48" Featherweight Camelback Machining Video

[dgfoster]

Here is a link a PM member who wishes to remain anonyous sent to me.

Machining a 48 Foster Featherweight

It shows a machinist (not a PMer) using an HBM to mill one of my 48's.

He makes some interesting observations regarding spindle length growth of an HBM.

His workholding methods are sound. I like the tattle tale indicators he placed at each end of the casting to watch for any shifting. He might have benefited from use of simple wood wedges to prevent ringing. He did a good job of 3-pointing it and paid close attention to minimizing cutter pressure on the casting.

I have made special fixtures for 3-pointing that have some advantages but probably would not bother with them for a one-off. For the one-off, blocking the casting in would not have induced flexion but might have also improved rigidity.

I thought the ideas presented might be helpful to anyone doing a similar job.

Denis

 

[M.B. Naegle]

I wouldn't recommend doing all this extra stuff. Think about all the hours and hours..... and hours.... of scraping you're missing out on!

 

[Tyrone Shoelaces]

Thermal expansion of the travelling spindle. That’s your enemy.

Regards Tyrone

 

[dgfoster]

Tyrone (or anyone who is inclined to chime in),

Are all the HBM's pretty much constructed the same with respect to the spindle and its bearing support and thus its tendency to lengthen with temperature rise? I run a medium-sized Lucas, but have never had occasion to be "in" the spindle/quill assembly. I suppose it has 5 feet or so of extension, though I only use the first 6 or 8 inches. Where is the "fixed" point located with respect to the to the cutter and between that fixed point and the cutter which thermal expansion becomes a factor? How long does a mchine need to be run until the spindle settles down in length. Under simple free rotation vs under load would heating be expected to change significantly---so, just turning the spindle on and letting it freewheel should warm it up about as much as it will warm under light load?

I have had the rare occasion to make my roughing cuts and shut down for the evening to return ready to go with the finish cut in the morning. That may be the circumstance that might be most likely to cause issues I am concluding---right?

Denis

 

[Tyrone Shoelaces]

I installed a 5” spindle machine years ago. I did the usual alignment tests. Big 4ft square on the ways, ran the spindle frame up the square with a DTI. 0 to plus 0.001” at the top of the travel. Everything checked out.
The operator set up a large steel component about 5 feet high on the table and started milling the front face. He was going across and down, across and down. Roughly 4” passes with a 6” cutter. It was taking about an hour to complete the face.
A day or two later I was called back to the machine. The component was on the inspection table with a big square up against it. The square was touching at the top and you could get a 0.004“ feeler in at the bottom.
” You’ve not set the machine up correctly, the column must be leaning !”
I told them the column was fine but I checked it again just to be sure. It was just as I left it.
Cutting a long story short the issue was thermal expansion of the travelling spindle. It was “ growing by “ 0.001” every 15 minutes. I proved this by putted a blank in the end of the spindle, placing a DTI against the blank, and just leaving the machine running for an hour.

Regards Tyrone

 

[hmw1972]

Hello Denis,

It was me who machined the 48" straight edge and posted the video on youtube. First off I would like to say the straight edge was the nicest piece of cast iron I have ever machined. No hard spots and flycut like butter. You are correct in how the boring mill spindle is configured. The back end is held axially by a set of thrust bearings (the fixed point) and any growth (or shrinkage) is at the cutter end. The distance from the fixed point to the cutter end on my boring mill is approximately 48". At this length the spindle will grow (or shrink) .0003" for every degree Fahrenheit change in temperature. Knowing this it becomes obvious very quickly that temperature control (i.e. temperature stability) during the cut is critical. I think at the low power levels a flycutter operates at there would be very little change in temperature from freewheeling to cutting and therefore just letting it run for a while at the flycutting speed would offer the best chance at thermal stability. I read somewhere (before I got my HBM running) that machinists running HBMs would leave the more accurate work for later in the day after the mill had run for several hours and stabilized. I now see the wisdom in that.

Ken

 

[TGTool]

It makes you wonder if heaters on the spindle housing could be used to compensate for the expansion. In theory, the housing would be growing backwards as the spindle itself is growing forwards.

 

[Tyrone Shoelaces]

It makes you wonder if heaters on the spindle housing could be used to compensate for the expansion. In theory, the housing would be growing backwards as the spindle itself is growing forwards.

Well thought out.
One of those machines I was referring to was installed in an ROF ( Royal Ordnance Factory ) in Nottingham I think. They were having similar problems with the travelling spindle. The ROF told the makers “ either fix it or take it back “. That’s how the makers solved the problem. They had variable heating elements attached to the tail bracket ( spindle housing to you ) that they could tune to maintain some sort of equilibrium.

Regards Tyrone.

 

[dgfoster]

Well thought out.
One of those machines I was referring to was installed in an ROF ( Royal Ordnance Factory ) in Nottingham I think. They were having similar problems with the travelling spindle. The ROF told the makers “ either fix it or take it back “. That’s how the makers solved the problem. They had variable heating elements attached to the tail bracket that they could tune to maintain some sort of equilibrium.

Regards Tyrone.

Establishing a reference to detect variation from "zero" must have been an interesting problem. I could see some sort of laser on the machine head reading a distance to a refelector attached to the cutter. A single mirror that would intermittently rerflect the laser with each rev would be workable. What method did they use to a esatblish a feedback loop?

Denis

 

[Tyrone Shoelaces]

Establishing a reference to detect variation from "zero" must have been an interesting problem. I could see some sort of laser on the machine head reading a distance to a refelector attached to the cutter. A single mirror that would intermittently rerflect the laser with each rev would be workable. What method did they use to a esatblish a feedback loop?

Denis

I think it was more like “ trial and error “ to be honest. They were concerned with one particular boring job from what I’m told. I got the story from one of the makers service engineers years ago.
I solved the issue with a cooling unit that extracted the oil in the spindle frame, chilled it, and then returned the oil to the spindle frame. It worked well enough to satisfy the customer but it was never perfect in my eyes. The tolerance on the job I was concerned with was 0.002” in about 5 ft and we got it down to within that.
The component was like a big, thick, steel angle plate ( 15 inch base by 5ft tall ) and they were milling “ a lot off a little “ . IE milling the foot and then milling the face stood on the foot. I was always taught to “ mill a little off a lot “. IE mill the face first and then mill the feet off the face. They weren’t helping themselves really.

Regards Tyrone

 

[hmw1972]

I was wondering if you could lock the boring spindle in the milling spindle in the front using one of the collet take up devices (subject of an earlier thread, someone posted a scan of a G&L brochure advertising them for sale). The idea of that device was to take up any radial clearance between the two spindles and make a more rigid set up for milling. However, if you could lock them together tightly and then let the back end float axially in the spindle frame any growth would be 'out the back' and the cutting tool wouldn't move. The difficult part would be finding a way to make the back end float axially. Ironically on my Wotan you would have an easier time of this if the boring spindle leadscrew and nut were worn out. Then you could 'back off' the feed handle and leave the lost motion on the side of the thread flank that would let the spindle grow back. If the machinist was really on the ball he could periodically back off the feed handle during the cut to give the spindle more room to grow.

Ken

 

[Tyrone Shoelaces]

I was wondering if you could lock the boring spindle in the milling spindle in the front using one of the collet take up devices (subject of an earlier thread, someone posted a scan of a G&L brochure advertising them for sale). The idea of that device was to take up any radial clearance between the two spindles and make a more rigid set up for milling. However, if you could lock them together tightly and then let the back end float axially in the spindle frame any growth would be 'out the back' and the cutting tool wouldn't move. The difficult part would be finding a way to make the back end float axially. Ironically on my Wotan you would have an easier time of this if the boring spindle leadscrew and nut were worn out. Then you could 'back off' the feed handle and leave the lost motion on the side of the thread flank that would let the spindle grow back. If the machinist was really on the ball he could periodically back off the feed handle during the cut to give the spindle more room to grow.

Ken

I’ve tried all that. The answer is “ No”. The expansion has to go somewhere and the arrangement at the rear end of the spindle is set up not to deflect more than a thou or two.

Regards Tyrone

 

[dgfoster]

I’ve tried all that. The answer is “ No”. The expansion has to go somewhere and the arrangement at the rear end of the spindle is set up not to deflect more than a thou or two.

Regards Tyrone

I've been thinking about means of measuring the distance from a "fixed" portion of the machine to the rotating head. A mirror and laser would work. Another method that might be more easily setup could be use of a magnet attached (glued) to the cutting head and a coil attached to the machine in pretty close in/out proximity so that a Hall-effect voltage spike would occur in the coil. The voltage would vary as the square of the separation distance of coil and magnet and that would be very desirable. It also would vary depending on the rotational speed of the spindle. So,that would have to be taken into account if, say, the final cut were made at higher RPM than roughing cuts. The voltage spikes would need to be measured with an oscilloscope or equivalent instrument that could accurately measure peak voltage. Maybe a circuit could be designed using a suitable capacitor that would smooth out the impulses and make them more easily read with a simple voltmeter.

Another physical change that is greater than linear would be air flow from an orifice in close proximity to the revolving head. I would need to review this but IIRC flow would be proportional to the cube or fourth power of separation. This would be good from a perspective of making the change in distance have a large change in flow. But, I do not know how it would be practical to measure air flow with precision. Maybe use a simple water column manometer?

Other ideas? Thoughts?

Denis

 

[Tyrone Shoelaces]

I've been thinking about means of measuring the distance from a "fixed" portion of the machine to the rotating head. A mirror and laser would work. Another method that might be more easily setup could be use of a magnet attached (glued) to the cutting head and a coil attached to the machine in pretty close in/out proximity so that a Hall-effect voltage spike would occur in the coil. The voltage would vary as the square of the separation distance of coil and magnet and that would be very desirable. It also would vary depending on the rotational speed of the spindle. So,that would have to be taken into account if, say, the final cut were made at higher RPM than roughing cuts. The voltage spikes would need to be measured with an oscilloscope or equivalent instrument that could accurately measure peak voltage. Maybe a circuit could be designed using a suitable capacitor that would smooth out the impulses and make them more easily read with a simple voltmeter.

Another physical change that is greater than linear would be air flow from an orifice in close proximity to the revolving head. I would need to review this but IIRC flow would be proportional to the cube or fourth power of separation. This would be good from a perspective of making the change in distance have a large change in flow. But, I do not know how it would be practical to measure air flow with precision. Maybe use a simple water column manometer?

Other ideas? Thoughts?

Denis

Or buy a CNC Hor bore were you can programme out any error. It might turn out cheaper !

Regards Tyrone.

 

[Garwood]

Don't you guys have an outer spindle on your HBM's? If you've got tight work to do why not use that spindle and avoid the quill?

 

[Tyrone Shoelaces]

Don't you guys have an outer spindle on your HBM's? If you've got tight work to do why not use that spindle and avoid the quill?

The one I was working on was a 1970’s era machine. Built in facing slide machine,fabulous boring machines but not great when it comes to milling. You could only really use the travelling spindle. The later machines were much better milling machines but the facing slides were nowhere near as good. It’s swings and round abouts.

Regards Tyrone

 

[hmw1972]

Not long after I posted my last reply I realized I could get exactly the same effect by mounting the cutter on the milling spindle instead of the boring spindle, as Garwood mentioned above. In the case of my Wotan I would still have the entire length of the milling spindle that could grow on me, as the thrust bearings are located in the back of the headstock. Still, it would cut the growth by more than half.

Ken

 

[dgfoster]

Hello Denis,

It was me who machined the 48" straight edge and posted the video on youtube. First off I would like to say the straight edge was the nicest piece of cast iron I have ever machined. No hard spots and flycut like butter. You are correct in how the boring mill spindle is configured. The back end is held axially by a set of thrust bearings (the fixed point) and any growth (or shrinkage) is at the cutter end. The distance from the fixed point to the cutter end on my boring mill is approximately 48". At this length the spindle will grow (or shrink) .0003" for every degree Fahrenheit change in temperature. Knowing this it becomes obvious very quickly that temperature control (i.e. temperature stability) during the cut is critical. I think at the low power levels a flycutter operates at there would be very little change in temperature from freewheeling to cutting and therefore just letting it run for a while at the flycutting speed would offer the best chance at thermal stability. I read somewhere (before I got my HBM running) that machinists running HBMs would leave the more accurate work for later in the day after the mill had run for several hours and stabilized. I now see the wisdom in that.

Ken

Ken, thank you for the kind words regarding the casting. I do everything I can to maintain optimum machining and scraping quality of my castings.

I was talking with Forrest Addy last night and he reported using an oil circulation system at Bremerton Naval Shipyards to keep the spindle and quill at equal though not constant temperatures so that growth/contraction in the spindle was offset by the same growth/contraction in the quill. Setting up such a system would not be trivial but sounds elegant.

Myself, I intend to look at the application of a sensor system to detect and compensate for change in cutter stickout. It might not be that complicated.
Of course, it may also be unworkable. So much to do, so little time...

Denis

 

[sfriedberg]

Denis, and anybody else thinking about adding thermal compensation to a machine, don't overthink the sensing. A simple accurate temperature probe (or several mounted at different locations) is probably all you need for input. If you hold the housing and spindle/quill at a constant temperature, you do not need fancy magnetic or hydrodynamic micro-clearance sensors. The thermal mass of your machine (and probably less-than-ideal placement of both sensors and heaters) is going to prevent anything that remotely resembles dynamic control.
By all means, add oil coolers and similar solutions first. Establish how hot your machine gets when it is working hard for an extended period. Choose a temperature somewhat higher than that. Use your thermo probes, heaters and simple PID control to get your machine up to that chosen temperature and hold it. Needs at least a simple UI to tell you when it's stabilized. As your machine starts generating heat through work, the control will slack off the heaters.

 

[dgfoster]

Denis, and anybody else thinking about adding thermal compensation to a machine, don't overthink the sensing. A simple accurate temperature probe (or several mounted at different locations) is probably all you need for input. If you hold the housing and spindle/quill at a constant temperature, you do not need fancy magnetic or hydrodynamic micro-clearance sensors. The thermal mass of your machine (and probably less-than-ideal placement of both sensors and heaters) is going to prevent anything that remotely resembles dynamic control.
By all means, add oil coolers and similar solutions first. Establish how hot your machine gets when it is working hard for an extended period. Choose a temperature somewhat higher than that. Use your thermo probes, heaters and simple PID control to get your machine up to that chosen temperature and hold it. Needs at least a simple UI to tell you when it's stabilized. As your machine starts generating heat through work, the control will slack off the heaters.

sf,

I guess I was not thinking of such a complex system of measurement and heaters with feedback loops to establish a steady-state stickout of the cutter. All I really would like is a way to measure stickout with the machine running so that I could, first of all, simply observe length changes and measure them. Then it would be possible to compensate for them manually by simply adjusting the z manually during a cut or maybe just being able to determine when the stickout "settles down" to negligible variation during a cut and therefore know it is safe to continue the cut and not expect error introduced by spindle growth during the cut.

Heaters, feedback loops, PID or Arduino controls, and algorithms to keep a constant effective spindle length would sure be cool though.

Denis