Reverse engineering tips? CMM? Use the bridgeport DRO?

I will be reverse engineering lots of stuff in the coming months. Mostly simple hole locations and sizes with diameters that are +/-.004 on 2.5 axis parts. I was thinking a fast method for stuff like this would be gage pins in a drill chuck and trying to find the centers of holes then just writing down the coordinates as I get them on the DRO.

What are your tips or instructions for finding hole and radii locations quickly? At what point is it worth buying a CMM? If I have never used or even seen in person a CMM before but I am very highly mechanically inclined and about half way through a BSME. Could I figure it out pretty easily most likely? When should I consider buying one?

Any other quick reverse engineering tips?

Depends how accurate you need to be. Putting flat part on a scanner and scanning them as .pdfs is a low tech way of "digitizing". You can convert the .pdf to a .dxf and bring it into CAD.

A CMM is very expensive and most use a proprietary programming language. Something like a Faro arm would be cheaper, but still pricey. For high accuracy you can send things out to be laser scanned.

Another option would be scanning with a probe on a CNC machine. Great for hole locations, kinda clumsy for profiles or 3D work.

Or the old school methods of transfer punches or measuring between gage pins for hole locations. Again won't help you with profiles or 3D.

Can you recommend a scan to cad software? It would work for most of what I want to do but I have found that most leave lots of artifacts and are poor at making smooth arcs. I tried Mastercam X4's raster to vector with poor results and a few years ago I tried Scan2CAD but didn't really figure it out nor did I need to at the time. Might give that a look again.

It really depends on how complex the parts are. The more complex the more creative you have to be and the longer it takes unless you invest in sophisticated tools. Generally we buy tools to simplify and increase productivity.

I RE'd these castings and created solid models with nothing more than a surface plate, a height gauge, a digital caliper, some gauge blocks, radius gauges and cad software. It took allot of triangulation to get the hole and bore positions. It was allot of work and would have been much simpler to dial them in on a mill and record the coordinates.

I RE'd an entire differential for the original Ferrari Testa Rossa prototype using most of the tools mentioned above however I couldn't figure out a way to do the cam profiles so I sent them out to a vendor for CMM. I imported the data into my cad software created polar duplicates and compared flipped profiles to remove any anomalies since the profile should have been symmetrical from the center lines of the lobes and valleys. Once I had the cams modeled it was easy to measure them against the models for accuracy.

My point in all this is you have to be creative and wrap your mind around the job. I did an apprenticeship studying under the famed Swiss Designer Jo Marquart in the UK and I was absolutely amazed at what the guys in the shop could do with the least amount of tools. It really is true, "Necessity is the mother of invention".

We live in a time when we have so many advanced tools to prevent the workforce from thinking. It really is a shame how much is being lost. If we ever have an EMP who is going to dig us out of the mire and get industry moving again? I'm no different, I know I've been dumbed down by all the tools and gadgets. I see my lack when talking with experienced manual machinists. I've seen some of them do stuff that would be difficult to program on a CNC. Unfortunately they are becoming few and far between. Thankfully we have a hoard of them here, now all I have to do is ask questions and listen.

disclaimer: After reading my post I wanted to make sure all you CNC machinists out there don't think I'm disrespecting you. If your good at what you do I admire and respect you. If your not good at what you do and your trying and learning I admire and respect you too. My comment was based on a computer less society from the aftermath of an EMP.

For me, it is always educational to listen to "old guys".

Chip
... I, for one, welcome our slide rule overlords...

If you do it commercially get a haimer 3D sensor and use it together with your DRO to find hole locations, centers, edges. Then, draw in a 3D CAD and check again against the part.

If you have to reproduce 3D freeforms you could generate a cloud of points on the surface. CAD can lock these points together, e.g. Rhino will do it. But that really is a PITA.

Like I said these are 2.5 axis parts for the most part. So far Ive got a height gage, some radius gages, some calipers of various sizes and a 6" with long jaws. I was also thinking maybe like an optical thing in the bridgeport spindle could be used to find hole centers pretty quickly. Like a microscope type thing you look through that has a cross hair with indicated marks that I could zoom in and out. I could use these to have an equal number of ticks on both side and then zoom to where they're precisely even.

Did I just invent something or does such a thing exist?

I dare-say that a centring scope could be fitted with suitable graticules and, possibly, zoom optics to come pretty close to your "optical thing". Not familiar with the machinist versions as they have always been a bit too pricey for my potential needs but I have used such things in the lab. Construction is virtually identical so it should be possible. I've got a (very) cheap microscope about the place waiting to be converted. That one came with a video camera attachment too which could be fitted in place of the standard eyepiece. I have seen references to inexpensive software allowing a video camera to be used in centring scope applications. Memory says hole finding and centre location was one capability illustrated. This would have been about 3 years + back so odds are things have improved since so the electronic method sounds a better way of going about things.

Some of the DRO boxes have the ability to send the readout to a computer either as a stream or when commanded to do so. This seems a bit less tedious than physically recording the data. This sort of locating is easier if you can just push things into alignment rather than having to wind the handles. Better still if the camera system can calculate real centre from a pretty close position instead of relying on you to get it just so. Sorting the data out isn't too hard. Actually an idea starter task for Lab-View wannabes. In retrospect I'm surprised there isn't an affordable commercial kit out there, £2,000 looks possible which for a CMM, albeit limited, would be a steal.

Clive

Johann Ohnesorg said:

If you do it commercially get a haimer 3D sensor and use it together with your DRO to find hole locations, centers, edges. Then, draw in a 3D CAD and check again against the part.

Click to expand...

+1 for the 3D taster, better than a test indicator and edgefinder. That was how I settled on doing it before I bought my CMM. I'f you are going to do a lot of this find a cheap working manual CMM, it will make life much easier. I found an 18 X 24 for $2500.

Hmm, this is the first I have heard of centering scopes but yea to find hole centers they would need little tick marks and the ability to zoom in and out. Part of me wants to just buy a chinese digital microscope and mount it in the center of the spindle and use it like an optical comparitor. The only problem is I wouldn't know that it is definitely on spindle centerline....actually now that I think about it it would all be off by the same amount so that may not matter. I couldnt use it to indicate work in of course, but to find coordinates using the DRO using one of the ole centers as a datum....yea. should work, right? I could lay the crosshairs over the computer screen.



Thoughts? Something like this: ThinkGeek :: Wireless USB Digital Microscope

I might have to give this a try. They only thing I would have to be sure of is that it is perfectly perpendicular with the table.

I could even use something I know is round and perpendicular to true it if need be.

I run a CMM for my day job (as well as program and run 3 axis VMC's). You'd be hard pressed to beat somebody who knows what they're doing, and knows the final product measuring the given parts and giving you a model using a CMM. I say knows what they're doing because it's very easy to give more useless information than is needed with a CMM (been there done that). That said, depending on your tolerances and the quality and shape of your parts, gauge pins with a test indicator, and a mill with a (trusted) DRO can and may give you the accuracy you need.

It's all in context though. Multiple tools can be required to give you the accuracy you need. You might find all holes with the DRO, but know a certain group of holes has a tighter relationship, so you can use hard gauging (calipers, mic's, gauge blocks/pins, etc...) to further check, and double check the size and relationship to one another. A solid understanding of the part, and it's function is needed in order to get the information you require.

An example is a hole that measures at 1.013" from another hole on the DRO, and another hole it's in relation too, but you you know the pitch is supposed to be 1.000" you wouldn't create your geometry at 1.013" and then apply your manufacturing tolerance to the measured hole, you'd apply it at the "nominal" hole location (1.000"). It's a delicate game of looking at, and interpreting the numbers.

I'm pretty quick and good with a CMM, and if there's one thing I've learned is that the raw numbers only tell one side of the story. Designers of parts are lazy (I know, I are one from a past life). We did things to whole numbers unless necessary by part design, so if a part measures a weird number, double check it's right. An open line of communication is paramount between you and the customer (isn't it always?)

Most of the reverse engineering jobs I've done have been motorcycle parts to keep my race bike running straight and true (and fancy). A typical rundown of me reverse engineering a part, has me using the CMM for positional relationships, gauge pins for hole sizes, and telescoping gauges for bore sizes, and mics', calipers for stock sizes. Another handy tool which I've used before is a flatbed scanner. I've printed out a transparancy with a 0.25"grid. I've only used this for gaskets (i have no doubt it could be used for more.) I scan the part with the grid sheet on the bottom, then bring it into autocad as a jpeg, and scale it properly until the grid is the same as the transparent (can't remember but mine was a bit over .25"). Then I trace over the image with lines and arcs until I have defined geometry. Your eye is the best tool on the job. If it isn't, train it till it is. When I was done creating the geometry I printed it out and glued it to a sheet of gasket materiel then cut it out. Worked better than expected. Gasket aren't a high tolerance thing, but after using that method I wouldn't doubt I could hold less than 0.005" (probably much tighter) using the scanner and a transparent grid. I just haven't had the need to try it again.

I like doing this sort of work, as it's breaks up the monotony of checking things, and feeding machines. It's also gets the brain working, and I find it's kind of interesting. If you were closer I'd be in touch as I have a feeling I'd be able to offer a competitive alternative to the mill and DRO, but the border and shipping presents a whole 'nother set of problems.

Best of luck, and If I can be of any help, post up. Sorry for the long rambling post. I tend to do that.....

A digicam on the spindle system was written up in the UK magazine Model Engineers Workshop 3 or 4 years back. With a follow up improvements article. As I recall things alignment to centre was said to be fairy straightforward and the whole thing worked pretty well. Could probably find the articles if you are really interested.

On reflection one of the cheap Chinee CMC router kits might make a good starting point for a home brew CMM. Put a camera, proximity sensor or touch sensor on the bridge and set it scanning. Hole detection should be pretty easy if you arrange a good contrast between work-piece and background. Heck I wrote a routine for a similar job back in 2001 using the LabView vision analysis stuff so if I can do it.... Some of the not too expensive proximity sensors have a useful measuring range for smaller jobs, coupla inches or so. The still unused Taig "toy" CNC mill I got back in 2004 (seemed a good idea at the time but don't ask!) came with a touch probe unit allegedly useable with Mach 2 for profiling. Simple micro switch thing working in drive the axes till it touches and switches then pull away mode. Guy who sold it to me said you could get drawings out of the point cloud. Dunno how. (But maybe he just wanted his £8,000 for the new machine and his suggested software. Disappeared PDQ.)

Clive

Reverse Engineering

Go to the Tormach web site and under products click on scanner. There is a video. See if it is anything you can use.

Ernie

Who's power tongs are you reverse engineering?

For hole locations why not just use a test dial indicator and find the center of the holes on your machine? Been there, done that, and I can't think of anything faster that delivers the same level of accuracy.

TurboFabSupply said:

Thoughts? Something like this: ThinkGeek :: Wireless USB Digital Microscope

I might have to give this a try. They only thing I would have to be sure of is that it is perfectly perpendicular with the table.

I could even use something I know is round and perpendicular to true it if need be.

Click to expand...

Works very well for flat parts. Don't even bother mounting it in the spindle, attach it to the side of the head.
Tape two pieces of thread to your screen for a crosshair.

You don't need to be on center to find hole locations, pick two edges in X, average them for the X value, do the same in Y. If you need diameters I'd suggest picking three points and putting the data into a CAD.

Do not touch the zoom once you have started and take points only using the intersection of the crosshairs. These cheapo lenses have significant distortion off center.

If you have to move in Z then you have got to align the optical axis in X and Y which can be pretty tedious and on a unit like this with a movable lens it would most likely have to be done every time you remount the camera.

Two considerations are field of view and how accurately you trust your machine to be square. You will need to be in the range of a .250 wide picture to get within a thou with a megapixel color camera.

Mounting one of these on a B-port will make you appreciate how much a table "kicks" when you reverse directions. Just like an optical comparator you should take all points moving to the crosshair from the same direction.
Bob

4GSR said:

Who's power tongs are you reverse engineering?

Click to expand...

A new customer brought it to me, they mentioned it was used in the oil industry. There were no identifying marks so I searched the web until I found pictures to compare. Nearest I could tell it was made by Weatherford - Oil Country. I still don't know what they are used for - some kind of power wrench I suspect.

I burned allot of midnight oil on that job - it was rush rush rush. I built solid models and 2d drawings in 5 days while still working the normal day job. It was a tough project with weird angles, draft, split parting lines and allot of blends. The blend intersections were so complex it was really pushing the limits of my modeling software. I had to do allot of the blends subtractively. When they gave the drawings to the pattern maker he didn't want anything to do with them and insisted on making the patterns from the original castings. The customer was living in the dark ages, I couldn't get him to understand how simple and quick it would be to have the patterns made from the models. The customer was out of his element and I knew the job was heading for problems so I stepped away when they wanted me to do the machining drawings, I didn't even charge for my time. About a year later I did hear about the problems from another vendor and how they had to weld the castings up and rework the patterns several times. That was back in 2006 and I've never heard from them again - no doubt I was blamed for the problems. I think people have a tendency to blame who ever isn't around to defend themselves and what they can't understand.

Dale,

Yea, that is from one of the companies that Weatherford bought in their buying spree of the late 90's. I remember us borrowing one to make up and break out some premium connections way back when i work there on some equipment we were testing.
Those power tongs would "slap" you in the face if you wasn't paying attention to what you were doing!

Have fun!

Ken

4GSR said:

Those power tongs would "slap" you in the face if you wasn't paying attention to what you were doing!

Click to expand...

Lets talk the reality of the drilling platform. Tongs are used to "make and break" the screwed together connection of a length of drill stem. The floor of the drill rig has a segmented tapered collet (imagine a TG100 with a 6" bore). The drill stem is either being lifted out of the hole (or returning back into the hole). The joint is brought to about waist high and a new length of pipe is threaded into the threads, and the tongs are set to hold the loose drill stem against turning while the Driller (the guy operating the draw works) spins the collet to tighten the joint.
The tongs are free to move around quite a bit, but the outside end of them has a large hole through which a 1/2 to 3/4" wire cable is threaded through it with 'U' able clamps. The other end of the cable is secured to some portion of the rig itself.
As long as everything is working properly, the spinning collet runs the threads onto the non-moving drill stem (held by the tongs) and when complete, the tongs are removed, and the drill stem is lowered until the next joint is ready for another length of pipe.
Now, lets let things get a bit sloppy as things wear. The cable starts to fray, and someone puts some tape on it to prevent having the broken ends poking into you. You are running behind time and the Driller is in a hurry. He speeds up and slams into the joint make-up, really jerking the cable again and again.
Remember that the joint face is waist high, and that there are three guys working at the roundtable.
The driller gives one last pulse to the turntable, and the cable breaks!
The tongs are themselves very solid pieces of steel with the arm about 3-4 feet long, and the cable is another 5-10 feet long that has just become a massive weed whacker.
Back in 1953, when I was doing this stupid shit, we were "offsetting" another rig about 1/4 mile away, and we saw the lights flicker on and off repeatedly (a signal for help).
I will never forget the scene that I saw when I got up onto the rig floor. Body parts were everywhere, and the only persons still alive were the Driller and the man up in the draw works. A few days later, I got my leg amputated above the knee, and I decided that the oil field was not for me.

The above scenario was the way things were done back then. I really don't know what they are like nowadays.

Some of this led to OSHA.

Lee (the sawe guy)