SpectruMedix LLC., State College, PA
Mar 2001 - Nov 2002 Senior Electronics Design Engineer
Spectrumedix was one of the best experiences an engineer could possibly have. We were using Agile techniques for software and hardware development before the Agile approach was formalized. That is, there were 'user stories', the wish list items, scrums, we didn't do waterfall charts to my knowledge, but Tom Kane, the scrummaster in all likelyhood kept something similar to report to Ilan Reich, we had stand-ups, pair (peer) programming (Brian Quay - Sean O'Connel), and other Agile tactics. Retrospectively, and I think without selective memory, this was a potent team and despite occasional and understandable diagreement, accomplished Herculaen task of not only developing a functional system, but a user-friendly, safe and UL/CE certified system! To fully appreciate the magnitude of this accomplishment, read below to the section describing the 'red flag' hardware that had to be reliably controlled. Wow!
The To Be Continued.....
Before getting into the reason I was hired (the SCE9610 genetic sequencer), let me note a number of smaller concept-to-deployment projects which were also under the job description of 'other duties as necessary'.
These projects involved:
- Formalization of design and performance requirements aligned with general 'V-model' validation.
- Presentation of conceptual designs, hardware and software interconnectivity, general PCB and interconnect component selection and general algorithms for review & critique.
- Schematic capture or supervision of schematic capture.
- BOM generation, generating purchasing information for any new components.
- Board layout, or 'layout mentoring' to preclude incorporation of layout weaknesses which might affect performance or reliability.
- First unit population, performance testing, design validation at lowest V level.
- Writing and otherwise documentation assembly/test plans for production if needed.
- Providing training if desireable to ease first article production anxiety. ;-) (hand holding)
The following is a 'placeholder list', and I can detail these little projects if requested.
- PIC based PID Vacuum controller for Ksenia Krylova's protein sequencing system, a derivative of the SCE9610
- PIC based PID temperature controller for SCE9610 replaceable cartridge testing
- PLC based coltrol electronics (PCB, operator panel, and motor driver) for polymer gel dispensing pump to improve production efficiency. Mechanical design and fabrication by John Kernan.
- 'Arc detector', an RF detector front end (direct conversion broadband receiver circuit) coupled to a microcontroller to detect high voltage arcing within the SCE9610 system.
- Probably 5 or 10 more that I'm just not thinking about right now! Chck back after I've reconnected with some of the team named throughout.
Background: Spectrumedix arose from the buyout of a State College PA company called 'Nuclide'.
Founded in 1961 by Leonard Hertzog (obituary), Nuclide was instrumental in the development of mass spectrometry. In subsequent decades Agilent and other companies with extreme resources cam into competition with Nuclide.
By 1992, Joe Adlerstein was able to buy out Nuclide to form Spectrumedix. Adlerstein at some point acquired intelledctual property from the Iowa State University to allow Spectrumedix to begin development of a new product, the SCE9610 genetics sequencing system. Tom Kane, a PhD Chemist took charge of building the first prototypes, and with help from a talented mechanical designer, John Kernan, brought together the first prototypes. The electrical infrastructure at this time was a 'wire as you go' result of adding National Instrument control cards as needed, adding software features, and hardware control features via a main interface PCB using 4 Basic Stamp modules to serve as keyboard handlers, I/O controllers, etc. Essentially they served as PLC's.
While I stand by my statement that "Spectrumedix was one of the best experiences an engineer could possibly have.", outside the view of us engineers and chemists who were "in the trenches" of getting the product more robustly re-designed, at higher levels the seeds were being sewn for Joe Adlersteins replacement.. If you haven't decided to skip down to my responsibilities and accomplishments yet, I'll proceed with describing why Specrumedix is now defunct, as this is the reason I can go into details which would otherwise have to be kept confidential.
Joe Adlerstein was replaced by new majority shareholder Ilan Reich, who had inhereted a high risk proposition from day one. Reich entrusted Thomas E. Kane PhD Chemistry as VP of Research and Development, to carry the deveopment forward. Tom was my immediate advisor, supervisor, nemesis, and general fountain of all knowledge outside electrical engineering. ;-) to Tom. Tom similarly entrusted me with responsibility to continue hardware development in a direction that would solve existing reliability problems and prepare the system design for CE certification.
Tom rightly insisted that changes be well justified and validated. On occasion this caused the need to excercize great diplomacy and perhaps a bit of slightly tense interplay between myself and Tom, as we were approaching design validation from a Scientists view (Tom) vs. an Engineering method (myself). In hindsight, neither of us realized the fundamentally different reasons the ways of validation rightly are different in these fields! This may seem like a bit of 'Waxing Philosophically', but in hindsight, it was part of a great experience in getting the task accomplished and working with very bright but very different people. Tom and I and several others on the team, Brian Quay, Sean McConnel, Scott Goodman, John Kern, Brandon Tarr, Steve Lauver, Aaron Gilbert, and many others outside the Engineering department soldiered on and made the required changes creating the new 9610, which passed independent CE certification by independant testing and certification by MET LABS, on the first attempt, and went on to sell more than 100 units.
Before we discuss the original (pre-CE) 9610, let's pause and acknowledge the discomfort of discussing the flaws the pre-CE prototypes contained.
I feel it is vitally important to emphasize that the accomplishment of building the first couple of prototypes was significant. The 'flaws'
(new word please?) were perfectly understandable in an environment of rapid paced development with emphasis on data quality
(macro view of system operation) while instrumentation issues (ground loops, parasitic coupling, inadequate HF power supply, etc) were just outside
the experience of the early team. If we use the 'V' model of validation used in modern system validation, we would see the practice before I arrived on the scene was to validate at the top level, the user or application level.
I do not claim to have implemented or formally used the V model, but we did use something akin to a V model by a combination of common sense, necessity and serendipity. My contribution was in what I've called 'internal' validations, which were both at the system level and board level, and in low level electronics measurements with which the user would never be concerned. Indeed these were the lower levels of a V model. Note that though th V model is often applied to software validation, there are many derivitives and it seems to work well with hardware design and mixed hardware/software/firmware systems. Perhaps now we need a 'W' model. ;-)
In any case, the first 10 systems had a few 'reasonable oversights', and there was an enormous amount of excellent work. So I do not criticize my predecessors in any way, but the techniques used to design a custom system for one-time PhD thesis research in a University environment (litterally only having to operate on a lab bench) and the product design techniques engineers employ for robust, safe, field (perhaps even consumer) deployment are not the same. The pre-CE design team did great work though and the first generation of systems (about ten), produced good data. I took over the task of making the corrections with the benefit of much good work and learning to build on.
That said, the task entailed making fundamental changes in the internal interconnectivity, isolation, power distribution, shielding and other areas, and naturally, it required diplomacy and perseverance to convince parts of the original design team that these seemingly big changes were needed. So, after a thorough system review, documenting the key portions of the system current flow net, and careful explanation as to root causes, failure modes, and remedies, the team got behind the proposed changes. We made the changes, and as you know from the preceeding paragraph, the changes did their job.
It should be noted that in addition to system level (top level of 'V') validation (by examining end-user sequencing data against standards) that internal electrical validations were also performed to confirm that 'trouble currents' were now being routed properly and not creating common-mode or other induced parasitic effects. This second 'internal' electrical validation at the net and component level was something I had done on many previous occasions, with ELMDAS Co., and on all new designs at LeMont Scientific, Inc. These validations provided me and a few others elecronics oriented people with a signinficant amount of confidence that we would be deploying a very robust solution. This additional internal electrical validation also had a great side benefit in providing many 'teaching moments' in which junior engineers and technicians could see subtle electronic parasitics and learn more sophisticated instrumentation techniques in setting up an oscilliscope, logic analyzer, and other instrumentation. Again, these issues afecting robustness and reliability really seem to be best visible at the lower levels of the 'V' model.
The specific hardware design issues were:
- Analog control signal to 30KV programmable power supply was referenced through the system frame, being anodized Bosch strutt
- not having a clear, low impedance reference return to the NI D-A card which generated the laser power and high voltage supply control signals
- Lack of suppression diodes and solenoid return current path controls were causing PC to hang through NI cards DC coupled grounds and the unisolated power supplies used to power the solenoids.
- Other stuff I'm not remembering.......*
- *There were about 8-10 issues uncovered in the assessment phase, and my lab notebok is not available to me.
- System design responsibility for CE/UL/FCC certification of genetic sequencing systems using class IV lasers and high voltage
- Significantly modified design of prototype genetic sequencer to pass CE (Compliance European) EMC standards System passed on first try
- Project leader for UL/CE approval of support components
- Design, PCB layout, build, test and validation of support devices such as polymer dispensing pumps, temperature controller calibration system, and sub-systems to the SCE9610. These sub-systems included optical isolation and high power control sub-systems, arc-over detection system, motor control logic, etc.
- Microcontroller programming, embedded uP programming, C
- Mechanical design, PCB design, layout, documentation and production implementation
- Engineering support interface for Marketing, Production and Service departments
- Supervised activities of Electronics engineer, Mechanical designer and Documentarian