CamBam
News:
 
*
Welcome, Guest. Please login or register.
Did you miss your activation email?
April 25, 2019, 12:47:58 pm


Login with username, password and session length


Pages: 1 2 3 [4] 5 6 7
  Print  
Author Topic: Automatic tool changer for the lathe  (Read 18782 times)
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1083


View Profile
« Reply #45 on: March 01, 2018, 23:20:04 pm »

Yes they are overbuilt by some margin (which is poor engineering) really.

I had to calculate the the force required to keep the overload securing bolt tight enough for normal
operation during  turning, but would release during an overload situation like running it into the chuck.

I calculated the cutting force to be no more than 200 Newtons even with some extra safety margin.
These bearings (a pair of 68 mm OD  Angular Contact) have a Static Load Rating  at 30900 Newtons each!
The thing is, these bearings are cheaper to buy as they are more popular, and so many more are manufactured than the correct sized ones for this job.

The spindles are Cro/Molly and the ratchet wheels are 1045 medium carbon steel.
The only upside (with being so heavily built) is that the surface finish and my ability to get much better sizing (first time round) of the of the lathe has greatly improved on both counts as compared to the original manual tool-post.

I greatly under-estimated the battery life as well, as after watching BigClive on youtube speaking
about the real as opposed to actual battery capacity of the Chinese L/ion batteries, I rated mine at 2ah they have 4ah printed on them.

Last Saturday I pulled them out (for the first time since installing them in the turret) and placed them in my fancy battery charger which gives you a blow by blow description of what's it's doing and the battery voltage's were 3.75,3.75 and 3.73 volts respectively.

This to some degree, is because if I have a simple job of turning to a diameter and no tool changes
I don't even turn the turret on and just select the tool required by rotating the turret by hand like a using a quick  change tool post.

After watching the guy on ExplainingComputers.com I was leaning towards using a Raspberry Pi
instead of the BBB because of the built in hardware 264 Video decoders, but the power requirements 2.5 amps seemed a little high for the battery capacity I thought I had, but now I think this might be Ok.

Dave 
Logged
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1083


View Profile
« Reply #46 on: March 06, 2018, 12:37:52 pm »

Another small update
I've finished the bearing housings and am just waiting on the tap to finish the spindles.
Here's a pic of the comparison between the original tool post and the new turrets.

The next and last this to do is adapt a disk brake to the spindles before adding the electronics.

I did order a Chinese one, however it may be a little big and so I might have to reverse engineer it and make a smaller one based on it's design. The Irony of this hasn't passed unnoticed Roll Eyes.
 
I've looked at some Li ion battery packs that have built in battery management, just have to find one that is the right dimensions to fit.

Dave


* Turrets Main Bearings Finished.png (1256.68 KB, 1000x563 - viewed 126 times.)

* Size compasison with origional tool post.png (1247.74 KB, 1000x563 - viewed 116 times.)
Logged
Bob La Londe
CNC Jedi
*****
Offline Offline

Posts: 3538


^ 8.5 pounds on my own hand poured bait.


View Profile WWW
« Reply #47 on: March 06, 2018, 16:16:58 pm »

I am sure you have thought of this (and I think I read that you did to).  Why not use a cordless drill battery that has a manual charge indicator on the front like my Milwaukee drill batteries?  Then use a voltage regulator to get a lower voltage.  Start the machine, check the battery, swap to the one on the charger if its low.  They are modestly expensive, but very good batteries.  If you want higher voltage string two together in series with two locking cradles.  Maybe a pair of 20s or 24 regulated down to 36 for a nice solid higher torque modest speed stepper application.  

I have 18s, but I would go higher so you can maintain constant voltage for longer by using a regulator.  36V is a decent voltage for a small stepper controller and stepper.  Its not like the tool changer has to be fast.  Gear it down, be modestly slow, and rock solid for position plus or minus half a microstep. With that approach you might be able to run even lower voltage.

Sorry if you have already thought of all of this.  I admit I have not read the whole thread.  

I still question the wireless approach, but I am anxious and curious to see it work.  
Logged

Getting started on CNC?  In or passing through my area?
If I have the time I'll be glad to show you a little in my shop. 

Some Stuff I Make with CamBam
http://www.CNCMOLDS.com
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1083


View Profile
« Reply #48 on: March 07, 2018, 04:21:55 am »

Hi Bob

With the wireless link, I've been using the original turret to machine the others (and other things) for 3 months now, and the wireless link has worked faultlessly, mainly because of my industry experiences it's designed to fail safe. That's why I've never mentioned it as there was nothing to report.

I did look at the power tool battery pack idea and was very keen on this too,and as you rightly pointed out it has some great advantages and appealed a lot to me.

For example I could remote the battery management electronics and save space which is tight inside the turret and use off the shelf proven and robust components rather than something that I've cobbled together.

The “fly in the ointment”  here was the actual battery connection to the turret which would have to 
keep coolant and chips out of the interface between the battery and the  connection. I think that over time this would be a point of failure.
   
 Dave
Logged
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1083


View Profile
« Reply #49 on: March 29, 2018, 01:08:02 am »

Another small Update.
I've managed to get both new turrets finished up to this stage where they have run connected to the computer. So all that's left to do, is fabricate a Floating disk calliper for each, and then machine the top and rear covers which have some features that may prove interesting to machine, so I'll leave these to the very last.

The solenoid's for the brake activation have arrived, so I've started to cad them up and hope to have a working version late next week. (still waiting on the sensors for position and brake activation)

I've had jump through some hoops to get the Main Bearing Housings machined, as the fixing of the Z axis to the column, was found wanting (rigidity wise) under heavy loading, so I've ordered and received these Linear rails HGM25's the heavily pre-loaded types used for Z axis's (you should only use these with ground surfaces) which I have on the column.

So as soon as the Ball-nut fixing block arrives, I'll install them.
The only thing to do then, is perhaps change the spindle shaft to except standard tooling at a later date.

I've found the most time consuming thing is the upkeep of the drawing files to reflect what changes I've made in the shop, there's well over 60 Freecad/CamBam/Camotics/Visual Studio/Atmel Studio files as well as a slew of interchange files(.dxf's and .step) and it's nearly a full time job just  to keep on top of them.

Dave


* Test Run of Turret 2.png (1210.45 KB, 1000x563 - viewed 92 times.)

* Mechanicals Done except floating Caliper.PNG (67.87 KB, 585x542 - viewed 100 times.)
Logged
lloydsp
CNC Jedi
*****
Online Online

Posts: 7821



View Profile
« Reply #50 on: March 29, 2018, 01:40:35 am »

Yeah... Keeping the docs' is almost a bigger job than making the parts.

I'm about to embark on two new projects for the shop... in another thread.  I don't want to hijack this one.

Lloyd
Logged

"Pyro for Fun and Profit for More Than Fifty Years"
Garyhlucas
CNC Jedi
*****
Offline Offline

Posts: 1225


View Profile
« Reply #51 on: March 29, 2018, 18:44:53 pm »

Our products have about 800 parts sourced from about 60 vendors for each job. Our SolidWorks library has about 4000 parts currently. We use a program called AllOrders that is linked to SolidWorks by our part number and a custom Excel macro. Export the SolidWorks BOM. Import to Excel and export the macro results whichs roll up all material lengths and quantities and import to AllOrders. Print POs for the materials, Picklists for inventory, Workorders for the shop, Packing List, and get a very accurate cost including labor for the whole job!
Logged

Gary H. Lucas

Have you read my blog?
 http://a-little-business.blogspot.com/
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1083


View Profile
« Reply #52 on: March 30, 2018, 02:39:40 am »

Hi Garry

In 1982 I was the production clerk at RipCurl (the wetsuit people) where we instituted a similar scheme using
A Diablo Mini and several Osbourne portable computers and  (with the front cover which was the keyboard) clipped on, they very much resembled a sewing machine.

So I could do something similar, but the time to upkeep the "RipCurl System" required a full time position and was one of the busiest times of my life, and being a one man band,  doing this right now is not high on the priority list.

If I were to make the turrets in quantity, I would farm some of the 2 D work out  to a water-jet cutter and assemble the turrets in house were I could finish machine the parts and do the final assembly.

Dave
Logged
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1083


View Profile
« Reply #53 on: April 12, 2018, 14:17:33 pm »

Another small update.

Well the Ballnut mounting block finally came yesterday, so I'll install the new Z axis rails and slides
over the week end and am looking forward to seeing it run.

In the meantime, I scoured the internet and my engineering design books for a suitable design that would meet my requirements, but no joy as they either wouldn't scale or had too many moving parts.
So I broke out the drawing pad and doodled for a couple of nights trying to find something that would be simple and reliable, but had to be a floating design (self adjusting).

I narrowed it down to a couple of likely candidates and cadded up the best prospect.

What I needed was a negated brake (the solenoid when energised released the brake) which only happened when the turret was changing tools so as to save power.
 
I wanted it to be simple and this mechanism has three moving parts that move only 0.5 mm for the pads and 3 mm for the actuator.

The first pic is of the initial mock up in CB.
The second pic is the CB geometry imported into Freecad  and turned into a 3D model.
The third pic is the calliper bodies and a substitute rotor showing the pockets for the pads.

I've got some time tomorrow to finish machining the bodies and hopefully get the pads and actuator done.
I've already programmed the new parking manoeuvre and added a two stage variable power
method for the brake. (separate energising and holding stages) again to save power.

Dave


* proposed disk brake mechanism.PNG (66.55 KB, 414x503 - viewed 73 times.)

* Prototype Disk brake For turret.PNG (39.57 KB, 837x538 - viewed 82 times.)

* mock up of caliper bodies without brake pads.png (973.26 KB, 1000x563 - viewed 76 times.)
Logged
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1083


View Profile
« Reply #54 on: May 24, 2018, 04:32:20 am »

Another Update on the progress so far.

Well, I machined up the first set of  Calliper bodies using the slope property in the profile mop. (17.32 deg)  with  a 3 mm cutter and  0.1 mm steps.
I also left a 0.25 mm clearance on each end of the pocket so that the pad would float freely.
(this would come back to bite me later)

I then cut out the Brake pads, which I found very fiddly and hard to set up in my present vice,
and  I new I had to order a  5 in machinist’s vice to get any form of repeatability in my work.

I then assembled the calliper and tested it in the bench vice with a dummy rotor.
With bright led torch and a magnifying glass I actuated the mechanism with a pair of pliers.
And everything looked good, the pads opened and closed, but worryingly needed more force than I expected to operate.

The second problem arose when I rotated the rotor, there seemed to be a bit of backlash, and on investigation found that the pads were firmly attached to the rotor but moving in the pockets (with that extra clearance I gave them).

To address the first problem I machined up a new set with the minimum wedge angle the geometry would allow (to get the most mechanical advantage) and also took the clearance in the pockets down to 0.1 mm).

Ultimately this was a fail, cleaning up the pockets did reduce the lash to a very small level, but the force required to operate it 6 kilogram region and as  I'm committed to using as lest power as possible so I had to come up with another solution that only required a 3 kg force solenoid.

See next post
Dave


* First Caliper construction.png (994 KB, 1000x563 - viewed 66 times.)
Logged
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1083


View Profile
« Reply #55 on: May 24, 2018, 04:39:12 am »

continued...
I could see now, that I need some force multiplication if I was going to use the 3kg solenoid.
On next attempt, I used a button Actuator which braked on the periphery of the rotor.
The picture shows the button “Centre Shaft” that the brake material would sit in.
To do this, I had to semi manual program an R70 radius in the ZX plane.
This went well. Pic 1

Next I made a capstan that fitted to the button housing and provided the necessary torque to activate the brake. Due to the space constraints I could only get 3 to 1 multiplication and I had to make quite a few of these to get it to operate smoothly (originally I  used a bush in the capstan but had to replace it with a roller bearing, which made all the difference).

This turned out to be a mini project all on it's own!


I then tested this out (ran it in one of the turrets) but  disappointingly it didn't have enough braking capacity, which meant that I'd have to install a heavier spring in the button but I wasn't sure how much extra torque there was left to do this.

So I conducted a little experiment and took the brake out and sat it on the bench, then placed a small plastic jug graduated in ml and filled it with water until the  spring popped. I did this a few times to get an average.

It turned out the force was 25 newton, the same as the solenoid and so with capstan multiplying the force 3 to 1,  I really could increase the spring pressure but  for various reasons I didn't like the idea.

So  I set that button aside (I'd made a couple) and ordered a 5kg solenoid  for it.

Over the next week or so I read some engineering and Mechanisms books and haunted the web to see what other people had done with braking mechanisms.

Then it dawned on me that a braking mechanism requires a fairly high force to operate that's the nature of the device, I now realised that I needed a Indexing Mechanism instead as they don't require anywhere near the same force as a brake to operate.
So was there another way to achieve what I wanted with the components I had already made.

It turned out there was, I modified the Disk Rotor by cutting 120 (3mm scallops) out of the   periphery.

I then installed some drill rod in the shaft of the button (this engages with the scallops) so I didn't have to make any new components only modify the ones I already had.

This finally did the trick I could now use the 3 kg solenoid  and a comparatively small spring to get the job done. Pic 2

Dave 


* Finished Brake Pad R70 Radius ZX Axis.png (900.86 KB, 1000x563 - viewed 75 times.)

* Moved from brake to indexer.png (1200.01 KB, 1000x563 - viewed 77 times.)
Logged
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1083


View Profile
« Reply #56 on: June 05, 2018, 02:37:06 am »

Another  progress update....
Once I got the Design of the Indexer sorted out (many attempts with  different  geometry and materials)                  I then set out, to test it out!

I wanted to know these things:

How long would the battery's last ?

After many consecutive tool changes, would the Solenoid Duty Cycle be exceeded ?

Would the Capstan Cable hold up?

What would happen if the Indexer failed ?    Software wise.

What things were likely to fail in the indexer, and have some limited self diagnosis and reporting of the error.

To make the tests I made up a Gcode program consisting of only tool changes,it was supposed to be 100 but I stuffed up the cut and pasting and only had 96 , which I only found out during the course of the tests.


I added a couple of radio buttons Auto-recovery and Manual recovery.

What happens in manual mode, is the turret will signal a tool change abort (and hi-light's the tool number) in purple:
 For safety you now have to toggle the “Restart After Tool Change” button which will turn orange and indicate “No Restart after Tool Change” to Mach3
You now have access to the tool change buttons, and if you press the hi-lighted tool number an      tool change re-attempt will occur.

If  the  re-attempt is successful and a tool change begins or is in progress, then toggle the “No Restart after Tool Change” button and Mach3 will continue on executing the Gcode from where it left off.

Auto-recovery Mode
If the Auto-recovery radio button is checked:
If the turret senses a Indexer failure It will first assume that the duty cycle of the solenoid is exceeded, and do a re-attempt every 10 seconds until successful or 10 attempts have elapsed.
It will then do a  solenoid activation check  and if this passes, reports a cable failure\Capstan securing bolt lose error, or  else it will report a solenoid failure.

Dave
Logged
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1083


View Profile
« Reply #57 on: June 05, 2018, 02:39:53 am »

I started the tests without the battery's as I discovered that they need to be topped up and so used the
bench power supply.

For the first test I set Mach3 running and did 1064 consecutive tool changes keeping an eye on the solenoid temp and it barley got over the ambient temp which was (15 Celsius)  and the cable had held up fine so far.
I took some current draw measurements, and on average (using a two stage modulated signal) for the solenoid, the draw was an extra 250ma to 300ma (this jumped around a bit as I measured it with a multimeter which was a poor choice ) while the tool selection was in progress and at stand still the current draw was obviously the same as before.

So far the results were encouraging, but I was left a few questions:

I hadn't tested the solenoid in a high ambient temp environment, so if I heated up the solenoid 
(with the other half's hair dryer)  could I saturate the solenoid heat sink, and would it fail, and in what way, and how would the turret respond to this.

I set the turret off for another 1000 consecutive tool changes and harassed the solenoid with the hair dryer and got it too hot to touch for long but nothing happened and Mach3 finished and rewound.

I then did what I should have at the start, and calculated the rise in temperature above the ambient that the current I was putting through it would cause, turns out that the energising current would cause an 8 Deg C rise but this only lasts milliseconds and the holding current would cause a temperature rise of 3 Deg C.

So even though it seems to be  very unlikely that the solenoid would ever fail due to heat stress, I coded for this event anyway by letting the turret re-attempt a tool change every 10 seconds for 10 times then giving up and reporting a unrecoverable error.

I've also added a Homing Function that works by using a fast homing method (if the tooling is already indexed) and, if this sequence fails (because the tooling wasn’t Indexed and was at some random angle), a slower but more comprehensive (auto backlash compensation re-calculated) method is used.

So in short the battery life, which still has to be checked more thoroughly (so from the re-calculated value going on the measured current) I estimated that it should be around 650. and will test this as soon as I work out where to put them (the batteries) in the turret.

The Indexer (and thus the turret)  has an 100 % Duty Cycle, so continuous running is assured.
The Turret can change 900 tools an hour Indefinitely and has on occasion (last Sunday) ran 3600 in a row during the bench  testing.(with the bench power supply)

The indexer cable (which is synthetic not ferrous)  has lasted, north of 35,000 tool changes so far and shows no sign or wear.

So now I think I'm going to call the Indexer Done and the basic form of the turret (dimension and material wise) is settled.

What I intend to do now is make an Indexer for the other turret, to mostly check and clean up the
 Cad and Cam files (I had changed some dims) while actually making the Indexer at the machine's.

Dave


* 1064 continuious Random tool changes.png (237.38 KB, 1000x534 - viewed 66 times.)

* Turret Indexer Failure Mode and Recovery.PNG (259.38 KB, 1335x680 - viewed 72 times.)
Logged
Bubba
CNC Jedi
*****
Online Online

Posts: 2590



View Profile
« Reply #58 on: June 05, 2018, 12:38:21 pm »

I'm impressed. Nice work Dave.
Logged

My 2¢
dave benson
CNC Jedi
*****
Offline Offline

Posts: 1083


View Profile
« Reply #59 on: June 06, 2018, 15:24:44 pm »

Hi Bubba

Well I'm not counting my chickens just yet, I think I'm over the Hump and most of the difficult stuff is done.
I still have to Make the Top Case which has to have a water tight seal and an access port to change the batteries.
Also a Rear end case with a portion removed and a piece of lexan inserted to allow the bluetooth module to function.
Because I used up some precious space inside the turret with the Indexer solenoid I had to order some small format
versions of the micro-controller, which are only 50 mm x 38 mm.

I made the other indexer base today, and while I was at it I thought I may as well make the other 11 as well, which from flat bar stock was a nice straight forward job.
Dave


* Indexer Base Plates 12 .png (1042.33 KB, 562x1000 - viewed 71 times.)
Logged
Pages: 1 2 3 [4] 5 6 7
  Print  
 
Jump to:  

Powered by MySQL Powered by PHP Powered by SMF 1.1.21 | SMF © 2015, Simple Machines

Valid XHTML 1.0! Valid CSS! Dilber MC Theme by HarzeM
Page created in 0.167 seconds with 19 queries.

Copyright © 2018 HexRay Ltd. | Sitemap