Drive to Knowledge Newsletter 23-2a

The Drive-to-Knowledge Newsletter?is designed to investigate the benefits and methods of a?Knowledge-Driven Product Development process for medical devices.

This is the second issue of our monthly LinkedIn newsletter with our thoughts on topics that we think are interesting and relevant in the domain of medical device development.

We hope that you enjoy it. We welcome feedback and comments.

?---Peter

The CAD is perfect!?So why doesn’t it work?

Today’s MCAD tools are great and enable development of complex, highly detailed models relatively quickly.?The tools enable things like interference and clearance checking in 3-space and allow for easy spin-off models for use in FEA tools.?(Note that there are numerous instances of engineers and designers doing amazing things in amazingly little time with much less refined tools – like slide rules!.?The development of the WWII P-51 Mustang is legendary – less than 4 months from contract award to first flight!)

Are the CAD models sometimes too perfect??Do they lull us into overconfidence??In a CAD environment orthogonal surfaces are perfectly orthogonal, cylinders are perfectly cylindrical, holes are perfectly centered on pins or screws.?Bodies are perfectly rigid and infinitely strong.?What could possibly go wrong?

At FPrin we follow a model-based design approach because it helps us to assess response to inputs and variation in those inputs.?Typically, we want a system to “not care” about component variation within tolerances, materials variation within spec limits, use case variation within limits of requirements, etc.?And for a measurement system or something that needs to respond to inputs we’d like that response to be precise, true, and accurate.?For something anticipated to be high volume we invest significantly in analytical modeling and simulation, including the response to variation.

But suppose we are building a one-off test fixture.?In this case we still need to think about how the system responds to imperfection, and some modeling of that response is appropriate.

If this surface is not perfectly flat will we care??Depending on if we care a little or a lot then we may do something differently – like make 3 raised features to define a plane.

?If this thing is not perfectly stiff will we care??Suppose it might deflect 50 microns under the anticipated load.?If we’re trying to measure 20 microns we care, but if we are measuring millimeters we may not care much or at all.?If the former, then we’ll want a model to help us define the spatial characteristics and make material and process choices.

Will we care if this thing creeps or relaxes under stress??If we’re trying to maintain a force or pressure, and are doing this with an otherwise stiff system, then we might care.?And if we do care, then we’ll want a model to help us avoid materials that are subject to stress or creep relaxation – like most materials used in additive manufacturing.

To adequately consider these things, we need to have some requirements, and some kind of analytical model. In other words, we need to know if we care (based on the requirement).?Does it even matter (as informed by the model)??If we cannot say what is good enough, then we can’t say what is not good enough.

Bottom line, we need to make sure that we know what is good enough for each component in the “load” or measurement path (thermal, mechanical, whatever it is). A CAD model is often not enough. An appropriately complete analytical model allows us to choose process, material, and components appropriately to be on the right side of good enough.?

PS – remember, test fixtures are the means to an end.?Often it is better to over design the fixture, and maybe design in some flexibility for “what-if” scenarios, so that we can spend our time on the real value-added work – investigating the characteristics of the device under test that are relevant to the design!