Tips/Techniques DRO Glass vs. Magnetic Scale Accuracy Etc.

[DavidR8]

I "believe" what he means is that this is the error (prolly +/- 2.25 thou that results as the signal is decoded throughout a full sweep pole to pole of one magnetic length of the scale. Different scales have different magnetic lengths.

If you look at his graphs, you can see the cyclical nature of the output and its error.

I don't quibble with the fact that there is cyclical error. It has to be there. However, I'm not at all sure how that error translates to positional error. If it's 1 to 1 then I'm on your page - it's not trivial at all! But that may not be the case (and in fact, I doubt that it is). Hence my questions to Yuriy. If he answers, I'll assess that answer, give him some feedback, and comment here.

Till then, I don't think we should get too concerned.

Bottom line is that my magnetic scales are very linear and do not jump around like that at all. It is a very smooth constant rate of change which suggests that the error is not 1 to 1 but much much less.

Thanks John, I appreciate the analysis.
I think that I will go ahead and order mag scales for my mill.

 

[Susquatch]

Thanks John, I appreciate the analysis.
I think that I will go ahead and order mag scales for my mill.

Hold off just a bit David.

I heard back from Yuriy. He explained his method which was the same as mine. There is no reason that the results should be so different. He is going to do some video of the testing and he also plans to try it with a 1 micron indicator. I further suggested that testing with a finer resolution indicator might identify cyclicity in his reference indicator. It does seem odd that all his results are so cyclical.

I plan to repeat my own testing more rigorously to see if I missed something.

We will see.

 

[Rauce]

Hold off just a bit David.

I heard back from Yuriy. He explained his method which was the same as mine. There is no reason that the results should be so different. He is going to do some video of the testing and he also plans to try it with a 1 micron indicator. I further suggested that testing with a finer resolution indicator might identify cyclicity in his reference indicator. It does seem odd that all his results are so cyclical.

I plan to repeat my own testing more rigorously to see if I missed something.

We will see.

What’s the method?

 

[Susquatch]

What’s the method?

You are right to ask. Very simple. He just used a two inch dial indicator on tenths of an inch intervals (0.010). Every hundredth on the indicator he recorded the DRO output. I didn't do it quite like that, I simple watched for correlation between the two readings on a 3 inch indicator. Then repeated with a one micron indicator over a much smaller interval. Then I tested linearity over 24 inches on my X axis against a 24" bar. But I wasn't looking for errors, I was only qualifying it quickly. He didn't do the linearity test simply because he knew it could be dialed out. I did it to see if I had to dial it out - but didn't have to.

 

[Rauce]

You are right to ask. Very simple. He just used a two inch dial indicator on tenths of an inch intervals (0.010). Every hundredth on the indicator he recorded the DRO output. I didn't do it quite like that, I simple watched for correlation between the two readings on a 3 inch indicator. Then repeated with a one micron indicator over a much smaller interval. Then I tested linearity over 24 inches on my X axis against a 24" bar. But I wasn't looking for errors, I was only qualifying it quickly. He didn't do the linearity test simply because he knew it could be dialed out. I did it to see if I had to dial it out - but didn't have to.

That’s kind of what I was afraid of. A 2 inch dial indicator is no where near precise enough for the task in my opinion.

If it was me I would use an indicator (ideally .0001” or 1 micron resolution) only to zero against a reference surface and then against standards of known dimensions like gage blocks against the reference. A good set of traceable gage blocks will have a certificate with the deviation of each block in millionths of an inch as measured at the factory to further reduce error. Repeat measurements of each gage block and plot the results.

 

[DavidR8]

Hold off just a bit David.

I heard back from Yuriy. He explained his method which was the same as mine. There is no reason that the results should be so different. He is going to do some video of the testing and he also plans to try it with a 1 micron indicator. I further suggested that testing with a finer resolution indicator might identify cyclicity in his reference indicator. It does seem odd that all his results are so cyclical.

I plan to repeat my own testing more rigorously to see if I missed something.

We will see.

<holding off/>

 

[a smile]

I really like the atmosphere of this forum to discuss the problem, and it is possible to find the real root cause of the problem only when there is controversy.

It happened that I also encountered the problem of digital measuring tools. When I started to repair and clean the Heidehan raster scale used by my DECKEL3, the Y-axis was installed in the wrong direction, which caused physical damage to the over-range, and the reading-head stand was physically damaged. After I repaired it with strong glue, it could display numbers, but it would stop reading when you did not notice. So, it was a big headache, which caused me to scrap several parts when I made them. So, just showing is not necessarily correct! I recently found a compatible ruler replacement, and now the problem is solved.

Also, the glass ruler you rely on or magnetic ruler, it shows whether the value is accurate you need to have a very accurate measuring tool to check, and this value is only in the absence of temperature, humidity and other trace factors under the premise, and these conditions are not ordinary enthusiasts can have, Therefore, it is a simple and efficient method to use the measure block with the highest accuracy level as possible.

I repeat: Please stay away from Chinese measuring tools because no one cares about standards and no one follows the rules.

When you have very precise requirements for the size, you will find that your measurement method will make your measurement value have a big jump, which will make you at a loss.

My lathe uses newall brand, the ball grating ruler is 0.01mm for the large pallet, and the micro grating ruler is 0.001mm for the medium pallet. Most of the time, I can ignore the value of the last digit, unless there are some parts that require strict coordination, so I increase the deceleration of 1:10 for the lead screw of the medium pallet to facilitate accurate adjustment of the feed value. However, mechanical vibrations can easily cause the numbers to drift, so no matter what type of gauge you use, the key factor is always the mechanical structure itself.

 

[whydontu]

Silly question - wouldn’t comparison against a good quality vernier caliper be a reasonable way to judge? For example, I have a glass scale DRO on my lathe. I have a 24” Starrett vernier, should be easy enough to clamp the Starrett beam parallel to the ways and compare the DRO to the caliper. Clamp the Starrett fixed anvil to the headstock, clamp the sliding anvil to the crossslide.

 

[Rauce]

Silly question - wouldn’t comparison against a good quality vernier caliper be a reasonable way to judge? For example, I have a glass scale DRO on my lathe. I have a 24” Starrett vernier, should be easy enough to clamp the Starrett beam parallel to the ways and compare the DRO to the caliper. Clamp the Starrett fixed anvil to the headstock, clamp the sliding anvil to the crossslide.

That might tell you if the glass scale needs to calibrated for linear error on the DRO. A 24” vernier caliper doesn’t have the resolution to tell you how accurate or repeatable the scale is.

 

[Susquatch]

That’s kind of what I was afraid of. A 2 inch dial indicator is no where near precise enough for the task in my opinion.

Generally I agree. But we are talking several thou here. It "should be" better than that. Nonetheless I did tell Yuriy that we don't really know if the cyclisticy is in the indicator or in the scales.

If it was me I would use an indicator (ideally .0001” or 1 micron resolution) only to zero against a reference surface and then against standards of known dimensions like gage blocks against the reference. A good set of traceable gage blocks will have a certificate with the deviation of each block in millionths of an inch as measured at the factory to further reduce error. Repeat measurements of each gage block and plot the results.

He does plan to retest using a 1 micron indicator.

But not everyone has gauge blocks like that. Mine are only class B.

 

[Susquatch]

Silly question - wouldn’t comparison against a good quality vernier caliper be a reasonable way to judge? For example, I have a glass scale DRO on my lathe. I have a 24” Starrett vernier, should be easy enough to clamp the Starrett beam parallel to the ways and compare the DRO to the caliper. Clamp the Starrett fixed anvil to the headstock, clamp the sliding anvil to the crossslide.

You have a 24in Starret Vernier? Wow! I'm feeling some tool jealousy over here.......

But Yuriy is finding several thou errors over short distances. No need for a long vernier.

I do agree with @Rauce though, it would be excellent for calibrating the scale linearity provided the temperature is within the specs for both pieces of equipment.

 

[whydontu]

You have a 24in Starret Vernier? Wow! I'm feeling some tool jealousy over here.......

But Yuriy is finding several thou errors over short distances. No need for a long vernier.

I do agree with @Rauce though, it would be excellent for calibrating the scale linearity provided the temperature is within the specs for both pieces of equipment.

I inherited some tools (and a tiny bit of skill) from Bob, a machinist who got his journeyman ticket in 1943. He was the original test subject for my theory that cutting oil makes you grumpy. He was <GRUMPY>.

Bob story. Bob was missing most of his index finger on his left hand. One of the tasks we would often do was drill 13/32“ holes through 6x6x3/8” steel square tubing. Bob would centre punch the hole location, hold the chunk of tubing in place on the drill press table using his left hand, right hand on the quill feed lever, and drill the holes in one pass with the 13/32” bit. No vise. After watching him do that, I decided that it would not be advisable to get on the bad side of Bob.

 

[Tom Kitta]

I also have a 24" Vernier caliper. But this is not accurate enough for DRO work. I used a dial test indicator on one end of the DRO scale and moved 20 inches then back to zero. Repeated perfectly. I do not know if it indeed moved 20 in and 20 in back - but it repeated well.

If one were to find a 20 in precision block then one could place it between two dial indicators. Find out their deflection - record it and then do the movement with a DRO.

The problem with DRO testing is we do not exactly have a good way to test the DRO. Sure there are no standards in China for this BUT I assume making a bad ruler and a good one is about as much effort so they may as well make stuff that actually work. This is for glass scales.

 

[Tecnico]

Good info @Tecnico. Leave it to you to find and point this out.

LOL! Always enjoy provoking discussion about interesting topics.

I’m impressed by the discussion this topic has generated, plenty of smart people in the room. I also see that Yuriy has become involved courtesy of @Susquatch. Interesting to see the video Yuriy posted on doing the calibration process and the data results.

Not that I understand all of the finer electronic details but it’s informative to see the discussion on the logic of how the measurement is derived/interpreted from the pattern embedded in the tape or glass. Plenty to learn.

One aspect that I’ve seen mentioned peripherally is accuracy of calibrated standards used. One of the factors in arriving at an accurate measurement is doing all of the measurements at a consistent temperature because calibration standards are defined as accurate at just one temperature. When comparing results from various scale samples, that implies the data is only accurate and comparable at the defined standard temperature.

Where am I going with this? In general, for the use that we'll (me) make of our tools we won't usually need to hold dimensions to something like .001" but it appears that the potential is there to experience that much or more error due to temperature differential.

One of the factors which makes a glass scale more attractive to me is the fact that glass has a much lower Thermal Coefficient of Expansion (TCE) than the material used for the magnetic tape. I don't have specifics of the exact glass and mag tape materials but picking data off the web shows Pyrex runs about 4.0 x 10E-6 mm/mm per deg C (or inch/inch since it's just a ratio) while according to the manufacturer, Polyester mag tape (fridge tape) runs about 48 or about a factor of 10 times. This reminds me of one of my lives where I played around with glass for a sensor application because it is thermally stable.

Put practically, if the ambient temp changes by 20C from when the tape was magnetized, the length of the tape could have changed by about .001" per inch. That also means if you're in @David_R8 's shop when it's 3000C outside like the other day and laying out a 10" dimension with the DRO then it could be off by something like .010".

I don't know how to quantify simply the effects of the exact composition of the tape, the effects of the stainless strips on either face of it or the 3 adhesive interfaces (whether you've mounted tape in your cross slide or it's on the factory aluminum strongback) or the part of the machine the scale is mounted to but the potential is there to affect the distance between the magnetic "stripes" by temperature in a difficult to quantify manner. It sounds like a detailed FEA that I'll leave to others who have more time to play!

Keeping the scales (machines) at a constant temperature is the obvious way to avoid inducing errors but I'm also thankful that any of the scales I could practically use glass for I did. Only my lathe embedded cross slide and mill quill have magnetic scales. My mill lives in a shop that will probably have 20C temp swings.

So, where does this lead the discussion?

D

 

[DavidR8]

Magnetic on the mill is attractive for two reasons:
1) Painless install. Putting the DRO on my Clausing was a colossal pain in the @$$ because the cast iron was so hard. I'm under no illusions that import cast iron is the same quality as 1945 cast iron but I still dislike working with it.
2) Low profile. The X scale will go on the front of the table to preserve Y travel so the low profile of the mag scale is attractive.

That said I'm experienced putting glass scales on so maybe that's the easy path.

 

[Tecnico]

Magnetic on the mill is attractive for two reasons:
1) Painless install. Putting the DRO on my Clausing was a colossal pain in the @$$ because the cast iron was so hard. I'm under no illusions that import cast iron is the same quality as 1945 cast iron but I still dislike working with it.
2) Low profile. The X scale will go on the front of the table to preserve Y travel so the low profile of the mag scale is attractive.

That said I'm experienced putting glass scales on so maybe that's the easy path.

LOL! It's never easy is it!

FWIW neither my Myford or First gave me problems with working the CI.

I presume you've looked at the lower profile mag scales? That's what I used on the carriage of my lathe.

D

 

[DavidR8]

My mill is that same as a CX601 and Joe's install is what makes me lean in the mag scale direction.

CX601 Mill - Adding DRO's

Just completed adding X,Y and Z axis DRO's to a CX601 mill. Tried to keep things very simple but still needed to maintain accuracy. Used self adhesive 10mm wide magnetic tape (1 micron) in a stainless/mag tape/stainless sandwich. Brackets are 3D printed and I may make from aluminum at a later...

canadianhobbymetalworkers.com

 

[Dabbler]

There is so much to unpack here.

I read that post. I was all set to order magnetic scales and now I’m not sure what I’m going to do.

I think that I will go ahead and order mag scales for my mill.

He just used a two inch dial indicator on tenths of an inch intervals (0.010)

That’s kind of what I was afraid of. A 2 inch dial indicator is no where near precise enough for the task in my opinion.

I also have a 24" Vernier caliper. But this is not accurate enough for DRO work.

First let me open with the thought that Yiuily's opinion in this matter is somewhat against the grain. This makes it both interesting, and somewhat inviting scrutiny. Firstly even the highest priced Mitutoyo 2" indicator is only accurate to about 1.7 thou over 2 inches *under ideal circumstances*, at 23C, 10% humidity. But I'm not really critical here... I would love to discuss his methods with him. No; a vernier is not nearly accurate enough to verify or dispute Yurily's claims with any authority at all.

One starting place is to start with what is previously believed about the two scale types in industry. For that I'll suggest Sony and Mazak, which I'm familiar with. For many years Sony never offered glass scales for their DROs because of accuracy problems. Sony was one of the early pioneers and innovators in building high precision DROs. Mazak, with their ultra precision milling centres exclusively use magnetic scales, and their positioning and repeatability is below a third of a tenth, by last reckoning. @Alexander do you have a current figure on this?

In no way am I dismissing glass scales. Both seem to be reliable. Jim Dawson, who has made a large portion of his income over the last 40 years installing DROs as an adjunct to doing CNC conversions has tried to debate the merits of Yurily's claims, without satisfaction. That makes me more concerned.

Lastly, the best way I know to measure an axis is to indicate in a long parallel on that axis, and align a stack of gauge blocks with a step at one end, and use a sub tenths indicator to get the zero at each end, and compare the gauge stack to the digital readings. Using a plunge indicator, even off by a degree or two will give you way over 2 thou error over 2 inches.

------

All that being said, I agree above comments that you can get 'high' accuracy out of either scales. But how you can assure yourself that you have the correct readings takes some metrology skill. The readings have to be temperature compensated, with provable references for anyone to make sense in this debate. Also, relying on the DRO only to ensure accuracy is not how any professional achieves high accuracy machining, so buy what you feel is the better install for your machine. and the learn how to use it well, quirks or not.

 

[Dabbler]

I really need to add something here. The Mitutoyo digital calipers that many of us use and trust to around .001, has capacitive domains on about a 7 mm pitch with capacitive traces about every 2.5mms. By doing complex math concerning multiple quadrature cycles/readings, these calipers reliabily perform to about [updated] .001, even after decades of use.

The same is true for the magnetic scales. The math is identical.

The glass scales use a different technique that Yurily describes accurately. What he doesn't say is that to get any reading at all the width of the scribe and unscribed parts of the glass has to be a 50% duty cycle. The second grating has lines that have to be the same width, but with slightly longer unscribed widths. If they are not aligned in parallel to one another, then count errors occur.

Both systems have checks to eliminate errors - and they both seem to work well. I have a 2 tenths magnetic scale DRO that is over 40 years old, and it still reads against the gauge block stack to 3 tenths (over any 6" interval), even after all these years. I haven't yet bought glass scales, but one day I will certainly give them a try also.

 

[Tecnico]

My mill is that same as a CX601 and Joe's install is what makes me lean in the mag scale direction.

CX601 Mill - Adding DRO's

Just completed adding X,Y and Z axis DRO's to a CX601 mill. Tried to keep things very simple but still needed to maintain accuracy. Used self adhesive 10mm wide magnetic tape (1 micron) in a stainless/mag tape/stainless sandwich. Brackets are 3D printed and I may make from aluminum at a later...

canadianhobbymetalworkers.com

That is a very clean looking installation!

D