Clear hard anodize finishing (mythical)

This entry is part 2 of 4 in the series Mythical Specifications

Once in awhile, I’ll run across a requirement to use a specification that isn’t physically possible.  Something I see from time to time is the request to apply the specification of a clear hard anodize finish to the drawing of an aluminum part.  This makes me chuckle (unless the requester is thoroughly convinced that this mythical beast really exists).

A hard anodize (Type III) finish is intended to provide wear and abrasion resistant surfaces with improved corrosion protection due to greater thickness and weight than common anodizing (Types I and II).  The goal when using hard anodizing is to have a wear index of 1.5mg/1000 cycles, according to MIL-A-8625F.

An anodize finish on an aluminum part is achieved by growing an aluminum oxide layer on its surface using direct current through an electrolytic solution, with the aluminum object serving as the anode.  The current releases hydrogen at the cathode and oxygen at the surface of the aluminum anode, creating a build-up of the aluminum oxide.   The voltage required by various solutions may range from 1 to 300 V DC, although most fall in the range of 15 to 21 V.

Common methods apply an aluminum oxide layer that is .00002 – .001″ thick.  A clear appearance remains as the thickness approaches .0006″ thick.  Thicker than that, the layer darkens to a bronze, gray, or black color (depending on the purity of the aluminum substrate).  At .0017″, the color is very pronounced.  Hard anodize specification calls for .0020″ (+/-20%) thickness.  This is far above the point where the anodizing process produces a colored finish.   Another factor is that of temperature during the process.  Hard anodizing requires the process to occur as a much lower temperature for the harder surface (higher process temperature = softer surface).  Additional coloration occurs due to the lower temperatures required by the hard anodize process.

Although a balance may be struck between hardness and clearness, the specification of clear hard anodize is not an achievable specification in a strictly technical sense.  Any compromise to get close to this specification is going to have some color and reduction in hardness or durability.

Reference: MIL-A-8625F (.pdf)

Mythical creations

This entry is part 1 of 4 in the series Mythical Specifications

Engineering sometimes calls for the use of the specification called unattainium.  In other words, at times there is just no easy way to find the balance between design requirements and reality.  Other times, someone isn’t knowledgeable enough to make certain specifications, so they come up with specs that may sound right, but aren’t real.  Anyone ever run into a set of parts that were designed with all the mating features being line-to-line?  How many of us have searched high and low for a “black alodine” finish?  Another mythical metal finish is “clear hard anodize”.  I’m going to cover some of these points in future articles.  For now, I’d like to see other misspecifications that people have experienced in the engineering field.  Please comment about what you’ve seen.

I was walking through the metric jungle

Today was a wonderfully sunny day, so I said to myself, “Hey, why not take a stroll through the metric garden.”  (Why do I ask myself such things? Don’t ask me.) The metric stroll should be easy enough, with its scientifically simple base ten measures.  The simple meter is wonderfully divided up into 100’s and 1000’s for convenient lengths of measurement.  It also quickly multiplies into…umm, kilometers.  No one uses hectometers? Oh wait, the French kind of do to derive their hectare.  It’s funny, that hectare isn’t listed in SI.

It’s arbitary

Sure, the imperial foot may have been based on someone’s actual foot length, but it is a useful length for some industries. Its not nearly as arbitary as the meter, which is defined as the distrance that light travels within a vacuum in 1/299792458th of a second. Why does SI use a bizarre fraction to define the core unit of measure for their decimal system?

Maybe base ten numbers aren’t all they are cracked up to be. What, never heard of metric foot? Or for that matter, metric ton, metric inch, or metric mile. Why do all of these units exist? For all the berating that the imperial system gets, the measures within it are based on real world needs. Since ancient times, units very similar to the modern imperial system have been common place. That said, it may be important to note that both SI metric system and imperial system have goofy offshoots.

Missing units!

Hey, what happened to the liter? An entire unit of measure for volume is missing from the international standard! Did aliens abduct the liter for use on their alien world? Well, no. It’s actually very common in the US, if that doesn’t seem ironic. Oh, and don’t get diehards started on a discussion about the correct spelling of meter or liter!

Psst, USA is metric!

Something else that is ironic, the USA has been legally metric since 1866.  So, it’s not true that the USA isn’t metric.  Neither is it true that the rest of the world is 100% metric.  Specific industries, companies, populations and individuals still have the right to choose their standards and measures, both in the USA and elsewhere.  Pipe threads in France are NPT, not the ISO sizes that were meant to replace them.  Pints of bitter are still actual pints in Britain.  Speaking of Britain, I’m reminded of the TV show Top Gear.  Miles, horsepower and inches are so commonly used on that show, I forget that the UK is supposedly metric.

This stroll through the metric garden is starting to look more like a forced hike through the metric jungle.

Malware and your USB stick (USB Safety part 1)

For us old-timers, there is a memory of the old days when we passed around floppy disks (floppies) to share computer programs, data files or images.  Floppies where great because users could easily read and write to them.  It didn’t take long for viruses and other malware to begin spreading through the sharing of floppies.  In fact, floppies from unknown sources where often handled with suspicion.  People would frequently scan floppies using anti-virus software.  These days, the floppy disk has almost completely disappeared, along with the issue spreading malware on them.

A whole generation of people has grown up without an easy-to-use read/write exchange media similar to the floppy disk.  For a long time, sharing data was in the form of CD-ROM and DVD.  These require special software to create and read.  They also don’t allow new information to be added to them once they are created.   It is very difficult for malware to spread via these formats for that reason.

With the advent of the USB memory stick or thumb drive (sticks), we now have a new easy-to-use read/write exchange media.  Usage of sticks has increased drastically in the past couple of years.  Us old-timers have looked upon these sticks with the same suspicion we use to reserve for floppies.  I never allow sticks from unknown sources to be plugged into my computer.  People who are new to the realities of the Information Age (particularly, younger people or others who have just started using computers within the past few years) don’t have this same prohibition.  I’ve witnessed people gleefully passing around sticks to share files or run software on various systems.  I’ve watched as people would use their sticks for both work and personal purposes (back in the old days with floppies, this was an absolute no-no).  As a result, industry is now witnessing computers and networks get infected, just like the old days.

Such usage of sticks has rebirthed the spread of malware.  According to Trend Micro, sticks and other types of external drives are highly common sources for the spread computer virus infection.  The problem is getting worse very quickly.  Companies and private users are now faced with this new onslaught of malware.

Rounding of numbers

On most computer systems, decimal numbers that have 5 as the last digit are automatically rounded up when removing a decimal place. This may create a problem.

Some people have a rule that SolidWorks drawings should not have overridden dimension values (Override values).  I generally agree.  Yet, there are several legitimate reasons to use Override values.  One major reason is for proper rounding of linear dimensions for removed digits after the decimal.  Currently, SolidWorks offers no option that allows the user to automatically round dimension numbers in a way that is consistent with current industry standards and practices.

SOLIDWORKS 2015 now has several rounding options that follow the rules below.  More information, please see SOLIDWORKS What’s New Rounding article.

Rounding rule for dimensions

On most computer systems, decimal numbers that have 5 as the last digit are automatically rounded up when removing a decimal place.  For example, the number 1.425 rounds up to 1.43.  This creates a problem.  Most standards require that such numbers are rounded to the nearest even number in the last decimal place.  For example, that number 1.425 should be rounded to 1.42, and 1.435 should be rounded to 1.44.

ASTM E 29 states:

6.4.3 When the digit next beyond the last place to be retained is 5, and there are no digits beyond this 5, or only zeros, increase by 1 the digit in the last place retained if it is odd, leave the digit unchanged if it is even. Increase by 1 the digit in the last place retained, if there are digits beyond this 5.

NASA’s Engineering Drawing Standards Manual states:

When the first digit discarded is exactly 5, followed only by zeros, the last digit retained (i.e., the digit preceding the 5…) should be rounded upward if it is an odd number, but no adjustment made if it is an even number. For example, 4.365, when rounded to three significant digits, becomes 4.36. The number 4.355 would also round to the same value, 4.36, if rounded to three significant digits.  This procedure is known as odd-even rounding.

It is my understanding that this rule helps reduce statistical bias by allowing different numbers to be rounded up or down.  Using the computer default rule (5 is always rounded up) only allows for the upward rounding of such numbers.  This can create greater statistical errors, particularly when compounding rounded numbers to derive further rounded numbers.

Rounding as it affects tolerances

No rule is absolute.  There are other considerations when rounding.  A number should never be rounded so that it increases the original limits of a dimension.  Although this rule mostly applies to inspection techniques, it can also apply to specification.  For example, if there is a feature whose size limits are 1.255-1.275, the specification cannot be rounded so its limits are 1.25-1.28.  In such a case where rounding occurs, the specification limits should be 1.26-1.27.  Fortunately, this isn’t something that often occurs in mechanical design (though it does pop up when trying to apply dual dimensions).

Usually, rounding the limits is something that more often happens in quality assurance during incoming inspection of products.  In such cases, Interpretation of Limits rule from ASME Y14.5 declares limits are absolute.  For example, 12.25 MAX is the same as 12.2500000000000000 MAX.  If the feature measurement is 12.2540, that measurement should not be rounded to 12.25, as it is still out of tolerance because it exceeded 12.25.

SolidWorks should supports more rounding options

Right now, SolidWorks does offer one rounding option for dimensions.  In documents options, there is a setting to round numbers to the nearest fraction, but only if fractional numbers are in use.  I would like to see other rounding options supported, but not a document option.  SolidWorks should have a setting added to the dimension PropertyManager that allows the user to establish a rounding rule for a particular dimension.  For each dimension, users should have a choice to use the odd-even rounding rule, nearest fraction rounding rule (only when fractional numbers are in use) or always round 5 up rule.  This shouldn’t just be for drawings.  It should also be available in the model because they are often used as part of the product definition and because dimensions in the model can be inserted into a drawing.

For now, one can use Override values on the drawing.  The drawback to this is that Override values do not automatically update if there is a change to the associated model geometry.

So, this sounds like this issue should be yet another Enhancement Request.

As of SOLIDWORKS 2015, there are several options for numerical rounding which are available.

  • Round half away from zero, where the only digit being removed is 5, then round the last remaining digit away from zero.
  • Round half towards zero, where the only digit being removed is 5, then round the last remaining digit towards zero.
  • Round half to even, where the only digit being removed is 5, then round the last remaining digit so that it is an even number.
  • Truncate without rounding, where any and all digits being removed have no effect on the last remaining digit.

There is also an option to only apply alternative round methods to dimensions, with the setting Only apply rounding method to dimensions.  When this setting is checked, round half away from zero method is applied to all system and properties values, but the alternative rounding method (round half towards zero, round half to even or truncate without rounding) is applied specifically to dimensions.  Without this option checked, the chosen rounding method applies everywhere in SOLIDWORKS.

To account for dual dimensioning issues, tolerance rounding includes an option to fit the secondary unit’s tolerance range so that it does not extend outside of the primary unit’s tolerance range.  To use this capability, goto Tools > Options > Document Properties > Dimensions and click on Tolerance button. In the Tolerance dialog, check the option Inward rounding of secondary unit tolerance extents

Drawings represent final product

One comment I’ve seen about ASME suggests that it is geared towards fully detailing product definition.   One trap that rookie designers and engineers will often fall into is over-specifying their parts by placing manufacturing process information on the drawing.

The new designer may do this because maybe a machine shop made the part wrong and was trying to work the rookie’s inexperience to weasel out of their responsibility.  Maybe someone in Quality Control was confused by a drawing because they don’t have adequate blueprint reading skills, so they come to the new designer to ask that more information be spelled out on the drawing (when it is already fully specified).  These are just a couple of examples.  Often, new designers don’t know why manufacturing processes are not included on drawings, nor even that there exists standards that forbid it.

ASME Y14.5-2009 (and previous versions) states:

1.4(d)The drawing should define a part without specifying manufacturing methods.  …However, in those instances where manufacturing, processing, quality assurance, or environmental information is essential to the definition of engineering requirements, it shall be specified on the drawing or in a document referenced on the drawing.

It is usually pretty obvious when manufacturing methods are necessary to the engineering requirements, even to the individuals new to the field.  Unless one is in particular industries, manufacturing methods are almost never required.  A drawing should fully detail the final product without over specification.

ASME Y14.5-2009 adds as an example:

Thus, only the diameter of a hole is given without indicating whether it is to be drilled, reamed, punched, or made by any other operation.

The manufacturer is responsible to provide a final product that complies with the drawing regardless to the processes they use.  It is still important for designers to know the processes that will most likely be employed, so they know that the product is economically manufacturable.  This does not mean that they should unnecessarily limit the manufacturer to particular processes.