5 This is a photo of our original board with serial port for off-board GPS,
6 without the big off-board 1000uF cap from the original ejection circuit. All
7 v0.1 boards were hand-assembled by Bdale. This is more significant than it
8 sounds... the CC1111F32 is a 36-pin QFN package, which necessitates reflow
9 soldering. Since we needed to reflow solder anyway, and because TI used them
10 in their reference design, we went a little crazy and used 0402 passive parts
11 everywhere. That means working under a microscope to place parts! Without an
12 inspection microscope, hand loading and testing might be impossible.
14 <a href="v0.1/loadedpair.jpg"> <img src="v0.1/loadedpair-thumb.jpg"></a>
17 significantly different than later versions, in both mechanical and
20 * 2.5 x 1 inch board with all parts mounted on one side
21 * 4-pin PicoBlade serial port connector for attachment of external GPS module
22 * USB connector projected approximately 3mm over the edge of the board
23 * Debug connector used 4 IC socket pins on 100 mil centers
24 * 50ma LDO regulator on early boards, later boards used a 100ma part
25 * [Microchip 25LC1024](http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en520389) CMOS serial EEPROM instead of DataFlash
28 * [Microchip MCP9700A](http://www.microchip.com/wwwproducts/Devices.aspx?dDocName=en027103) discrete temperature sensor
29 * used 1000uF electrolytic capacitor charged to 5V for pyro supply
30 * a dual LED instead of just one attached to the CPU
32 The v0.1 artwork had three issues, two of which required physical rework
33 on each board, all of which were fixed in v0.2.
35 * The USB connector footprint was placed wrong, so that the
36 connector hung out over the edge of the board instead of being
38 * We needed chip select on the SPI memory. To fix that, we gave up the
39 ability to put the accelerometer into self-test mode and used that
40 GPIO line to pull chip select on the memory, which required two
42 * The igniter sense circuits each needed a second resistor to
43 complete the voltage divider so our 3.3V CPU ADC could read the
44 5V ejection voltage. This was fixed by changing two resistor
45 values, and tacking two additional resistors onto the board
46 with jumpers to ground.
48 The schematics and PCB artwork for this version as of the working-v0.1 tag
49 are available here as pdf copies for easy reference:
51 * [schematic](v0.1/telemetrum.pdf)
52 * [pcb artwork](v0.1/telemetrum.pcb.pdf)
53 * [bill of materials](v0.1/telemetrum.bom)
57 These are photos of our second version. All v0.2 boards were hand assembled
58 by Bdale. The board featured in these photos also used the Tyco header for
59 pyro connections that we used to offer as an option. Nobody ever bought a
60 board with anything but screw terminals, so we no longer offer this option.
62 <a href="v0.2/cimg5164-cropped.jpg"> <img src="v0.2/cimg5164-thumb.jpg"></a>
63 <a href="v0.2/cimg5171-cropped.jpg"> <img src="v0.2/cimg5171-thumb.jpg"></a>
65 Lots of things were different from v0.1. Perhaps the biggest change was
66 integrating a Venus GPS receiver and a passive patch GPS antenna. This
67 required stretching the board length to 2.75 inches, and changing the layout
68 to put through-hole parts on the opposite side from
69 the surface mount parts. While we were at it, things were rearranged to put
70 tall parts near the center and not on the board edges so that boards could
71 be mounted in smaller diameter tubes.
73 Another significant change was adding a "companion" port, intended to support
74 expansion boards but also possible to use as a programming interface. With
75 a suitable cable, the presence of the companion port allowed any TeleMetrum
76 board to be used to re-program any other TeleMetrum or TeleDongle
77 board. Elimination of the discrete temperature sensor and second LED were
78 necessary to free up the pins needed for the companion interface.
80 We also changed to a different flash memory chip with 1 megabyte of storage.
82 Issues with the v0.2 boards included:
84 * The passive GPS antenna turned out to have disappointing performance due
85 to our many PCB geometry constraints.
86 * The initial voltage regulator on v0.2 boards was a 100mA part. Once we
87 realized how much power the GPS receiver needed to achieve initial lock,
88 we switched to a 150mA part.
89 * With the LiPo battery charging rate from USB set at approximately 100mA,
90 total power consumption can exceed the rate at which we draw power from
91 the USB interface, particularly when the GPS is in cold start mode. This
92 means a battery must be attached during operation, and also that the
93 battery will only charge effectively from USB when the board is turned off.
94 * The reset circuit worked well for the cc1111 but was marginal for the GPS
95 chip .. sometimes the board would have to be power cycled several times
96 to get the GPS to come up correctly.
98 The schematics and PCB artwork for this version are on the v0.2 branch in
99 our git repository, here are pdf copies for easy reference:
101 * [schematic](v0.2/telemetrum.pdf)
102 * [pcb artwork](v0.2/telemetrum.pcb.pdf)
103 * [bill of materials](v0.2/partslist.csv)
107 These are photos of our third version, which was our first production
108 build. By the time we understood what we wanted to change in v0.2, things
110 well enough that every time we went to a launch someone would ask about
111 buying boards from us. So in January 2010, Keith and Bdale made the decision
112 to produce enough boards to sell some to others. These were sold between April
113 and December 2010, and performed very well.
115 <a href="v1.0/cimg5972-crop.jpg"> <img src="v1.0/cimg5972-thumb.jpg"></a>
116 <a href="v1.0/cimg5974-crop.jpg"> <img src="v1.0/cimg5974-thumb.jpg"></a>
118 The changes from v0.2 were fairly small. The value of the reset capacitor
119 changed to increase the probability of the GPS chip coming up correctly, and
120 we moved to using a Taoglas amplified patch antenna attached to the U.FL
121 connector instead of the passive GPS patch antennas.
123 There were really only two issues with the v1.0 boards that annoyed us. The
125 that the reset circuit still wasn't satisfying. The GPS chip would sometimes
126 fail to initialize, particularly in hot weather. The second was that
127 fluctuation of the 3.3 volt power rail correlated with radio transmission
128 induced noise patterns in the accelerometer data.
130 The schematics and PCB artwork for this version are on the v1.0 branch in
131 our git repository, here are pdf copies for easy reference:
133 * [schematic](v1.0/telemetrum.pdf)
134 * [pcb artwork](v1.0/telemetrum.pcb.pdf)
135 * [bill of materials](v1.0/partslist.csv)
139 These are photos of v1.1, which rightfully looks very similar to v1.0. These
140 boards were sold through most of 2011.
142 <a href="v1.1/telemetrum-v1.1-thside.jpg"> <img src="v1.1/telemetrum-v1.1-thside-thumb.jpg"></a>
143 <a href="v1.1/telemetrum-v1.1-smtside.jpg"> <img src="v1.1/telemetrum-v1.1-smtside-thumb.jpg"></a>
145 There were several changes in v1.1:
147 * different flash memory part due to supplier availability problems. It turned
148 out that 2 megabyte chips were cheaper than 1 megabyte, so the size doubled.
149 * updated reset circuit to improve reliability at temperature extremes
150 * changed the GPS antenna footprint to eliminate the large through-hole
151 originally intended for use with a passive patch
152 * irq line eliminated from the companion port
153 * an additional resistor divider added to allow sampling the 5V supply
155 With an associated software change, being able to sample the 5V and 3.3V rails
156 made it possible to factor out almost all of the coherent noise from the
157 accelerometer data seen with v1.0 boards.
159 The one big problem with v1.1 boards turned out to be that they would
160 sometimes reset during flight, usually when firing pyro charges. Bdale
161 wrote a [blog entry](http://www.gag.com/bdale/blog/posts/RF_Immunity.html)
162 that explains in some detail what we finally learned was causing this.
164 For those who don't have ready access to the gEDA suite, here are pdf snapshots
165 of the files for Production PCB version 1.1 in more easily readable form.
167 * [schematic](v1.1/telemetrum.pdf)
168 * [pcb artwork](v1.1/telemetrum.pcb.pdf)
169 * [bill of materials](v1.1/partslist.csv)
173 This is our current production version, first made available for sale on
176 Changes from v1.1 include:
178 * changed to 70g accelerometer from Analog Devices since Freescale MEMS parts
179 are still unavailable following the Japanese earthquake and tsunami.
180 * reset controller changed from 3.15 to 3.00 volt trip point
181 * bypass cap added to LDO input
182 * pyro circuit resistor values tweaked to eliminate a glitch that at least in
183 theory made it possible for pyro charges to fire at power on. We never saw
184 a problem, but the change was easy and obvious.
projects / web / altusmetrum / blob