Thread: Restoring a Jones and Shipman 520 cylindrical grinder II

Michael, I'd respectfully suggest that it doesn't matter whether it's a lathe or a CG, the principles remain the same, and no I wasn't thinking of a lathe (and incidentally you can run a lathe tool way off centre and it will still turn accurately). If you're machining a round piece of stock it doesn't matter how that material is being removed, if your TS and HS are different heights you will get either waisting or simply taper. They can be above the wheel, below the wheel, or on wheel centre, but they must remain the same. If you try to compensate incorrect height with offset, you'll finish up in all sorts of grief and a machine that will grind accurately at only one job length. That height may vary from a static position (ie they're simply the wrong height, and if your centres are the same diameter that's an easy check. However it may also be more of a dynamic situation where the heights vary at different lengths due to bed twist.

I'd suggest that putting a business card in like you have is far too coarse for this type of check. You would need to do a much finer check than this, and as RC has suggested, get a ground rod and put it between the centres and indicate it roughly around where the wheel would contact. Use an elephant's foot tip and indicate both the height and lateral position of the rod as it's traversed past. Trying to do it by grinding finished work will be very difficult without more fundamental checks.

G'day Michael,
Firstly, I love the work you have done to date, your attention to detail making the billet tail stock is a credit to both your ability and aesthetic appreciation.

My feeling is that your grinder would be more sensitive to alignment issues because of the small diameter of the grinding wheel. If the workpiece moves from above to below the grinder spindle plane as it traverses there will be more deviation in diameter with a small wheel than with a large wheel.

You mention that when you got the waisting down to 3/10ths you had a thou taper which could be fixed by tweaking the table angle. That indicates to me an error in both the X & Y alignment between the head and tail stock allowing the work to traverse across the grinding spindle plane causing the wasting while also moving into (or away from) the spindle causing the taper. The fact that both effects occur simultaneously means that they will also interact and compound the error and resulting effect.

I don't think the work centre being above the spindle centre matters as the material removal occurs tangentially to the grinding wheel. The problem is that the X & Y planes of the grinding spindle are not parallel to the X & Y planes of the work.

When you shim the tail stock, are you shimming both the front and back (grinder and operator sides) so the base remains parallel to the table thus only changing the Y position leaving the the X position unchanged? If so try shimming just the front or back so you get an offset in both the X & Y position. This will still be a crap shoot but short of indicating a test mandrel might be worth a try.

Actually in an earlier photo you used a cylindrical master, can you mount that between the work centres to indicate off?
Keep up the good work.
Cheers,
Greg.

Originally Posted by kwijibo99

My feeling is that your grinder would be more sensitive to alignment issues because of the small diameter of the grinding wheel. If the workpiece moves from above to below the grinder spindle plane as it traverses there will be more deviation in diameter with a small wheel than with a large wheel.

Yes I'd agree with that Greg, good observation. If you had a CG with an infinitely large wheel, variations in centre height wouldn't matter, but as the wheel gets smaller the variations would increase. It's why one should always use a round probe to indicate a flat part, but a flat probe to indicate a round part. Hence my comment about using an "elephant's foot" probe above (I simply made one, they're easy to make) as they will isolate the axis you're trying to measure. However on second thought, in this particular instance it may actually be better to go against that doctrine, and indicate with a conventional round probe. It would be hopeless to try to get measurements from, but may better indicate that there is a problem in that area.

I've been very busy, so didn't see that cylindrical master, but that's also what I was suggesting above. The issue with doing so is that it's possible to get a combination of height and offset that will indicate relatively accurately, but will not grind consistently. The reason is the combination is only good for that length of standard, and when the length of the centres changes, so will the taper. It's similar in some ways to offsetting a lathe's tailstock to grind a shallow taper. That's fine, but if you change the length of the work, the taper will change.

At the end of the day, it may well be that the machine is deemed close enough and you just run with it. However when head scratching occurs I personally believe it's best to follow a logical fault finding process, hence why I keep harping on about dragging out the Schlesinger book and simply going through the steps specified in the order specified. It will almost certainly be much faster than just trying constant crap-shoots to try to nail the problem.

Originally Posted by Pete F

The issue with doing so is that it's possible to get a combination of height and offset that will indicate relatively accurately, but will not grind consistently. The reason is the combination is only good for that length of standard, and when the length of the centres changes, so will the taper. It's similar in some ways to offsetting a lathe's tailstock to grind a shallow taper. That's fine, but if you change the length of the work, the taper will change.

G'day Pete,
I'm not sure I understand what you mean here, if the table ways to which the work centres mount are running parallel to the spindle (set for zero taper) which can be checked with an indicator, and a parallel test bar mounted between the work centres shows no deviation in the X and Y plane on an indicator (stationary relative to the grinder spindle) when traversed does not this verify that there is no offset in either plane of the tailstock relative to the head stock or the grinder spindle? Given this, any subsequent repositioning of the tail stock to accommodate longer or shorter work should not introduce any error. This assumes of course there is no localised damage or warping of the dovetail to which the tailstock is anchored or the ways on which the table itself travels which again can be checked with an indicator I would think.

The only error a single iteration of this test would not show is if the tail and head stock bores are not coaxial. If this is the case then the amount to which the centre point itself protrudes from the tailstock will change the offset, this too might be something that is contributing to the variation Michael is seeing. Performing the test with the tailstock centre fully withdrawn then repeating it with the centre fully extended should however reveal any error.

I think the problem Michael is having is that he is trying to reproduce a tailstock that is precisely coaxial with his existing head stock and this is no mean feat. I imagine that the original tail and head stock would have been mounted together and line bored at the same time thus ensuring the bores were coaxial but when trying to reverse engineer this I don't think you have any choice but to fudge things until they work.
Cheers,
Greg.

Yes that's correct, and exactly my point.

This assumes of course there is no localised damage or warping of the dovetail to which the tailstock is anchored or the ways on which the table itself travels which again can be checked with an indicator I would think.

That's a big assumption, especially on a used machine of unknown pedigree. It's why, particularly when issues arise, that it's generally best to go back and do the tests in logical order, as then you're not assuming anything. IF, the bed is not warped, and IF the ways are parallel to table movement, and IF the headstock spindle is parallel with the bed/table movement, etc etc etc. then if you stick a ground bar between centres you should be able to accurately check for tailstock offset, and move the TS won't affect things.

Now in fact it's unlikely that the errors would align such that you could get a theoretically perfect result, it could even be impossible, I'm not sure. However what you may find is that you're down in the level of measurement error at that work length. However since you're just compensating for errors and not correcting them as they're found to occur, they become obvious at a different work length. Hopefully that makes sense.

Edit: sorry just to reiterate. Most of the time the above assumptions are possibly fair enough, especially if you were dealing with a new machine of quality pedigree. However when things just aren't making sense, then it's generally far more efficient, at least in my experience, to go back and check the basics are right and never assume anything.

Managed to get out to the shed this evening to test the Greg/ Pete theory. I'd fogotten about the cylindrical square but then was not even sure that it would fit. It does (just), so I guess that means I don't need a test bar (at least not immediately). One annoying thing I found is that I have an elephant's foot contact point, but only for imperial indicators. I have a couple of micron level indicators but not the feet. I may have a tenth's indicator but I'm not sure
P1030646 (Medium).JPG
I put a bit of tape along the side of the cylinder to give myself a bit of an indication of position.
P1030647 (Medium).JPG
With the indicator in the position shown, from the tailstock in 20mm increments (measurements in thou, with sub thou increments estimated) - 32, 31.75, 31.5, 31.25, 31, 31, 31. So the TS is high, (only by a thou) but there is a dip in the bed?
With the indicator mounted horizontally and 11 thou of shim, the readings were 3, 2, 1, 0, 99, 98.25, 98. With the shim removed, the readings were 5, 5.25, 6.5, 7.5, 8.75, 10, 11, so horizontally about 5 thou of shim should get me parallel?

I was hoping to leave scraping and all that stuff until the summer refit bit it looks like I will need to find out what is happening before then. It's a small bed and the ways are all of around 12" long so it should not take too long...

Michael

Michael I've made a dozen or so indicator tips and they're very easy to make. Normally I've made round ones, but the elephants foot is no harder. Just turn one up on the lathe. The only difference with the elephants foot is to make sure the thread is cut concentric so that when its screwed in the face is truly perpendicular to the indicator shaft. I've found just accurately chasing them with a die is more than close enough. I've never worried about hardening them, but there's no reason they couldn't be made from silver steel and hardened. I just lapped the elephants foot and polished the round ones, and they're difficult to tell from a commercial version once blackened.

Let me know if you need any levels etc as I expect to be over there in the next month. Is that bed hardened? I'm assuming it's not and if so it wouldn't take long to scrape a bed like that at all and, as I think you've gathered by now, I feel it would be worth your while Grinders lead a hell life and their beds take a hammering. All surfaces in fact. Personally I would indicate the table movement and confirm the errors follow in the table. Since the table hasn't yet been proven, it may not be conclusive, but if the errors follow there's a good chance it's the bed. The table should be indicated anyway so it's not wasted time, just all going backwards.

Originally Posted by Michael G

Managed to get out to the shed this evening to test the Greg/ Pete theory. I'd fogotten about the cylindrical square but then was not even sure that it would fit. It does (just), so I guess that means I don't need a test bar (at least not immediately). One annoying thing I found is that I have an elephant's foot contact point, but only for imperial indicators. I have a couple of micron level indicators but not the feet. I may have a tenth's indicator but I'm not sure

I expect nothing less then a supramess to be used

The silence over the last few months has been me thinking about Pete's advice. I finally started doing something about it and read Connelly to see how to approach this. In a separate thread I've made up a spotting guide and will need to scrape up the bottom of the table to use as a spotting guide for the machine body.(//metalworkforums.com/f65/t199887-fun-horizontal)
To start with though I needed to scrape up my camel back (actually, more of a swayback... - see the picture below). I'd done so once to a borrowed straight edge, only to find that that had a bow to it - so I had a beautiful scraped finish on a not flat straight edge. I got that machined back to flat but have always wondered what to scrape it to as it is slightly longer than the diagonal of my surface plate.
I bit the bullet and decided to scrape to what I had, alternating ends on the plate (it was really only the last 100mm of 600 that was in free space). The biggest problem with this straight edge is that it has some porosity in it - not bad but at the same time annoying.
P1030696 (Medium).JPG
Finally I got that to a stage where it was reasonably uniformly spotted and went on to check the underside of the table. While it did have a banana to it, it was not too bad. The biggest issue was that it had worn enough that (as per the photo) there was a ridge in the centre of the flat way from the groove in the middle of the corresponding machine way.
P1030697 (Medium).JPG
According to Connelly, I now have to alternate spotting with my guide and a straight edge to get the table good enough to spot the machine base and then do the machine base before going back to the table...
From the preliminary work that I've done, the flat way on the machine is relatively flat (intuitively, the table moves from side to side so the middle (in continuous contact) will wear more than the ends (intermittent contact). The base is in contact all the time so I would expect it to be relatively flat because of that contact.

Michael

Getting there slowly. Last post I said that the flat way was reasonably flat. Not so the V way. The first photo shows it at first blue - showing spots on the end but that's it. After ten or so cycles, I'm finally starting to get the spots extending (second photo). This is all with a straight edge - my V spotter shows that the angle is pretty good but there is no point in perfecting the angle if the V is worn in the middle.
P1030724 (Medium).JPG P1030728 (Medium).JPG
The blue I'm using is a commercial type but when it gets thin it fades to nothing so trying to get better spots per square inch gets difficult. Four years after my appearance at Scrapefest I've finally made up a pot of Marco blue to try so hopefully tomorrow I'll see if that is any better.
Stuffing around with grease is messy. Stuffing around with paint is messy. Anyone want to guess what mixing paint into grease is classified as?

The other thing I have discovered is that the lights in my shed cast shadows so another diversion that I have going at the moment is a frame to mount the two spare LED fluros from the batch I bought that can be suspended and provide even lighting (I hope). Normally it will live over the SG but I want to put up hooks in several places so that it can be moved where needed for things like scraping.

Michael

Nice work Michael. How are you testing to see if your Vee is even on both sides and parallel to the flat? Ive always wondered about that part. Or are you justing working on getting it flat for now?

Cheers

Simon

Sent from my SM-G900I using Tapatalk

Most of the tools required for the J&S are standard sorts of things (Whitworth spanners for example) but to remove the hub from the spindle and tighten the wheel nut on the hub arbor, a couple of pin type spanners/ wrenches are required and I can't find them from my usual sources.
The answer is (of course) to make them, which I spent the last shed session doing. The small version is for the nut holding the hub onto the spindle (that has square slots for the spanner lugs) and the larger is a pin wrench for the wheel nut that secures the wheel on the hub (Yet to have round pins fitted).
P1030765 (Medium).JPG
Machining was done on the newly completed false table for the R/T (separate write-up coming)
P1030763 (Medium).JPG
Now this diversion is done it's back to the scraping.

Michael

To answer Simon's question, I was just trying to get to flat but now I'm doing the Vs. According to Connelly, the way it is done is to make up a spotting template and then alternate between the template and a straight edge. That way you know the V is the right angle and is also straight. In practice I found that getting the V right was pointless if it was not straight, so scraped the V ways until flat and then worried about getting the V angle right.
My spotting guide is only around 4" long - typically they are 6 to 8" but it is only a small machine. You may be able to see how even it is with the second photo (and then again, maybe not).
P1030776 (Medium).JPG P1030775 (Medium).JPG
Once you have the V good and the flat relatively flat, you use a DTI to check that the flat is the same distance as the guide is pushed along the V. Connelly suggests better than 2 tenths variation, but I found a change in hand pressure as it was pushed along was enough to give much more than that so I'm not really sure how skewed the two surfaces are. I think they are alright as the mean of the needle flicker stayed about the same.
P1030778 (Medium).JPG
The mating part is done in a similar fashion, except that the table is alternated with a straight edge. At the end of play today this is where I am up to. The first photo is with the straight edge and the second is with the table after a cycle.
P1030779 (Medium).JPG P1030781 (Medium).JPG

Michael

I was reminded today that I had this thread hanging, so thought I would at least finish it off. While the grinder is not completely finished, the remaining bits are primarily cosmetic/ ease of use stuff (like paint), so probably not worth posting. However, this is the clutch mechanism and as Ash had a hand in the design (//metalworkforums.com/f65/t1965...60#post1880060), I thought it only proper to post something.
The basic 'working bit' looks like this
P1030961.JPG
In words, there is a sleeve that slides on a rod up the top and it supports a clevis, which has two phos bronze shoes in it that will pull on the clutch cone. Force is applied with a couple of rod ends, one on the clevis and one on the end of the clutch lever. To get the lever swinging in the right way it runs in a ball joint so that it can be moved past a retaining hook and lock the clutch in the disengaged position.
The ball was turned with my radiusing tool. I started with a length of bar stock of the finished size (diameter 40mm) and drilled a through hole so it could be mounted on a mandrel. It sits in a pair of washers with a 45 degree chamfer in them. To line the holes up for drilling through I used a gauge pin and a hose clamp. The ball was secured to the lever with a roll pin - not strictly period, but then the rest of the top section isn't anyway.

P1030946.JPG P1030947.JPG P1030965.JPG
The shoes were bored to the correct radius by simply mounting them in the clevis and then boring them out to size. The first photo shows the incredibly precise set up used while centreing on the faceplate. I did have a plan B if that did not get the result I wanted but it was fine
P1030939.JPG P1030940.JPG
The lever uses an old knob that arrived with my lathe (it was the wrong size, colour and thread type for the lathe so it was changed). It is set up so that the lever is off to the right when the grinder is in use but comes across when the clutch is disengaged as a visual reminder.
All up it looks like this -
P1030958c.jpg
Michael

Good to see my ideas helped. You do some good work, better than some of the machinists I know.