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

[awemawson]

To give me a means of bringing on the turret indexing 'fault' I wrote this little diddy to exercise it - the intention being to give it long and short moves in both directions. It selects tools in this sequence 1,12,2,11,3,10,4,9,5,8,6,7,6,8,5,9,4,10,3,11,2,12 and then it cycles back ad infinitum


[youtube]e5bHWOefifQ[/youtube]


Of course I left it running for ages and it never failed 

[WeldingRod]

Excel spreadsheet.  Weigh all your tools, do the combinatorial thing (12 factorial or 12 × 11 x 10 x...) for lines, 12 columns for.positions, then do a net moment calculation for each opposite pair, and sum it up.  Lowest net moment is your best choice.
Or, just take the weights and match the heaviest ones across the disc.

Sent from my SAMSUNG-SM-G891A using Tapatalk

[Pete.]

If that thing needs 40 amps to operate why should a little imbalance bother it?

[mc]

Have you noticed it always rotates clockwise slightly when locking?
Surely if it's on position, it should pull straigt in onto the coupling with no rotation?

I can't remember how you said the position is measured, but I think I'd be nudging whatever it is around so it's pulling in straight.

[awemawson]

Moray yes I think that you're on to something there. I have long thought that there MUST be some way to set up the relative rotational positions of the Index pulse and the two pairs of Curvic Couplings.

The index position is sensed by a proximity sensor looking for a notch. When the turret is initialised the SMCC card seeks the index position and homes on it - you can see the drive dog for the powered tooling rotate, then finding the position go off it and back on for accuracy.

Now it's rather difficult to see what is attached to what within the turret, but certainly at the point that it is seeking Index the front curvic is fully engaged and the rear one dis-engaged. The listing of the SMCC program that I have shows that there are 2000 encoder points between tool positions so when it is asked to change tools it seeks Tn x 2000 pulses.

But what sets the relative rotational positions of the  Index to the other bits I am not clear. The Index proximity sensor is in a tapping in the turret casing so not rotationally adjustable.

This morning with the machine cold it resolutely refused  to run the Turret Exerciser program without failing. It was OK on anti-clockwise moves but failed to seat the curvic on clockwise moves.

There is a setting ('Offset') on the servo card that so far I have failed to adjust by the book - it requires that certain wires be removed and links made, but the terminal numbers in the book don't all correspond to the printed numbers on the card. Basically you enable the drive, and also it's 'inch' mode and apply 0v to the analogue input, and tweak the multi-turn offset pot until the servo motor stops turning. Now just using the system rather than links, for 0v input the servo motor IS stationary.

However this morning with the program failing every time, I thought - what the heck - tweak it. I gave it 5 turns anti-clockwise (I think that it's a 20 turn pot) and guess what, the program instantly started working. So somehow I need to bottom out the wiring and get the offset accurately set.

It is very difficult to see if the clockwise last minute rotation that Moray noticed is still as consistent as it was - the camera angle is such that the tool disk is moving away at an angle anyway as it seats so it looks to rotate anyway - I shot another video so I'll let you decide as it's easier to see in the video than in real life.

[youtube]MlefohURqWU[/youtube]

I've also included the Turret cross section for reference.

[awemawson]

Well that last video was pretty useless - out of focus and still from a difficult angle. Hopefully this one is somewhat better:

I get the impression that there is still a slight settling clockwise on clockwise tool changes, but not so much on anticlockwise moves 

[youtube]kmdd6ztPV3M[/youtube]

Proof of the pudding will be tomorrow morning when it's cold again I suppose

[awemawson]

Excel spreadsheet.  Weigh all your tools, do the combinatorial thing (12 factorial or 12 × 11 x 10 x...) for lines, 12 columns for.positions, then do a net moment calculation for each opposite pair, and sum it up.  Lowest net moment is your best choice.
Or, just take the weights and match the heaviest ones across the disc.

Sent from my SAMSUNG-SM-G891A using Tapatalk

Not quite so easy Rod, as the mass of the tools act from different points due to their differing shapes, which is why I went on to consider the 'analogue' approach using a mocked up disk.

[awemawson]

Proof of the pudding will be tomorrow morning when it's cold again I suppose

I'm pleased to report that it's passed the 'cold test' this morning, behaving right from the get go 

However last night I went through the various circuit diagrams and worked out the anomalies in the numbering scheme for the 'offset' adjustment, how to disable the servo brake (not mentioned in the book procedure!), and to the disable the X & Z axis so they cannot move. So if I get time this afternoon I'll see if I can perform the procedure.

[hermetic]

I agree, that little extra move to get final alignment doesnt look right, partly as you say, because the movement is unequal in the different rotations, from my simplistic, completely non cnc point of view, could there be some small amount of slop in the servo drive system, like a slightly loose pulley, or a partially sheared key, caused by the machine crashing, which the alignment system is compensating for, I suggest this because many times in my life, looking for a complex answer to a relatively simple problem has always come back to bite me in the arse, especially if I have assumed that the fault always lies in the most complex part of the system, ie, the damned electronics!
Phil

[awemawson]

Now I've roughly tweaked the 'offset' the Curvic Coupling seats nicely each time, so although I agree it could easily have been something 'a bit loose' it almost certainly was that the offset was wrong.

If you watch the last short video "Turret Exerciser Better Angle" you'll see that the servo hunts either side of set point then when on station fires the lock solenoid pushing the two parts of the curvic into mesh, and at that point there is very little (if any) relative movement implying that not only was  it at set point, but that the set point is correct.

[awemawson]

Getting so far with the Curvic Coupling seating nicely, I thought I'd push my luck, and re-install the 90 degree powered tool, which weighs literally twice the weight of the other tools.

Well, the answer is, yes, my turret exercise program still runs ok, but you can see the previous symptom of a bit of movement as the coupling goes home. What would you do with an unbalanced car wheel - you'd fix lead weights to balance it - so why not?

First need to work out how much by weighing the tools

[awemawson]

Then we need a wooden mould to pour the lead in. The lead I had lying about had been flashing from a roof, so it had some pitch on it, and the melting was a bit messy producing lots of dross, but we got there, poured the mould and it worked out bang on the right weight

Trying the machine with the turret now balanced, bingo the curvic coupling is again seating nicely, so a successful experiment

[WeldingRod]

That must be one heck of a pocket for that scale!

Sent from my SAMSUNG-SM-G891A using Tapatalk

[awemawson]

We don't do 'small' round here 

[cnc-it]

It looks like the servo is overshooting?..I've had this on the older Siemens dC drives..it looked just like this symptom. I increased the gain with the proportional gain pot until it settled down.

If I'm correct it's pot R125 (x axis) , R225 (Yaxis) , R325 (ZAxis)  Kp (n) on the AC drives. The manual says turn the gain up if you have inaccurate positioning.



 

 

[cnc-it]

Here is a picture from the Siemens DC drive manual. Not sure if it's relevant here but worth a try 


https://drive.google.com/file/d/1_fheJGFmyK5layWUKvz-G2y88A2-c6_o/view?usp=sharing

[awemawson]

Thanks old chap, but if you look 'up thread' I've already fixed this issue - it was the servo 'offset' adjustment.

[cnc-it]

No problem Andrew. It's just that I had an Hitachi Seiki lathe that used a proximity sensor to position the turret like the Beaver here but it didn't back off and re position on the notch like yours does, it just stopped when it reached a tool position so that got me confused for a while    

[JeffK]

Hi Andrew

I have been following your progress and applying some of your fixes / checks to a Beaver TC20F I bought around Xmas (fool I hear you say!!)
I have got so far but have run into a few problems. Unlike your machine mine came to life fairly easily as it had been kept in a workshop but had not
turned for a couple of years. I can now jog the x and z axis and also the spindle but have not managed to do a zero return on it yet. I am not very
PLC proficient but did have some success with various overtravel issues etc.
 
My main problem is when I start everything up I get a x axis and z axis Pulco error which will clear when reset then following that I get the following:
'EX05 - The T-word is illegal, only t-words between 100 and 1299 are valid, present tool no. is 11'.
I jog in the negative direction as per instructions in the basic handbook I have but when I switch to zero return I get a second message:
'EX30 Axis on reference trip, select 'jog mode' then move axis in the negative direction.
With the mode set to zero return I can not jog the axis (though the manual reads as though I should be able to to return to zero). If I change back to jog mode
I can jog manually but with no zero return indication.

Initially - do you have a manual which explains the Beaver specific EX codes at all? If so would you send me the code explanations as they are not in my handbook.

Have you any ideas about the zero return? Should it jog back to zero (x-530, z-660) manually and give me some indication when in zero return mode or am I missing something?

Any idea how to reset the T-word error / message?

I noted that you spent some time initially getting the program format correct for moving the turret - can you let me know what the t-word format is as I have tried all sorts of formats and just get a 'start reject' message when I press the cycle start button. I was thinking this may also be because I haven't managed to do a zero return yet....

I would very much appreciate any help or ideas to get me a bit further... I have resorted to cleaning the machine now to make forwards progress though looking at the photos of yours I have a long way to go !!! There have been a few 'maintenance additions' to the machine by the looks of things in the back but I get the feeling I am not far away from getting things working properly - just limited by my own lack of experience with these machines....

Thanks in anticipation....
cheers
Jeff

[awemawson]

Jeff - not fool at all - they are good machines.

My machine has no EX commands nor have I heard the term Pulco. Tool number have to be between 1 and 12 followed by a D term that defines which tool offset to use.

For turret operation on my machine it first has to pass over it's X and Z references (done by selecting the reference symbol on the mode switch and pressing AND HOLDING  the appropriate jog button), and then the turret subsystem has to be initialised in MDI mode by executing an M80 command, followed when it has finished (takes 10 seconds) by "Tn" where 'n' is the currently presented tool. (Commands terminated by 'LF') The control won't do anything other than jog until this procedure is completed. Now your system may well be very different . . .

The problem is that your machine is the "20F" therefore has the Fanuc control, so very few of the commands will apply at low levels. Nor do I have any information on the Fanuc equipped machines.

Beaver fitted several different turrets to these machines, and the turrets were controlled by different servo systems. Mine uses Beaver made turret, and the Siemens AC servos, but controlled by a Baldor 'SMCC' card that is a stand alone CNC two axis controller subservient to the Siemens 820T. But I suspect that your system probably uses Fanuc AC servos controlling a Baruffaldi turret that Beaver bought in as a cost cutting exercise.

I suspect that apart from the mechanics of the machine and possibly the main spindle drive little is the same between our machines.

Can I suggest that you start a new thread logging your progress, and detail what units you have by way of controls and servos. Put up loads of pictures of the items and internal of the rear cabinets.

[JeffK]

Thanks for the quick reply Andrew. From the photos you have posted your machine does look somewhat different to mine. I will do as you suggest and take
plenty of pictures tomorrow to post in a separate thread.
I think that some of the issues I am having have been due to the axes being manually moved by turning the ballscrews with the power switched off on the machine.
This is resulting in errors in the pulse encoders and possibly preventing the zero return. Not sure how to fix this just now.
Anyway I will post a new thread and hopefully people can give me pointers.
I will continue to follow your thread with interest as I too hate to see good equipment scrapped just because it needs a bit of attention and time spent on it.

cheers

Jeff

[awemawson]

Jeff, so long as your axis's are on the 'inside' of your travel limit switches and reference sensors, then winding by hand with the power off should not be an issue. After all that is why you have to pass over the reference sensors to reset the controllers measuring system.

I can put you in touch with an ex-Beaver Engineer who undoubtedly can assist, but he is extremely good at writing invoices
I should warn you 

You appreciate that Beaver went to the wall in the early 1990's so not many are about who know these machines, and as far as I can tell they only made something of the order of 250 of these lathes.

I look forward to seeing your new thread! (Can you please limit the picture size to 800 x 600 pixel to save on Eric's server space)

[JeffK]

Thanks Andrew - it would be good to have his contact details - though I think I will battle on myself just now as I am still making slow forwards progress.
I believe quite a few of the machines went for the export market so likely hard to find anyone in the know in the UK. In addition I am up north of Aberdeen so
even harder to get anyone to actually work on it....

cheers

Jeff

[RodW]

Geez, I finally gathered up the courage to venture into this thread and waded through from beginning to end. What a mission Andrew! Congratulations!

I kinda wondered part way through if it would have been possible and easier to retrofit it with LinuxCNC? Some of the hardware from Pico Systems and Mesa (which I use) might have provided a way forward without being constrained by lack of factory parts.   And I also wondered if it would have produced a superior result? (Mainly becasue one day I might try and do that to an old machine).

It kinda nice to open up Google Drive in Chrome on your machine controller so you can download the cad files you want to process....

[awemawson]

Rod,

The motivation was to get it back 'as was' and anyway there are some massively powerful servos and spindle drives on this beast that would cost a fortune to replace.

I know of a chap who has done what you propose to a Beaver TC10 the smaller younger brother of this machine, but it's not what I wanted to do.