Sensitive dial indicators

First of all, this thead I'm starting may be of interest to some but few have a need to delve into measurements finer than 0.001." So follow the discussion but unless you have a need to calibrate or certify mechanical masurements most of my commentary is intended for the theoretical minded and the accuracy junkies.

I refer to "dial indicators" and here I mean those with a conventional round face, radial plunger, and clockwork guts. Indicators that sense with a lever arm or electronics are not included in my comments.

The on-going thread about straight edges and flats and the need for highly sensitive dial indicators got me looking over the market. There's plenty of 0.0001" graduated indicators available both in the new and used (read auction) market. Some have quite a long range of travel. Those who intend to read dial indicators down into the millionths range will need to pay very close attention to short coupled stiff set-ups and temperature differences.

Anyone who's ever dealt with repeatable, certifiable, highly accurate measurement knows to treat dial indicators as comparative instruments. You set a comparative instrument by adjusting the zero on a dial indicator to a reference. The reference standard may be a certified gage or a stack of Jo blocks. It may also be used to register departure from the linear axis of a machine tool provided reversal error cancelling technique is used to separate work error from machine tracking error. Regardless the zero on the indicator dial is the target and all departures of the needle from zero representing tolerance band or out of tolerance are taken on faith in the indicator's short range accuracy or by setting to another tolerance reference.

So the message is: don't read a long travel 0.0001" graduated dial indicator and expect the dial reading to correspond to the actual distance traveled without checking it by means of length standards or other references. 0.0001" is an almighty small distance and taking readings from dial indicator's for actual data collection of linear motion is taking a lot for granted.

I gathered my 0.0001" reading dial indicators and set up a test rig. Some of these indicators were frankly beaters (one a dud I did not test because of a stuck plunger), most were in good shape, and two were new in the box. Using the measuring head of a large micrometer as a means of offsetting accurate distances, a hunk of iron for a framework, and a welded tab drilled and set-screwed to accept an indicator barrel I made up a quick and dirty dial indicator inspection station.

My shop was at 54 F degrees for the duration of my experiments. Most measurements and data points were short distance where environmental temperature variations of less than a few degrees represented error in the millionths and thus negligable.

Most all my indicators performed well when cycled within their range although reverse repeatability seemed to suffer in proportion to neglect and the time since the last attention from the instrument tech. If I dithered the set-up with a tapping pencil the reluctant movements settled into conformance with their accuracy specs.

Most of my 0.0001" indicators were honest in that if I dialed off 0.0250," the dial reading was 0.0250." Size of the indicator dial (I have AGD 1 thru 4 indicators in my collection) doesn't seem to have an effect on accuracy although naturally the larger dials are more easily read from a distance. The jeweled mevements were definitely smoother and more responsive and required less dithering to settle the movement.

Two of my beater indicators had hesitations on their movements. A flush with lighter fluid (my fav solvent for all things small in scale) and a blow out with canned air fixed one but the other (a very nice Federal E3BS-R2) seemed to have damage to the plunger rack.

I'm not getting into brand names except to say that Starrett, Mitutoyo, and Federal are represented. All performed equally well having regard for state of neglect.

One new in the box dial indicator was a cheap ChiCom copy of the famous German built indicator sold under a variety of names for many years, most recently, Mahr. These indicators sell for $250 to $600 over the counter and are distinctive because of their semi-circular case shape and 180 degree reading dial. This ChiCom copy indicator is sold by www.shars.com and has a +/- 0.002" range graduated in 0.00005." It is jeweled and the works inside were of brass.

If I'm any judge, the Shars is a fairly well executed knock-off of its prototype. It sells for $115 from the catalog. It a features shock-proof movement and low plunger force. If any are looking for a budget sensitive dial indicator I would give this one a qualified plug.

My day of monkeying around has shown me that 0.0001" indicators are quite rugged if fitted with a shock resistant mechanism, quite fragile if not. If they work at all they hold their accuracy well.

When using sensitive indicators one should reckon with all the factors that haunt precision measurement and take steps necessary to reduce errors to a minimum (dithering, watching environmental temperature, handling heat input, stiction, deflection (particularly in the set-up), cosine error, repeat zeros, and the others). A nice stiff dial bore gage might seem to offer bullet proof readings tempting one to split "tenths" but trust me: take the final readings with a reference check to the gage block stack you have next to your work station - and don't forget temperature difference between the work, the bore gage, and the setting reference.

[ 09-22-2006, 01:13 PM: Message edited by: Forrest Addy ]

Hi Forrest

Thanks for the nice write-up. I personally try to avoid overly sensitive indicators (> 0.0005) whenever possible because of the generally steeper actuation forces required due to higher gearing ratios. While I do have a 0.0001 unit mounted on a stout little gran-check, I've just never felt very comfortable using something that sensitive in a less rigid set-up, say in a typical mag base - even a Noga - due to somewhat unpredictable flex issues. That said, I'm just curious as to weather digital indicators might require less plunger force due to lack of need for mechanical advantage. I've never actually seen one up close and personal so I really have no idea.

Kind regards,

Jack

every time i smack my mits .0001 with a hammer
it gives me a totaly differant reading
wuwt

seriously
i'm allways amazed when i'll walk up on a guy saying this machine is junk
he got a $20 1 inch travel indicator with the plunger at 45 degrees to his work.
thanks forrest for reminding me there is inteligence on the planet

I personally try to avoid overly sensitive indicators (> 0.0005) whenever possible because of the generally steeper actuation forces required due to higher gearing ratios.

Click to expand...

This is one place where electronic DIs demonstrate superior performance. The actuating force is independent of the scale, so you can obtain very high resolution without excessive pressure on the work.

Leigh

Forrest, In the surface plate thread you mentioned a "reed hinged motion multiplier". While I think I get the general idea, I'll confess to ignorance as to the math you mentioned in it's design. Spent about an hour googling it in various ways, and apparently it's not a real hot topic of conversation.

Any ideas on where I could get a look at such an animal?

And thanks for the above post; gives me hope for this box-load of Federals I've collected.

Having fun trying to measure stuff I got no business measuring,
Rob

Perhaps the german one you were thinking of
was Mahr?

Jim

Jim, Mahr is it. My bad I will fix it.

I've actually got one of those at work.

I think it's a "millmess" which seemd to be
an unfortunate name to me. Sort of like the
chevy nova.

But it's been sitting in a lab where
everything seems to take on a patina (the *bad*
kind, not the good kind - is there any sort
of acceptable patina on a dial gage??) after
a few months. Somthing in the air I guess.

In honor of this thread I will dismantle it
again and put it right.

Jim

forrest you are right about most people not needing mearurments finer than .001. i work alot in "tenths" and quite frequently in "millionths" but i use lasers and auto collimators for most of these measurements. i have a selection of indicators like you describe and most are .0005" some are .0001 and i even have one of the mahrs that you talk about, i bought it years ago for something i was working on and basically once that project was done i had no more use for it, i keep looking but every time i think i have found a use for it someone opens the door to the shop and the weather changes all my readings. these things are too touchy to just grab and make quick checks on things in a machine or even with a surface guage on a surface plate. usually when i find myself using on of these types of indicators i am using it in purpose built fixture to measure specific features on specific parts and with care and good guage practice they are very accurate and very reliable.

I have a love/hate relationship with my indicators. I only occasionally need to work as carefully as tenths so just got by with an Interapid half thou reading test indicator with 1-1/2" dial. I can see small movements with it to the point that if I zero carefully I can check which edge of the 0 numeral the pointer is on.

So I finally got a real tenth test indicator (not major brand) but find I have to do the tapping routine to keep it honest about whether it actually moved or not. So in practice I often wind up still using the Interapid because I have more confidence in it, and I humor myself that I can see a tenth or two movement on it if I need to.

Jan

I have a hobby shop and mostly make fixed blade knives, so I don't really need anyting much more precise than a set of Harbor Freight calipers too often in my shop. Even considering that, I will admit to being obsessed with precision measuring stuff.Even if I start making folding knives, anything less than .001 is pretty meaningless. So why do I have a set of Starrett Weber gauge blocks? I have the big set, but was sorely tempted to buy the 21 piece set that was in the for sale area here a few weeks ago. I have no idea why. My home shop's not enough of a controlled environment that I could measure to those tolerances. Besides, if I really needed something that precise, I could visit my wife at work in her temp and humidity controlled metrology lab There's something sexy about a tall woman in safety glasses bent over a SuperMic

Great post Forrest. I always enjoy your stuff.

I'll second the kudos.

This makes me feel better about saying "Eff It" when I can't get work in the 4-jaw lined up to better than .001 even with the DTI.

It's also timely as I have recently decided to upgrade my indicators from the above-mentioned $20 import. I guess there is no sense in a modestly-skilled home-shop guy to get all worked up over the higher resolution gauges.

This might be a good time for some folks to mention what could be some 'typical' jobs in a normal home shop, requiring better than .001" measurement?

Seems that making bearing bores is tight work, yes? I think I've read that .0005" on diameter is in the range here. For instance - ya scooped up some nice ABEC 7 bearings on eBay and would like to make a spindle unit of some sort, maybe for a small CNC mill or such. I can imagine a homeshop guy doing this. Complete, small spindles aren't all that plentiful on eBay, but bearings come by the boatload.

Anything else that a guy with typical manual equipment might find himself doing that requires such tolerances? I've always thought tenth work was best left to job shops. But I'd love to hear that I'm wrong - that there are some normal tight-tolerance jobs one might tackle with a modest shop, and do them well with the right technique.

Maybe it is worth mentioning that an 0.001 indicator "reads" to tenths anyhow.... That's any wiggles less than a full division....

You can probably "see" down to at least 3 "tenths" that way, and you get to do it with an indicator that doesn't jump out of its case when a gnat hiccups next door.

Maybe the friction is larger than it should be for unltimate accuracy with the 0.001 indicator.... but the "reflected inertia" of the gear train in a very sensitive indicator is significant also.

This applies to levels also.........

forgive my ignorance

but what kind of parts need to be worked to an accuracy of , lets say, 25 millionths or better ?

( I once made gyro parts to .00002, but all they really needed was .001, the greater accuracy was only to make the print look good (I got that from the horses mouth))

some shafts to +.0002 - 0


I am sure some of the other folks would be interested as well

make that .0002 on the gyro parts

We used to call that extra zero behind the decimal point that popped up in prints out of the blue 'the zero of ignorance'.
As good as it sounds... taking the tolerance on a part made on a Bridgeport from .001" to .0001" MIGHT be doable, but it'll cost ya!

Obsessive/compulsive rant alert!!

Small tenths are quite important in a number of common applications.

Take surface plates and measuring instruments for example. Generally you have to think of total tolerance as something like a budget where so much of a limited resource goes for rent, so much for lights, telephone, food etc. When it comes to error the total amount has to be apportioned in a number of ways. In measuring instruments errors cannot be compensated for. Thus the instrument error has to be much smaller than the tolerances. I've long used 1/4 to 1/20 (depending on application) instrument or measuring error as the maximum fraction of the smallest total part tolerence.

Thus successful inspection of precision parts from a surface plate or a granite flat depends on the flatness of the plane represented by the reference surface. If you're using a height gage and a test indicator to determine plane surfaces and their parallel offset, horizontal bores from a reference plane, etc you need the surface plate to be flat to a small part of the smallest part tolerence.

In automotive manufacturing, parts are inspected and grouped by their tolerances so they can be selectively assembled with their mating parts for optimum performance, life, or what ever attribute that the design requires that selective assembly will enhance. Wrist pin bores in piston and wrist pins are commonly clearanced to 0.0002" to 0.0003" at room temp. At engine operating temp the aluminum piston expands at a greater rate then the pin, increasing the clearance. The bores are honed and the wrist pins centerless ground to very close tolerances. Larger installed clearances shortens the life of the wrist pin/bore significantly. Integer thousandths in size control in this case does not provide sufficiently refined resolution whereas tenths of thousandths does.

Rolling element bearings especially ball bearings have limited internal clearance. If the inner race is expanded by an oversized bearing seat on a shaft it may take up the internal clearance and force the balls into the races. Here again ten thousandths tolerance resolution is essential to tolerance to fit to acheive longevity and proper function of the machine elements in question.

Hydraulic valve spools are selectively fitted to small clearances in their bores in the bvalve body. The clearance between spool and body bore is more like a restriction than an actual seal but this is all there is to prevent leakdown. Coarser tolerances may result in leakers or immovable spools.

Diesel injector parts have clearances measured in millionths.

0.000020 tolerance in some parts of a gyro are essential if it is to run correctly and provide reliable pitch/yaw/roll rotation sensing to pilots and instruments without precessing or drifting.

High performance machine tool spindles and their bearings have tolerances in the low tenths and even to the millionths for they bore the holes and turn the diameters for parts whose tolerances may be expressed in tenths of thousandths.

It is true that any ignoramus can write a string of zeroes to the left of a significant figure and then place a decimal point. It's also possible that there may be a real design requirement for a close tolerance. Any machinist who tampers with part tolerences on his own initiative may be risking lives should they depend on the part in question. Scoff all you want but make it like the plan.

So before you make wholesale declarations about an alleged silly tolerance you must consider its application.

As a young fellow, I used to listen to lunchtable "experts" sneering at things they never bothered to research. I soon learned their expertise was self-assumed. They were like the blind men examining the elephant who made the bad calls (It's a wall/it's a tree/it's a snake) based on insufficient information. They drew an illusory authority from the witless by blathering unsubstantiated criticism. Although such behavior is a common human trait, anyone who took these guys seriously was asking for it: cynicism is the eighth deadly sin. That said, anyone who discounts all their assertions on the basis of a few stupid remarks may discover that expertise can be real spotty. A fellow workier may be a raving moron in some topics but a real authority in his specialty.

I recall Bob Love, a misanthrope, misogynist, and political reactionary if ever there was one. He'd scoff at anthing he saw and sneer at the probable intelligence and motivations of passers by. However he was an expert on balancing large shafting and propellors on the Yakimoff balancer, a process requiring minimum equipment and much expertise.

An apprentice could safely ignore Bob's remarks about Senator Goldwater, the plan file lady's figure, and the need for better bomb shelters. OTH an apprentice ignored Bob's remarks on dynamic balancing at his peril. Bob really knew his stuff in that department.

As for home shop uses.... Just had the 0.0001 indicator out for a good use of it...

Scraped the crosslide on the lathe recently. When deciding when I was "done", I had two things to satisfy as far as the compound base surface.

One, the round area the compound sits on had to show up with good bearing when marking on the blued granite flat.

Second, it had to be parallel to the under surface that slides on the crosslide ways.

The second was determined by putting the indicator on a base at the correct height, and then moving the crosslide around under it on teh flat, with the plunger bearing on the surface in question. (I need a bigger flat... tight space)

I was "done" when I had a couple tenths variation and good bearing.

I probably could not have effectively used an 0.001 indicator for that. Even though it was a comparison measurement, and not dependent on absolute error, it just would not have been sensitive enough to see the errors.

A few hundred years ago I overhauled air driven gyros (DC-6 horizons) and 3-axis rates for 720's
As I recolect, the tolerences were "loose" to almost "too loose".