1. Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.
    Dismiss Notice

Right smack in the running of this mess

Discussion in 'Members Lounge' started by KeepItSimpleStupid, Apr 10, 2017.

  1. KeepItSimpleStupid

    KeepItSimpleStupid Well-Known Member Most Helpful Member

    Joined:
    Oct 30, 2010
    Messages:
    9,964
    Likes:
    1,099
    ONLINE
    My journey in electronics/instrumentation was somewhat mentored by National Instruments.

    This http://electronics360.globalspec.co...g-ni-ceo-reflects-on-the-past-and-looks-ahead article sums it up on their side.

    When you go a "little earlier" than the start of the article, NI developed hardware and software drivers for the PDP-11 and oddly enough, I had aPDP-11 background from High school and college, just not at the real-time level.

    The "growing pains" started when, the boss said we had to modernize our testing from BCD interfaced meters and x-y recorders. A minor jump was eliminating the BCD instruments and convert to IEEE-488. So, guess what? The testing process got SLOWER. But, the BCD meters were getting obsolete.

    The next part of the process was eliminating the X-Y recorders. My comment was "The software we need isn't here yet, but boss said the money was".

    NI had introduced LabView on the MAC and with IEEE-488 we cobbled together what parts we had and made a platform change. I wanted an SMU based system, but was told no. This was basically cost and we were building two systems, one to be a spare and mobile. I think we swapped out one item while the other was being repaired during it's 17 year life. The MAC never lost a hard disk. Just one floppy drive, some dust and a fan. A server for data was nixed.

    A MAC server maintained by us and a Linux based SMB server (maintained and backed up by our IT department) did show it's face.

    Slow as usual. The conception was to print tests as they were done, but that turned out to be impractical. The printer did better spitting out stuff continuously rather than warming up. Laser printers were new too.

    Trying to be a manager and developing some custom hardware (all related) for yet another system was basically too much for me. I had one inexperienced person as myself basically working under me, So, I was overwhelmed.

    So, me choosing LabVIEW and the MAC paved the department's use of Mac computers for primarily desktop publishing. I liked the MAC's flat memory model and long fie names (32 characters at the time vs 8.3 for Windows). Microsoft Word was developed for the Mac market until the PC caught up.

    The PC caught up and other instrument now had PC Centric interfaces/programs. In another system I was TOLD to use cards in a MacIIx based system to computerize something. I did not want to do that, I was over-ruled, Bad move. Slots from Macs and PC's disappeared.

    The 17 YO system got overhauled again with SMU's like I wanted. I did not do the programming. An interface to Excel now existed rather than comma separated data, so individual tests and summary data was available for each test. This one was PC based. Still slow.

    Unlike the previous system where data was displayed as acquired (which was also a slow process), you had to wait 2 minutes to see if the data was good.
    I guess in the grand scheme of things, one could use meters with a graphical display now.

    labview made it impossible to plot a single point on a graph. You had to plot 1, then for point 2, 1&2; for point 999, point 0-998 for 999 points 0-999; real slow on say a 66 Mhz machine.

    NI kept updating stuff during the development and went from a MAC only model to a Linux/MAC and PC model. Then the PC became the standard for both publishing and laboratory equipment.

    It was weird to make decisions in a snapshot of time where stuff didn't exist, was too slow or was constantly changing (software) and being overwelmed with technology.
    One decision I was FORCED to make was to design an interface to an existing monochometer and to build electric shutters and a filter wheel. There were no drivers for the stepper drivers and it probably would have been best if I wrote them. The stepper driver could be programmed with it's own basic-like program, so it could handle the filters and even moving the wheel.

    At the very end, a big empty box controls one shutter and "they" probably have no idea where the code is. The world developed an IEEE-488 based monochrometer, BUT our old one was better. It had a real-time interface that controlled start/stop and counts using a PDP-11. So, the intent was to control the stepper directly with a programmable interface. I learned a lot. The software got unreadable on a Syquest drive, but it wasn't complete anyway.

    In terms of "process control" type stuff, it seems that the "research variety" is much harder to do. So, I never had training in the say PLC/Wonderware type of approach, but did stuff with the real-time approach.

    To advance a synchonous motor was "cake" to do with the PDP-11 whereas the PC required counters and hardware to do it. e.g. move for x mS.

    Research gizmos are really hard when management doesn't know what's required and the vendors aren't clear. One example is that it was unknown that you needed to purchase a separate programming package for the stepper controllers to program sequences in a basic like language. Or that I needed a special type of converter to make RS422 signals compatible with the higher level RS232 signals. All of the stepper stuff I had was really useful information.
     
  2. Tony Stewart

    Tony Stewart Well-Known Member Most Helpful Member

    Joined:
    Aug 31, 2012
    Messages:
    3,152
    Likes:
    281
    Location:
    Richmond Hill , ON Canada near Toronto
    Your boss was an idiot for not listening to you with the Keithley adopted SMU bus.

    I successfully used HP's parallel bus card cage and some others with HPIB programming using HP9825's in the 70's for anything from a 1 day test program to measure all the phase frequency for Tiros N Tx UHF Tx's on startup and 1 second burst to a 3 month project making a end-toend HP9825 calc with DMA , Rs422 1km link at a screaming 9600 baud with 96 ADC ports and 96 DAC options and 96 digital ports for 15A to 25A Relay box for I guess what now would be called Real Time SCADA for Black Brandt rocket launcher umbilicals.

    Then later I got into designing DS1 BER test sets for worlds 1st ISDN Broadband WAN to the home and after company went titsup with a 1week design for Gandalf Muxing 1024 ports at 9600 baud , went on to HDD ST506 testing then dual port SCSI and up .Spent many years <3 doing that and one supplier Syed Ifticar offered me his job <mgr/owner> at Syquest but my wife at the time didn't want to move from Winnipeg to Silicon Valley. ( or Scott's Valley or Fremont). Also had offers at Seagate and Micropolis for similar high level, but same problem.

    Then got into TE mgr role with 10 ATE's, unit test, System test and about 30 mainframes scatter around production also 3 Clean rooms for servo writers , HDA testing ... all in Winnipeg at Burroughs/Sperry\Memorex/Unisys.mThat lasted 10 years and got into Operations Mgr at wireless 2 way meter reading and a everything custom RF testing, later Iridium satphone 6GHz test design and once did the worlds first working Kopin LCD headset on VGA and ultrasound monitor signals. I forget all my Forth Programming skills and VBA and HPIB, since I work with the brightest SW gurus, but spent part time in FoxPro doing all the BOM, Master Registry, Pick List for fast proto build in SM and warehouse database stuff.

    I didn't have a need for NI since we had a Computer Automation Functional tester, but later a colleague used to test every parameter on projector TV's in production. circa '99 with virtual design using visual programming.

    All in all I had fun.

    I see all these kids trying to design stepper motors and all 5.25" HDD's started with fast seek steppers and Hitachi/NPL had. a slick tiny design with full accelerated slew rates and viscous dampers for overshoot control in early 80's.

    They don't even teach this stuff yet.
     
    Last edited: Apr 10, 2017
  3. KeepItSimpleStupid

    KeepItSimpleStupid Well-Known Member Most Helpful Member

    Joined:
    Oct 30, 2010
    Messages:
    9,964
    Likes:
    1,099
    ONLINE
    The SMU acronym, I referred to a Source Measure Unit which Keithly, Opps Tektronix, and Keysight offer.

    http://www.tek.com/keithley-source-measure-units

    http://www.keysight.com/en/pc-1862522/source-measure-units?cc=US&lc=eng

    The eventual replacement was the 2400 SMU series, but I would have had one of these http://www.testequipmenthq.com/datasheets/KEITHLEY-236-Datasheet.pdf to choose from.

    My boss at the time, said it used feedback and therefore wasn't OK for testing solar cells. I built an analog version, of sorts. What I would call a 4-terminal I-V converter biasable to +-10V and could handle +-100, 10, 1 and 0.1 mA full scale at 10V out. When using suppression, bias was limited to +-5V.. I only had to work between +-2V and 0V was common. A really fun project. Worked OK for DC and great for AC (it's primary purpose). A trapezoidal signal made from chopping a monochromatic light source. In the DC world, it had about 40 pA of offset current. Voltage was a few mV. The nulling system didn;t work. Still, not bad.
     
  4. dave

    Dave New Member

    Joined:
    Jan 12, 1997
    Messages:
    -
    Likes:
    0


     
  5. Tony Stewart

    Tony Stewart Well-Known Member Most Helpful Member

    Joined:
    Aug 31, 2012
    Messages:
    3,152
    Likes:
    281
    Location:
    Richmond Hill , ON Canada near Toronto

    I designed a Solar panel tester one day and it used transistors didoes and R's to match the Panel design specs exactly and then without need for a PMT hunting algorithm would give PMT within 5%.

    The feedback needed of course is voltage and current, so I still think your boss made some wrong assumptions.

    The ESR of the active BJT and Diode load is inverse to its power rating just as the slope of the power generation is inverse with it's equivalent source resistance with -ΔV/ΔI

    The other feedback I would use is a pulses or secondary PD sensor and PV temp to adjust the load knee voltage relative to for Voc to predict the Vmp point and determine efficacy from the Req.

    http://obrazki.elektroda.pl/2675845500_1491871876.jpg

    I designed it with Excel ! and simulated on Falstad.
     
    Last edited: Apr 10, 2017
  6. KeepItSimpleStupid

    KeepItSimpleStupid Well-Known Member Most Helpful Member

    Joined:
    Oct 30, 2010
    Messages:
    9,964
    Likes:
    1,099
    ONLINE
    Yep, I do to, since after I got a different boss (old one retired, now one nano-managed), we used the SMU
    micro-managed doesn't fit. Either nano, pico or femto-managed.

    These J (current density)-V curves and values were an NREL traceable. NREL gave us calibration values for a few either filtered or unfiltered, non-AR coated Crystalline Silicon 1 sqcm cells that I think were made by Solarex that we used for calibration. Then we cross-calibrated like 10 more. Each daily calibration was with the last calibration cell you used and one new one. If the two didn't get the right values with a range, you had to find 3 that did or figure out what went wrong.

    Efficiency goals had to verified by NREL (National Renewable Energy Laboratories)

    We never had the equipment to measure the spectrum of our light source, but we could integrate the QE(lamda)*AM1.5G(Lambda) to get the AM1.5 global Spectrum short circuit current.

    Never did panel stuff, just basic research on low-cost materials that might make a solar cell like https://en.wikipedia.org/wiki/Zinc_phosphide (Rat poison) at one time.

    Many of the research devices, 12 cells would fit on a 1"x1" square substrate.
     

Share This Page