might be asking too ;-)

Sent from my SAMSUNG-SM-G891A using Tapatalk

[Pete W.]

Hi there, Andrew,

I've been following this project and finding it interesting.

I hope that his dog and yours will get along OK.  Shall you have to teach his dog how to behave in proximity to sheep?

I had a 'switch it on and it works' moment here yesterday.  Very QRP by comparison with your project - a mere 350 milliamps at 12 Volts!  I won't go into detail here but will try to recount some of my activity in a new thread.  Still, there was that grateful and relieved release of breath that I'm sure you've experienced too!!! 

[awemawson]

Ok, I've gotta ask... my sole reference for chuffed-ness come from the British Baking show.  My kids are asking; what is chuffed, and why should someone else need to say.it about you?
I'm might be asking too ;-)

Sent from my SAMSUNG-SM-G891A using Tapatalk

"pleased, happy," c.1860, British dialect, from obsolete chuff "swollen with fat" (1520s). A second British dialectal chuff has an opposite meaning, "displeased, gruff" (1832), from chuff "rude fellow," or, as Johnson has it, "a coarse, fat-headed, blunt clown" (mid-15c.), of unknown origin.

[awemawson]

Pete the dogs seem fine together  Remember we frequently have people with dogs staying in the cottages, in fact the other one has two at the moment, so three extra dogs on site !

Yes switch on moments and trying new operations are definitely stomach clenching moments. Trouble is that this machine has so much power behind all it's movements that it won't take prisoners if you make an error, and metal will be bent!

This morning I've being getting my head round work offsets (G54 etc) as implemented on this machine, and trying to get things set up so things I create in FeatureCAM are all located within the 'safe zone' where spinning chucks and tool turrets DON'T get into intimate contact with each other!

Believe me a 10" chuck spinning at 3500 rpm with 26 kW behind it has an awful lot of stored energy just waiting to leap out and catch you unawares, it also kicks up quite a wind with the jaws acting as fan blades.

The hinge for the Tail Stock access door arrived this morning, but it is really too flimsy for this application. I've ordered a far heavier duty one (2 mm thick)  from RS Components that should arrive tomorrow morning. This one has the advantage also that it is un-drilled, so I can pick up and reuse the existing tapped holes in the door frame.

[DICKEYBIRD]

This morning I've being getting my head round work offsets (G54 etc) as implemented on this machine...

When you get work offsets working well, could you please take a few extra minutes to describe what you did & how you did it?  I read every word in all your threads to trying to learn how to do things right but work offsets coincidentally are especially important to me right now.  I've always muddled my way along in Mach3 Turn & never learned how to use work offsets.  I just programmed everything in an (what I called) absolute manner.  It worked & many parts were made successfully but tool changes were difficult.  After finally getting a decent controller & software now, I'm starting over & trying to learn how to do things correctly.

[awemawson]

Milton,

When my lathe seeks it's reference points, it finds them at X= 530, Z=660 (millimetres of course) so it then sets up it's co-ordinate system with those positions having those values.

Now on this lathe it uses it's zero reference as the face of the chuck - (the Traub used the chuck mounting face ie the back of the chuck). So if you use a CAD program to produce your G code, which will conventionally use the outermost end of the stock (away from the chuck) as it's Zero Point for Z, then you need some method of 'shifting the zero' . (For a lathe it is really only the Z axis you are worried about until you come to tool length compensation but let's ignore that at the moment)

So this is where G54 'work offset' comes in. There is a screen on the 820T controller under 'settings' that lets me pre-set an X and Z value for the G54 offset command (also for G55,G56 & G57) - I have loaded mine with a Z value of 300 at the moment to keep things well away from the chuck as I am testing.

OK back to the plot - when the controller gets a Z address to move to, it adds it to the Z value stored in the currently active work offset command (Also the tool length compensation value) so if I tell the machine to "G00  Z10.00" it actually moves to Z300 +10 = Z310

By issuing a G53 you can cancel any offsets that are in effect.

Now today I've been using both those commands to ensure that the FeatureCAM post processor that I am tweaking always moves the tool turret to a safe place before a tool change. It has a place that you can pre-set it before running the PP, but then the value it assigns will be massaged by the G54 that is in effect and move to the wrong place.

My solution is to have a little macro always available in the controller thus :

%SPF  54                                        (Sub Program L54)
G0 G53 G71 G90 G40 D0 X450 Z400 (full speed move to X450,Z400, cancel G54, Metric measurements, Absolute move,cancel cutter radius compensation, no tool offset)
G54                                                (Reactivate G54 work offset)
M17                                                (Sub-Program end

Which is called by the main program that the PP generates thus:

%MPF 8
( CUST                 PART#               )
( 8-6-2018 )
N25 G71 G90
N30 L54 P1        (Call Sub Program L54)               
N35 T3 D3 (  TOOL 3 EN_TURN_55  )
N40 @714
N45 G92 S3500
N50 G96 S170 M4
N55 M8 (MSG, ROUGH TURN TURN1 )
N60 L54 P1 (Go to safe place)

etc
etc

[DICKEYBIRD]

Thanks a lot Andrew, I believe it's finally starting to sink in.  When I finish the latest homing sequence update, your tutorial & actually keying it in at the lathe a few times should sort me out!