A Testing Machine Range Has a Bottom, and It Is Set by the Resolution

This page gives no advice on calibration, measurement or choosing equipment, and it gives no error limits for any class, because the table that holds them sits past the end of the preview we could read. It did not come from a forum question; it comes from pulling every standard number this site has ever cited and counting them, which is how we found that the site had written about calibration fifty one times and about the testing machine standard not at all. The browser tooling we normally use remains unavailable.

Which document, and how much of it we could read

The standard is ISO 7500-1, on the calibration and verification of static uniaxial testing machines, fifth edition dated 2018, prepared by ISO/TC 164 on mechanical testing of metals, subcommittee 1 on uniaxial testing. This is the first time this site has used it, and the first time it has used anything from that committee. The three articles before this one all came from the geometrical specification committee; this is a different corner of the same problem.

The free preview covers the front matter and the body as far as subclause 6.4.8. That matters more than usual here, because Table 2, which holds the permitted errors for each class, is in the subclause after the preview stops. The text we can read refers to that table repeatedly and we cannot see it. So this page can tell you that classes exist, what they are called, and how they set the bottom of a range. It cannot tell you what error any class permits, and it does not guess.

We also did not read the clause on the class of a range, the reporting requirements, the rule on intervals between verifications, or any annex, and we did not read the separate standard on calibrating the force-proving instruments themselves, which this one leans on throughout. We did not check whether the document is current.

One detail from the foreword is worth keeping. The fifth edition is described as a minor revision of the fourth, and the only change listed is that the definitions of seven symbols in a table were editorially revised. A whole edition, for the wording of seven definitions.

What verification actually is

The scope splits it into three parts: a general inspection of the machine including the accessories that apply force, a calibration of the force-measuring system, and a confirmation that the machine achieves the limits given for a specified class. The third is the one that produces the word on the certificate.

The scope also carries a note that is easy to skim past and hard to unsee afterwards. This is static calibration, and the calibration values are “not necessarily valid for high-speed or dynamic testing applications”. A machine verified to a class has been verified standing still.

The floor, and where it comes from

The four multipliers are the interesting part. A range does not extend down as far as the indicator can display; it stops at the resolution times four hundred, two hundred, one hundred or sixty seven, according to class. The finer the class, the higher the multiplier, so the better machine gives up more of the bottom of its range.

Multiply each pair and the four numbers nearly collapse into one. Four hundred times nought point five is two hundred. Two hundred times one is two hundred. One hundred times two is two hundred. Sixty seven times three is two hundred and one. The rule underneath is two hundred divided by the class number, with the last value rounded to a whole number. That arithmetic is ours, and the standard prints the four values rather than the formula.

This is the third time this site has divided a printed column to see whether it was one rule. Once it was, once it was not, and here it is one rule with a rounded tail. We are flagging the repetition rather than presenting the method as new; the earlier pages are linked rather than restated.

And what the resolution itself is

Since the floor is built on the resolution, it is worth knowing how the resolution is arrived at. Where the reading fluctuates, the standard says the resolution is half the range of fluctuation plus one increment.

And then a note that limits it: this “only determines the resolution due to system noise” and “does not account for control errors”, hydraulic machines being the example given. The number that sets the floor is a noise measurement, and the standard says so.

The relative resolution, the resolution as a percentage of the indicated force, has to be worked out at every calibration point, not once for the machine. That is the shape of the whole document: a machine is not accurate, a machine is accurate at a force.

The conditions the number comes with

A calibration figure is only meaningful with its circumstances attached, and this standard attaches a lot of them.

• Calibration is done at an ambient temperature between 10 °C and 35 °C, and the temperature is recorded
• The force-proving instrument shall not change by more than 2 °C from the beginning to the end of a run
• Immediately before calibrating, the instrument in position is preloaded at least three times between zero and the maximum force
• The zero reading is taken about thirty seconds after the force is completely removed
• On an analogue indicator it is checked that the pointer balances freely around zero; on a digital one, that a sub zero value is clearly shown, for example by a negative sign

One of these is a fastener in disguise. Where the force-proving instrument is attached to the machine with threaded studs, the instrument must have been calibrated the same way, and it has to be rotated through 120 degrees between each series of measurements. The calibration fixture is a threaded joint, and the standard treats its orientation as something that can change the answer.

Friction is a property under test

Two more clauses are worth having for the same reason: they name things that are usually thought of as incidental and make them part of the verification.

The mechanical accessories, a pointer or a recorder, are checked for good working order and for “resistance due to friction”, with series of measurements run both with them connected and without. The friction of the readout is a measured property of the machine.

And for hydraulic machines where the pressure at the actuator is what measures the force, the effect of the position of the piston has to be verified, with the piston in a different position for each of the three series. Where there are two pistons, both are considered. Where the piston is sitting is treated as capable of moving the reading.

What this settles and what it does not

• The lower limit of a range is the resolution multiplied by 400, 200, 100 or 67 for classes 0,5, 1, 2 and 3 respectively
• By our arithmetic those four values are two hundred divided by the class number, with the last rounded
• Where readings fluctuate, the resolution is half the fluctuation plus one increment, and a note says this covers system noise only and not control errors
• Relative resolution is determined at every calibration point
• Verification is three things: general inspection, calibration of the force-measuring system, and confirmation against the limits for a class
• The calibration values are not necessarily valid for high-speed or dynamic testing
• Calibration runs between 10 and 35 degrees, with under 2 degrees of drift, after at least three preloads
• Where a threaded stud attachment is used, the instrument is rotated 120 degrees between series
• Friction in the pointer or recorder is verified, and for hydraulic machines the piston position is varied deliberately
• The fifth edition is a minor revision whose only listed change is the wording of seven symbol definitions
• We could not read the table of permitted errors, so no class tolerance appears here, nor the clause on the class of a range, the report requirements, the verification interval, any annex, or the standard on the reference instruments
• We did not check whether the document is current, and no advice on calibration or measurement is given

The habit worth taking is to ask where in its range a number was measured. A machine is not a single accuracy; it is an accuracy at a force, with a floor underneath it that is set by how finely the indicator can resolve and how good the machine is claiming to be. A small load read on a large machine is the case the floor exists for.

The neighbouring subcommittee writes the hardness test, where the number after HV turns out to be a force in kilograms.

This is not one of the six steps. It shows up across them, or after assembly. Where the decisions that lead here were made is in specifying a screw, which sets out the order and why doing it out of order is rework.

Common questions

Does a testing machine range go down to zero?

No. The standard says the lower limit of the range shall not be less than the resolution multiplied by 400 for class 0,5, 200 for class 1, 100 for class 2 and 67 for class 3.

Why does the better class have the higher multiplier?

The standard states the values without explaining them, and this page does not guess. The effect is that the finer class gives up more of the bottom of its range.

Is there a formula behind those four numbers?

By our arithmetic each multiplier times its class number gives two hundred, or two hundred and one in the last case, so the values behave like two hundred divided by the class with the final one rounded. The standard prints the four values, not a formula.

What error does each class allow?

This page does not say. The table holding the permitted errors is past the end of the preview we could read, so no class tolerance appears here.

How is the resolution determined?

Where the reading fluctuates, the standard takes the resolution as half the range of fluctuation plus one increment. A note says that this determines only the resolution due to system noise and does not account for control errors.

Does a calibration certificate apply to a fast test?

The scope carries a note saying the calibration values are not necessarily valid for high-speed or dynamic testing applications. The document addresses static calibration.

What temperature is calibration done at?

At an ambient temperature between 10 and 35 degrees Celsius, recorded in the report, with the force-proving instrument changing by no more than 2 degrees from the beginning to the end of a run.

Why would the instrument be rotated between measurements?

Where it is attached with threaded studs, the standard requires rotation through 120 degrees between each series of measurements, and requires that the instrument itself was calibrated the same way.

Is friction checked?

Yes. The mechanical accessories such as a pointer or recorder are verified for good working order and resistance due to friction, using series of measurements taken with and without them connected.

References

• ISO 7500-1:2018, Metallic materials, calibration and verification of static uniaxial testing machines, part 1, tension and compression testing machines, fifth edition, prepared by ISO/TC 164/SC 1. Free preview of eleven pages, covering the front matter and the body as far as subclause 6.4.8

This is a digital document rather than a scan, so text extraction is reliable on it; every quotation here was still checked word by word against the file. The preview stops at the end of subclause 6.4.8, and Table 2, which holds the permitted errors for each class, is in the subclause after that. The readable text refers to that table repeatedly, so this page names the classes and reports how they set the bottom of a range but gives no error limit for any class, and does not infer one. We did not read the clause on the class of a testing machine range, the reporting requirements, the rule on intervals between verifications, any annex, or the separate standard on calibrating force-proving instruments, which this one depends on throughout. We did not check whether the document is current or superseded, because the catalogue page refuses requests from this site. This is the first time this site has used this standard, and the first time it has used anything from that technical committee. The following is our own arithmetic, not a statement in the standard: that each multiplier times its class number gives two hundred, or two hundred and one in the last case. This is the third article in which this site has divided a printed column to see whether it resolved to a single rule, which is disclosed here as a repetition rather than offered as a method, with an earlier page linked. This article did not come from a forum question; it comes from extracting every standard number the site had cited and counting them. No advice on calibration, measurement or equipment selection is given, and no brand is named. The browser tooling this site normally uses remains unavailable.