Jump to content
Forum sending old emails Read more... ×


  • Content Count

  • Joined

  • Last visited

  • Days Won

  1. kff2

    MTK3339 GPS

    The old version of MTK3339 has built-in datalogging also (a total of 128K of data, about 8000 positions if I remember correctly). I tried it a few weeks ago, worked fine. See http://www.auroramobile.ru/content/files/pdf/gtop/locus/gtop_locus_library_user_manualv12.pdf
  2. kff2

    cc112x RF chips

    Has anyone else noticed the relatively new cc112x RF chips? They are similar to cc1101, but have much better specs: 129dbm receiver sensitivity for cc1125, and 16dbm output power. http://www.ti.com/product/cc1125 TI claims > 25 km line of sight range at 1.2kbps, all while using under 50mA of TX power and with non-directional antennas. The post below hints that the range could be improved to over 100 km by adding cc1190 (a PA / LNA). http://e2e.ti.com/support/low_power_rf/f/155/t/206584.aspx The chips are reasonably priced ($2 for 1120 and $5 for 1125). Unfortunately, I do not see modules based on these newer chips on either ebay or aliexpress. TI sells a kit for $299 that includes a pair of 1125 modules, but those work @ 869mhz only, which I believe disqualifies their use in the US. Modules based on 1120 do come in the 420-470mhz variety though. http://www.ti.com/tool/cc1125dk http://www.ti.com/tool/cc1120emk-420-470 Finally, there is another chip launching later this year: cc1200. http://www.ti.com/product/cc1200 It appears to be similar to cc1125, but has higher maximum throughput (1000kbps vs 200kbps).
  3. For those of you considering getting a Chronos watch, please keep in mind that 1) This watch is not waterproof. Apparently, the "30m waterproof" classification, as advertised in TI's literature, really means "can get an occasional splash". The watch will leak if you swim or take a shower in it. 2) This being an experimental device, there is absolutely no warranty provided by either TI or DigiKey. See this post for more details: http://forum.43oh.com/topic/2142-bad-experience-with-a-ti-deals-purchase/
  4. kff2

    CC2500 Breakout/BoosterPack

    I finally figured out why the board could not be programmed with the radio attached to it. I previously removed all jumpers between the two halves of the Launchpad, assuming that the 0.05 mm header was connected directly to the programmer. This was wrong. It turns out that the VCC jumper needs to stay on. Surprisingly the board could still be programmed with VCC disconnected -- I am guessing there is enough current leakage from other pins. With a radio module attached, and with the additional current that it needed, programming no longer worked. To summarize -- when using the 0.05 mm programming header, keep the VCC jumper on. The other jumpers can be removed.
  5. There is a newer version of MTK3339 available from the same seller on AliExpress: http://www.aliexpress.com/item/FGPMMOPA6H-utilizes-the-MediaTek-new-generation-GPS-Chipset-MT3339/623742154.html The main differences are size (new unit is 20% thinner: 4.7mm vs 6.2mm), pinout, and presence of an input for an external antenna. The module automatically switches between the built-in and external antennas. As far as I can tell, everything else is the same. This is what Adafruit now sells as their "Ultimate GPS module": http://www.adafruit.com/products/790 Datasheet: http://www.adafruit.com/datasheets/GlobalTop-FGPMMOPA6H-Datasheet-V0A.pdf
  6. kff2

    FTDI FT230X USB-Serial BoosterPack

    Does anyone have an opinion on whether the ferrite bead can be safely omitted from the board? What about the other passives (there are 9 in all!)? I have a cheap breakout board from ebay based on FT232RL that does away with most suggested components (only has two small caps), and it seems to work perfectly fine. I wonder where to draw the line in simplifying the design.
  7. The results that you are seeing are not necessarily wrong. Signed integer overflow is not well defined in C or C++. It looks like gcc wraps around overflown integers in some cases and saturates them in others. Inconsistent, but not really a bug either. See http://thiemonagel.de/2010/01/signed-integer-overflow/ You can try the -fwrapv option mentioned in the article.
  8. kff2

    CC2500 Breakout/BoosterPack

    So I think I have figured it out. When resting the board on the launchpad header, the radio comes into close contact (~1") with the 12mhz crystal on the launchpad used for USB. The radio somehow causes interference and USB communication goes haywire. My workaround is to to solder a female 0.05 header to the board and a male 0.05 header to the radio, and to remove the radio whenever I need to program the board. I guess I could also make a cable to connect the board to the launchpad.
  9. kff2

    CC2500 Breakout/BoosterPack

    I finally got around to assembling the smaller stand-alone board. The problem I have is that the board does not program with the radio attached. When I remove the radio, programming completes successfully. Any idea why that would happen? I use the 0.05 header on the launchpad for programming. The board only has C1, C2, and R1 in addition to the microprocessor. I double checked all connections. Perhaps the radio is defective?
  10. kff2

    [S]IV-18 VFD Clock Booster Pack

    I assembled my kit today. It works beautifully. Many thanks to Cubeberg for designing the board, sourcing the components, and making the booster pack available to all of us. I managed to compile the source code for the clock with mspgcc. If anyone is interested, I only had to make a few changes to the original code: 1) Change isr declarations: //#pragma vector=TIMER0_A0_VECTOR //__interrupt void TIMER0_A0_ISR(void) ISR(TIMER0_A0, ISR1) //#pragma vector=USCIAB0RX_VECTOR //__interrupt void USCI0RX_ISR(void) ISR(USCIAB0RX, ISR2) //#pragma vector=TIMER1_A0_VECTOR //__interrupt void TIMER1_A0_ISR(void) ISR(TIMER1_A0, ISR3) You will need to include for the ISR macro. 2) Replace the two occurrences of _delay_cycles with __delay_cycles (note the extra underscore). 3) Since there is no _bcd_add_short() intrinsic in mspgcc, I added one manually to time.h: static inline unsigned short _bcd_add_short(register unsigned short bcd_a, register unsigned short bcd_ { __asm__( " clrc \n" " dadd %[bcd_b],%[bcd_a] \n" : [bcd_a] "+ro" (bcd_a) : [bcd_b] "g" (bcd_); return bcd_a; } 4) In write(), send1 and send2 variables need to be unsigned char instead of char. Otherwise the line send1 = 1<<(7 -digit); causes an overflow for the first digit (PM dot always stays on). To compile: msp430-gcc -mmcu=msp430g2553 -oS -o clock.elf main.c You can then upload the binary with mspdebug rf2500 "prog clock.elf" Hope this helps somebody.
  11. kff2

    [Group Buy-7]IV-18 VFD Clock - $20

    Bluehash ------------ 2 viperspd--------------1 cubeberg-------------2 Mtlevine0------------2 Georg----------------2[international] GeekDoc-------------2 Adrian----------------2 DSFLORES------------1 arashi-----------------2 SugarAddict----------2 kff2----------------------1
  12. kff2

    DCO calibration

    I am using gcc version 4.6.3 20120301, compiled from sources. Without optimizations, __delay_cycles(100) is c042: 3f 40 20 00 mov #32, r15 ;#0x0020 c046: 1f 83 dec r15 c048: 0f 93 tst r15 c04a: fd 23 jnz $-4 ;abs 0xc046 c04c: 03 43 nop c04e: 03 43 nop WIth -O2, I am seeing c03e: 3f 40 20 00 mov #32, r15 ;#0x0020 c042: 1f 83 dec r15 c044: fe 23 jnz $-2 ;abs 0xc042 c046: 03 43 nop c048: 03 43 nop Looks like the loop is one instruction shorter, as the tst is not really needed.
  13. kff2

    DCO calibration

    I think I have figured out why the timings were wrong. Under mspgcc, __delay_cycles() appears to depend on compiler optimizations to work correctly. By default, __delay_cycles(100) at 1Mhz takes about 160 us. With the -O2 flag, this goes down to 110 us. This looks like a bug in mspgcc.
  14. kff2

    Volt/Amp/Watt meter

    Thank you for sharing this oPossum. Is there a reason not to use hardware SPI when talking to 5110 besides compatibility with older chips? Or are there other issues, such as the D/C signal somehow getting in the way? Also, am I correct to assume that 5110 is no longer manufactured and that the displays that we buy from ebay come from recycled phones? Is there a similarly cheap / easy to use display that is still in production and can be bought outside of ebay (e.g. from Mouser)?
  15. kff2

    Uni-T UT61E for ~$50

    I got this multimeter today and I must say I am impressed. Very decent build quality, and the accuracy is excellent. 2.499V on Maxim 6225 precision reference (2.5V) and 5.000 on 6250 (5.0V). 0.1% tolerance resistors all read spot on. Capacitance was about 2% lower compared to my Fluke 87V. I did have to tweak calibration a little bit to get these values. There is an easily accessible trimpot inside for adjusting voltage. Originally I was getting 2.497 for 2.5V (which is still within spec). There is a serial port that spits out about 4 values per second. I used this open source program https://github.com/stv0g/dmm_ut61e to read the data. Overall, it's really not bad for $50. I still prefer my Fluke, of course, but it also costs 6 times as much. This multimeter is definitely better than Extech 330, which I bought from Amazon about a month ago and promptly sent back. If anyone has a question about the multimeter, please let me know.