A New Dividing Plate for my Dividing Head

[Michael GilliganParticipant @michaelgilligan61133]

Posted by John Olsen on 17/03/2015 02:00:43:

If you make a plate by any method that comes to hand, then use it with geared dividing head or rotary table to make another plate, the errors will be divided by the ratio of the worm. Practically this means that you could mark out the first plate by almost any means and the second plate would be pretty good. …

.

John,

You make an excellent point; and this method is entirely adequate for most general engineering work. [We live in wonderful times, when we can leverage the precision of mass-produced items to serve our craft.]

I think it worth just noting that in some fields [Astronomy and Surveying spring to mind], the requirements for accuracy and repeatability of circular dividing far exceed our own. … and their quest for perfection almost defines the concept of 'Diminishing Returns'. I don't want to labour the point here, but anyone interested could investigate "Periodic Error Correction" as applied to the tracking of telescope mounts.

MichaelG.

[Neil WyattModerator @neilwyatt]

> Practically this means that you could mark out the first plate by almost any means and the second plate would be pretty good. ..

The accuracy of the second plate would be largely limited by the accuracy of the gears. The maximum error would be:

max error in worm + max error in gear + (max error in plate/number of teeth in gear)

After one cycle of making plates, the error in the gear is likely to be the largest of these. The worm would have to be very poor to have a large error as you are always using the same short section of the thread, but a gear can have a significant maximum angular error between two teeth without it being very noticeable. The chart below shows errors in a hobbed gear I made –  a maximum of 1.1 degrees error!

This was made by indexing the gear around while attached to a digital level, my (rash?) assumption is that the level is more accurate than the gear.

Neil

 

Edited By Neil Wyatt on 17/03/2015 09:17:32

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Neil Wyatt on 17/03/2015 09:07:36:

… shows errors in a hobbed gear I made – a maximum of 1.1 degrees error!

…

This was made by indexing the gear around while attached to a digital level, my (rash?) assumption is that the level is more accurate than the gear.

.

A very rash assumption, I would say, Neil

[unless, of course, you have an extraordinarily precise digital level]

MichaelG.

[John OlsenParticipant @johnolsen79199]

Well, I just have to hope that the one in my Vertex is better than that since it is kind of tricky to check. I don''t even have a digital level. All I know is that the gears I have cut have run OK together for the sort of service I have required. Nothing all that arduous, just 20DP wheels for a gearbox for the Myford. Incidently I started that project because the thousand pounds or so for the real one sounded expensive…having finished the project it looks more like a bargain! People worry about gearcutting but that was the easy part of the project.

John

[Michael GilliganParticipant @michaelgilligan61133]

Posted by John Olsen on 17/03/2015 12:56:08:

… I don''t even have a digital level.

.

That probably puts you at a considerable advantage, John

MichaelG.

[John Stevenson 1Participant @johnstevenson1]

Posted by Chris Pattison 1 on 16/03/2015 19:33:09:

To have a plate with suitable hole pitches, I would start looking at having one made in MDF using a laser cutter. I would expect that these holes would be precisely spaced. This would then be used as a template to manufacture the plate in steel.

.

Perspex would be better as the quality is higher. Lasers just love perspex.

Nothing special, just an old file I found and translated to the laser. Right hand half is clean as the bit of scrap still had some of the backing paper on. LH half shows some discolouration due to the paper being missing but the quality is good and consistent.

Cutting time was about 4 to 5 minutes, didn't check as walked away to do something else.

[Neil WyattModerator @neilwyatt]

> A very rash assumption, I would say, Neil

Well the Wixey website says:

Resolution:	0.1 degrees
Accuracy:	+/- 0.2 degrees
Repeatability :	+/- 0.1 degrees

I'm happy to accept that mine is equally accurate. I can vouch for the repeatability and +/- 0.2 degrees is better than +/1 0.6 degrees.

Neil

[Mike PooleParticipant @mikepoole82104]

I thought this was a pretty clever idea

Mike

[John OlsenParticipant @johnolsen79199]

I was told once that a somewhat related technique was used to construct accurate leadscrews in the early days. They wound wire around long mandrel to form a kind of thread, and then cast a lead nut around it. This can apparently give a very accurate pitch since wire can be drawn in long lengths with a very consistent diameter. The diameter is determined by the size of the hole in the die, so provided that does not wear significantly while drawing the wire, the diameter will be constant. The pitch will only depend on the diameter of the drawn wire, although for mechanical reasons the mandrel should have a constant diameter too. Such an arrangement is not ideal for long term use but is fine for making some master leadscrews.

John

[Michael GilliganParticipant @michaelgilligan61133]

Posted by Neil Wyatt on 17/03/2015 17:32:51:

> A very rash assumption, I would say, Neil

Well the Wixey website says:

Resolution:	0.1 degrees
Accuracy:	+/- 0.2 degrees
Repeatability :	+/- 0.1 degrees

I'm happy to accept that mine is equally accurate. I can vouch for the repeatability and +/- 0.2 degrees is better than +/1 0.6 degrees.

Neil

.

Neil,

I can see no reason why a hobbed gear should be that bad. The hobbing process should automatically even-out most errors.

Did you re-mount the gear, between cutting it and measuring it ?

… If so, then I suspect that some of your recorded errors might be from eccentricity in the mounting [it's surprising how much angular error can be produced by a few thou' offset].

MichaelG.

[Nigel McBurney 1Participant @nigelmcburney1]

Reading back through this topic,which is quite old ,one post mentioned a straight tooth gear slipping and producing a thin tooth, gear slippage on the mandrel can be caused by a cutter being sharper on side, than the other. Usually caused by the cutter rubbing on a hard spot on a previous job (,a lot of cutters used today second hand bought from auction or other sources ,most are good but some can be damaged,) the blunt side pushes away from the work and the sharp side cuts deeper, a can move the workpiece particularly if the gear is on maximum capacity of the dividing head. It does pay to mark the first tooth ,and check the gear when finished ,a gear tooth vernier is ideal for his task. Very often an awful lot is discussed about theory of dividing methods and accuracy ,but very little about the work set up accuracy ie mandrels and work holding ,correct cutter sharpening,and checking of the finished product.

[Neil WyattModerator @neilwyatt]

> They wound wire around long mandrel to form a kind of thread, and then cast a lead nut around it.

Two wires, then you unwind one of them to leave a basic, but reasonably accurate, thread.

Neil

[Neil WyattModerator @neilwyatt]

> I can see no reason why a hobbed gear should be that bad.

I can – I should have pointed out that it was free hobbed.

The first one had an extra tooth, despite being gashed.

I think the cause was insufficiently deep gashing.

Neil

[Harold Hall 1Participant @haroldhall1]

I have used the method suggested by Maurice (Mo) with satisfactory results, actually providing an article in issue 8 of MEW regarding the method. It's first use was to produce a plate having 125 holes. However, I did not have a computer and asked the draughtsman in my department to produce a printout. It was printed using pen plotter onto film and was very accurate.

Subsequently, when I had acquired a PC with a CAD program I printed out my own using dot matrix, inkjet and laser printers. With these, I have always found that the large circle on which the divisions are placed is very slightly oval, typically an error of 0.5mm on a 100mm circle. Whilst this would have minimal effect on the accuracy of the divisions, the detent on my dividing heads could not cope with this. Rather than go into detail here, my solution to this can be seen in Photo 16 and associated text.

I have also used the method of producing the plate on the milling machine using the coordinates for each hole to position them, as implied by the use of CNC and DRO's mentioned by some above. Unfortunately, most will not have these facilities available and for them my pages on placing holes on a PCD may be helpful.

These basically use the method implied by John S when he started one of his comments by saying

Rough and ready Excel spreadsheet."

 

For my coordinate source I use the minimum X and Y values making it outside the circle on which the divisions are being set. This can be seen on __SK 2.__ and ensures that all calculated values are positive making it easier to work with than if using the method in ___SK 1.__

 

See the article "Holes on a PCD, The Mathematical Method" for more detail

 

Harold

[davebParticipant @daveb17630]

Posted by John Stevenson on 11/07/2011 21:52:29:

Posted by Andrew Johnston on 11/07/2011 19:47:42:

 

 

the starting angle (with 9 o'clock being zero)

 

Regards,

 

Andrew

 

 

Mine must be on British summer time as zero is at 3 o'clock.

John S

 

I wonder if this is caused by the X axis designations (- +) being reversed, on many DRO systems these are selected on initial set-up.

Dave

 

 

 

Edited By daveb on 18/03/2015 23:44:04

[Dave MartinParticipant @davemartin29320]

Posted by John Stevenson on 17/03/2015 16:02:24:

By'eck – if that's the size of your tea strainer John, what size HobNobs are you on now?

[Author]

Posts