Hello! I will preface this post by saying that I have no experience whatsoever with this kind of project. I am looking to build an apparatus that will allow me to determine acceleration of a body after impact. Specifically, i am a student investigating traumatic brain injury and concussion and need a way to measure acceleration of an (anesthetized) rat head that will range from 15Gs to almost 200Gs.
So far, I intend to mount an ADXL377 (High-G Triple Axis) accelerometer to a PCB with 0.1 uF power supply caps and a 10 nF cap for filtering. From there I will need to run approximately 5 feet of braided wire (to span the distance between the concussion apparatus to a laptop) to 4.7 uF caps (for filtering the noise introduced through the 5 feet of wiring), and then to a 16 bit ADC connected to an Arduino Uno which will plug into a laptop through USB.
I have a couple of questions regarding the feasibility of this setup:
Will I need an amplifier before the 5 feet of wire to prevent loading the ADXL377 output and attenuating the signal from it? Or will signal drop over the 5 feet be insignificant?
Will the 4.7 uF caps I intend to place on the 5 feet of wire carrying data from the ADXL377 to the ADC actually function to reduce noise the way I want? If not, what kind of filter (if any?) should I use to accomplish noise reduction?
How can I calibrate the high G accelerometer? For accelerometers with a smaller range, using the accelerometer at rest and gravity (i.e. just dropping the accelerometer) to set the 0G and 1G voltages are enough to calibrate. However, for my purposes that range of 1G represents less than 1% of the accelerometer's range. The reason I am using a 16 bit ADC instead of the Arduino's built in 10 bit ADC is because 10 bits gives me a ~0.39G resolution, whereas 16 bits gives me 3 mGs of resolution. If I am to use 1G as a calibration range (I cannot think of another reliable method to get more acceleration), 0.4Gs of error definitely won't be sufficient to even get a reliable reading of that 1G range - hence the 16 bit ADC.\
The ADC (and amplifier?) should be wired as close as possible to the sensor, in order to keep stray pick-up as low as possible.
Don't add any capacitors to the signal lines. The shock signal that you try to measure contains high frequencies, which must not be suppressed.
For calibration of high accelerations you can use the centrifugal force. It's not easy, but doable. Mount your entire calibration circuit at the center of a turn table, and the sensor near the edge of the table. Power it by a battery, that withstands the moderate force near the center of the table. If you want to transmit data from that setup, add a wireless module, else store the measured values in EEPROM for later readout, or add a connector to transfer the data after the table has halted.
Many years ago I built a test circuit and later a rotating multiplexer, that withstands 50,000g of centrifugal force, using ordinary (soldered) IC and other components. Most critical are components in metal case, which possibly have bonding wires inside, which will break and fly away at some centrifugal force. Use tantalum caps instead of electrolytic caps, if required, and possibly the internal RC oscillator instead of a quartz or ceramic resonator. In your final device consider to separate the sensor and ADC from the controller and remaining circuitry, so that only the sensor will have to withstand the shock. Or pack your controller board into foam, that reduces the shock acceleration considerably.
I'm not sure what exactly you want to measure. Hair and skin will reduce the shock, applied to the skull, what may be wanted on the impact side, but not if the sensor is mounted at the opposite side of the head. Also the effect on the skull and brain may differ very much. In crash test dummies the sensors are mounted inside the dummy, and a special coating is applied that simulates natural skin.
I don't think it will be possible for me to wire the ADC very close to the sensor because when the rats are hit by the impactor, their entire body travels in quite a large arc inside the apparatus. To have the ADC inside the apparatus would mean I'd have to attach it to the body of the rat (their heads are so tiny that it's a small miracle that I'll be able to attach just the sensor and PCB). The rats wake up from anaesthesia within ~1 minute of being off it, in which time I have to attach the sensor to their heads, conduct the brain trauma, and apply lidocaine to the impact site, so I'm near the upper limits of the time that they are unconscious.
I really like your centrifuge idea - i'll definitely try that. My one question is that does it matter that all three axes of the accelerometer undergo acceleration at once during the calibration, instead of linear acceleration of one axis at a time?
Ideally, i would like to measure the acceleration/ deceleration of the brain inside the skull as that is what is thought to generate concussion and other closed-head brain traumas, but short of implanting the PCB onto the brain, I think that having the sensor mounted to the centre of the top of the head (with an impact to the lateral aspect of the head) is the next best thing to measure.
With the linear centrifugal force you'll have to calibrate each axis separately.
Will it be possible to measure the brain acceleration with a dead rat and incorporated sensor, or a rat dummy, and to check the impact on the brain with rats without sensors?
Yes, that's what I thought - I think just rotating the sensor through a couple different trials so that each axis of the accelerometer is parallel to the direction of centripetal acceleration should accomplish this.
Yes, I would like to use a dead rat at first to test the apparatus to make sure it works and also to kind of create a sort of "cheat sheet" standard curve for rat head acceleration as a function of impactor velocity. However, I'd still like to have the ability to hook the setup up to a live (but anesthetized) rat for the future (if different variables, such as rat age/ size, breed, impactor size, etc.) change, and also so that we can report exact accelerations if required when submitted papers to journals. Hence the desire to have a small sense with leads to the interpreting hardware.
Various government organisations have done research into head trauma injuries for over 50 years - back to when the USAF was designing the first ejection seats for aircraft.
I would be astounded if there isn't already standard testing processes, procedures and equipment to do this kind of investigation.
Before you start injuring and killing animals I would suggest that you talk to researchers that have worked in this field and check to see if your proposed methodology is going to be able to give results that will pass peer review.