Recently an American-based German manufacturer was awarded a contract to produce some parts for a Japanese company. According to the contract, critical surface areas of the parts were to meet roughness criteria as defined by the R_z parameter. This appeared to be no problem. However, it turned out that the R_z specified was not the current ISO or U.S. standard version they were used to, but a much older Japanese domestic standard dating from 1982.

Recently an American-based German manufacturer was awarded a contract to produce some parts for a Japanese company. According to the contract, critical surface areas of the parts were to meet roughness criteria as defined by the R_z parameter. This appeared to be no problem, as R_z-mean roughness depth-is the most commonly specified surface finish parameter in Germany and several other European countries, instead of the U.S. preferred choice, R_a.

However, it turned out that the R_z specified was not the current ISO or U.S. standard version they were used to, but a much older Japanese domestic standard dating from 1982. Measurements based on this standard would not be fully compatible with current standards. None of the German manufacturer’s instruments were set up to measure to it, and it took a virtual search party to even find a copy of the standard document.

This situation is not unusual and as manufacturing becomes increasingly internationalized, instances like this occur with increasing frequency and with potential costly results. In Germany, as in several other European countries, as well as the United States, it is expected that manufacturers will switch to new standards as soon as they are published. However, in Japan and some other countries, introduction of a new standard does not prevent the use of the old edition, as long as there is a reference to it. If a company has been successfully making parts for a number of years using an existing standard, there is no requirement for them to change just because a new standard has been issued.

A Profusion of Rz Parameters

In the case of R_z, the situation is particularly complex because it is one of the oldest surface finish parameters. Unlike R_a, or average roughness, which considers all the peaks and valleys in a sample, R_z typically looks only at the five highest and five lowest points in a sampling length and averages peak to valley distances. Thus, many users consider it more sensitive to actual changes in surface finish than R_a. What made R_z so useful when it originated back in the 1920s and ‘30s is that it can be evaluated manually from a graph, or through the eyepiece of an optical instrument and does not require electronics.

But as technology has evolved around the world, a number of different ways to evaluate mean roughness depth have become codified in various national and international standards. Since the R_z parameter was first standardized, five different definitions have evolved, based on different algorithms or formulas, different sampling or evaluation lengths, and by using different filters. These five definitions can be summarized as follows:

• Definition 1. 10 point average distance between the five highest peaks and the five deepest valleys within an evaluation (reference) length, from an unfiltered profile.
  
  
• Definition 2. Distance between two lines, parallel to the mean line, that touch the unfiltered profile at the third highest peak and the third deepest valley, within the sampling length. This is an approximation of the 10-point evaluation, but the parameter was named R_z in several old JIS standards.
  
  
• Definition 3. 10 point average distance between the five highest peaks and the five deepest valleys, within a sampling length, from a roughness profile.
  
  
• Definition 4. Average maximum peak to valley distance within five sampling lengths, from a roughness profile.
  
  
• Definition 5. Sum of the largest peak height and the largest valley depth, within a sampling length, from a roughness profile.

This situation presents numerous opportunities for confusion and costly error. The specifying engineer might write a specification based on an old or new standard. The machinist may be working from an old or new part print. The print may have originated in an engineering department operating under a different national or international standard. The measuring instrument may incorporate any or all of the algorithms under the same or different names, or it may not.

Thus, when manufacturing engineers, machinists or parts inspectors encounter R_z on a part print, they need to know which R_z is referred to. Because there are significant differences between standards from different countries, and also between different editions or versions of a standard within a country, they need to know where the standard was issued and when.

One of the differences in R_z evaluation is in the type of filter. The first electrical filters used to separate roughness from waviness data were 2RC high-pass analog filters. These were used until new Phase Correct Gaussian filters (50% G) were introduced in ASME B46.1 in 1995 and in ISO standards in 1996.

Thus, if a part specification is based on a standard written before 50% Gaussian filters were introduced, evaluation should be based on 2RC filtering. In the real world today, many instruments are offering only 50% Gaussian filters for evaluation of all parameters, whenever they were standardized.

Comparing Results

The table, “Comparison of R_z Definitions,” illustrates the various definitions of the R_z parameter and lists the different national and international standards associated with each method. As might be expected, using different definitions of the parameter yields different results.

For example, compare results of measurements performed in accordance with the latest editions of these standards:

• R_z JIS-94, based on definition 3
• R_z ASME-02, based on definition 4
• R_z ISO-97, based on definition 5

Each of these measurements is based on different mathematical equations or algorithms. Evaluation is performed over different lengths, relative to different base, or reference lines, and using different filters. Thus, results show the following ratio:

R_z ISO-97

Accurate Results

Comparing apples to oranges is rarely helpful even when making fruit salad. When applied to R_z measurements, it can lead to untold frustration and considerable cost. Clearly, ensuring the part designer, manufacturer and customer all understand which R_z method is indicated is the best way to ensure accurate results. Conversion factors should be used only when necessary and even then with an abundance of caution.

Of the five different methods of R_z evaluation discussed, the most commonly specified and used are R_z ASME-02 and R_z JIS-94. However, some surface finish instruments today are capable of accommodating any R_z parameter or filtering method, even if they are not already set up to do so. The only real stumbling block is recognizing when there may be a difference in the first place. Q

Tech Tips

In some countries introduction of a new standard does not prevent the use of the old edition, as long as there is a reference to it.

Unlike R_a, which considers all the peaks and valleys in a sample, R_z typically looks only at the five highest and five lowest points in a sampling length and averages peak to valley distances.

Since the R_z parameter was first standardized, five different definitions have evolved.