So Electronic Design has this troubleshooting contest where participants have to describe in 750 words or less a compelling example of something that the participant fixed, improved, rebooted...etc. 
All rules and guidelines are linked here. The winner gets $500, and a nostalgic trip down memory lane. 

To my dismay, I'm not allowed to enter because..

"Eligibility: Entrant must be a legal resident of the 50 United States and the District of Columbia, who is 18 or older as of 11/21/2018."

... I'm not 18 years old or older. At heart, I'm still 13 so I'm not eligible. (Bye bye $500.)

But in good spirit, I would like to share my entry had I been eligible below. Enjoy!





"Optimizing Evaluation Costs by Analysis between Datasheet Timing Parameters and Standard Protocols"

There was this problem where we had to brainstorm on hastening the evaluation of an IC's timing parameters (can't tell what kind of IC or whether it's I2C or SPI, but it's either of the two). Keeping up with modern trends in technology demands continuous improvement, and we couldn't stay satisfied with the current efficiency of our established procedures. After some morning meetings and pain-staking systems analysis, I managed to integrate the measurement method to a proprietary automation system recently developed by our validation automation department, that cut costs by around 50%. However, I had another trick up my sleeve to take this solution even further.

Timing parameters are evaluated by sending a distinct pattern of pulses - commands to the device under test (D.U.T.) that change the D.U.T.'s performance - through a standard interface such as I2C or SPI. The positive pulse widths are varied until such time that a response can no longer be detected from the D.U.T. (typically an ACK (acknowledge) bit or some toggling in the logic level of a DMON (digital monitor) or testmode pin). The distinct pattern of pulses vary from product to product AND the kind of parameter that is evaluated. For example, the starting point in moving the rising/falling edge toward the other channel's rising or falling edge will differ based on the D.U.T.'s address.

A closer look at these starting points reveals a striking solution that becomes more conspicuous the more times the evaluation is done. Making distinct patterns of pulses is what consumes most of  preparation time in such measurements, and realizing the resemblances between datasheet timing parameters and the sweep's starting point makes preparation of a full set of patterns feasible. This ultimately skips the 'manual' task of every timing evaluation - making it fully automated.

In addition, evaluation no longer requires the analysis of an engineer. A set of instructions can be laid out to, say a technician, and the measurement can proceed - also reducing labor costs.

Unfortunately, such a solution is not immune to flaws. The problem with fixating the pattern is that it won't be flexible to changes or upgrades to the protocol (or if there's a new parameter introduced to the family of timing characteristics already described in a product's datasheet). For example, I2C is lately being overshadowed by I3C, which is an emerging standard for sensor interfaces.  


Even if I3C were backward compatible to I2C, the parameters surfacing from I3C's new features may not be supported by pre-built patterns for I2C.  Use of different instruments involved in measurement also need to be considered.

This concludes my personal account on improving evaluation of datasheet timing parameters. I'm certainly looking forward too to the other stories and adventures shared by participants of this contest.