Fitting a 5 micron DRO to Myford ML7

[choochoo_balooParticipant @choochoo_baloo]

I'm doing the research to buy a DRO for my Myford ML7. However I just read that the generic DRO are, unsurprising basically two board manufacturers both of which are based on metric.

Thus those of us, me included, who work in Imperial, reckon we must be aware that a 5 micron read-head can produce tangible rounding errors in the tenths digit, since all metric to Imperial conversion is done in software.

Obviously it's trivial to show 5 micron = 0.0002 " (1 sf).

My question is, practically, do 5 micron read-heads matter for an ambitious ML7 user hoping to aim for sub 1 thou tolerance? (from time-to-time; not always!)

[choochoo_balooParticipant @choochoo_baloo]

[SillyOldDufferModerator @sillyoldduffer]

Well the conversion part is likely to be a red-herring. Chances are the DRO will use floating point arithmetic where a a micron is 3.93701E-5 inches making 5 microns = 1.968504E-4 inches in single precision. This is considerably better than 0.0002. As double precision is more likely than single these days, I think conversion error can be ignored.

5 micron resolution fundamentally restricts the accuracy of what can be measured. It's roughly two tenths meaning the system is plenty good enough to measure a thou, and fairly trustworthy to about half a thou, but not better than that.

So not bad. I'd say the scale is well matched to the machine because I doubt a Myford and average operator are capable of working to better than half a thou in ordinary circumstances. Measurements in the tenths region require extra special care and attention: too easy to believe an accurate job has been done when someone else would get embarrassingly different measurements.

Mustn't mark your own homework in that region unless absolutely certain everything is being done right, and most people can't be sure. More to it than a good machine and DRO.

Dave

[Robert ButlerParticipant @robertbutler92161]

Posted by choochoo_baloo on 11/04/2022 16:15:16:

My question is, practically, do 5 micron read-heads matter for an ambitious ML7 user hoping to aim for sub 1 thou tolerance? (from time-to-time; not always!)

I would say a bit ambitious for a "Newbie"

Robert Butler

[Ady1Participant @ady1]

5 microns on a hobby lathe??

more like 30-50

[Bob WorsleyParticipant @bobworsley31976]

Don't forget that the measuring beams, spars, whatever, are an analogue to digital device, with either light or magnetic bars and spaces doing the measuring. As such, there is always a +/-1 count error in this analogue to digital conversion. So, if the spar has 5 micron lines and spaces then there will be an additional +/-5 micron error. Of course if the spar has a =/-2 micron spacing then this conversion error will be 2 microns and mostly hidden in the conversion to a 5 micron readout. There might also be hysteresis error from moving one way then the other, this could be far more than 5 microns. Spars I have seen have some pretty stiff rubber seals, and the drag from them could be any ones guess.

As usual, do repeated, 20 or so, measurements and machining using something that will resolve to 1 micron, not accurate, but will resolve as a precision reading. A fiducial micrometer is the thing, not a digital micrometer where you have the same conversion error, analogue wins here. The difference between accuracy and precision, also repeatability and resolution. Easiest to machine a diameter on a lathe, from one direction, then the opposite. Measure up, bit of statistics and you have your answer. Read Right First Time by Price.

[John HaineParticipant @johnhaine32865]

Posted by SillyOldDuffer on 11/04/2022 16:53:22:

Well the conversion part is likely to be a red-herring. Chances are the DRO will use floating point arithmetic where a a micron is 3.93701E-5 inches making 5 microns = 1.968504E-4 inches in single precision. This is considerably better than 0.0002. As double precision is more likely than single these days, I think conversion error can be ignored.

Um, 0.0002 = 2 x 1E-4, so why is 1.968504E-4 significantly better?

[John HaineParticipant @johnhaine32865]

Here are some numbers off the scale on my milling machine. This is the M-DRO readout with their standard magnetic scales. I zeroed the X scale then moved it in increments of 5 micron indicated and for each reading switched to inch to see what it thought the reading was.

mm    inch                  increment
0.000    0.0000000    
0.005    0.0001900    0.0001900
0.010    0.0003900    0.0002000
0.015    0.0005900    0.0002000
0.020    0.0007800    0.0001900
0.025    0.0009800    0.0002000
0.030    0.0011800    0.0002000
0.035    0.0013700    0.0001900
25.400    1.0000000    
19.050    0.7500000    
12.700    0.5000000    
6.350    0.2500000    

Of course what it does is actually count pulses from the scales, and it looks like it multiplies the count by 0.005 to get the mm number but rather than convert that to inches it accumulates an inch measure by an algorithm designed to minimise the error which is why it seems to add .00019 once then 0.0002 twice then repeat. No sign of accumulating error or hysteresis.

I'd conclude that given you take all the usual care, use sharp tools, have your lathe properly adjusted, I don't think a 5 micron scale would stand in your way.  If it does, then just work in metric!!

Edited By John Haine on 11/04/2022 18:18:20

Edited By John Haine on 11/04/2022 18:23:39

[SillyOldDufferModerator @sillyoldduffer]

Posted by John Haine on 11/04/2022 17:55:12:

Posted by SillyOldDuffer on 11/04/2022 16:53:22:

Well the conversion part is likely to be a red-herring. Chances are the DRO will use floating point arithmetic where a a micron is 3.93701E-5 inches making 5 microns = 1.968504E-4 inches in single precision. This is considerably better than 0.0002. As double precision is more likely than single these days, I think conversion error can be ignored.

Um, 0.0002 = 2 x 1E-4, so why is 1.968504E-4 significantly better?

I took it ChooChoo was worried about rounding error in the maths going from metric to imperial. My 'considerably better than 0.0002' is a poor choice of words…

[Neil LickfoldParticipant @neillickfold44316]

I am looking at fitting a cheap magnetic tape DRO to my S7. It is mainly so that I can see approximately where the tool is and finial sizing is done my measuring or fitting to the other parts. In my case the final Z movement will be from the end hand wheel and cross slide dial. Being only 0.01mm resolution, on Z is going to be close enough anyway, and 0.02 mm resolution for X. When I am wanting a better X resolution, I switch out the standard cross feed nut assembly since getting a new casting, and replace it with a unit I made that uses a M8X0.5pitch thread. It has the imperial thread dial that makes each division 0.01mm on diameter. Just 4 times the winding of the standard 2mm pitch one.

I think that as long as the unit you get has metric to inch conversion on it , will be fine. We have very good DRO on the work machines, and I still don't trust them for anything that needs to be sized. But that is just the old way of doing things coming out.

I think that once you get the DRO installed , you will be thinking I should have done this earlier.

In my case, I don't want the back of the lathe dedicated to a linear scale, when I still have a taper turning attachment that I like to use once in 10 years or so it seems. I just like having the use of them, So I am getting the contactless magnetic tape system to try, and to mount the scale in the front of the lathe below the saddle on to an Ali extrusion that will make a bit of a cover to help keep the unit clean.

Neil

[Jon LawesParticipant @jonlawes51698]

What are you making that demands that accuracy?!

[Neil LickfoldParticipant @neillickfold44316]

Pistons for F3D pylon engines and pistons or F2C diesel engines. Normally these are trimmed on Sub micron lathes to suite the individual liner. Some do trim on std cnc lathes with 1um resolution and then sometimes do some form of manual alteration afterwards. The fastest engines are those that are fitted correctly after being diamond turned. The diesel engines, the crown to pin height can be longer than the factory height , but can't be shorter, as the liners have shims to get the compression setting correct with the contra piston in the right place. The F3D engines are almost the reverse, the pin to crown height needs to be on size or ever so slightly less, unless you make new shim sets for the liner for that setup.

The other thing that I make is bearing outer races to be replacements for older F2A speed engines that I still like to use, as they are no longer made in those outside diameters any more. I also recondition and resize the bearing race that is on the crankshaft, then make a outer race with a smaller to suite race for the bearing and the final sizing is done in the ceramic ball selection.

Non race engines do not require the above precision to be good running engines.

[Anonymous]

Posted by choochoo_baloo on 11/04/2022 16:15:16:

…do 5 micron read-heads matter for an ambitious ML7 user…

Almost certainly not; the resolution may be 5um but the accuracy will be less, possibly far less. My Newall microsyn scales are 5um resolution, but 10um accuracy. Each scale comes with a card showing measured accuracy versus length.

Andrew

[BazParticipant @baz89810]

As Jon Lawes says what are you making that demands such accuracy and more to the point how do you keep your workshop at a constant 20 degrees centigrade.

[HopperParticipant @hopper]

Doesn't anyone use emery paper anymore to get that last half a thou off when striving for sub one-thou tolerances?

[Ady1Participant @ady1]

I find that aluminium oxide paper very good

[JasonBModerator @jasonb]

And remember the cross slide scale will only be measuring the position of the slide relative to the carriage. Any wear or deflection won't be taken into account as the tool gets pushed away from the work.

[Y C LuiParticipant @yclui16187]

Sub one-thou accuracy is actually not that ambitious. one-thou is 0.025 mm. If the resolution of the DRO is limiting, you will get an error of 0.005 mm in radius or 0.01 mm in diameter which is well within one thou. I will be worrying about other sources of errors in the machine.

It's not a lathe but my Emco FB2 mill will have the Y position off by 0.03 mm due to thermal expansion of the head after running for a while. For this reason the 5 micron DRO on the machine is an overkill.

Edited By Y C Lui on 12/04/2022 08:14:05

[Martin KyteParticipant @martinkyte99762]

Posted by Y C Lui on 12/04/2022 08:05:00:

Sub one-thou accuracy is actually not that ambitious.

Edited By Y C Lui on 12/04/2022 08:14:05

I can certainly do it on the Myford just with the slide micrometers and a mic.

regards Martin

[John HaineParticipant @johnhaine32865]

I've been worrying about this for days. It seems obvious that the display just counts pulses and multiplies the count by 0.005 to get the metric reading, but how does it get the inches? Then I realised that every 127 pulses comes exactly every .025" and .635mm. So finding the right increments to add to the inch display for minimum error is similar to arranging a digital dividing head to make 127 teeth. Our old friend Bresenham to the rescue I think, no need for floating point conversions.

[Michael GilliganParticipant @michaelgilligan61133]

I am not familiar with the devices in question … so this may be a ‘Red Herring’

Before plunging into Metric to Imperial conversion issues … May I suggest reading the links to Hans U. Meyer, that I posted long, long ago :

**LINK**

https://www.model-engineer.co.uk/forums/postings.asp?th=57456&p=3

The scales could be natively Imperial

Just a thought … I will leave you to it.

MichaelG.

[not done it yetParticipant @notdoneityet]

If the OP is worried about that last umpteenth signicant figure, I suggest he uses a suitable micrometer to make his final measurements.

DROs are good enough for most things, but clearly the OP is not wanting to make confirmation measurements with a standard measuring device which can be checked against sufficiently accurate standards – like a set of gauge blocks.

My cheap DRO set indicates to 0.01mm and obviously has tolerances on accuracy and precision, but is good enough for most machining operations. For most modellers, I reckon the ‘fit’ is more important than the absolute accuracy. They are not making serial machines or parts for production items. How many drawings even require dimensions to those tolerances? That is outside of science/NASA/etc requirements.

The 0.001mm set is clearly better than the cheaper set – but would I always actually trust that last digit to be absolutely accurate? No, because there are always errors involved in making any readings.

A spec of sawdust and a plank comes to mind…

[John HaineParticipant @johnhaine32865]

Posted by Michael Gilligan on 16/04/2022 09:28:50:

The scales could be natively Imperial

…

MichaelG.

Could be, except they aren't!

[Michael GilliganParticipant @michaelgilligan61133]

Posted by John Haine on 16/04/2022 12:58:30:

Posted by Michael Gilligan on 16/04/2022 09:28:50:

The scales could be natively Imperial

…

MichaelG.

Could be, except they aren't!

.

Then, evidently, I chose my word, and its underlining appropriately.

MichaelG.

[bernard towersParticipant @bernardtowers37738]

Im with NDIY its not the size its the fit!

[Author]

Posts