What's new
What's new

Time to build a machine to automate a process, Anyone want to help my first journey?

as9100d

Stainless
Joined
Apr 19, 2019
Location
Paris, Arkansas
What I have is a part thats roughly 4"x4" x 1.15" Made of Acetal. The outside 4 edges get 1 hole dead center of all axis. The drilled hole is all the way to the center of the part where they converge into a 1/2" center hole.

We make some 5000 of these a month and while op1/2 arent that bad, the last op for the 4 holes on the 4th axis take up the most real spindle time because of the load/unload and rotations.

What I want to do is remove this last op from the $100,000+ cnc machines and move them to a machine I can build for about $4000 in materials not taking into account the labor, I have several projects that could use this type of automation and will look at the labor as a trade off for learning.

What im struggling with is wrapping my head around a way to load and position the material via conveyor or other cheap method other than a person manually loading and unloading this machine.

I will purchase 4 er11 1.5kw spindles and vfd drives to power the 4 drills. Then I will make some ball screw driven axis that move those spindles.

Any of you fine folks ever gone down this road and have photos to share?
 
You want something like this from Dessouter.

DESOUTTER AIR DRILL PNEUMATIC DRILL SELF FEEDING 2300 RPM AFDL 5500 | eBay

Not sure what size drill you want drill on the faces, the Dessoutter drills maybe short on torque, but there are other brands out there that will accomplish what you want.

I designed an automatic feed drill a few years ago that had a pistol grip air drill mounted to linear rails. The drill had a 1/2" chuck and plenty of torque at 500 rpm. The assembly was driven by an air cylinder. If I did it again I would use air/oil pnuematic/hydraulic system to provide smoother control. It used a simple air valve to advance and retract the mechanism. If you use solenoids then you could have limit switches that would reverse the carriage when it reached the end of it's travel.

For the $4000 I don't see any easy automated method of loading/unloading, I suspect it will require a human to do the loading/unloading
 
I gave those a thought but the price is just too crazy. The 1.5kw spindle with vfd is $350 each, PLC to handle it all would be $250, another $500 for the air solenoid system and then id use clearpath servo motors for the axis control.

Think two drills facing eachother a known distance apart with space for the part to be . Ill have that one axis move in the plus then in the negative to accomplish the holes on both sides. then repeat for the side at 90 deg.

I'm halfway thinking of a small part size table that raises up, conveyor drops a part on this small table and then the table retracts down. once its down have a pneumatic cylinder push from a corner to align the part on two work stops then have a pneumatic cylinder from the top put pressure on the part to keep it from moving.

once the cycle is done drilling, both workholding clamps retract, table raises up, and an injector cylinder pushes the part down a tray and into a plastic bin.
 
A 4 drill commander head, mounted on a snow automated drill ?

Logic switch on side of snow allows multiple different ways to auto close/open vise/fixture,
via air solenoid.
 
Snow machine looks neat, not sure how to automate one of those but every used one I found was still pretty crazy expensive for what they are.


attached is the photo. The outside holes are what need to be drilled. They are stepped holes that require a custom drill bit.

Right now we do them on the 4th axis and they just eat up too much expensive spindle time.

block1.jpg
 
Why 4 spindles? Seems you could built an indexer cheaper. My first glance at loading would be queue them up in a stack in something to keeps them oriented and kick the bottom one out into a fixture, etc.
 
A dial type machine, also called a rotary transfer machine, would do the job.

This is a turntable about 16 inches in diameter with 6 four inch diameter holes each with indexing pins meshing with the four U shaped slots in the part.

The first process station is for loading. The next four stations are for drilling. The last station is for unload.

Loading is done by pushing the part into a hole from below. A stack of parts, for example, 30 parts,. is loaded on a elevator driven by a screw. At every index, the screw raises 1.15inch, to position the next part. The rotary transfer disk then rotates one position sliding the disk off the stack and into the first drill station. The disk makes its way through the four drill stations and drops out by gravity at the sixth unload station into a bucket.

This is easier to implement with a set of cheap stepper motors and matching drive/step counter boxes than it is with pneumatic pistons. The rotary transfer disk index in the past was frequently done with a modified gear drive.
It was called a Geneva mechanism.

A stepper motor driving a timing belt through a speed reducer would work just as well. There is a problem doing this with a servo motor with position feedback. The motor control will be unstable if there is too large of a mismatch between the rotary table inertia and the effective rotary inertia of the motor/ speed reducer combination. The stepper motor has no position feedback and does not care as much.

The difficulty is that the part held in the transfer disk also needs to rotate 90 degrees at each index to give the drill head access. That can be arranged with a gear/ bearing assembly in each hole driven by a stationary ring gear around the perimeter. Or a pneumatic pusher that indexes the part holder 90 degrees at each station.

The rotary transfer would be fast. The rate determining step is the drill time. Maybe two seconds per hole. You also need to provide a vacuum cleaner at each drill station to keep the chips from piling up.

The trouble with this is that a stack of 30 disks would be completed in 60 seconds assuming a 2 second drill time. The machine requires a full time operator.

A less expensive machine would only have one drill station. The part would be indexed four times within the single process station. The stack of 30 disks would take 240 seconds to complete assuming the same 2 second drill time per hole.

If you decide to make a rotary transfer machine you might as well do all of the face drilling operations at the same time the side holes are being drilled.
 
I dont have an injection mold machine and know nothing about them. That may be in the future though as we have 6 other parts that we currently mill but could mold.

Anyone have a small injection mold machine they want to give me?
 
Why don't you mold it?

Agreed, at ~60K a year this would be a good molded part.

If it has to be machined, I'd want to setup a drill slide and indexer in the machine so that you'd profile the part, then drill using a side-oriented ballscrew-driven method. Done in one, less handling, access to coolant (I'm worried about an uncooled/lubed drill melting the delrin), even if that requires programmed pecking.

Keeps all the chips in the machine, so less cleanup too.
 
My only fear with this kind of setup is that the holes need to be perfect aligned and concentric. How accurate would the positioning be?

A indexing mechanism that does not meet the accuracy requirements can be assisted by using a shot pin engaging in index holes in the part turntable. The indexing mechanism would need to have a flexible drive or some free play to allow the shot pin to fix the turn table location.

The recommendation to mold the part is not a recommendation that you mold the part. The idea is that you have the molds made by a mold maker and then have the injection molding done by a plastic injection company that works with acetal. The holes would still need to be drilled if accuracy is a concern. This could be a expensive mistake if your customer decides to go elsewhere after you have purchased the molds.
 
I dont have an injection mold machine and know nothing about them. That may be in the future though as we have 6 other parts that we currently mill but could mold.

Anyone have a small injection mold machine they want to give me?
Molding would give you a part every 30-45 seconds, maybe even 20 seconds, with no one standing there running the machine, with much cheaper material and with virtually no waste. The problem is if you don't know anything about it you would have to have someone else do it, but you already are having someone else turn the pellets into bar, or plate, for you.
 
Id love to injection mold them but...we keep buying cnc mills at $15-20k each for used robodrills and they pump out thousands of parts in no time.
 
IMG_20191006_170625 (1).jpg

Here is the part as manufactured. Id like to pull the rotary off and add 2 more vises then have the standalone machine in a cell next to this one.
 
If you modified the jig on your 4th so that you could stack 2 or 4 of them at a time you could save a big chunk of the time you're trying to eliminate? Push your vises as far to the left as you can go and then run the op on the 4th only every second run. I don't have any advice for building an automated machine but it could save you some time while you work on that big project.
 
Id love to injection mold them but...we keep buying cnc mills at $15-20k each for used robodrills and they pump out thousands of parts in no time.
IF you penciled it out I am pretty sure molding would look pretty damn rosy. You would probably pay close to the same for finished parts as you do the extruded plate, if you don't count the cost of the mold.
 
I haven't heard any numbers as far as how accurate the holes have to be, size, etc. As far as a mold is concerned, I see an expensive mold with all the side draws. What is it costing now with robodrills compared to a custom machine? Were is the biggest cost as in handling or what? Do you need five drilling stations on the machine (dial table). Is this to be in the robodrill with indexing or what do you propose?

Tom
 
If you modified the jig on your 4th so that you could stack 2 or 4 of them at a time you could save a big chunk of the time you're trying to eliminate? Push your vises as far to the left as you can go and then run the op on the 4th only every second run. I don't have any advice for building an automated machine but it could save you some time while you work on that big project.


I'd love to but im out of table space and the holes have to be on the money in position. Thats why im just going to free up that op completely to get more parts on the table and drill outside the machine.
 
I haven't heard any numbers as far as how accurate the holes have to be, size, etc. As far as a mold is concerned, I see an expensive mold with all the side draws. What is it costing now with robodrills compared to a custom machine? Were is the biggest cost as in handling or what? Do you need five drilling stations on the machine (dial table). Is this to be in the robodrill with indexing or what do you propose?

Tom

Holes have to be concentric within .001 and hole size is .196 through with a .206 to .800 deep on 3 sides. Mold is only as expensive as the person making it..which will be me and I guess I work for free. Part is 1.11" thick.

Do you think I'd need to have a draft on the walls for that kind of part or can I keep the walls square?

The drill cycle with an operator (currently open and close doors but can automate that of course), is 1.5 minutes per part using TSC and carbide drills with a toolchange. It adds operator fatigue and more chance to miss something since there are 3 ops in the machine going on at the same time. Had a few mishaps putting the part in the wrong direction already and no way to dummy proof the rotation.

The automated machine I figure I can build materials only no labor, for under $5000 complete. That will save me 100 hours a month which equates to roughly $21600 per year.Flip side of this is...my customer will buy more, however im limited to available spindle time. Freeing up 100 hours a month will let me make an additional 1000 pieces per month for a total of 5000.
 








 
Back
Top