Fixture zero point: avoiding errors between shifts

Why the zero drifts between shifts

Zero rarely disappears by itself. Usually people lose it when the same setup exists in three places at once: the operator’s memory, the control offsets and verbal agreements between shifts. While the machine is run by one person this is almost invisible. As soon as another shift takes over, the difference often shows up on the first part.

The failure starts with small things. One operator measures from the stop, another from the edge of the plate. One wrote an offset in the control, the other quickly checked and slightly changed the value. In the end everyone is sure they’re working from the same zero, while in fact there are two.

The setup usually gets lost in several ways: the fixture is removed for cleaning or rework and put back slightly differently, an operator probes a different point, someone changes an offset in the control and doesn’t record it at the machine, and the next shift receives an oral explanation and understands it their own way.

Oral handover almost always causes discrepancies. One says “zero at the left edge” but doesn’t specify which element. The other hears “at the stop” and chooses the nearest convenient point. A few hours later no one remembers the original reference and what offset was added after the first good part.

Even a small shift quickly ruins the dimension. If the fixture moved 0.03–0.05 mm, you can already see it in the position of a hole, face or groove. On a production part the error repeats across the batch until someone starts looking for the cause in the tool, material, or program. Often the problem is simpler: the base shifted and no one checked the zero.

This happens especially after weekends, on night shifts and after a rushed changeover. People want to start faster and rely on memory. Memory is a poor helper here. The zero on the fixture must live not in words but in one clear place that everyone sees the same way.

The task is simple: one zero for all shifts. If every shift takes its reference from the same base and looks at the same record, the first part becomes a check, not a new search for zero.

How to choose a base on the fixture

A fixture base must give the same part position every time. If the base is unstable, the zero will wander even if the machine is fine.

First, it helps to separate three things people often mix. A support holds the part and sets height. A stop positions it along an axis or at a face. A control surface is used for measuring and referencing: it’s convenient to check size from it, touch it with a probe and set the zero in the setup sheet.

In practice it’s better to pick as a base not just any convenient surface but one that’s already machined and changes little over time. This can be a ground support pad, the face of a rigid stop, the axis of a locating pin or a fixture body plane that’s easy to check with an indicator. On CNC lathes and mills such choices usually give less scatter than referencing a random edge of the part.

A good base doesn’t cause arguments between operators. One person sets the part, the next shift does the same and the position matches without long searching. So the base should be simple and unambiguous: clear geometry, accessible for cleaning and probing, not dependent on clamp force and without temporary shims.

If you’re unsure between two options, choose the one you can verify in 10 seconds. For series production this is a practical rule.

A bad base is easy to spot. Don’t rely on painted areas, rough cast surfaces, edges with burrs, worn jaws or places where chips constantly accumulate. Moving elements also create problems: today they sit tight, tomorrow there may be play and the zero moves.

The base doesn’t have to be perfect on the drawing. It must be clean, rigid and repeatable in real work. If after cleaning and reinstallation the indicator shows the same position, the base is chosen correctly.

When to set the zero on the part and when on the fixture

The choice depends not on an operator’s habit but on what should remain constant: the part or the tooling. If a batch is made in series and the fixture isn’t changed, it’s more convenient to take the zero from the fixture. If each blank varies a little and the size must reference its actual surface, it’s better to set the zero on the part.

A simple example is a bushing on a CNC lathe. If the Z zero is taken from the face of each blank, the operator after a batch change or blank replacement checks the face, verifies the stickout and adjusts the offset. This works but costs time. If a stop in the fixture is rigidly fixed and the zero is tied to that stop, the machine gets the same starting point every time.

For series this is more convenient. The fixture stays in one place, the referencing scheme doesn’t change, and between shifts you don’t need to find the base again by the first part. The operator only needs to check that the part sits on the stop and is clamped correctly. On production machines and automated lines this approach often eliminates extra stops for rechecking.

Referencing to the part becomes a problem when batches change quickly. Suppose the day shift ran one length of bushing, at night they load a similar blank with different allowance. If the zero is tied to the blank face, the new shift re-finds the zero, checks stickout and does a test cut. A single incorrect touch shifts the whole batch. With fixture-based zero, changeover is simpler: you change the program or offset for the new size while the reference point stays the same.

But you can’t always move the zero to the fixture. If blanks vary in length or shape, if a critical dimension must be measured from the actual face or an already machined surface, if cast or forged blanks have varying allowance, or if the part is set without a rigid stop, it’s safer to reference from the part.

The logic is simple: the machine should measure from the place that actually exists on every blank, not from a hypothetical point in the tooling.

In short: fixture zero works where the tooling consistently reproduces the same position between shifts. Part-based zero is needed where the part itself defines the datum. The most common mistake is moving the zero to the fixture just for speed when the drawing calls for measuring from the part surface.

How to document the referencing scheme

The referencing scheme should answer in 10 seconds: where is the zero and what must the operator check before starting. If you need to read a lot of notes in the margins, the scheme doesn’t work.

Make one clear document on an A4 sheet or a single file. Don’t mix a history of edits, old sizes and personal abbreviations. When the scheme contains only the necessary data, the shift spends less time and offsets drift less after changeovers.

Show three things on the drawing: the zero itself, the axes and the part orientation. Mark the zero point with a noticeable sign and use the same style on all fixtures. Draw X, Y and Z axes as the operator sees them at the machine, not as the designer prefers.

If the part is only set one way, show that too. Sometimes an arrow with a note is enough: which is the front face, which side rests on the stop and on which surface the control touch is made.

A good scheme usually includes minimal information: fixture number, mounting place on the table or chuck, active coordinate system, the control point for checking zero and the actual offsets. Everything else can go to the setup card so the main sheet isn’t overloaded.

The most common mistake is making the scheme understandable only to its author. If a new operator cannot immediately find the base and the zero, the document needs rewriting.

How to hand the zero to the next shift

If each shift searches for the zero anew, the machine loses time and parts gain unnecessary risk. The zero should be transferred between operators like any standard working parameter, not as knowledge held by one person.

Start with basics. The operator cleans the fixture surfaces, wipes the seating place and checks stops. Even a small chip under a support easily causes a few hundredths of a millimeter of drift, and the next shift will look for the error in program or tooling.

After cleaning, place the fixture back in the same position as before. If there is a permanent mounting scheme on the table, don’t change it without a note in the setup card. When the position is slightly different each time, the zero on paper may look the same while in reality it isn’t.

Next, confirm the zero at the chosen point. Usually take one clear control surface or locating pin and verify the offset in the work coordinate system. It’s important not only to get a value but to record it immediately.

The setup card needs a short set of data: fixture number, mounting location, active coordinate system, actual offsets by axes, control check point, date, shift and operator signature. Such a record helps more than long marginal comments.

It’s useful to add a short verbal note: whether the fixture was removed, a stop changed, or the table cleaned after the batch. One concrete phrase is often enough so the next shift doesn’t spend half an hour searching.

The new shift then checks the first part by two or three dimensions that best show base drift. For example, measure the distance from the datum face to a hole and the X dimension between two machined faces. If both dimensions are within tolerance, the fixturing and zero were handed over correctly. If at least one is off, find the cause immediately before running the whole batch.

Example for a production part across two shifts

Take a simple production part for a CNC lathe: a steel bushing with two setups. In the first setup the operator machines the OD and first face. In the second setup they flip the part, clamp it on a finished face and machine the second face, ID and a chamfer.

The first shift starts from the fixture. The zero is set not approximately but at the same point on the stop and along the same clamp axis. After setup the operator checks the base with a probe or indicator, records offsets in the machine work system and writes them into the referencing scheme. They also note which surface X is taken from and where Z is relative to the stop face.

You can use G54 for the first setup and G55 for the second. That’s convenient: each setup lives in its own offset and shifts don’t mix them up.

The second shift does not search for the base again. They perform a short check: clean the jaws, stop and seating places of chips, verify the fixture number and offset number in the setup card, check the control dimension on the first part after start-up and confirm tool offsets weren’t changed after the previous batch.

If everything matches, the operator runs the first part at a reduced feed and measures two dimensions that show base drift fastest. For a bushing these are usually the length from the face and the OD after reclamping. If both are in tolerance, the batch continues without a new zero search.

The difference becomes visible quickly. If previously each shift spent 15–20 minutes finding the base, two shifts a day easily add up to 40 minutes of downtime. That’s not all: the risk of producing several drifted parts at the start of a shift falls, and on a series that’s direct savings in time and material.

Errors that cause drift

Drift between shifts is usually caused not by the machine but by working habits. One operator measures from a stop, another from the chuck face, a third trusts an old entry in the setup card. As a result the first part in the morning already falls out of tolerance.

Once the zero on the fixture is chosen, you can’t change it locally without updating the scheme and the record. Otherwise the day shift works by one logic and the night shift by another. Even a few hundredths quickly turn into scrap in a series.

A common mistake is setting the zero on a surface that regularly collects chips or takes knocks during part removal. Such a base seems convenient but is unstable. Better choose a stop, pin, a clean fixture plane or another rigid element that doesn’t change position after each part.

Confusion also starts when old and new offsets live in the same card. Jaws were replaced, the stop moved, but the old coordinates were left nearby “just in case.” The next shift sees two sets of numbers and easily picks the wrong one.

Loose notes mislead too. A phrase like “zero moved slightly, check the first part” explains nothing. You need a short, consistent form: what changed, which base is accepted, what offset was entered, who did it and after which operation.

After replacing jaws or stops don’t assume everything stayed the same. A new jaw can differ in fit, height or projection. A stop might not be fully seated. Checking takes a few minutes but often saves the whole shift.

After such replacements a simple control is enough: compare installed elements with the referencing scheme, clean base surfaces and stops, confirm the correct offset is active in the control, take a control touch on the base and check the first part by two dimensions, not one.

Usually drift is born not from one big mistake but from several small ones. An old sheet, a dirty base and oral handover—that’s already enough for the dimension to wander.

Quick checks before the first part

The first part after a changeover usually shows not a program error but a small thing someone missed. A chip under a support, a burr on the base or an old manual correction in the offsets table causes drift that people then hunt for half a shift.

Start by looking at the setup, not the screen. If the fixture zero is chosen correctly it should repeat without guessing. But this works only when the base is clean, the scheme is nearby and offsets weren’t adjusted by eye.

Before running, four checks are enough. Wipe base surfaces on the part and fixture. Compare the actual setup with the referencing scheme, including stop position, part orientation and touch point. Check where the current offsets are stored so there aren’t two versions of the same zero. Then measure the first part by the dimensions that reveal drift in X, Z or height fastest.

In practice the link between paper and machine breaks most often. The scheme shows one stop but another is on the fixture. The table contains an old tool offset. The part is clamped correctly but the base is different. At this moment the operator starts changing sizes, while the first step should be to restore the correct referencing.

A good sign of order is simple: another person comes on shift and in a few minutes understands where the base is, which zero is accepted and which offsets are active. If nobody can figure it out without the original setup author, the zero handover system still needs work.

Another useful habit: don’t rush to make a manual correction after the first measurement. If a control dimension is off, first check base cleanliness, the scheme and the offset record in one place. Only then change the offset. This order often prevents a double error where the shift first loses the zero and then moves the size further by manual correction.

What to do next

If you must find the zero from scratch after every shift, the problem is usually not the operator. Most often the shop simply lacks one rule: everyone should work by the same scheme and someone must be responsible for keeping it current.

Start with documents. For similar jobs create one standard form that contains the referencing scheme, the setup card, a photo or simple sketch of the fixture, zero coordinates and the date of the last change. When the form is the same, a new shift spends less time guessing and sees faster what changed.

Then assign responsibility. If a setter moved a stop, replaced jaws, added a spacer or changed the base, they must immediately update the scheme and the setup card. The shift supervisor or lead operator should ensure the new version is passed on and the old one isn’t left nearby to confuse people.

Next, introduce a simple routine: take the latest scheme and setup card, verify base surfaces and stops on the fixture, check zero coordinates by the control point and make a record after any tooling change.

After that review old fixtures where the zero is re-found almost every time. This usually means the base is inconvenient or the scheme is disconnected from the real fixture. Sometimes it’s enough to move the control point, add a permanent stop or rewrite the card in plain language to remove 15–20 minutes from every changeover.

A small example. For a production part the day shift sets the zero on the fixture in 5 minutes, while the night shift spends 25 because it uses a different sheet with last month’s notes. After introducing a single form and an update rule, times usually equalize within days.

If you’re choosing a new CNC lathe or changing tooling, discuss datuming before launching the series. At EAST CNC such issues are usually resolved together with equipment selection, commissioning and service, so the scheme between shifts becomes clear from the start, not after the first mistakes.