Is possible to make a part “too good”? The answer was yes in the following case and I suspect there are many other examples out there. Often the requirements placed on engineering drawings are copied from previous designs. When done with little or no understanding of why the original design had such a callout, unexpected results will likely take place. The situation most often experienced is too tight of a tolerance that results in unnecessary cost to manufacture and additional cost due to scrap or rework but not poor performance.
Why would engineers and designers use specifications they do not understand? There is always the pressing schedule and time to market. Not reinventing the wheel is also very popular. If it was good enough for design 1 why not for design 2? If design number 1 had been truly understood, carrying the requirement to the next design could be valid.
The case at hand had to do with synchronized manual transaxles. The synchronizer consists of a few parts that allow the gears to mesh without grinding when shifting from one gear to another. Without going into the theory of synchronizers suffice to say there are surfaces with teeth and others smooth. The smooth surfaces had a surface finish callout that had been successfully used for years perhaps decades. So again, why reinvent the wheel?
A new and similar design was created and the callout for surface finish was copied. It had worked for millions of parts so why look any deeper. The new design was expected to be in high demand so it was decided to have the parts made both in-house and at a Japanese company. Parts coming from this Japanese supplier were always dead nuts nominal with little variation. When the transaxles containing the subject parts were tested, the manual shifting did not feel right. The shifter seemed to stick and shifting was very poor.
As I described in previous post the first item checked is the drawing callouts. This of course was OK. Next check the “offending” parts and they too were to specification. Someone then decided to check the in-house parts made to the same specification that worked successfully for years.
Feature callouts such as holes or linear dimensions have upper and lower limits. Surface finish is often specified as a maximum with less roughness than specified expected to be a benefit. As we learned this was not the case for this design.
When the in-house parts were inspected in the lab vs. the production QC the parts were not to specification. They likely never where to specification. They were rougher than the surface callout on the drawings. The parts that were not to specification worked and the parts out of specification did not. How can that be possible?
If you have ever experience rubbing Jo Blocks together you know how they stick together as if magnetic. The reality of the attraction is the elimination of air between the two surfaces. The supplier had made the parts to specification for the first time, which resulted in the Jo Block effect - squeezing out air and much of the lubricant. When it was time to shift gears the parts on the synchronizers that were to separate required an unacceptable amount of force to get them apart. The solution was an additive to the lubricant that altered the surface tension. That addressed the transaxles in the field. Going forward the surface callout was relaxed to provide sufficient roughness to maintain a film of oil and not result in a Jo Block effect.
The case here points to a lack of QC at that time. Had the parts been made correctly or inspected correctly a better understanding of the effect of surface callout would have resulted. To rely on CQ is not a good approach to design.
There is a deeper lesson to learn here about the nature of design. There are the things we know; the unknowns, and the bugaboo - the unknown unknowns. I plan to post some thoughts on this aspect of design in the next few weeks.
Why would engineers and designers use specifications they do not understand? There is always the pressing schedule and time to market. Not reinventing the wheel is also very popular. If it was good enough for design 1 why not for design 2? If design number 1 had been truly understood, carrying the requirement to the next design could be valid.
The case at hand had to do with synchronized manual transaxles. The synchronizer consists of a few parts that allow the gears to mesh without grinding when shifting from one gear to another. Without going into the theory of synchronizers suffice to say there are surfaces with teeth and others smooth. The smooth surfaces had a surface finish callout that had been successfully used for years perhaps decades. So again, why reinvent the wheel?
A new and similar design was created and the callout for surface finish was copied. It had worked for millions of parts so why look any deeper. The new design was expected to be in high demand so it was decided to have the parts made both in-house and at a Japanese company. Parts coming from this Japanese supplier were always dead nuts nominal with little variation. When the transaxles containing the subject parts were tested, the manual shifting did not feel right. The shifter seemed to stick and shifting was very poor.
As I described in previous post the first item checked is the drawing callouts. This of course was OK. Next check the “offending” parts and they too were to specification. Someone then decided to check the in-house parts made to the same specification that worked successfully for years.
Feature callouts such as holes or linear dimensions have upper and lower limits. Surface finish is often specified as a maximum with less roughness than specified expected to be a benefit. As we learned this was not the case for this design.
When the in-house parts were inspected in the lab vs. the production QC the parts were not to specification. They likely never where to specification. They were rougher than the surface callout on the drawings. The parts that were not to specification worked and the parts out of specification did not. How can that be possible?
If you have ever experience rubbing Jo Blocks together you know how they stick together as if magnetic. The reality of the attraction is the elimination of air between the two surfaces. The supplier had made the parts to specification for the first time, which resulted in the Jo Block effect - squeezing out air and much of the lubricant. When it was time to shift gears the parts on the synchronizers that were to separate required an unacceptable amount of force to get them apart. The solution was an additive to the lubricant that altered the surface tension. That addressed the transaxles in the field. Going forward the surface callout was relaxed to provide sufficient roughness to maintain a film of oil and not result in a Jo Block effect.
The case here points to a lack of QC at that time. Had the parts been made correctly or inspected correctly a better understanding of the effect of surface callout would have resulted. To rely on CQ is not a good approach to design.
There is a deeper lesson to learn here about the nature of design. There are the things we know; the unknowns, and the bugaboo - the unknown unknowns. I plan to post some thoughts on this aspect of design in the next few weeks.