Prototyping for Biosensors

General project help for Adafruit customers

Moderators: adafruit_support_bill, adafruit

Please be positive and constructive with your questions and comments.
Locked
User avatar
rajankum99
 
Posts: 1
Joined: Sat Oct 04, 2014 4:18 am

Prototyping for Biosensors

Post by rajankum99 »

Hi,

So I am engineer working on wearable biosensors. I would like to develop ultra-low power electronics, that can do electrochemical measurements and relay the info by bluetooth.

I know this won't be possible with the Arduino or any other systems, but I wanted to know which system (beaglebone, PI, or arduino) would help me get a grasp of basic analog circuit design and programing, so I may develop my own sensor.

Thank you for your time.

User avatar
Franklin97355
 
Posts: 23912
Joined: Mon Apr 21, 2008 2:33 pm

Re: Prototyping for Biosensors

Post by Franklin97355 »

Depending on the complexity of the sensor any of those would work but the Arduino will let you get to the raw guts of the sensor easier. Also as a wearable you can get Arduino compatable devices in much smaller packages.

User avatar
adafruit_support_mike
 
Posts: 67454
Joined: Thu Feb 11, 2010 2:51 pm

Re: Prototyping for Biosensors

Post by adafruit_support_mike »

Analog circuits and sensors live in a different family from microcontrollers like the Arduino.

You often use both in a data acquisition system, but there's usually an analog side, a digital side, and a layer of hardware that moves information between the two.

On the analog side, you need to be concerned with input impedance and noise.

All electrical signals contain noise from random motion of the electrons. The amount of noise is constant (called the 'noise floor') and is a fraction of the overall signal. The less power you have in your signal, the larger the noise fraction becomes.

In very general terms, the amount of useful information you can get out of a measurement depends on the ratio of the signal to the noise. Reducing it to digital form (since that's what happens when you move information from the analog side to the digital side), the resolution of a system can't be better than log2(S+N/N) bits deep. If your noise voltage is 1% of your signal voltage, that comes to log2(101/1)=6.6 bits worth of useful information in your signal.

Not only do those limits apply to your basic input system, they also apply to the amplifier that boosts the raw input to a level where it's usable by the rest of your circuit. The less power an amplifier uses to generate a signal, the harder it is to tell whether that signal came from the input or is an artifact of thermal noise within the amplifier's components.

Silicon also imposes a relationship between the amount of power an amplifier uses while idle and the amount of distortion that will appear in a signal passing through it. Basically, any PN junction is a diode, and the voltage across a diode rises by 60mV when the current through it increases by a factor of 10. To turn that into a distortion value, you find the ratio of the quiescent current to the signal current.

If you have a diode with no bias current and send a signal that ranges from 0mA to 1mA through it, the voltage across the diode will range from 0v to about 0.65v.. two thirds of a volt of distortion.

If you bias the diode with 1 microamp and send the same signal through it, the voltage across the diode will range between about 0.47v and 0.65v.

If you bias the diode with 1mA and send the same signal through it, the diode voltage will range between about 0.65v and 0.66v.

If you bias the diode with 10mA and send the same signal through it, the diode voltage will range between about 0.71v and 0.713v.

Those changes in diode voltage translate to distortion as the signal goes through a transistor, so to keep distortion low, you want the amplifier's quiescent current to be 10-100 times as large as the current in the signal you want to measure.

Fortunately the noise and distortion through a series of amplifiers is dominated by the noise/distortion of the first one, so you can burn energy for low noise/distortion where the signals are smallest. Spending 100uA on quiescent current to measure a 1uA signal doesn't hurt much when the signal you actually process is in the 1-10mA range.

Start by learning to work with high impedances and 1mV/1uA signals at DC, then learn how to work with them at higher frequencies, which is a reasonable challenge in a world of 10M oscilloscope probes. Then learn how to extend those techniques to the nanoamp/microvolt range. Once you can do that, you can make informed decisions about how far you can reduce the power consumption of a circuit and still get decent signals to feed into an ADC.

Locked
Please be positive and constructive with your questions and comments.

Return to “General Project help”