Mating pins & holes

[esolberg]

Another novice question, thanks for bearing with me. This really has to be basic mechanical engineering, hopefully you folks here don't mind too much. I'm a software guy lol.

I'm pocketing out some 3/16" holes for some pins used for registration/alignment. The pins I'm using are precision ground stainless steel dowel pins accurate down to something ridiculous like 0.00001. They are listed as "standard", not "oversized" pins.

Now, being a software guy- I should just be able to mill a 3/16 hole and it'd be a perfect fit right?

I ended up having to go 0.004" bigger on the radius (so actually .008 on the diameter) before the pins would press fit in. After that they fit pretty snug but still easy to insert/remove. This seems excessive doesn't it? I hate to know from that start that my registration is +/- .004.

Would I get a closer fit if I milled under size and finished off with a reamer? I know this must be machining 101 regarding pins and matching hole sizes and how much to under/over cut for the fit and material. Is this info listed somewhere?

Thanks in advance for the advice.

[jerry_stanek]

Have you checked that the bit is the diameter it said it is. also there could be bit deflection try running the holes in the opposite direction.

[MogulTx]

Jerry beat me to it. It is likely a bit diameter thing. If you caliper your bit it probably is not an 1/8" bit. I have a couple Onsrud bits that say they are 1/8 but caliper out at .116. That is four and a half thousandths off of what I would expect.

THEN, you have to have room to get it to slip into the hole... ( A few thousandths is good) ....and if you are "registering" that implies at least two pins, which means you may have some lack of accuracy on your CNC that you have to account for just in the spacing between the two holes...

Add it all up, and you need a little extra to make things work.

I have never seen stainless pins turned to the hundred thousandths size category. They must be some snazzy pins if they are that accurate. I often use a dowel that I first caliper, and then caliper my bit to make sure I am creating a toolpath that is believable. I also have been known to turn a dowel down after inserting it into the bed of the table, and to register it on top of that... kind of cool to be able to do it either way....

[Rob Gunn]

Eric, This is the rule of thumb that I used for 25+ years in the toolroom. "Slip fit" for a pin that you want to remove I would ream to .001" over pin diameter. "Press fit" for a pin that you dont want to remove would be .001" under pin diameter. Drill your pin hole on location with a drill 1/64th" undersize and then finish ream to the press or slip fit reamer. Most any toolmaker will have a full set of reamers in his box that will do what is needed. Maybe you know someone in the trade that could help and lend you the correct reamer?

[dana_swift]

Eric- I'm a software geek myself with a strong EE background. What you are experiencing is "the real world". In a computer a 0.250 pin will exactly fit a 0.250 hole in a model of a material. Oh how nice, and unrealistic.

The real world has bit flexure, thermal expansion, material springback, gear backlash, dull bits, metal in the CCP (Cheap Chinese Plywood), structural flexing, operator errors, and other such fine issues to help distinguish practical people from theoretical ones.

All that said, its possible to get 0.001 from a shopbot, but not easily, and only for a given setup. Usually the best possible is 0.002 in repeatability. If you dont know what you are doing accuracy is going to be more like 0.015.

Depending on what I am doing I change the setup, calibrate the bit, use elaborate hold-down, etc. Or if 0.015 accuracy is acceptable just cut and not worry about feed rates, and the real world problems listed above.

When it matters, like trying to exactly fit a pin in a hole, Cut some test material. Measure the hole size created. Redefine a bit whose diameter is the effective diameter you computed based on the actual hole size cut.

Now it possible to get 0.001. But only for cutting holes in that material, with that bit, with those feeds and speeds. And even then you may find variation of 0.001 or 0.002. If your rack and pinions are not tight you will get whatever accuracy that is no matter what you do. And they wont stay tight! So that bit calibration only applies for hours after your measure it, otherwise temperature changes throw everything off.

Software is nice because it is so repeatable, but in the real world a person needs to know when their design software should have been titled "solid fantasy", "scratch up", or "firewood works", etc..

Welcome to the real world!

D

[steve_g]

Eric

Last night I was mulling over how to say what Dana said... I was going to say "it's a router, not a Bridgeport mill..." But I like how Dana said it better!

SG

[esolberg]

Another novice question, thanks for bearing with me. This really has to be basic mechanical engineering, hopefully you folks here don't mind too much. I'm a software guy lol.

I'm pocketing out some 3/16" holes for some pins used for registration/alignment. The pins I'm using are precision ground stainless steel dowel pins accurate down to something ridiculous like 0.00001. They are listed as "standard", not "oversized" pins.

Now, being a software guy- I should just be able to mill a 3/16 hole and it'd be a perfect fit right?

I ended up having to go 0.004" bigger on the radius (so actually .008 on the diameter) before the pins would press fit in. After that they fit pretty snug but still easy to insert/remove. This seems excessive doesn't it? I hate to know from that start that my registration is +/- .004.

Would I get a closer fit if I milled under size and finished off with a reamer? I know this must be machining 101 regarding pins and matching hole sizes and how much to under/over cut for the fit and material. Is this info listed somewhere?

Thanks in advance for the advice.

Thanks everyone for the advice. Somehow I left out a very important piece of information. I am cutting the holes in aluminum.

I agree probably my router bit was off. As you may have seen in my other thread, this bit is not with me any more.

[esolberg]

Eric- I'm a software geek myself with a strong EE background. What you are experiencing is "the real world". In a computer a 0.250 pin will exactly fit a 0.250 hole in a model of a material. Oh how nice, and unrealistic.

The real world has bit flexure, thermal expansion, material springback, gear backlash, dull bits, metal in the CCP (Cheap Chinese Plywood), structural flexing, operator errors, and other such fine issues to help distinguish practical people from theoretical ones.

All that said, its possible to get 0.001 from a shopbot, but not easily, and only for a given setup. Usually the best possible is 0.002 in repeatability. If you dont know what you are doing accuracy is going to be more like 0.015.

Depending on what I am doing I change the setup, calibrate the bit, use elaborate hold-down, etc. Or if 0.015 accuracy is acceptable just cut and not worry about feed rates, and the real world problems listed above.

When it matters, like trying to exactly fit a pin in a hole, Cut some test material. Measure the hole size created. Redefine a bit whose diameter is the effective diameter you computed based on the actual hole size cut.

Now it possible to get 0.001. But only for cutting holes in that material, with that bit, with those feeds and speeds. And even then you may find variation of 0.001 or 0.002. If your rack and pinions are not tight you will get whatever accuracy that is no matter what you do. And they wont stay tight! So that bit calibration only applies for hours after your measure it, otherwise temperature changes throw everything off.

Software is nice because it is so repeatable, but in the real world a person needs to know when their design software should have been titled "solid fantasy", "scratch up", or "firewood works", etc..

Welcome to the real world!

DThanks Dana you hit the nail on the head for sure. I have a funny story. In 7th grade I took woodshop. Absolutely loved it but to be honest my skill with the tools was not so great (thank goodness for CNC). But I designed some pretty interesting projects that included angles, curves, inlays, intermediate joints etc. I did this all on paper and in my head- this was before CAD or PCs. The projects turned out passable, but I got A's for the designs. The next year the teacher used all 3 of these as class assigned projects. I got to watch all my friends make my stuff better than I could haha!

[esolberg]

Eric, This is the rule of thumb that I used for 25+ years in the toolroom. "Slip fit" for a pin that you want to remove I would ream to .001" over pin diameter. "Press fit" for a pin that you dont want to remove would be .001" under pin diameter. Drill your pin hole on location with a drill 1/64th" undersize and then finish ream to the press or slip fit reamer. Most any toolmaker will have a full set of reamers in his box that will do what is needed. Maybe you know someone in the trade that could help and lend you the correct reamer?

Thanks Rob - I own 3 reamers from a project I did a few years ago and one of them just happens to be the exact slip fit size 3/16 + .001. I'll use that!

[esolberg]

I've been asking a bunch of rookie questions here, and everybody's answers have been a big help. I don't know why it didn't dawn on me that milling out holes wasn't the best way to go - based on the advice above I drilled 11/64 holes then reamed them to 3/16 + .001 (doing both on the shopbot). The fit and alignment were perfect.

The project is a jig to hold a circuit board while applying solder paste through a laser cut stencil. I had the same pin holes put on the stencil for registration & holding in place during use. Fortunately I had undersized them slightly so I was able to ream them to a perfect fit too. I figured I was over engineering it a bit -- people usually do these by aligning them manually for small jobs. But it came out so perfect that it really was worth learning.

I attached a few pics

The solder stencil for my PCB

http://i37.tinypic.com/2h4975u.jpg

The jig I cut out on an aluminum plate on the shopbot. The holes on the corner were just to hold down the aluminum while cutting- I countersunk them so I could surface the entire plate flat to the shopbot. The jig has little pins in it that align with M1.6 screw holes on the PCB as well. Initially I had cut the jig out of plastic but these pins were too small and broke off so I went with aluminum.

http://i35.tinypic.com/343m0qg.jpg

Jig with PCB inserted

http://i33.tinypic.com/25f6c0x.jpg

Jig with PCB and stencil

http://i33.tinypic.com/jfw0f9.jpg

The boards came out perfect and applying the solder paste was a breeze as I didn't have to think about holding the stencil still and aligning it manually.

Anyway, thanks for the great advice. Always enjoy learning new stuff.

[dana_swift]

Eric- that is way cool!

Making PCB's with a shopbot is an ongoing development, although a very slow one! I would like to find a way to have the bot deposit resist on a copper clad board (CCB). I have found dot matrix printheads that could be attached to the Z axis and used as a very large format printer. The problem is inks..

I am looking forward to putting a 3D extruder head on my bot, one of the things it might be able to do is "print" resist on a CCB. After etching, it could "print" the solder mask and "silk screen" artwork. So my 3D model would be only one or two layers of extrusion.

Making custom PCB's quickly for one off projects would really be useful.

Oh.. and being the lazy cuss I am, I would like to adapt a 5th axis to allow pick and place orientation of components. (My fourth axis is dedicated to the indexer.) Ideally it should be able to place 0201 parts as easily as it can place 144 TQFP's. Or devices with 0.5mm lead densities, those are a real pain in the a$$ to do by hand.

Building a fixture to hold component tape with a probe to lift parts out using vacuum and putting them on the board. The hard part is the motion system which the shopbot already provides. Then populating a SMT board can be done with the same machine that created it! The bot can also make a dandy case, that has been done many times now. Running the drill file is understood..

Turn times for complete 2 layer boards should be in the hours, and that is all the way to populated ready to smoke boards. (A little EE humor there.)

You get my thinking going showing mask jig's! I see the bot as a big motion control system with a router attached as ONE option, not its only configuration.

Do you use a toaster oven for reflow? I still hand solder all the individual SMT components, so I dont need masks. That pick and place project just NEEDS to happen tho. Most days my bot pushes a bit through wood.

How quaint eh?

Nice work, and as you can see- inspirational!

D

[esolberg]

Greg that sounds awesome. The inspiration is infectious!

I looked at using a toaster oven for reflow, but the guys at sparkfun (who appear to know what they're doing) recommended a skillet instead. I went with that, so I don't have any comparison to the toaster oven but its nice to be able to watch it reflow up close. But since you got me going, I'll tell you my whole story lol...

I'm actually a software engineer- but I had a passion for electronics as early as 8 years old. I used to read books in the library not understanding what I was reading, but still accumulating bits of knowledge. In my early teens i used to design lots of digital circuits. Digital is easy- its just logic & timing. But whenever I tried prototyping - in addition to difficulty affording and acquiring parts- I didn't understand the analog enough to make my stuff work.

When I finally took a class where I learned how ohms law is applied (not just i=e/r but how it is used practically) suddenly everything I learned made sense to me. I was surprised how much I had retained - I wish my brain cells still worked that good.

I planned to go into EE in college, but computers were picking up and I took to software like a fish to water, so much so that I actually quit college to go to work and haven't looked back.

In one job in the late 90s I had to design an entire system- I did the software but also had to design hardware. It was basically a fancy credit card reader that would communicate with other readers in the area and process through a back office system. I built that with a PIC microcontroller but I had never successfully made anything other than trivial circuits work before. And I didn't have the proper tools like an oscilloscope- if something didn't work I had no way of knowing if my oscillator was working or if it was some other problem in my circuit or even the software. Lo and behold, my little LED blink program worked the first time I turned it on and I was able to build up from that. Then I designed the circuit board (using Corel Draw haha) and sent that off to make a simple through hole board. It came back and I learned the hard way about double checking your board layout. There was one circuit trace that was wrong. I was mad at myself for that, but it wasn't difficult to remedy with a jumper wire.

That was the last electronics I did for about 10 years, until recently I got a wild hair to do this little project I'm working on. I began to explore what was available and simply couldn't believe it. I couldn't have imagined using SMT parts and re-flowing my own board. I couldn't even imagine being able to afford professionally made boards. So I was pretty certain this wouldn't work, there was just too much new that I'd never done before. The nice thing, though, using Eagle is it really made it hard to screw up the circuit board since the PCB and the circuit diagram came from the same file. That probably existed back then too, but not affordably. Anyway, I had never built anything SMT before so when I selected components I tended to pick the first one in the list in eagle. Presumably these were the most popular so I thought. I didn't have any perspective on how big (or small) things were. When my parts arrived and I realized how small 0402 parts were I just laughed. There's no way in the world I'll be able to make this work!

So I got everything together along with a mylar solder stencil and proceeded to crank out 6 prototype PCBs, placing those little 0402 grains of sand with a magnifying glass and tweezers. To reflow I didn't use a skillet but rather a burner that cost $15 on amazon (I used a 3/8" aluminum plate on top of it to avoid putting the boards right on the burner). I wanted to get all high-tech with the temperature control so I could match the solder manufacturers temp profile. I invested a bunch in a PID controller, thermocouples, figured out how to attach the thermocouple to the aluminum etc. etc, RS485 interface & the whole 9 yards. I fired it up to reflow the first board and realized that the burner could not heat fast enough to even come close to the manufacturers reflow profile. So I quickly detached the PID controller and just let it heat up manually and removed the boards a few seconds after seeing the last pad reflow.

To my complete shock, all six boards came out perfect and actually worked.

Now I've revised the project some and the board you see there is my "Rev B". The mylar solder stencil was a bit of a pain to work with so this time I spent a little more and got a laser cut steel stencil. As you can see that is much better and aligning it the way I did made it unbelievably easy. This doesn't seem to be a common approach because the place that made the stencils didn't seem to understand what I was trying to do. I think most people just align them manually which doesn't make a lot of sense to me. I also used slightly larger parts this time (smallest 0603 haha). One thing I would do different next time is have the board stick up a little bit above the top of the aluminum, like 0.01". As it is, I cut down to .062 and the board lies flush. But I think it was depositing just a little too much solder paste, and I think having the board stick up just a little would remedy that.

Even though it was a bit too much solder paste in places, it still reflowed perfectly with surface tension magically compensating for all my imperfections. I did 8 boards and 2 of them were messed up simply because I bumped them and knocked a couple parts out of place and didn't notice before I cooked em. Another one has an LED that doesnt work - not sure yet whether it got too hot or was just a dud. Other than that, I'm again very happy with the outcome and just amazed at what is possible now.

I love your idea of pick & place. I did it manually (photo attached haha) and that is the most time consuming part.

Let me know if you get anywhere with that!

Cheers,
Eric

http://i40.tinypic.com/2bcgmt.jpg

[dana_swift]

If you are hand locating 0402 size components with a magnifying glass, I am impressed indeed!

I rarely use anything smaller than 0603. (For the casual reader, that means the component measures 0.060" x 0.030".) Most of my components are 0805, as they are easier to handle, and FIND when they go zing.. I hand solder them to the PCBs.

The solder profiles are nice when the facility is available to follow that, but I just use the temperature regulated soldering iron until the solder flows and that's it. I have never lost a semiconductor, so that seems to be good enough.

As to MPU's I like the ARM Cortex M3's, they pack a lot of punch for about $10. Making one-off projects is something I would like the bot to help with. When the components cost about $30 for a board, I would like to be able to make the board itself for $5 to $10.

I cant make an Arduino for less than what it costs to buy one assembled.. I realize I am benefiting from economies of scale and mass production, but most retailers get 100% markups. So I should be able to get a lot closer than what I do, and look to the ShopBot to help lower that cost.

It will happen, I'm well down the road, just not at my destination :)

D