nravanelli
Posts: 20
Joined: Mon Dec 08, 2014 4:05 am

Capacitors

Fri Jan 30, 2015 2:09 am

I have been reading up on capacitors and I think they may be a good inclusion into my 8 channel thermistor ADC unit... My question is - where should I put them? My reading from the 8 channels is every 5-10 seconds... would I require the decoupling capacitors between my analog input-ground or between 5V and ground? Or both for all 8 channels?

I currently have my 8 thermistors wired like this:

5V ----- thermistor ----- 10k Resistor ----- analog input + ground

ame
Posts: 3172
Joined: Sat Aug 18, 2012 1:21 am
Location: Korea

Re: Capacitors

Fri Jan 30, 2015 2:24 am

Moar capacitors!

Capacitors are always good! Just solder a few in here and there. You can't go wrong.

Seriously though, if your circuit is working you probably don't need to add any. It's always good practice though to have a few decoupling capacitors next to any chips you are using, and on your power supply.

Ravenous
Posts: 1956
Joined: Fri Feb 24, 2012 1:01 pm
Location: UK

Re: Capacitors

Fri Jan 30, 2015 8:24 am

As Ame suggested, I'd put a decoupling cap between 5V and ground, close to the A-D chip itself. If these are, say, 0.1 uF (100nF) simple ceramic capacitors, one there will hopefully knock out any switching noise caused by the chip.

If the wiring out to each of the individual thermistors is long, then a capacitor between 5V and ground at the thermistor itself might help too, but that's just a guess.

By the way if you have problems with the readings themselves varying, that might just be general noise and nothing to do with decoupling. Tidy wiring might help if that's that case.

User avatar
aTao
Posts: 1087
Joined: Wed Dec 12, 2012 10:41 am
Location: Howlin Eigg

Re: Capacitors

Fri Jan 30, 2015 9:43 am

Ravenous wrote: If the wiring out to each of the individual thermistors is long, then a capacitor between 5V and ground at the thermistor itself might help too, but that's just a guess.
Also for long cable runs he type of cable is very important. shielded star quad (preferably with centre gnd conductor) will reject almost all noise.
>)))'><'(((<

User avatar
SpazzTechTom
Posts: 15
Joined: Sat Jan 31, 2015 12:50 am
Location: Glen Rock, PA. USA
Contact: Website

Re: Capacitors

Sat Jan 31, 2015 2:24 am

A capacitor is an energy storage device that blocks instantaneous changes in potential between the two points it is attached to. More simply put, it slows down changes in voltage. This is useful to remove unwanted ripple (noise) in voltage at the point it is attached to. So if you think you need to protect against noise, it's true that its not a bad idea to have one placed near the point where you want a stable voltage. However, if you are trying to measure fast changing signals with your ADC (it doesn't sound like you are), the capacitor can actually cause errors in your measurement due to its delaying effect, and it could be a very bad idea.

Probably a better use of the capacitor is near the power input and reference voltage for your ADC device because noise there can cause your ADC to obtain an incorrect value. It is comparing your signal input to that reference voltage, so it the reference is dancing around, your measurement can be wrong.

You should placing capacitors close to the device because noise can reenter into the circuit after the capacitor. There are different types of capacitors as well. Some, like electrolytic, have a polarity. If these will fail if you connect them backwards.

How much a capacitor slows down the change in voltage, often called "transient behavior", can be easily calculated. You might want to do this if you are actually going to connect a capacitor to your signal input. In general it takes about 5 time constants for the change in voltage to be made. A time constant, usually represented by the Greek letter, tau, is simply equal to Resistance * Capacitance:

t = RC

Where:
t = tau / time constant
R = resistance in Ohms
C = capacitance in farads (F) (note most capacitors are rated in uF, nF, or pF.
SpazzTechTom
Having fun with technology!

RaspISteve
Posts: 98
Joined: Wed Oct 24, 2012 9:15 pm
Location: Cheltenham, UK

Re: Capacitors

Sat Jan 31, 2015 5:09 pm

Perhaps worth adding is that although electrolytic capacitors are good at steadying a power rail they tend to be a little inductive meaning they aren't good for absorbing fast spikes. Quite often you will see an electrolytic in parallel with a disc ceramic begging the question: "what's the point of making a big value capacitor bigger by just 0.1uF?"

The aim being each cap does what its good at. One absorbs slow spikes the other fast spikes.
Share and Enjoy.

Mark_T
Posts: 149
Joined: Sat Dec 27, 2014 10:54 am

Re: Capacitors

Sat Jan 31, 2015 5:52 pm

All logic chips need decoupling to guarantee correct behaviour - if in doubt 0.1uF (must be ceramic)
on each supply pin to ground. And right at the chip, not an inch away.

A lack of decoupling can show up as bizarre and unexpected erroneous behaviour of a device that
is pattern sensitive, and which might only show up once in a blue moon - decouple properly and avoid such
debugging nightmares! If switching lots of current at logic speeds (LED arrays / displays for instance), much
more decoupling is needed, so 0.1uF + 10uF ceramic + 220uF electrolytic might be used then.

Whenever you use a high speed ADC to measure a slow signal, you have what is termed "aliasing" -
all the frequencies the ADC can respond to will contribute noise to the reading. Unless you filter
out the unwanted frequencies before sampling you will have noisy readings. Thus, yes, a capacitor
from ADC input to ground can be used (or more generally an RC filter). If you multiplex several
signals to one ADC then you need a capacitor per multiplexer input, not on the ADC pin.

Alternatively you can read at the full rate of the ADC and do digital filtering to remove the unwanted noise.

Return to “Automation, sensing and robotics”