5
\$\begingroup\$

Elec Eng student here, I've joined a project at my university's mechanical eng department who are building a rather large hybrid rocket. Unfortunately they've been having issues with the microcontroller (an atmega32) that's driving a hobby servo connected to the oxidizer valve. when testing in the lab everything works fine, but in two static tests in the field the valve closes shortly after opening. I suspect it's a power issue causing the MCU to randomly reset as they are running the MCU from a cheap switchmode 240VAC > 5VDC PSU with the only smoothing being a 100nF cap across the MCU! The problem I think is that in the lab they use normal mains but in the field it's powered by a generator..

If I can recreate the issue in the lab I'm going to look at beefing up the power-supply to the MCU, perhaps running from the 12VDC line into a linear reg (7805? ) with lots of input and output smoothing. what else could I do to make the power-supply bulletproof? common mode choke on the input? any suggestions or cool app-notes are welcome :P I'm also considering a grounded metal box for the electronics to sit in as opposed to the current plastic one. good idea?

\$\endgroup\$
3
  • \$\begingroup\$ Interesting problem you have! What you plan to do is right path regardless of whether it solves this specific problem. You do not say where this MCU sits. Is it on rocket or outside? Can you ad weight to it (seems like you can, you are planning to add metal box). How are you planning to get 12VDC? It will be goo if you can provide electrical isolation from generator using step-down transformer. \$\endgroup\$
    – mj6174
    Commented May 2, 2013 at 21:58
  • 2
    \$\begingroup\$ Maybe battery power it \$\endgroup\$
    – Andy aka
    Commented May 2, 2013 at 22:22
  • \$\begingroup\$ The MCU sits on the rocket body, the rocket is 4.5m long, the oxidizer alone weighs 45kg's!! so a metal box for electronics is not a big deal. For the test flight in the future at some point it will run entirely from battery's of course, but hopefully I'll redesign the whole board by then. In the mean time for the static tests it's powered with the rest of the system. there is also a 12VDC and 24VAC line for the other valve's / sensors etc. I think I'm going to try a big 7805 today with big smoothing caps :D \$\endgroup\$
    – Rambo
    Commented May 3, 2013 at 6:13

2 Answers 2

3
\$\begingroup\$

One way to "bullet proof" your power supply is to add transient voltage suppresors (TVS). See this application note for instance: http://www.vishay.com/docs/88490/tvs.pdf

Also check the fuse configuration of the ATmega32. Some of them may help to make your system more robust. In particular, check the Brown-out detection module and the start-up time (SUT0 and SUT1 fuses).

\$\endgroup\$
1
\$\begingroup\$

For powering the MCU in general: you want at least 1uF of bulk capacitance on the (I assume) 5V supply into the microcontroller, plus usually 0.1uF ceramic capacitance per pin on the device. Your plan of a 7805 plus large amounts of bulk capacitance is a good plan. At least 10uF of electrolytic capacitor on input and output is a good starting point. The linear regulator sound appropriate for this application because the microcontroller and servo input won't require too much 5V power.

For this testing, I don't think you're yet need to be worried about 'bulletproof,' but the fact that this is going to be a critical system on a substantial rocket indicates that the ATmega32 and hobby servo are not reliable enough to be safe in life applications. To engineer this to be safe, reliable, and effective, you should consider an MCU intended for at least automotive applications and ensure that the electromechanical and mechanical systems will fail-safe. 45kg of oxidizer is something to treat with appropriate engineering caution and care.

The packaging/mounting of your MCU control system will also be very important. The g-forces experienced on the rocket need to be considered as will shock+vibration characteristics/reliability. For example, double stick tape and poke-in wire connector are definitely poor choices for this implementation. Speak with your PI or project advisor and see if they have anything to contribute for safely implementing this system. Do not take safety for granted in engineering this, or any other project.

General advice: power the device in the lab in the same or a similar manner as to the intended application. Thus, use a battery/bench supply and a regulator to produce the MCU supply in all testing, avoiding using the 5V wall-wart type supply.

\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.