I'm looking into designing a board for a robotics application that is going to be spinning very quickly, and am wondering if there is anything special to take into account. It's going to be in the ballpark of 6000rpm, with the edge of the board 50mm from the center. That's a constant acceleration of around 2000g, with shocks of a couple times that. If I shock mount the board, that would reduce the sudden acceleration but not the rotational one.

What should I do differently from normal PCB design? Would through hole be better than surface mount? Which way should I orient the board, so that the acceleration is parallel to the board (preferable) or perpendicular? Which way should I orient components? Would conformal coating or potting be a good idea? I'd prefer not to, both for repairability and because of the weight. At what kind of accelerations would the component internals themselves start to break?

Also, maybe this is the wrong place to ask this, but what's the best orientation for a lithium polymer pouch battery in this environment?

  • \$\begingroup\$ I found this question helpful, though it relates to sudden shock as opposed to a constant acceleration and does not answer many of the questions. \$\endgroup\$
    – ahalekelly
    Commented Mar 11, 2014 at 19:16
  • \$\begingroup\$ The less momentum components have, the better. Make sure your substrate doesn't flex. CALCE at UMD calce.umd.edu is a whole department about electronics testing and more! \$\endgroup\$
    – HL-SDK
    Commented Mar 11, 2014 at 19:23
  • \$\begingroup\$ I know of research projects at my university that use almost exclusively flex PCB for high-stress situations like this; mostly because of the lower mass and better compliance to flexing of component leads. \$\endgroup\$
    – user36129
    Commented Mar 11, 2014 at 19:36
  • \$\begingroup\$ Mount only the parts you must have, on the board. A perpendicular coil can form a transformer replacing the battery. \$\endgroup\$ Commented Mar 23, 2014 at 16:59

1 Answer 1


The best advice I can give you is keep heavy stuff along the centreline. This also applies to xtal oscillators you may be using. You might also want to ensure that the mass of elements you use are balanced about the centreline of spin or you might get some heavy-duty vibration.

For small/low mass SMT components surface mount is fine but it might be worth considering glue to add extra bonding - not too much of course because the glue has mass.

Over about 10,000 g we tend to solid pot the electronics and we've not had a problem with jobs shipped over many years up to g forces over 20,000. Connectors (inter-board PCB stuff) can be a bit flaky so use those with turned pins and multi-leaf contacts (at least three).

There is a company in the US called Statek that make hi-G xtals - not cheap but if you want peace of mind I'd consider using them. Alternatively, look at Linear Technology's array of silicon oscillators - they not bad (and no moving parts other than the electrons) - I've been testing one across temperature recently and it stayed within 10 ppm/degC across 60degC.

I've never known the internal wires on a chip to fail. We always try to make sure the acceleration pushes the components onto the PCB but sometimes this is not feasible and then we pot stuff. I believe, at 6000g you should be OK not potting.

Also, It might be worth doing a spin-test on the finished article to check for vibration and anything going wrong. We do that with all our kit. Vibration we fix just like a tyre fitter does - small weights aren't preferred so we usually drill small amounts of metal off the outer casings.

  • \$\begingroup\$ That's an interesting point about XTAL's. If spun (or otherwise accelerated) fast enough, is there measurable physical distortion of the crystal? If so, how does it affect piezo-electric behaviour - variations in frequency, amplitude, phase, duty cycle? \$\endgroup\$
    – markt
    Commented Mar 12, 2014 at 1:43
  • \$\begingroup\$ Artillery shells with smart fuses experience 30,000+ g's. "RADAR" fuses go back to the tube days. Vibration (Jerk in physics terms - the derivative of acceleration da/dt) is a bigger problem over time. The instantaneous forces from da/dt can be crazy big. #andy aka what are you making :-) \$\endgroup\$ Commented Mar 12, 2014 at 3:31
  • \$\begingroup\$ @C.TowneSpringer - telemetry monitoring modules to go on rotating machines like when testing jet engines. Lots of strain gauge inputs and thermocouple inputs. All digitized and transmitted off the rotor at stupidly high data rates! \$\endgroup\$
    – Andy aka
    Commented Mar 12, 2014 at 8:15
  • \$\begingroup\$ @Andyaka Very cool. I worked at Boeing for a while in the old days. All the test wire was orange - so it didn't get left behind I think. Also very long sapphire light pipes to look at insides of GE engines. Neat stuff. \$\endgroup\$ Commented Mar 13, 2014 at 0:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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