LEARNINGS FROM CAPE
Well defined, workable, and reasonable mission need to be defined with benchmarks in place.
Affects entire project, clear rules, mission definitions and work schedule need to be in place.
Set definite goals and deadlines, but realize they may/will change.
Do not let obsolescence bite you. The world of electronic components manufacturing is a vicious one. In the course of CAPE1 development, both the Communications processor and the power amplifier were obsoleted and phased out by their manufacturers. After choosing a device in a design, order enough to last through the prototyping, testing and deployment stages.
Expect several revisions. Nothing ever works the first time. It is likely it will not work the second or third time either. In the real world of design and development, products often reach several revisions before released to production.
Every time a deadline was pushed back, the lab was vacant and a last minute push happened when the pushed deadline got close.
Towards the end, we were short on available personnel, particularly during finals week.
Funding was done well, from now on, there will be only limited personnel authorized to make any CAPE purchases.
Redundancy in connectors. If interboard connectors are used again, then I propose that 3 row PC104 connectors are used on both sides, instead of 2 row connectors. This will allow the entire interboard bus duplicated. This will prevent a loose connection from halting the performance of the satellite. Also take a look at PC104 interboard connector alternatives from Samtec.
Limited design of multiple layer boards. I suggest using multiple layer PC boards (MLB), however limited in function. Understanding the difficulty in moving to MLB design, I suggest that running bus lines from the 2 connectors on each end would allow ease of creating a redundant connector design without having to have so many traces where components are needed on the top and bottom.
Solar panel surface mount connectors. It turned out having wires running from the power board to the solar panel boards created a connection nightmare with broken wires and loose solder joints. Also, it created a mess when putting the sat together and getting fingers into tight places. I suggest using a surface mount connector that allows the solar panels, when screwed in place on the sides, automatically “snap” into the power board. CalPoly has done this with impressive connector design. See Micro Card for details.
Bettter antenna deployment. If antenna deployment is considered on the next sat, I suggest having a design that does NOT involve taking apart the satellite in any way. I suggest micro solenoids for catch and release mechanism or external wrapping of the antenna
Design an easy structure. Get with a mechanical team early to design an easy structure to take apart and put back together which doesn’t take 3 hours. Look into CalPoly designs more closely.
Understand filtering. Batteries didn’t seem to cause much of a problem other than the large amount of noise created on the line. Understand how to filter the noise generated by the battery on the power line early on. A clean power supply solves lots of issues in other places, especially with the radio.
Find out if your regulator has an “efficiency mode”. If it does, know what effects switching in and out of this mode causes. Hours were spent troubleshooting the radio at certain voltages when the problems were caused by a noisy power supply caused by the regulator switching out of efficiency mode. The solution flying in CAPE1 was to prevent the regulator from ever going into “efficiency mode.” Another solution could have been to let a processor control the efficiency mode and bring the regulator out of the “efficiency mode” with enough time to stabilize prior to transmitting.
Use Linear Voltage Regulators when possible. New linear voltage regulators with extremely low dropouts and high efficiency make it very feasible to use linear voltage regulation, especially in the RF circuitry. There are now linear regulators such as the Analog Devices ADP3338 with 190mV dropout and nearly 90% efficiency at 3V and 1A.
Test Solar Panels on High Altitude Balloons. Our team used the 28% improved triple junction cells by Spectrolab. However, these cells produced far more current in sunlight in space than the light here on earth causing our monitors to shutoff electronics. Have a plan to monitor higher than tested solar power.
More accurate ADC. Since the software libraries have been built for the 8bit http://www.ti.com TI ADC ADS7830, the 12bit ADC ADS7828 (which matches the same package type and pinout) can be used. I propose using a 12bit ADC from TI which will make voltage and current reading more accurate or allow for larger ranges to be measured with same accuracy. BTW, I’ll upgrade my ADC library for the new team if they’d like in order to use the chip.
Be wary on communication between processors. The OBC and TNC had a “simple” 2 wire USART bus for communicating back and forth. However, even with this “simple” connection, it took months just for the 2 processors to handshake and talk properly all the time. I suggest if multiple processors are used, take extreme care on developing the handshaking and comm. protocol between all processors for this is a tricky dance for the processors to learn.
External watchdog timer for all processors. Each processor should have a dedicated watchdog timer (WDT), any of which can completely reboot the satellite. Do not rely on one processor to “babysit” other processors and try to determine if they are locked up. Each processor should have its own WDT watching for a lockup.
Lack of OBC control of the batteries in case of overcharging from solar panels.
Diagnostic USART. When the OBC was having software issues, nothing helped more than having diagnostic information posted to a PC screen during trial runs. This allowed the OBC to be hardened for autonomous running. I suggest having the comm. processor to contain a USART port solely for commands and diagnostic reporting to PC. This will help tremendously.
Get a receive to work. Don’t assume because your code transmits fine that it will receive fine also. Be able to receive commands from your ground station early on.
Find out if your radio has an “efficiency mode”. This is something that caused a huge headache when the radio seemed to “not operate correctly” during certain voltages. It turned out that the radio would oscillate between high and low efficiency around a certain voltage.
Operate within the limits of your device. More time was spent getting the Chipcon CC1020 to operate outside its designed parameters and inside COTS HAM equipment (G3RUH 9600 baud modem). The transmitter portion of the device turned out to be more flexible than the receiver. Still, transmitting at a [http://en.wikipedia.org/wiki/Frequency_modulation#Modulation_index frequency modulation index] greater than one(1). If you cannot operate within the limits of your device, either find a new device/solution or adjust your limits.
Documentations/Design Keep an online log. Keeping a better log online should have been done. I suggest having each team member keep a log of all their activities and post them online somehow.
ExpressPCB doublecheck. RULE: The person double-checking should never be the person who designed it! There were way too many times the team had to reorder the same board because someone overlooked a connection or misplaced a component. Someone who hasn’t designed the board catches mistakes easier than someone who has been looking at it for too long.
When placing a component on the “backside”, remember, pins are reversed. Ask the power team about placing components on the backside. When staring at the PCB design, the pins are backwards which can be confusing.
Weekly meetings need more structure, should be for continuing project, round-robin help and brainstorming sessions need to be separate and not necessarily part of normal weekly meeting.
Lack of thorough, centralized, and catalogued documentation resulted in incorrect pin-outs of powerboard, and other resulting mismatches of hard copy documentation to actual components and trace layouts.
Need to have a standard documentation naming procedure and centralized storage of all docs.
Print the filename and version on your PCB. I don’t know how many times we’ve picked up a board and asked “What version is this? What PCB file is on this board? Does this board here have those changes we made? Is this board older than this board?” etc. Be sure before you order the board, the PCB filename and the version number is printed on the board
Keep a copy of all old file versions. Never know when you have to revert to what has been done before.
Establish a filename convention including date and version. Simple.
Be sure to keep an updated schematic as well as PCB file for your board. Its very hard to go back and look at your PC board to see where a certain component connects to. Its way easier to look at the schematic and see what’s connected to what.
Take good pictures often of the team and it’s work. One, you never know when a reporter suddenly needs pictures of your sat or your team for tomorrows front page story. You want to make sure you have that ready. Two, its kinda cool to go look back at where you started versus where you are now.
Need to work on SOP’s for different aspects of the project.
Written protocols need to be implemented. Also, Jason will put together an ESD training program and help with protocols. The lack of protocols affected proper operation of many sensitive components, including flight model.
Written protocols need to be written and observed. Also, being careful of equipment and components not belonging to the CAPE project in the clean lab.