Experiment Engraving Machine Dials

[awemawson]

Bill it's a long story but covered in depth here:

http://madmodder.net/index.php/topic,9439.0.html


The axis is travelling too far - the only possibilities that I see are:

a/ gear ratios wrong (seems very unlikely)

b/ obscured 'gap' in encoder disk so loosing one or more pulses per rev.

I was about to delve into that thread on building the 4th axis to refer back to the ratios to see if b/ is mathematically likely 

[awemawson]

OK so the encoder is 125 counts per rev
It is tooth belt driven by the servo motor 20:25 reducing the count to 100 per servo motor rev
The servo motor is geared to the worm gear 4:1 in a low backlash set up
The worm to the worm wheel is 90:1 - again low back lash set up

so 90 x 4 x 100 is 36,000 encoder pulses per axis rev

Now to be out 1/2% seems very strange unless the worm / wormwheel isn't exactly 90:1 and I can't believe Nikken would make something with an odd ratio - it was originally a stepper motor driven rotary axis with the same 4:1 gearing

0.5% (approx) is 180 steps being lost if that's the cause 

... I'm befuddled - none of this makes sense 


Short of pulling it apart and physically checking each ratio in the drive chain accurately I don't see a way forwards 

(I re-did the test and it is absolutely consistent, so unlikely to be a loose coupling or something like that - I also, having run forwards 3 revs to 1080 degrees, ran back to -10, then forwards to zero and it was bang on)

[John Stevenson]

PM or email your address aGain [ best Forrest Gump voice ] and I'll stick another 125 count Heidenhain encoder on the carrier for you.
That should then take one option out of the scenario [ unless this encoder is pants as well ? ]

[RussellT]

The worm can't be out by half a percent - the wheel would need a half tooth.  If it was consistently missing one gap in the encoder disc it would be out by 0.8 percent - but could it miss one gap about 60 percent of the time.

Can you try it with slower movement and see if that's the same?

Russell

[BillTodd]

does the Heidenhain encoder have a zero output?  (i know some do )

Can you confirm using the zero output that you are getting 125 counts per rev ?

[awemawson]

Bill, no I don't think that this one does 

Anyway the plot thickens

I wanted to prove if the error was actually consistent, and measure it more accurately. So I set up a 'zero pointer' (*) and wrote a diddy program to turn the 4th axis 20 turns or 7200 degrees so that the error would be magnified and more easily measured.

Odd results: first two runs at 2000 degrees per minute gave errors of 37 and 38 degrees added to where we should be - so close that I thought that I'd perhaps bumped the scriber holder when taking the picture.

But I then repeated the test of 7200 degrees but halved the speed to 1000 degrees per minute, and got a result of 31 degrees 


... so it's NOT consistent and could well be a dim bulb in the encoder (I think that they use an under run pea bulb) or it could conceivably be a pulley or shaft slipping.

A bit of dismantling tomorrow if I can find time

Turns out to be about 0.527% so the original guess at 1/2% wasn't too far off


(* I'd have used the inbuilt fidicule marker if I'd remembered that there was one  )

[John Stevenson]

It's probably on last years British summer time.

Encoder leaving here on Saturday, clearly marked "Please throw underarm "

[awemawson]

John you are a hero 

 

[awemawson]

so bright and early this morning I set it going at a crawling 100 degrees a minute for the 7200 degree test, then went and cleared out the pigs, cooked my porridge, ate it, and returning to the workshop it was ALMOST there  (an hour and 12 mins of course)

... how did it do, well it still over travelled but by less at 29 degrees.

So no choice but to start pulling the thing apart, a diversion I could well do without. Swarf cover off, and even so getting the end cover off the belt drive was a pain ... who designed this ....    (me!)

Eventually got the bits obscuring the belt cover off to reveal a toothed pulley on the end of the servo motor with a grub screw only finger tight  I was tempted to put the pulleys back with Loctite as well as grub screws, but future dismantling would be almost impossible without destroying the alloy pulleys.

So I decided to drill and tap them both for a second grub screw at 90 degrees to the first - this is where things started to go awry  Trying to undo the (hard of access) grub screw on the encoder, the ball end of my (cheap and nasty Chinese) ball ended allen key decided this was the time to break off inside the grub and jam there forever 

However I managed to gently tap the pulley off the shaft, then as the grub screw / broken ball end proved too hard to drill out, put two new grubs in and ignored the old one. I put a reamer through to take the slight protrusion of the old grub screw which would have stopped the pulley going back on.

(These are tiny M3 grubs with 1.5 mm hex sockets so a bit delicate.)

So having jumped through those hoops it was just a case of putting it back together making sure the pulleys were aligned and the belt tensioned and trying it out.

..did it work ..... ...you bet 

SORTED


Now perhaps I can get back to making that bally dial that revealed this issue 

[John Stevenson]

Clumsy bastard.

Still sending the encoder as it's no good to me

[awemawson]

Thanks John, you're a brick 

[awemawson]

So with no more ado I skimmed that doughnut down yet again (48 mm now!), set the mandrel in the 4th Axis and had yet another go.


... hoo blooming ray ... it worked perfectly. The scale wraps around the part seamlessly (the join is between the  95 and 96 lines, and I suspect that the characters are better formed - one side of the zero's previously looked a little straight.
 

So now I can go ahead and make the real thing complete with recesses for O rings to add friction in place on the drill grinder

[Pete W.]

Hi there, Andrew,

I'm pleased to read that you've got it sorted.                   

[John Stevenson]

Nope, 96 is greater than 97.

[ Should have gone to specsavers ] 

[Spurry]

Andrew
Very interesting write-up; I am pleased that it was you sorting out the problem...not me.
Pete

[mattinker]

 You were wondering whether this qualified as a project? Had us on tenterhooks for days!!

Glad it worked out, all the best Matthew.

[SwarfnStuff]

Not only kept us intrigued but it is always good to hear of a successful outcome where swarf, sweat and head scratching finally win out.

Gives us all hope that we too will get there eventually. 
      Now if I can only find where the workshop gremlins put that part I was working on a few minutes ago I too might have a win.

     Thanks for a good read,   

John B

[awemawson]

John S, Many thanks 
 
Spare encoder arrived this morning - much appreciated

 

[dawesy]

Glad this is sorted. Always good when you get to the bottom of an annoying problem

[John Stevenson]

It's not sorted though.
Seriously 96 is greater than 97, look at the picture.

 

[awemawson]

John I agree that that version looks slightly unequal. Not sure why  May be that the drag engraver, which is used non-rotating, turned a bit  - it won't be exactly concentric due to the screw holding the bit, and 97 was the first line engraved, and 96 the last.

I gave up with drag engraving as the depth and definition wasn't sufficient, so I dug out my 'Tapmatic Speeder' that gives a 6:1 speed increase on normal spindle speed, and engraved the dials using a conventional carbide 'Vee bit'. I've been putting off using it, as it requires a 'reaction bar' to stop it's body rotating, which was a bit of a pain to arrange.

So here they are temporarily mounted on the drill grinder - I'll clean them up and nickel plate them when I've enough bits waiting for plating for it to be worth setting the tank up.

[vtsteam]

Very enjoyable and useful reading, Andrew!

[John Stevenson]

I have been doing some rotary engraving using a diamond drag over the weekend.
Quite tiny letters in brass for something the Delectable Debs ™ ® © sells, This is new though as up until now she just stamped the model on with 1/8" stamps which was all she had, but now she has her own CNC, once perfected I can pass the job over.

Sooner than hi-jack Andrews post ' again ' I'll start a new one tonight after work.

[awemawson]

Hey what's a friend for if he can't hijack your thread 

[John Stevenson]

OK quick teaser.





Quite hard to photograph being shiny brass and so small.

The brass bar where the letters are is 8mm in diameter and the capital letters are 2.3 mm high.

More later tonight with the tool.