2008.12.21 - for packaging a board to attach to a computer on the ground station side, I have a Hammond box that looks good. Digi-Key part number HM380-ND, $2 in singles. 2009.03.16 - rewire P1_2 to be CS* for the SPI EEPROM instead of accel_st, ground ST pin - USB connector shouldn't hang over the edge of the board - reorder debug connector pinout to match what actually works with cp2103 - make USB connector a dual footprint to allow remoting using another connector like the one we're using for serial port from the tiny series? - change debug header to 5 pins so we can use one board to program another by bringing a gpio over to drive reset? 5-pin connector from tiny series. 2009.04.21 - Keith would like to use SPI to talk to TelePyro and TeleTerra boards, so we're thinking of giving up an LED to get a second chip select line, and reusing the existing SPI channel to the memory chip. Now, how do we do an input interrupt from the PIC to flesh out a 7-pin connector? rename net cs0 to cs_mem to reduce confusion lose second LED attached to CPU, P1_1 becomes cs_ext lose R12 since LED is no longer there add J9, 7 pins, Molex pn footprint=530470710 digikey=WM1736-ND define 'interrupt' as a new net ... now, how to hook it up? So the problem is how to multiplex one cc1111 pin to be bidirectional. If the cc1111 asserts that pin, we need the PIC cs input to go low. If the cc1111 isn't asserting it (switched to input), then we want the state of the PIC interrupt output to be reflected there. But how can we drive the cc1111 input while simultaneously knowing when it's being asserted as an output? Doesn't seem possible. How does the Dallas 1-wire stuff work? 2009.05.27 - based on experiences and discussion at Mile High Mayhem, here's what we think we want to do at the schematic level: + known changes for two cuts and jumps and two added resistors + lose the temp sensor + lose the beeper + lose the 4-pin serial port connector + change the debug connector to 4 pins 1.25mm ala deleted serial port + add 7-pin 1.25mm connector that extends SPI for TT/TP and doubles as debug control port for flashing another board with 7-4 pin cable gnd, v_lipo, mo0, mi0, c0, cs1 (ex temp), irq1 (ex beeper) + lose the 1000uF cap + wire v_lipo to the igniter fets directly + include a connector for an external pyro battery + add one or two 1206 caps between +3.3V and gnd to ballast against power switch bounce and/or brown-out on igniter firing: DK 490-4512-1-ND is a 100uF 1206 part, 1.32 in 100 pieces DK 445-1428-1-ND is a 47uF 1206 part, 0.91 in 100 pieces + add MN5010HS GPS chip, chip antenna, and associated circuitry BAS16 is DK BAS16LT3GOSCT-ND, SOT-23-3, 0.05 in 100 pieces BSS138 is DK BSS138LT1GOSCT-ND, SOT-23-3, 0.14 in 100 pieces 33nF cap for antenna is DK 490-1315-1-ND, 0402, 0.03 in 100s 2009.06.01 - after more thinking about this, I want to add the serial connector back in but as an optionally loaded part ... if I can make it fit. - can I put the GPS chip antenna on the board *and* allow for an SMA or other connector if we want it off-board? - changing the memory part from the SOIJ8 to the DFN8 would save about 2mm in width, but that doesn't really help. - changing the 5V switcher from S8 to MS8 takes us from 5 tall to 3 tall, but costs $0.30 more per piece and isn't stocked by Digikey. Not worth it? - noticed the NXP SC18IS600, which is an SPI slave to I2C master interface with a few GPIO lines... mate that with an HMC5843, which is a 3-axis magnetic sensor, and we might have another interesting companion board for TM - To Do List - add serial connector back to schematic as optionally loaded part - move USB connector on PCB layout 80 mils in from board edge - move yellow LED and passives closer to LiPo connector + create PCB footprint for the GPS module + create PCB footprint for the GPS antenna 2009.10.09 - flew lots of stuff last weekend at O'fest. 100g accel worked ok in G-Spot, broke beeper presumably on landing with tangled chute. board seems ok otherwise, need to verify that baro sensor is still ok. - put out sample requests for gps passive patches from 3 vendors - when adding GPS, let's plan for passive patch loaded on the back of the board, plus a U.FL connector loaded on the top of the board with DC bias feed to support external active antennas. This probably means passive patch direct to gps chip, then a dc blocking cap, a chip inductor for dc bias injection, and the connector. A suitable connector is: WM5587CT-ND Molex 73412-0110 0.70 in single quantity Keith found gps active antennas with cable to U.FL connectors and Molex sells cables with U.FL to bare leads, so this should cover the bases. - TelePyro doesn't really need a CPU. How about an SPI to I2C converter chip with GPIO lines and an I2C ADC chip? Cheaper and easier than a PIC? PIC16F882 in SSOP-28 is $2.10 per, 1.41 @ 25 units -- PIC16F883 is $2.12, PIC16F886 is $2.32, have better memories CP2120 in QFN-20 is $2.41 per, 2.27 @ 25 units The Silabs CP2120 has 8 gpio lines but is only available in QFN20. The NXP part I discovered earlier, the SC18IS600, has 4 full gpio lines and 2 lame-o pins, but is available in a hand-solderable version. Add in a Maxim ADC, and something like this emerges: DigiKey 568-3232-5-ND NXP SC18IS600IPW,112 2.25 DigiKey MAX11603EEE+-ND Maxim MAX11603EEE+ 4.00 Mouser 595-CD4508BM TI CD4508BM 1.70 The 4508 has separate reset inputs and latch controls for each 4-bit latch. The reset forces all outputs low, and the reset is an active-high input. The NXP part has an active-low reset input, with 10-30k internal pullup. The ADC has no distinct reset input. A suitable SOT-23-5 reset controller is the MCP1319MT-29LE... it has a master reset input we can ignore, and both active high and active low reset outputs. DigiKey MCP1319MT-29LE/OTCT-ND 0.70 This plus 8 copies of the pyro circuit on TM (FET and 4 resistors) and we're good to go. Wonder if there are resistor networks it'd be worth trying to use to ease component placement? Yes... these are 8 x 0402, more or less: 100 ohm 741X163101JPCT-ND 0.16 10k 741X163103JPCT-ND 0.16 27k Y1273CT-ND 0.18 100k Y1104CT-ND 0.18 - another way to think about all this is that I could panelize a set of boards all done in 2 layers, then treat that as a sub-panel I put N copies down of to fill up a 33each.com max panel size. 2009.10.16 - move to AT45DB161D 16-megabit SPI "DataFlash" EEPROM from Atmel? AT45DB161D-SU-ND in SOIC-8 for $2.26 each appears to compare favorably against $4.40 for current 128kx8 part minor concern about write power, but probably red herring 2010.02.08 - looking at LiPo charger alternatives MAX8808 DFN8, not carried by DK MAX8814 DFN8, not carried by DK LTC4095 DFN8, LTC4095EDC#TRMPBFCT-ND 2.91 2.34@100 950mA, resistor current set, rate limit input, suspend input, NTC thermal control option, smart CHRG led with 4 states LTC4065 DFN6, LTC4065EDC#TRMPBFCT-ND 2.77 2.23@100 750mA, resistor current set, CHRG led, enable input LTC4065A DFN6, LTC4065AEDC#TRMPBFCT-ND 2.77 2.23@100 750mA, resistor current set, CHRG led, supply present out MCP73811 SOT23-5, MCP73811T-420I/OTCT-ND 0.63@1 0.39@100 input selects 85 vs 450mA charge current MCP73837 DFN10, MCP73837T-FCI/MFCT-ND 1.37@1 0.85@100 usb rate select, power good output, dual inputs, dual status LX2206 DFN10, LX2206CT-ND 2.68@1 2.01@100 rate select, power good output... needs more passives? LM3658 DFN10, dual input, dual rate, two status outputs .. not obvious win MCP73855 DFN10, MCP73855T-I/MFCT-ND 1.85@1 1.18@100 see Microchip app note 971 Talking to Keith, there's no reason we couldn't use an 8-pin PIC to augment a standalone LiPo charger, implementing a timer after vusb application before switching the regulator to high current, to allow the cc1111 time to finish the USB bus negotiation phase. We could also monitor various state bits and multiplex signaling onto one LED. A combo like the MCP73837 plus a PIC10F206 or 220 would net under $1.50 at 100 pieces and be pretty powerful. Since the PIC has avilable weak pull-ups on the gpio lines, the open-drain state outputs from the regulator don't need resistors to be read. 2011.01.17 - v1.0 and v1.1 boards have a minor error in that the v_lipo sampling resistors are on "the wrong side of the cut" if you try to use a separate pyro battery. The workaround is to cut that trace and put in a jumper, which pretty much requires "return to Bdale for rework". We should fix this for a future PCB revision. 2011.09.12 - chatted with Adrian some yesterday. He likes a Bosch pressure sensor that is DK 828-1005-1-ND .. i2c output, similar range to our current part but smaller, around $5 at our build quantities. 2011.12.10 - the Johanson balun/filter for 70cm use with cc1111 is DK 712-1540-1-ND uses less PCB area, but lower performance than the discrete approach 2013.01.04 - noodling over TM v2.0 .. baro sensor will be hard to handle on cc1111. So, how about 48-pin version of STM32L, cc110L radio, ms5607 and MMA665X accel. Move to the faster lipo charger, too. radio / flash on spi2, sensors on spi1, baro on spi1 alternate pins 2013.05.02 - serious stab at what should be on TeleMetrum v2 .. used cc115L section from TeleGPS since I didn't have a cc110L data sheet on the plane .. hope it's similar .. need to update to transceiver, not just transmitter!