KnownBugs

HomePage | RecentChanges | Preferences |

This page lists known bugs with the January 2006 Blackfin Handy Board design.

Battery and Charge Stuff

  • Battery + and - labels are wrong. New battery holes should be on 0.2" spacing and located so that a screw-terminal connector can be mounted there. Holes should be provided to allow routing battery wires from bottom of board and into screw-terminal connector.
  • Charge circuit sense-ground is incorrectly wired.
  • Verify charge circuit works correctly when bridge is properly wired.
  • Should we keep the bridge-rectifier on charge jack? This loses 1.2v or so and the charge circuit seems highly sensitive to voltage level. Maybe replace with a single reverse-polarity-protecting diode. I'd like to see a small chart of charge rate vs. input voltage to help decide whether or not to keep the bridge.

Other

  • FGPA Done LED is too dim.
  • Pullups on digital inputs missing.
  • Power supply circuit needs to be debugged. Still don't know what's wrong with it.
  • LED flicker in a tight loop using default values was slow (~1MHz) due to the processor booting up with very conservative (slow) defaults for timing. These long default delays aren't really a bug but something to be aware of.
  • As the battery gets low, the board will start acting weird. The USB and Reset lights will come on and flicker, and the speaker will make this groaning low-frequency noise. This would be a good time to plug in the charger (or unplug and re-plug) and go grab a snack or something ;)
  • Decreasing the clock cycles for writes can cause the display to either print garbage or go black. It was explained to me that the display requires a particular bus speeds in order to operate correctly.
  • Analog +5v supply and HB-connector +5v supply should be isolated.
  • This might not be true, but it works. If your computer does not see the USB-ICE device then go to the start menu and navigate to the Analog Devices -> CONFIGURATOR. Click test and choose the device. This will verify that everything works, but more importantly it seems to be the magic bullet that makes the device get recognized. Happened to me several times.
  • camera port should have 17x2 pins rather than 16x2, so that a standard 17x2 ribbon cable can plug in. Also, leave room on either side of the jack for the happy-gray style of pin header.
  • move male header so not over battery.
  • can LabVIEW Embedded or VDSP work w/o debug agent?
  • analog sensor noise issue, including rerouting +5v supply for external sensors.
  • is there a depopulated solution to use JTAG if no debug agent?
  • mounting pegs for DB9 should have solder pads for strain relief.
  • power switch ON/OFF swapped to match original HB.
  • right now there are 6 GPIO from BF. 4 go to Xilinx and 2 are wired as inputs for user. in rev, 1 of these will be Xilinx->Blackfin interrupt associated with 48 kHz DMA. what will we do with other 5 lines? ANSWER: FPGA program and done used with these GPIO. keep PPI high bit lines dedicated to PPI. remove U24 and RN7; have 6 pulled-up analogs and 6 floating. Keep 2 pins wired as inputs to GPIO.
  • Joe's 3 top-level HW issues: (1) debug agent (with 0-ohm resistors to allow leaving out of build); (2) charge circuit; (3) getting rid of noise on 10-bit ADCs.
  • Create dedicated OV connector (and remove Stamp connector). Use male pins for OV connector so standard IDE cable can use to connect to camera. Buffer inputs so +5v cams can be used. Make sure 16-bit camera mode can be used (thus reducing pixel clock to 10 MHz). Wire negative reset from reset chip.
  • Add missing two lines from ADC bus, so can use in word mode.
  • Remove U21 (i2c clamp chip), since BF's TWI pins are +5v tolerant.
  • Fix digital in pullups pinout.
  • Move system reset signal on FPGA to input-only pin. Are there other signals that can be moved to input-only pins?