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

[seadog]

Well done that man 

[Pete.]

RTFM failure

Well done Andrew!

[PekkaNF]

Very good job!

Interestin questions is why the servo amlifier had wrong parameter to start with?

Another interesting thing is what happens when you add inertia (large piece on the chuck). Will it overshoot more? In theorio it should not, because of the closed loop, but this "creep mode" sounds a little sensitive.

Pekka

[awemawson]

Thanks Pekka,

Yes very odd that just that one parameter on the Mentor was changed, and to do so you have to go through a procedure of setting other parameters to special codes. I very much doubt that the EEPROM had a one bit hissy fit so VERY puzzling.

Servo action seems pretty powerful now that the gain is back up  to reasonable figures - I've tried stopping it by hand as it's on it's way to set point, and it isn't possible (and I have big hands!!!)

But . . . life is full of mysteries . . .  such as what's happened to poor old PK, who seems to have vanished 

[cnc-it]

Perhaps the machine was supplied without the c axis option..ie. the parameter turned off in the Mentor . If the customer requested the c axis it would be an easy fix.. Beaver charges £££ to simply alter a parameter..I know I'm being cynical but this type of thing does happen especially on the Japanese machine tools!!

[awemawson]

Ya-but. . . . .

I KNOW that it worked when at Portsmouth University as the technician who swiped the probe and (bless him) had a copy of the parameters, sent me a dump of his back up including his programs. Some of these were for Lambretta crank flywheel balancing arbors that he was flogging on the side, and used the M19 spindle positioning to mill certain features 

Sadly it's not a true 'C' axis in that it can just go to a position and lock on and brace itself for machining at that angular position, where as the Traub had a 'proper' C axis that could interpolate so you could machine as it moved, co-ordinating C, X, and Z.

I've been in touch with Mawdsley BER today, (the firm that grew from the ashes of Mawdsley Dursley that made the 27 kW spindle motor) as I would like to finally resolve what the field excitation SHOULD be. Apparently they do still have some records from that time and may be able to pull up a specification of it's parameters. They are searching their archive.

[awemawson]

Well today is ANOTHER official GOOD DAY  

Mawdsley-BER have dug through the archives of the original Mawdsley Dursley company that folded in the mid 1990's and found the original test documentation FOR THIS VERY MOTOR. Not one like it, or a generic specification, but the actual motor. Now I'm amazed that  it's survived, and also very pleased as  I now can definitely resolve the field coil current  conundrum based on facts rather than guesses.

Copy attached

[nrml]

Now that is what I call excellent after sales service. Nice to see that they still take the time and effort to support their products after such a long time.

[cnc-it]

Just out of interest Andrew..is it possible to upgrade the C axis to full 4th axis or does that require replacing major components like the Mentor drive..?

[awemawson]

Well the spindle encoder is there (1024 lines) and the DC Drive would do it, but the firmware embodied in the controller isn't set up for it as far as I can tell.

Reading the Siemens manual for the 820T there seems no reason it couldn't have been a full on 'C' axis but that's not the way Beaver implemented it.

[cnc-it]

I see makes sense..maybe it was an option back in the day.

[awemawson]

Having got the Spindle Orientation now sorted the next barrier to knock down is getting the powered tooling working, after all that is why I need the Spindle Orientation in the first place.

But first I wanted to investigate a minor irritation in tool changing, where having issued a tool change command the tool disk would come forward unlocking it from its Curvic Coupling, and then pause for what seemed ages before rotating to the next tool position and re-clamping back on the curvic. now the PLC program, the code running in the SMCC turret control card, and of course the mechanics of the turret could all be possible culprits. I feared that the code in the SMCC card had possibly been corrupted. Having 'unclamped', the 'clamped' feedback switch opens and the 'unclamped' switch should close when the tool disk is far enough forwards for the Curvic to have disengaged, which triggers the SMCC card to using it's servo logic to rotate the turret to the next tool. Putting two channels of my 'scope on the 'Clamped' and the 'Unclamped' switches showed that despite the tool disk having travelled forwards it took a few seconds before the 'Unclamped' switch closed where upon the tool disk rotated as it should. This has eliminated the SMCC or PLC code as being an issue and either the switch needs adjustment, or maybe it's sticky.

I wrote a simple diddy to continuously change tools, and as it ran the delay reduced until it vanished, so either a sticky switch has now freed, or maybe warmed up. I'll have to let everything cool over night to see if it is still an issue. I had assumed that the two switches in question were proximity switches, but I don't now think that they are. The Patent Application refers to them as 'precision mechanical switches' - they are two wire, cylindrical and threaded like a proximity switch, but measure zero and infinity when closed and open under no applied power. In the circuit they are just in series with an opto coupler and current limiting resistor across 24 volts.

So on to the powered tooling. There are two issues:

A/ The Baruffaldi Powered tools have a different dog clutch shape than the machine drive spindle

B/ When a Powered tool is mounted, the faces of the two dog clutches are separated by 15 mm so wont drive anyway.

I had expected to have to dismantle the bearing housing for the drive spindle to remove the dog clutch and spindle, and make a complete new assembly, but it turns out that the machine dog clutch is just keyed to it's shaft and retained by a countersunk hex bolt, and I was able to withdraw it though an empty tool position.

Now the new dog clutch needs obviously to engage with the Baruffaldi ones, and be 15 mm longer (plus engagement depth). It would be a nice exercise to make on this lathe if the powered tooling was working, but that's a rather circular argument 

Have some pictures 

[awemawson]

So embryo replacement drive dog drawn up in Fusion 360 and now printing on the Cetus 3D printer. This is just to let me better visualise what I need before I start machining - be nice if a 3D printed part could just be used as is !

[awemawson]

So the 3D Printing has finished, and I've been able to use the model to measure how long it needs to be to engage, and yet have clearance when the tool disk rotates.

At the moment the drive dog is installed with it's shaft key but no retaining screw as I need to find a longer one. Amazingly it works. I suspect I could even do a  bit of light milling with it, but probably won't.

Next task is to work out how to make a metal one. The teeth have a 12.5 degree taper when in lies the machining problem.

[AdeV]

Next task is to work out how to make a metal one. The teeth have a 12.5 degree taper when in lies the machining problem.

Two options spring to mind:

1) Tapered end-mills.... if you can get one with the right taper.
2) Do it on a manual machine with a rotary table, and have the head tipped over by the correct angle.

3) Mill the bulk of it with straight sides, then use a die eroder to make the tapers (if they can do that sort of thing?)

I'll stop there before someone does a Spanish Inquisition joke...

[awemawson]

I'm tempted to make a 'D-Bit' to a 12.5 degree taper. The real problem is that the 'root' of the gash at the inner (narrow) end is only 2 mm wide, so any cutter will be fairly delicate.

The alternative but slightly more complicated way is to make a tapered slitting saw which would be more robust.

As a compromise I could make the protrusions slightly shorter radially thus increasing the gash root width.

. . . all suggestions welcome 

. . . of course a 3D printer that prints in Bronze would work nicely 

[nrml]

Can't you cast one in bronze or brass from a 3d print and do the fettle to final fit on manual machines and hand tools?

[WeldingRod]

Can you post a picture of the mating end?  I really wonder how it was made!
You could clearly 3d metal print them, probably under $100 each.  Ideally key them to a stub shaft to save print volume and $$

Sent from my SAMSUNG-SM-G891A using Tapatalk

[AdeV]

I really wonder how it was made!

I'd bet they're die cast. Not sure if they're machined afterwards, it would seem unnecessary, as the actual part-to-part wear would be minimal, since the parts aren't moving against each other once engaged; and having a super-precision fit isn't necessary.

[chipenter]

Tilt the rotab over 12.5 degrees and use a slitting saw  .

[Pete.]

Surely you could wire EDM the sides but I don't see how you could do the bottom.

[cnc-it]

I wonder if Baruffaldi has a part you could modify with the teeth already machined..it's a curvic coupling..these are hardened and ground to micron tolerance usually..larger ones are used on 4th axis horizontal machine centers to locate the pallets. Any slop would cause wear over time hence the need for a precision fit?

[awemawson]

I'm sure that the original Baruffaldi drive dog was steel, and machined. But of course I have never seen one, as my lathe was fitted with the dog to fit 'hens teeth rare' Beaver powered tools. So my 3D model is entirely guesswork but must be close. The dog shaft actually slides and is spring loaded into contact.

I've simplified and re-drawn the model this morning - hopefully this view will give a better idea of the tooth shape. I rather like the idea of lost wax casting it in bronze.

However it does occur to me that a plastic 3D printed one actually forms a useful safety 'mechanical fuse' in the case of jam ups, and it's dead easy to make another and re-fit it.

. . . I'm off now researching wax 3D printer filaments . . . .

[mattinker]

Andrew,

how about a three D printed core and cast in bronze using  a fine petrobond sand mold?

Regards, Matthew

[mattinker]

Me again! You could print on your 3D printer two or three piece mold to cast a wax version of your part?

A passing thought, cheers, Matthew