13 ms·

> it took two months before the workpieces were acceptable. For a new manual Bridgeport knee mill, it may be more than two months before the machinist who runs

> it took two months before the workpieces were acceptable. For a new manual Bridgeport knee mill, it may be more than two months before the machinist who runs it is happy with it. >There's a lot of money riding on machinery like this, imagine QA at the recipient of a load of parts deciding that they're worthless scrap How is that even possible for a competent/honest machine shop? As if the shop doesn't inspect its output. Is there some sort of CNC-vending-machine business model that I haven't heard of where clients email their CAD files to the machine and next week a freight truck arrives with the stuff?

jacquesm 12y ago
> How is that even possible for a competent/honest machine shop? Because measuring things is hard. > As if the shop doesn't inspect its output. Of course they do. And even then it can still happen that what passes inspection at the shop (especially in early runs) does not pass muster with the recipient. Which is why you don't want to be moving your equipment around if you can help it. Depending on the precision required it really doesn't take much to go from 'pass' to 'scrap' and it is not rare at all for the incoming inspection to reject parts that were accepted at the production facility. This is logical because there is money on the line. > Is there some sort of CNC-vending-machine business model that I haven't heard of where clients email their CAD files to the machine and next week a freight truck arrives with the stuff? You seen to have missed the internet revolution in machining, but yes that's exactly the model these days. Only I'd replace 'machine' with 'machine shop' or 'production facility'.
- fnordfnordfnord 12y ago
  >Because measuring things is hard. Harder than making things? No, absolutely not. >And even then it can still happen that what passes inspection at the shop (especially in early runs) does not pass muster with the recipient. That's a communication problem. Happens often enough, but DRM does nothing to fix it. >You seen to have missed the internet revolution in machining, but yes that's exactly the model these days. I know that may be what people want but that's still a fantasy for precision machining. I don't think anyone is near the point that you can fire the machinists and simply hire someone to sweep up the tailings, refill the oil tanks, and slap shipping labels on things, (and then I suppose cash the checks from your mountain chalet).
  - jacquesm 12y ago
    > Harder than making things? No, absolutely not. Measuring a shape in 3D with a given accuracy is immensely harder than making a shape in 3D with that same accuracy. Just one example that might convince you of this: Take any old lathe, and a micrometer. Run the lathe, cut yourself a nice chunk of hss shaft to 25 mm diameter, precision 1/100th of a mm, cut it off at 25 mm long. Now try to imagine all the ways in which that object that you just made can be non-cylindrical, in which the planes can be non-parallel and what kind of other deformities such as tip-radius artifacts there might be. If you're going to use your trusty micrometer to measure it you will not be able to detect much if any of that. It will take a very expensive setup to be able to determine within 1/100th of a mm what the exact shape of that thing you made is. > That's a communication problem. Happens often enough, but DRM does nothing to fix it. True, but specifically 'machine has not been tampered with' can make such communications problems a lot easier to solve. Having witnessed exactly that scenario here recently I'm fairly convinced that without the gyro to prove the machine had not been tampered with that it would have been a lot longer discussion, possibly even a court case. This happened post delivery and the machine was for interesting reasons unable to reach the advertised accuracy. It took months to get that one solved and the bill caused by the downtime was substantial. > I know that may be what people want but that's still a fantasy for precision machining. I don't think anyone is near the point that you can fire the machinists and simply hire someone to sweep up the tailings, refill the oil tanks, and slap shipping labels on things, (and then I suppose cash the checks from your mountain chalet). Who spoke about firing the machinists? The thing is that between the design department and the machining center there is a series of well defined steps and then a whole pile of machine specific knowledge to get a part to come out right. But the whole chain is digital now, in the interest of repeatability, including tool management, stand times, deformation correction and so on. It's not rare at all for client and supplier to be continents away from each other and they're definitely not going to visit each other in order to get some part produced. This also means the market got a lot more competitive which means that inbound quality control is a very important factor in keeping everybody in the chain reliable. Unattended operation including auto-feed has been done for years and it's not rare at all to see one operator/machinist run 5 machines.

jacquesm 12y ago

> How is that even possible for a competent/honest machine shop? Because measuring things is hard. > As if the shop doesn't inspect its output. Of course they do. And even then it can still happen that what passes inspection at the shop (especially in early runs) does not pass muster with the recipient. Which is why you don't want to be moving your equipment around if you can help it. Depending on the precision required it really doesn't take much to go from 'pass' to 'scrap' and it is not rare at all for the incoming inspection to reject parts that were accepted at the production facility. This is logical because there is money on the line. > Is there some sort of CNC-vending-machine business model that I haven't heard of where clients email their CAD files to the machine and next week a freight truck arrives with the stuff? You seen to have missed the internet revolution in machining, but yes that's exactly the model these days. Only I'd replace 'machine' with 'machine shop' or 'production facility'.

fnordfnordfnord 12y ago
>Because measuring things is hard. Harder than making things? No, absolutely not. >And even then it can still happen that what passes inspection at the shop (especially in early runs) does not pass muster with the recipient. That's a communication problem. Happens often enough, but DRM does nothing to fix it. >You seen to have missed the internet revolution in machining, but yes that's exactly the model these days. I know that may be what people want but that's still a fantasy for precision machining. I don't think anyone is near the point that you can fire the machinists and simply hire someone to sweep up the tailings, refill the oil tanks, and slap shipping labels on things, (and then I suppose cash the checks from your mountain chalet).
- jacquesm 12y ago
  > Harder than making things? No, absolutely not. Measuring a shape in 3D with a given accuracy is immensely harder than making a shape in 3D with that same accuracy. Just one example that might convince you of this: Take any old lathe, and a micrometer. Run the lathe, cut yourself a nice chunk of hss shaft to 25 mm diameter, precision 1/100th of a mm, cut it off at 25 mm long. Now try to imagine all the ways in which that object that you just made can be non-cylindrical, in which the planes can be non-parallel and what kind of other deformities such as tip-radius artifacts there might be. If you're going to use your trusty micrometer to measure it you will not be able to detect much if any of that. It will take a very expensive setup to be able to determine within 1/100th of a mm what the exact shape of that thing you made is. > That's a communication problem. Happens often enough, but DRM does nothing to fix it. True, but specifically 'machine has not been tampered with' can make such communications problems a lot easier to solve. Having witnessed exactly that scenario here recently I'm fairly convinced that without the gyro to prove the machine had not been tampered with that it would have been a lot longer discussion, possibly even a court case. This happened post delivery and the machine was for interesting reasons unable to reach the advertised accuracy. It took months to get that one solved and the bill caused by the downtime was substantial. > I know that may be what people want but that's still a fantasy for precision machining. I don't think anyone is near the point that you can fire the machinists and simply hire someone to sweep up the tailings, refill the oil tanks, and slap shipping labels on things, (and then I suppose cash the checks from your mountain chalet). Who spoke about firing the machinists? The thing is that between the design department and the machining center there is a series of well defined steps and then a whole pile of machine specific knowledge to get a part to come out right. But the whole chain is digital now, in the interest of repeatability, including tool management, stand times, deformation correction and so on. It's not rare at all for client and supplier to be continents away from each other and they're definitely not going to visit each other in order to get some part produced. This also means the market got a lot more competitive which means that inbound quality control is a very important factor in keeping everybody in the chain reliable. Unattended operation including auto-feed has been done for years and it's not rare at all to see one operator/machinist run 5 machines.