Andrew Zonenberg

@azonenberg

Infosec, HW/SW RE, high speed digital, test equipment, network hardware, microscopy, FPGA/ASIC, , KD2HKV, . Tweets are my own.

The lab
Vrijeme pridruživanja: rujan 2013.

Tweetovi

Blokirali ste korisnika/cu @azonenberg

Jeste li sigurni da želite vidjeti te tweetove? Time nećete deblokirati korisnika/cu @azonenberg

  1. Prikvačeni tweet
    6. ožu 2017.

    PGP keys: (work) BE79 FA02 16DC 14B5 9496 F216 CC04 AEFA C238 0256 (home) 859B A7BA DE9C 0BD5 EC01 FF36 3461 7AB9 B31C 7D7C

    Poništi
  2. 27. sij

    Slides from my S4x20 talk are up at !

    Poništi
  3. 26. sij

    No desoldering braid? No problem! Ok, machining solder off an 0402 pad is kinda stupid. But it was a good test run for some tooling changes. Also, at least for small areas, 5 μm per pass seems pretty doable on this mill.

    Poništi
  4. 26. sij

    Made a little jig for PCB work on my mill (on the same mill, of course). Machined from 1.5" x 1/4" 7075 aluminum bar stock. It clamps to the T-slots on the mill table and holds a 1.6mm PCB by the edges, held above table level to provide clearance for bottom-side components.

    Poništi
  5. 26. sij

    But I'm looking into options for using some sort of high speed mux to use a normal input channel for sync so I don't have to lose any. Even six channels is a respectable amount: enough to probe DRAM RAS/CAS/WE#CS#/CLK/DQS and leave your scope analog inputs for looking at DQ.

    Prikaži ovu nit
    Poništi
  6. 26. sij

    BOM wise, you're looking at a $386 @ qty 1 FPGA, a $113 @ qty 1 QDR-II+, eight $37 @ qty 10 comparators, and a $20 @ qty 1 PLL. So about $815 in silicon plus a 4-6 layer controlled impedance PCB. Might have to drop from 8 channels to 6 so I can steal one input per quad for sync

    Prikaži ovu nit
    Poništi
  7. 26. sij

    50 ohm SMA inputs intended for use with external transmission line probes. May also support 100 ohm differential inputs for direct input of LVDS/CML. Also a 10 MHz refclk input and trigger input/output for synchronizing to an oscilloscope.

    Prikaži ovu nit
    Poništi
  8. 26. sij

    Headless (1U?) system w/ 1000base-T interface to host PC. At full memory depth, 8Mx8 samples (64 Mb) over gig-e gives ~15 WFM/s uncompressed. Less deep captures will give proportionally higher performance.

    Prikaži ovu nit
    Poništi
  9. 26. sij

    Project idea: Kintex-7 GTX based logic analyzer. * 8 fast channels @ 10 Gsps * TBD slow channels @ 1.25 Gsps (1/8 rate) * 8 MB sample memory = 8M samples * 8 bits, more w/ RLE compression * Comparator based input stage w/ selectable threshold

    Prikaži ovu nit
    Poništi
  10. 20. sij

    Then the only remaining piece will be adding logging so I can collect trends over time and test effectiveness of outdoor drainage upgrades, and installing a 24V UPS so that I still get alarms even if the power goes out.

    Prikaži ovu nit
    Poništi
  11. 20. sij

    I plan to also track pump duty cycle over the long term (seconds on vs off over the past hour or something). Current/planned alarm conditions include water on the floor, tank level above the float, and flow rate approaching the pump's capacity.

    Prikaži ovu nit
    Poništi
  12. 20. sij

    The only remaining step was to code up a nice touchscreen HMI to crunch the data. It collects water level measurements at 1 Hz, does some averaging to reduce noise, then calculates tank volume and flow rate in L/hr.

    Prikaži ovu nit
    Poništi
  13. 20. sij

    Note the water on the floor in the bottom right corner of the previous pic. That's not spillage from the bucket I was testing the sensor on, that's an active leak into my lab. I also grabbed a Gentex horn/strobe and some water-sensing cables to place along walls in trouble spots

    Prikaži ovu nit
    Poništi
  14. 20. sij

    The parts I ordered after the Christmas flooding incident had come in, so I set to work building a proper monitoring system. I started with a DIN rail mount kit for a Raspberry Pi, a couple of I/O boards, and a 4-20 mA submersible liquid level sensor. Here's the dev setup.

    Prikaži ovu nit
    Poništi
  15. 20. sij

    After a bit more digging to direct sump pump runoff into the trench, it appeared that the immediate danger of flooding was gone. The next step was to improve monitoring and alerting so that if it came back, I'd be able to respond rapidly before too much flooding occurred.

    Prikaži ovu nit
    Poništi
  16. 20. sij

    Since the water coming off the roof couldn't get out to the sewer, it leaked out at this joint (a foot from the house), saturating the ground all along the exterior wall.

    Prikaži ovu nit
    Poništi
  17. 20. sij

    After the initial rush settled down I dug a bit deeper and found the cause: a shoddily installed ABS gutter downspot that was slipped over the end of a PVC pipe with no attempt at a seal. The ABS was cracked and the PVC was almost certainly clogged, so it all needs to be redone.

    Prikaži ovu nit
    Poništi
  18. 20. sij

    As I dug closer to the source, the water was coming so hard that it was actually squirting an inch or two in the air from the bottom of the excavation!

    Prikaži ovu nit
    Poništi
  19. 20. sij

    I grabbed a shovel and began digging a trench from the sidewalk toward several particularly wet spots of the front yard, in hopes that pulling water away from the house would help. When I got near the end of the driveway, the trench started rapidly filling with water.

    Prikaži ovu nit
    Poništi
  20. 20. sij

    I had previously RFQed for proper drainage upgrades, but the contractor was unable to do the work until the summer because significant excavations in saturated soil would collapse before any pipe could be laid. So it looked like I was on my own for the short term.

    Prikaži ovu nit
    Poništi
  21. 20. sij

    The water table in the entire property has always been high, and I planned to do drainage improvements outside in the near future, but it had not been a clear and present danger beforehand. This is my driveway last week. It's not even raining.

    Prikaži ovu nit
    Poništi

Čini se da učitavanje traje već neko vrijeme.

Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.

    Možda bi vam se svidjelo i ovo:

    ·