PDA

View Full Version : Digital scale accuracy



loose nut
04-11-2009, 05:46 PM
I have the generic digital scales, with the simple display, mounted on my RF 45 type mill.

If I mount a dial indicator on the mill and move the table in the x or y directions and check the movement with the indicator I get a bit of a discrepancy, maybe a couple of thou up to .005" and even .010" after a while.

Today I put in a wiggler, the type with the ball on the end of a shaft, and ran it into a parallel mounted in the vise, zeroed it and then ran it back and forth several inches, jogging it in both directions so there would be lots of backlash added into the movement, and back into the parallel again. I repeated this 5 or 6 times in the x and y directions and every time I got a 0.000 reading against the parallel.

These scales are suppose to be accurate to .001" which is good enough for my shop, mostly.

Now dial indicators are comparators and not really meant for measuring because they can hang up a bit when the plunger is returning to rest so there could be some error there and a bit of tangential error in the way it is mounted.

So which is right the scales or the indicator or is there something wrong with my methodology in checking this out. It would seem to me that if it returns to the same place with the same readings the scales should be right.:confused: :eek: :(:

doctor demo
04-11-2009, 05:57 PM
If You get the same reading every time on the scale, with the wiggler doing the same... then it would be the dial ind. that is misleading. Unless of course , the wiggler and indicator are in conspiracy against You.

Steve

tdkkart
04-11-2009, 06:13 PM
When was the last time you saw a CNC machine with a mechanical measurment system on it??
Most electronic scale systems use the same technology, the only differences being the resolution of the display system.

fasto
04-11-2009, 09:56 PM
If I mount a dial indicator on the mill and move the table in the x or y directions and check the movement with the indicator I get a bit of a discrepancy, maybe a couple of thou up to .005" and even .010" after a while.


I can't tell if you're asking about an error that happens when you :
1) Move the table a distance then move the table back to 0, the scale doesn't read 0 but reads 0.005. Repeating this adds a bit to the measurement each time;

or

2) Move the table 6.000 inches and the scale reads 5.995 inches

or

3) Move the table 6.000 inches and the scale read 6.005 inches

For 1, the batteries in the scales might be getting low. I have had this happen on my digital caliper. When the battery is low it starts missing counts. When the battery is really low it's wildly innacurate.

For 2, the scales aren't exactly parallel to the axis. The law of cosines is getting you.

For 3, I have no idea how this could happen, except for excessive backlash in the scale mountings.

More info, please.

Paul Alciatore
04-12-2009, 02:16 AM
Get a set of gauge blocks and use them to check the scales, indicators, etc. Then you will know.

Evan
04-12-2009, 02:40 AM
Now dial indicators are comparators and not really meant for measuring because they can hang up a bit when the plunger is returning to rest so there could be some error there and a bit of tangential error in the way it is mounted.


A dial indicator is not a comparator. It is incapable of making a comparative measurement. A DI provides an absolute measurement of variation. That is an entirely different matter. Even with a tangential contact the measurement is still absolute subject to a cosine error that is also absolute. If the reading varies then either the indicator is malfunctioning, the mounting is not secure or the position being measured is really varying.

It is easy enough to check the indicator. Set it up to measure to a non-varying reference surface. Introduce a known thickness of shim stock (spark plug gap gauge is good) between the indicator and the surface. Remove the shim stock and repeat several times. Try different thicknesses of shim stock, do not stack them. It will be quickly apparent if the indicator is accurate or not.

JCHannum
04-12-2009, 07:52 AM
A dial indicator will measure distance within it's range, as will a DRO scale. Both can have cumulative error within their limits, usually expressed as +-something per unit of measurement.

Dial indicator accuracies are usually +- one graduation over the full range. The full range is usually considered to be 2-1/2 turns of the pointer. Extended range indicators will have wider tolerances. Dial indicators are most accurate between the 10 o'clock and two o'clock positions from zero. A dial indicator, should be preloaded, about 1/8 to 1/4 of a turn before setting the zero point to ensure it will return to zero.

The accuracy of the DRO will depend upon the manufacturer, but probably approaches that of a good digital caliper, something on the order of +- 0.001" per 6 inches of travel or so.

Both can be out in different amounts and different directions and still be within their specification. If one is out in the plus direction and the other in the minus, the error will be doubled. Both can be out at some distance of travel and still return to the same zero point with dead accuracy. For these reasons, checking the two instruments against each other does not prove if either is out of spec. The best method of checking is with known standards such as gage blocks, assembling a series and checking the travel of the DRO against them.

loose nut
04-12-2009, 10:12 AM
The opinion seems to be that my experiment won't test the degree of accuracy of ether measuring device, which isn't what I intended, so I better rephrase the question.

If the digital scales come back to the same point with the exact same reading (within the accuracy level of the device) it would seem that they have a accurate degree of repeatability (not accurate measurement of distance traveled) where the DTI (Mitutoyo) which, for what ever reason, didn't have the same readings. It would appear that the scales are accurate for finding a set point in X,Y coordinates, IE: setting the center of a rotary table and being able to go back to that point after an operation is complete, on a mill table.

It would seem that they are repeatable unless the amount of error in my use of the wiggler and the error in the scales are exactly the same every time which would give the same erroneous readings. After repeating the trial half a dozen times that is almost impossible.

This is different than testing the accuracy of the distance that they traveled, my mistake, that's what I have to do next.

Hope this is less confusing, have I got it right this time.

A.K. Boomer
04-12-2009, 10:32 AM
I think the problemo lies in the way your using your DI, they are flimsy set-ups the way they mount - they just are - they also have fairly stiff little springing on the ball levers -- once you get them preloaded they can be somewhat accurate but only in the preloaded direction of travel, If your trying to measure fore and aft with one its a total Fuqe as now your releasing the tension from the entire "wet noodle" mounting rig, this is where your getting your discrepancy -- your seeing the battle between the ball lever "sticktion" and the mounting hardware flex, its a no-no, the hardware has no sticktion - its just flimsy length levers - when you back up its ALWAYS the first to move -- your DI ALWAYS takes more energies to compress than it does to spring back BUT its still an internal mechanism with frictional qualities -- your variance in readings is directly proportional to the difference between these two systems.

JCHannum
04-12-2009, 11:59 AM
Are you using a dial indicator (DI) or a dial test indicator (DTI)? They are two different instruments.

Regardless, either should return to the same zero point if applied properly. Mechanical indicators can get gummed up or mechanically damaged to the point that they will not return to zero reliably, this is not an inaccuracy in the gage, but the result of abuse or neglect. As AKB points out, the rigidity of the setup will also play a large part in the repeatability of the system.

loose nut
04-12-2009, 12:56 PM
It was mounted on one of the "mighty mag" mounts stuck onto the mill table pushing against the column. Could slip I suppose, I don't really care about the DI, I was only using it as a check against the scales (which isn't much good if the DI is moving), my concern is the repeatability of the scales which seems pretty good BUT????

Timleech
04-12-2009, 12:59 PM
A dial indicator is not a comparator. It is incapable of making a comparative measurement. A DI provides an absolute measurement of variation. That is an entirely different matter. Even with a tangential contact the measurement is still absolute subject to a cosine error that is also absolute. If the reading varies then either the indicator is malfunctioning, the mounting is not secure or the position being measured is really varying.


Maybe a dial gauge is not, in itself, a comparator but they are commonly used as part of a comparator.

see ebay 220390513653 or 250360000736 for examples.

Tim

Evan
04-12-2009, 01:05 PM
Maybe a dial gauge is not, in itself, a comparator but they are commonly used as part of a comparator.


The essential difference is that the comparator has a built in point of reference. A DI does not.

Timleech
04-12-2009, 01:27 PM
The essential difference is that the comparator has a built in point of reference. A DI does not.

But surely the OP's setup included a point of reference?
I think you're splitting hairs ;)

Tim

Evan
04-12-2009, 03:57 PM
Splitting hairs? Hardly. How can you use a DI as a comparator? The fact that it has a point of reference in the setup is of no value. It isn't dependable or portable or even referenced to the dial indicator, all of which must be true for it to become a comparator. A dial caliper is also a comparator and I have never seen anybody here confuse it with a dial indicator even though it has an indicator dial.

JCHannum
04-12-2009, 10:41 PM
A dial indicator or a dial test indicator by itself, lying in it's case, is incapable of doing anything.

Either can be used for comparative measurements when installed in the appropriate equipment. The comparators illustrated are set with gage blocks or other fixed standard, and the indicator is used to measure the deviation if any. The second eBay item is commonly referred to as a snap gage over here, but the function is the same. Either of the two gages would be useless without the dial gage.

Evan
04-12-2009, 10:52 PM
A car has wheels. That does not make a wheel a car. A comparator may have a dial indicator. That does not make a dial indicator a comparator, especially as that is not the common role for the instrument.

This is all I have to say on this matter as any attempt to argue otherwise is clearly an attemp to argue, not to inform.

Timleech
04-13-2009, 05:04 AM
This is all I have to say on this matter as any attempt to argue otherwise is clearly an attemp to argue, not to inform.

An extraordinary attitude.
"I'm right, don't bother disputing it" :rolleyes:

Tim

Peter.
04-13-2009, 05:24 AM
Evan how would you go about testing the acuracy of the DRO in the OP's circumstances? His test seemed fair to me.
I think constructive input is what'll do the guy most good not telling him how wrong his methods are.

Evan
04-13-2009, 06:06 AM
I did not say the op's methods were flawed, others did.

To test fix a micrometer to the table on an insulated block. In a drill chuck fix a piece of music wire. Connect an LED to the micrometer and the drill chuck (with appropriate resistor and battery) and very slowly wind the axis toward the moving anvil of the micrometer until the LED illuminates by touching the music wire to the micrometer. Zero the DRO and back the micrometer off a known amount and continue winding the axis of the machine in the same direction until the LED illuminates again. This test will be as accurate as the micrometer. The DRO should agree.

Then back off the touch with the micrometer until the LED extinguishes. That will characterize the back lash. The DRO reading should not change more than one least significant unit. If it does the mounting of the DRO is suspect.

Incidentally, I use this method to set the Z axis on my mill. My mill table is insulated from the rest of the machine by virtue of having PTFE linear bearings and an acetal leadscrew nut. It is repeatable to within 0.0002 or so.

oldtiffie
04-13-2009, 07:06 AM
Now to cut back to the chase and the topic of and question in the OP:


I have the generic digital scales, with the simple display, mounted on my RF 45 type mill.

If I mount a dial indicator on the mill and move the table in the x or y directions and check the movement with the indicator I get a bit of a discrepancy, maybe a couple of thou up to .005" and even .010" after a while.

Today I put in a wiggler, the type with the ball on the end of a shaft, and ran it into a parallel mounted in the vise, zeroed it and then ran it back and forth several inches, jogging it in both directions so there would be lots of backlash added into the movement, and back into the parallel again. I repeated this 5 or 6 times in the x and y directions and every time I got a 0.000 reading against the parallel.

These scales are suppose to be accurate to .001" which is good enough for my shop, mostly.

Now dial indicators are comparators and not really meant for measuring because they can hang up a bit when the plunger is returning to rest so there could be some error there and a bit of tangential error in the way it is mounted.

So which is right the scales or the indicator or is there something wrong with my methodology in checking this out. It would seem to me that if it returns to the same place with the same readings the scales should be right.:confused: :eek: :(:

Hi loose nut.

I would give odds that your indicator is at fault and that your DRO scales are fine.

The DRO seems to have performed very well in "X" and "Y" while the indicator was at best suspect in both.

Forget about the dial indicator and just use the dials on your "X" and "Y" hand-wheels. Turn the hand-wheel clock-wise, zero the dial and the DRO. Wind on 2", 4" 6" 9" 12" (all clockwise). Check the DRO at each point. Now do the same in an anti-clockwise direction.

Compare the DRO readings with the hand-wheel dials. I will be very surprised if there is a substantial difference between the DRO and the hand-wheel dial readings.

I think that you can be confident in the accuracy of your DRO scales to the degree of accuracy you require - probably a lot better.

I would either overhaul that dial indicator until it works satisfactorily, or if it cannot be repaired - scrap it.

JCHannum
04-13-2009, 07:54 AM
A comparator must have some means of measuring deviation from a set measurement. That comparative measurement can be made with a variety of devices, dial indicators and dial test indicators being among them.

A plunger type dial indicator measures linear dimensions and will not have cosine error. Cosine error can be present only in a lever type DTI and can occur when the probe is set at too great of an angle to the direction of movement.

I have not criticized loose nut's set up, as I do not know what it is, and I am not sure of which type of indicator he is using. His results would seem to indicate some mechanical problem with the indicator or the set up, as neither the indicator or the scale should have the degree of error he is seeing.

If it is a plunger type and he is not providing enough preload, the error described can result. I would recommend setting the indicator up so that there is one complete revolution of the dial before setting it to zero. If a lever type of DTI is being used, he could be seeing cosine error, but it is not usually that large of an amount.

Again, to prove the accuracy of the scale, the best method is to use gage blocks. The wiggler, or an electronic edge finder should repeat zero within 0.001". If you do not have gage blocks, accurately measure several parallels with a micrometer. Another parallel can be clamped to the table, it will have to be carefully trammed to ensure it is at 90 degrees to the travel of the scale to be checked. Locate the edge of the fixed parallel and set that as zero. The measured parallels can now be set in place and their width measured with the scale by locating their edges one at a time with the edge finder. Several parallels of known width can be put in place, and the distance measured incrementally as each is put in place. This will give a reasonably good idea of the accuracy of the digital display provided the parallels are truly parallel.

Evan
04-13-2009, 08:11 AM
A plunger type dial indicator measures linear dimensions and will not have cosine error

Of course it will if it cannot be placed square to the item measured. My Starret DI has a rounded button to allow such placement and the cosine error must be taken into account if an absolute measurement is desired.

JCHannum
04-13-2009, 08:44 AM
Cosine error is the term used in to describe the error introduced when using a lever type indicator with the probe at too great of an angle to the direction of movement. It is a common term in metrology.

A plunger indicator, if grossly misapplied will also display cosine error, but it is much more of a problem with the lever style of DTI.

Evan
04-13-2009, 08:58 AM
Cosine error is a general term used to describe the error produced when a linear measurement is made at an angle to the direction of motion and results in an error reading that is less than the degree of motion.


A plunger indicator, if grossly misapplied will also display cosine error, but it is much more of a problem with the lever style of DTI.

A plunger indicator will display cosine error in direct mathematical proportion to the degree that it is not square to the work measured. It will exist regardless of how small the angle departs from 90 degrees and has nothing to do with being "grossly misapplied" as a cause. Gross misapplication will more likely result in binding of the plunger than cosine error.

JCHannum
04-13-2009, 09:02 AM
Whatever........

The difference is in the design of the two instruments. The plunger of a plunger type dial indicator moves in a straight line, there is no cosine error inherent in the gage itself. Cosine error can be induced in the measurement taken by misapplication.

The probe of a lever type DTI moves in an arc, and cosine error is always present. It can be compensated for either mechanically or by jiggering the dial graduations, but if the lever is moved outside of the range of compensation, cosine error will rear it's head.

tdkkart
04-13-2009, 11:30 AM
Compare the DRO readings with the hand-wheel dials. I will be very surprised if there is a substantial difference between the DRO and the hand-wheel dial readings.



Don't be so sure.....
A couple years ago I was taking a test on a manual mill, was down to the point where I was about to start sneaking up on final dimension so my cuts had to be what I intended.

Measure, move the table, cut, measure again.....WTF??? hmmm??? Let's try that again......

Measure, move, cut, measure again.....WTF is going on here??
Something isn't right, I'm cutting about 25% more than I should be??

Scratch head, think, mutter, scratch some more etc.....

Hmmm, lets try this. Zero dial, zero DRO, turn dial .100", look at DRO..... .075"??

Huh??

Try again, re-zero, move dial .200", DRO reads .150"

WTF??

Call over the guy that's administering my test, demonstrate and ask him if he expects me to run the test with or without the DRO, and how the hell I'm to do this in the allotted time if I have to requalify the machine before continuing??

Test cancelled for today........

Turns out the shop was moving out some old equipment and the guys decided that the display on one of the machines that was being moved was better than on the machine that was staying, so they swapped the displays.
Problem was, the display didn't match the scales on the machine.

Handy part was, I got a 2nd chance at the test after now having seen it and knowing how to approach it better the 2nd time around.

I passed.

lazlo
04-13-2009, 01:50 PM
Turns out the shop was moving out some old equipment and the guys decided that the display on one of the machines that was being moved was better than on the machine that was staying, so they swapped the displays.
Problem was, the display didn't match the scales on the machine.

Man, that sucks! :)

Glass DRO scales are also relative measuring devices: the reader head is just counting hash marks on the glass scale as they go by. But that means the reader head needs to know the resolution of the hash marks it's counting :)

Twice I've seen DRO's reporting inaccurate results. Both times it was because the mount brackets were loose and I was either getting a cosine error (because the scale moved to a diagonal w.r.t. the table) or the reader head was flopping around with the leadscrew motion.

loose nut
04-13-2009, 07:30 PM
Boy I think I'll quite asking questions if it's going to stir up this kind of hornets nest.

When I said the DI was a comparator I didn't mean in the literal sense that Evan means only that it isn't a tool that should be use for linear measurement, which most of us are guilty of if for no other reason then they are convenient.

Now before someone jumps down my through for that statement, it's not mine, I have read it in many books and magazine articles on the subject and told this by "real" machinists (I'm holding on to my amateur standing). As someone said earlier in this thread DI's compress OK but not so much when they are returning back towards the at rest position, hence error can creep in.

When you put a piece of bar in a 4 jaw and set up a DI to center it, you take readings at points 180 degrees apart and compare the readings to find the amount of adjustment towards center, so you are using it to compare not to measure. Even if the plunger is at an angle to the work, the readings on both sides will be accurate even with a sine error because the actual distance doesn't matter, only the difference in the reading that you get compared to the reading from the opposite side, any sine error will be the same for both.

If one wants to be a real stickler, any measurement tool is a comparator, Mic's and verniers compare the thickness of a work piece against a scale, if a DI is mounted on a pillar set into a surface plate for inspection work is it not comparing the thickness of a part against a known standard. Everything else is just semantics.

All I really wanted to know is, if I get the same reading at the same point every time I try can I consider the repeatability of the scales to be adequate. I believe the predominant answer is yes. Thanks, now play nice.:D

JCHannum
04-13-2009, 07:55 PM
If you are referring to a plunger type dial indicator, you are mistaken in stating that they are not for linear measurement. They are indeed made expressly for that purpose. I have just finished installing bracketry for using dial indicators with 2" & 5" travel for table positioning on my milling machine. Dial indicators are regularly available in lengths from less than an inch up to 5" or so in length. They should and will return to zero repeatably if not damaged or allowed to become gummed up.

If you are referring to a lever type dial test indicator, the statement is only partially correct. A lever type dial test indicator will read dimensions accurately in a much smaller range as mentioned. They are limited by cosine error if used outside of that range. When used for tramming or dialing in in the lathe chuck, the error is of little consequence as the aim is for zero movement. However, these gages are used for accurate measurement in inspection and other applications and are capable of excellent accuracy if applied properly. They also will return to zero reliably if they are maintained properly.

You have referred to DI and DTI in your posts, and I have asked which instrument you are using. I have not seen a reply. The terms are not interchangeable and this makes it very difficult to make a meaningful reply to your question.

lane
04-13-2009, 08:20 PM
I read most of this but not all . Something to check is . I have seen the nut the screw rides in being loose in its housing and slipping back and forth are the screws are loose that clamp it down. This will give faults readings with a read out are what ever measuring system you are using. So check this before you go any farther.

lane
04-13-2009, 08:36 PM
And as far as this so called cosine area is concerned. Just make sure the stem of your indicator is parallel to the movement in both side to side and vertical and horizontal . If you have some mic standards set a 123 block on the table and clamp down indicated it square lay a long mic standard in a v block and push it up to the 123 block . Pick up end of standard with edge finder ZERO your measuring system remove mic standard and put in one shorter move till it zero out on edge finder check reading on system repeat with shorter standards . You can use any thing of known length . When done move back to 0 and see if numbers return to 0.

oldtiffie
04-13-2009, 08:56 PM
Boy I think I'll quite asking questions if it's going to stir up this kind of hornets nest.

...........................................
..............................................

All I really wanted to know is, if I get the same reading at the same point every time I try can I consider the repeatability of the scales to be adequate. I believe the predominant answer is yes. Thanks, now play nice.:D

Loose nut.

I think that you've got it right. Your set-up of your DRO scales seems to have no looseness in it. Your repeatability is OK.

Now just to set the "cosine error" bit to rest - as I think there is more quoting of it than knowledge of its importance or extent.

Cosine - aka Cos in a right-angle triangle is the ratio of the the length of the side adjacent to the angle to the length of the hypotenuse.

Or Cos angle = adjacent/hypotenuse

Or Cos = A/P

http://en.wikipedia.org/wiki/Cosine_(trigonometric_function)

Here is a good explanation:
http://www.cncexpo.com/IndCosine.aspx

In theory, the arm of a dial indicator (with a swing arm)should be at right angles to the surface/part being measured/checked/compared where-as the plunger of an axial indicator should be normal to the part being checked.

As Cos 0.00 deg = 1.0000 it follows that the adjacent side and the hypotenuse are the same.

Now let's have a look at the situation at varying angles:

Cos 0.25 deg = 0.99999
Error = 0.00001"/inch

Cos 0.50 deg = 0.999962
Error = 0.000032"/inch

Cos 0.75 deg = 0.999914
Error = 0.000086"/inch

Cos 1.00 deg = 0.99985
Error = 0.00015"/inch

Cos 2.00 deg = 0.9994
Error = 0.0006"/inch

Cos 3.00 deg = 0.9986
Error = 0.0014"/inch

Cos 4.00 deg = 0.9976
Error = 0.0024"/inch

Cos 5.00 deg = 0.9962
Error = 0.0038"/inch

Cos 10.00 deg = 0.9848
Error = 0.00152"/inch

The lever arm on my "adjustable finger" indicator is about 0.700" long so any so-called "Cosine error" is 0.700 x the values above.

So if my "arm" was off-set by 3 degrees from "square" or "normal", the cosine error at the end would be:
Cos 3.00 deg = 0.9986
Error = 0.0014"/inch x 0.700 = 0.0010" - jep, just a "thou" at 3 degrees.

For my axial indicator the error will be as above ie
Cos 3.00 deg = 0.9986
Error = 0.0014"/inch for each inch of plunger travel.

So a lot of this "cosine error" is not the problem it is made out to be in some or many circumstances.

It is easy to set a 90 degree/normal angle to within 2 degrees by eye and most times, that will be good enough.

So, loose nut, that is why I think you are doing OK, despite some of the adverse, and I suspect largely uninformed, comment.

I hope this has set your mind at rest.

JCHannum
04-13-2009, 09:41 PM
Tiff, I don't know whose comments you feel are uninformed, and would appreciate you clarifying that statement.

loose nut has asked a couple of questions, and there have been a couple of sugestions as to how to proceed. Lane suggested using 123 blocks, mic standards or something other of known dimension to measure at intervals, I suggested the same using measured parallels.

Simply checking that the scale returns to zero or the same number at different points will not tell the accuracy of the scale. Actual measurements of known dimension will have to be taken at several points along the travel of the scale to verify it's accuracy.

oldtiffie
04-14-2009, 12:48 AM
Jim,

I will answer in the negative here in that I do not consider your comments as being uninformed - but they should tell people the extent of the applicability of, in this case, "Cosine error" and by extension the relevance and use of either comparators and direct measurement.

I hear a lot of stuff quoted here that someone "heard" or "their mentor said" or some long-gone author of some renown said too - much of which may be either or both wrong or inapplicable.

Indicators versus comparators was a red-herring as well.

My main point was to show the OP that if his readings on his lead-screw dials was pretty close to that of his DRO scales, then it was near enough to both identify that the indicator was inadequate or faulty and to validate his DRO scales such that he could proceed onward to such other additional measures and measurements as he required to suit his purpose.

Further, the accuracy required of the mechanical set-up of DRO's relative to the machine are mainly a mechanical requirement. The "errors" of a set-up at or less than o.25 degrees (which if "FAR" better than a good DRO set-up to the manufacturer's specifications will be pretty well irrelevant as the cosine "error" will be pretty well insignificant and certainly within the +/- 0.0005" (or better) for most commonly used DRO's on mills and lathes. Surface grinder DRO's (down-feed) are quite another matter as they are an order of accuracy above those normally used.

loose nut
04-14-2009, 06:22 PM
The DI is the plunger type, and as for using them as a measurement device, we will have to agree to disagree, 'nuff said.

The DI is removed from the machine and I'm going to check the accuracy of the scales against gage blocks.

Tiffie you seem to have it right.

And that's all she wrote.

Have a nice day.