From df071c3398fed1aa6a3a64eea67b2d4b3a1f993d Mon Sep 17 00:00:00 2001 From: Werner Almesberger Date: Fri, 10 Sep 2010 00:06:33 -0300 Subject: [PATCH] New ECN about clock circuit of atusd boards. Measured (absence of) effect of adding an inductor to the clock line. - ecn/INDEX, ecn/ecn0005.txt: new ECN discussing the switch to a capacitative clock voltage divider. With frequency measurements. - atusd/tests/send-20100910: measurement of transceiver performance when placing an inductor in the 16 MHz clock line. --- atusd/tests/send-20100910 | 66 +++++++++++++++++++++++++++++++++++++++ ecn/INDEX | 1 + ecn/ecn0005.txt | 37 ++++++++++++++++++++++ 3 files changed, 104 insertions(+) create mode 100644 atusd/tests/send-20100910 create mode 100644 ecn/ecn0005.txt diff --git a/atusd/tests/send-20100910 b/atusd/tests/send-20100910 new file mode 100644 index 0000000..a5e2040 --- /dev/null +++ b/atusd/tests/send-20100910 @@ -0,0 +1,66 @@ +Same test setup as for send-20100909. Board 100908-A was modified to have +a 220 nH inductor in the C7 position, instead of a 0R resistor. + +Distant Local +--------------- --------------- --------------- +100813-red 100813-orange 100908-A (with inductor) + +TX 255, -91 255, -85 +TX 255, -90 255, -85 +TX 255, -91 - +TX 255, -90 255, -85 +TX 255, -91 garbled + +- TX 255, -47 +- TX 255, -47 +- TX 255, -47 +- TX 255, -47 +garbled (-90) TX 255, -47 + +255, -82 255, -43 TX +255, -82 255, -43 TX +255, -82 255, -43 TX +255, -82 255, -43 TX +255, -82 255, -43 TX + +Control experiment: + +100813-red 100813-orange 100908-B (unmodified) + +TX - 255, -83 +TX garbled (-91) 255, -86 +TX - 255, -86 +TX garbled (-91) 255, -86 +TX - 255, -86 + +255, -83 255, -43 TX +255, -84 255, -43 TX +255, -85 255, -43 TX +255, -84 255, -43 TX +255, -84 255, -43 TX + + +Summary +------- + +Direction No inductor (1) With ind. (2) Change +------------------------------- --------------- ------------ ------ +100813-red -> 100813-orange 255, -87 255, -91 -4 dB +100813-red -> 100908-A 255, -88 255, -85 +3 dB +100908-A -> 100813-red 255, -85 255, -82 +3 dB + +100813-red -> 100813-orange 255, -90 - ? +100813-red -> 100908-B 255, -90 255, -86 +4 dB +100908-B -> 100813-red 255, -87 255, -84 +3 dB + +(1) Data from send-20100909 +(2) Only 100908-A was modified. Changes in the measurements with + 100908-B show inaccuracies of the test setup. + + +Conclusion +---------- + +The differences observed are the same for the modified 100908-A board +and the unmodified 100908-B board. The inductor therefore does not seem +to have a significant effect on transceiver performance. diff --git a/ecn/INDEX b/ecn/INDEX index b3e2085..3781ccc 100644 --- a/ecn/INDEX +++ b/ecn/INDEX @@ -4,3 +4,4 @@ Number Status Description 0002 Done Add load capacitors to 16 MHz crystal 0003 Edit Replace balun and filter with integrated balun 0004 Edit Take into account layout considerations for RF +0005 Edit Correct atusd clock voltage divider diff --git a/ecn/ecn0005.txt b/ecn/ecn0005.txt new file mode 100644 index 0000000..d6225d6 --- /dev/null +++ b/ecn/ecn0005.txt @@ -0,0 +1,37 @@ +Correct atusd clock voltage divider + + +According to section 9.6.3 of the data sheet, an external clock supplied +to the AT86RF230 has a minimum peak-to-peak voltage of 400 mV and a +maximum of only 500 mV. Furthermore, the signal must be DC-free. + +The voltage divider in the 20100903 and 20100908 designs is too sensitive +to interference and the PLL constantly unlocks, rendering the device +dysfunctional. + +A more robust divider circuit can be obtained with the following +replacements: C7 -> 0R, R3 -> 33 pF, R4 -> 220 pF. After reworking the +20100908 boards, they no longer suffer PLL unlocks. + +A Qucs simulation of the circuit can be found in ../atusd/sim/cdiv.sim + +Frequency measurements yield the following results: + +Host Board Error Meas. accuracy + (f, ppm) (ppm, nom.) +------- --------------- --------------- --------------- +Ben #2 20100908-A +2 99.7 + +1 99.6 +Ben #1 20100908-A +23 99.7 + +24 99.9 +Ben #1 20100908-B +24 99.4 + +24 99.7 +Ben #2 20100908-B +2 99.8 + +2 99.9 + +The Ben's 12 MHz crystal has a tolerance of +/- 30 ppm, which is better +than the +/- 40 ppm required by IEEE 802.15.4. + +What remains to be verified is whether this change causes interferences +that may affect transceiver performance and may also violate emission +regulations.