Eulogy
Posted: November 23rd, 2020, 4:16 pm
I've lost three people so far this month - a neighbor, and two former coworkers. Steve T. was a calibration technician at the medical products manufacturer I retired from several years ago; Jeff B. was a quirkily brilliant mechanical engineer I worked with at the job before that.
I have a fond memory of Jeff from a Chistmas Eve about 20 years ago. He was struggling to learn a Finite Elements Analysis (FEA) software package in order to model our laser thermal-transfer process - a deeper understanding of the interactions among the laser beam, the ink, and the ink's target surface would prove useful in optimizing product performance. I was assigned to be his mathematical support, as I already had some experience creating computer models of the energy distribution within a moving laser beam.
Finite Element Analysis is used to model physical process too difficult to observe directly. It does this by breaking up your area of interest into a 3D grid and using the fundamental laws of physics - heat transfer, wave propagation, electrical capacitance, resistance, inductance, etc, etc, etc - to track the progress of each element in the grid over time. But to do this, the physical properties of the test region has to be defined in exhaustive detail - density, specific gravity, specific heat, surface energy, permitivity, Young's modulus.... Not all of these were in the almanacs for some of the materials we were using, and we had to make some guesses. And, the software manuals were written for an audience more familiar with FEA than we, and made for a steep learning curve. We made a lot of mistakes, getting crazy results - when we got any at all. But by the third week in December, we felt we were zeroing in on it. It was like flight-testing an airplane that flew a little further each time before crashing.
It was a long-standing tradition for the company's HR to issue a stern directive midway through December reminding all and sundry that we were a global company whose phones needed to be manned at all times to service our customers, so Christmas Eve would be a full working day. Every year we'd read the memo with a heavy roll of the eyes because we all knew that at around 10:00 am on December 24, the plant owners would announce over the PA that the place was shutting down at noon and for everyone to go home and enjoy a most merry Christmas. (The one year they didn't, no one came back from lunch anyway.)
Sure enough, Tom & Jim came on the PA that morning and sent everyone home at noon. But Jeff & I felt we were really close - we were fairly certain we knew what we'd done wrong in the last run, I'd made whatever recalculations were needed, he'd modified the model accordingly, and we REALLY wanted to find out whether or not we were right, so chose to stick it out. There were maybe five cars in the parking lot when I got back from lunch - and one of them was Jeff's. Mine made six.
So we set to work. We keyed in the changes, fixed all typos we could see, and hit Enter. "Illegal Operation in element 00D713A"... A little hunting showed I'd keyed a parameter into the wrong box. Enter. "Run Complete" - but it made no sense. It showed no changes at all to the system even though we were virtually blasting it with a 100-w laser beam. Back to the manuals...
We tinkered and typed and cursed and swore for the next hour or two, each time getting what was starting to look like a realistic movie of the process we were trying to model. FInally at around 3:00 we got what looked like a successful run. But as we began to interpret the results, it was showing us something unexpected. Maybe at this point, I should try to go into a little more detail about what we were trying to accomplish.
We manufactured a thermal-transfer printer for use in industrial product-branding applications. A thermal-transfer printer is a high-tech variant on the typewriter. In a typewriter, an inked ribbon passes between the paper and the individual print elements - the hammers at the ends of the keys. The ink is transferred to the paper by whacking the back side of the ribbon with one of the hammers. In thermal transfer, the hammers are replaced by a fine grid of heating elements. The grid is in constant contact with the ribbon, which in turn was in constant contact with the paper. Here the ribbon is a polyethylene film carrying a heat-sensitive ink that melts when heated above some point. By selectively turning the grid elements on and off, the ink melts and transfers to the paper.
Our machine, however, used a rapdily-scanning beam of light from a CO2 laser to heat the backside of the film. It was our assumption that the heat-transfer process was the same as that of a conventional thermal printer: the laser heated the back of the film, which then migrated through to the ink, melting it enough to bond to the paper below.
But what the FEA was telling us that the laser beam was passing completely through the carrier and the ink without affecting either, to be absorbed by the substrate on which we were printing. It was this absorbed heat in the substrate that was melting the ink from below, not the laser melting it from above! It was a moment of revelation that explained so many things we'd seen in the printers' performance and answered so many questions - we were literally screaming with excitement, and carrying on in a most unprofessional manner. Surprise breakthroughs in industrial research are rare, making them all the sweeter when they occur. I hope the other four people in the building could appreciate that and forgive our outburst.
(He's the same Jeff here - viewtopic.php?f=98&t=31480)
I have a fond memory of Jeff from a Chistmas Eve about 20 years ago. He was struggling to learn a Finite Elements Analysis (FEA) software package in order to model our laser thermal-transfer process - a deeper understanding of the interactions among the laser beam, the ink, and the ink's target surface would prove useful in optimizing product performance. I was assigned to be his mathematical support, as I already had some experience creating computer models of the energy distribution within a moving laser beam.
Finite Element Analysis is used to model physical process too difficult to observe directly. It does this by breaking up your area of interest into a 3D grid and using the fundamental laws of physics - heat transfer, wave propagation, electrical capacitance, resistance, inductance, etc, etc, etc - to track the progress of each element in the grid over time. But to do this, the physical properties of the test region has to be defined in exhaustive detail - density, specific gravity, specific heat, surface energy, permitivity, Young's modulus.... Not all of these were in the almanacs for some of the materials we were using, and we had to make some guesses. And, the software manuals were written for an audience more familiar with FEA than we, and made for a steep learning curve. We made a lot of mistakes, getting crazy results - when we got any at all. But by the third week in December, we felt we were zeroing in on it. It was like flight-testing an airplane that flew a little further each time before crashing.
It was a long-standing tradition for the company's HR to issue a stern directive midway through December reminding all and sundry that we were a global company whose phones needed to be manned at all times to service our customers, so Christmas Eve would be a full working day. Every year we'd read the memo with a heavy roll of the eyes because we all knew that at around 10:00 am on December 24, the plant owners would announce over the PA that the place was shutting down at noon and for everyone to go home and enjoy a most merry Christmas. (The one year they didn't, no one came back from lunch anyway.)
Sure enough, Tom & Jim came on the PA that morning and sent everyone home at noon. But Jeff & I felt we were really close - we were fairly certain we knew what we'd done wrong in the last run, I'd made whatever recalculations were needed, he'd modified the model accordingly, and we REALLY wanted to find out whether or not we were right, so chose to stick it out. There were maybe five cars in the parking lot when I got back from lunch - and one of them was Jeff's. Mine made six.
So we set to work. We keyed in the changes, fixed all typos we could see, and hit Enter. "Illegal Operation in element 00D713A"... A little hunting showed I'd keyed a parameter into the wrong box. Enter. "Run Complete" - but it made no sense. It showed no changes at all to the system even though we were virtually blasting it with a 100-w laser beam. Back to the manuals...
We tinkered and typed and cursed and swore for the next hour or two, each time getting what was starting to look like a realistic movie of the process we were trying to model. FInally at around 3:00 we got what looked like a successful run. But as we began to interpret the results, it was showing us something unexpected. Maybe at this point, I should try to go into a little more detail about what we were trying to accomplish.
We manufactured a thermal-transfer printer for use in industrial product-branding applications. A thermal-transfer printer is a high-tech variant on the typewriter. In a typewriter, an inked ribbon passes between the paper and the individual print elements - the hammers at the ends of the keys. The ink is transferred to the paper by whacking the back side of the ribbon with one of the hammers. In thermal transfer, the hammers are replaced by a fine grid of heating elements. The grid is in constant contact with the ribbon, which in turn was in constant contact with the paper. Here the ribbon is a polyethylene film carrying a heat-sensitive ink that melts when heated above some point. By selectively turning the grid elements on and off, the ink melts and transfers to the paper.
Our machine, however, used a rapdily-scanning beam of light from a CO2 laser to heat the backside of the film. It was our assumption that the heat-transfer process was the same as that of a conventional thermal printer: the laser heated the back of the film, which then migrated through to the ink, melting it enough to bond to the paper below.
But what the FEA was telling us that the laser beam was passing completely through the carrier and the ink without affecting either, to be absorbed by the substrate on which we were printing. It was this absorbed heat in the substrate that was melting the ink from below, not the laser melting it from above! It was a moment of revelation that explained so many things we'd seen in the printers' performance and answered so many questions - we were literally screaming with excitement, and carrying on in a most unprofessional manner. Surprise breakthroughs in industrial research are rare, making them all the sweeter when they occur. I hope the other four people in the building could appreciate that and forgive our outburst.
(He's the same Jeff here - viewtopic.php?f=98&t=31480)