Ra vs Rz: why is there no exact conversion, and which should you specify?

Why is there no exact conversion from Ra to Rz?

The ratio between Rz and Ra depends on the shape of the profile, not on how rough it is. A regular turned surface and a ground surface with the same Ra can differ in Rz by close to a factor of two.

The ratio can be worked out exactly for idealized profiles, and the table shows how wide the spread is. A pure sine wave gives a ratio of pi, a symmetric sawtooth gives four, and the cusp left by a round tool nose at constant feed gives close to four, which is where the popular rule of thumb comes from. Random profiles from grinding, blasting or polishing typically land higher, because their tallest peaks and deepest valleys sit several standard deviations from the mean line, while Ra on a near-Gaussian profile is only about 0.8 of one standard deviation. Add one deep scratch or a pull-out to an otherwise clean surface and Rz climbs while Ra hardly moves.

So when a chart says Rz equals some multiple of Ra, it is assuming a profile shape. Even the old German conversion chart published with DIN 4768-1 drew Ra against Rz as a scatter band, to be used with a safety margin, rather than as a single line. A chart value is useful for a rough sanity check but cannot decide whether a part conforms. If the drawing says Rz, the part has to be measured in Rz.

Which parameter should you specify for sealing surfaces?

For a dynamic seal counterface, specify Ra as a range and add an Rz limit, because the seal is damaged by peaks that Ra cannot see. For static O-ring and gasket faces, an Ra maximum is usually enough, with an Rz or Rz1max cap only where scratches have caused leaks before.

Seal standards already work this way. The national adoption of the PTFE rotary lip seal standard, IS 15545-1:2004 (identical to ISO 16589-1), gives the ground shaft counterface an Ra range of 0.2 to 0.63 µm or an Rz range of 0.8 to 2.5 µm, measured in the axial direction. Note the word "or": the two ranges are alternatives, not a conversion. Their limits sit in a ratio of about four, while a ground texture usually runs higher, so one shaft can fall inside one range and outside the other, and the drawing should say which one governs. The same standard adds that the seal contact surface shall normally be free of machining leads. For surfaces where single deep marks decide the function, sealing surfaces among them, the Mitutoyo quick guide names Rz1max as the preferred parameter, so that a single deep mark cannot pass on the strength of a good average.

Three habits make a sealing callout hold up at incoming inspection. Put the lower limit on the Ra range, because a mirror-polished shaft runs dry and wears the lip. Name the measurement direction, since a turned shaft measured around its circumference reads very differently from the same shaft measured along its axis. And say what lay is permitted: rotary lip seals generally want plunge-ground shafts with no helical lead, which neither Ra nor Rz detects, so it has to be written as a note.

For static faces, the texture that matters is usually directional. A radial scratch across an O-ring land leaks at any Ra. The callout should therefore control lay and defects first, with Ra as the general texture limit; the surface finish chart guide lists the typical Ra targets seal makers publish for glands.

How do you specify roughness when the drawing says Rz and the shop quotes Ra?

Keep the parameter the designer chose as the governing requirement and never replace it with a converted value. If the supplier or your own inspection works in Ra, add Ra as a second, separately toleranced requirement, not as an equivalent.

A converted value creates a part that can pass one requirement and fail the other. Suppose an original German print says Rz 6.3 on a seal land and someone rewrites it as an Ra limit of 1.6 µm, a typical chart pairing. A ground or blasted surface measuring an Ra of 1.4 µm per ISO 21920-2 methods can easily carry an Rz above the original limit, and the part now conforms to your drawing while failing the customer's. The callout examples below avoid that by keeping each parameter as its own requirement.

On the RFQ, say four things in words as well as on the print: which parameter governs on each critical surface, the standard and edition the callouts follow, the measurement direction, and what evidence you want with first articles. A profilometer report should name the parameter, the cutoff and the number of sampling lengths, because Rz averaged over fewer sections is not the same measurement as a full five-section value. The Mitutoyo guide notes that when a feature is too short for the standard stylus travel, the number of sampling lengths has to be reduced and recorded on the drawing.

Two of the documents Saola lists are relevant here: the first-article inspection report, which measures every toleranced dimension on the first piece from new tooling, and the dimensional inspection record sent with each lot, both described on the quality page. Those pages do not list a roughness instrument, so if a profilometer trace in Rz or Ra is a hard requirement, say so on the RFQ and it will be answered before tooling rather than discovered at your dock. Surfaces that get plated or anodized afterward need one more line, stating whether the roughness applies before or after coating; the coating sequence guide covers that, and the finishes our partner plants apply are listed on the surface finishing page, with the quote naming where each step is done. When the drawing is ready, send it for a quote.