Decoupling Capacitor Ground Selection

6 replies · 4K views · last reply
edegro
Messages
3
Reaction score
0
I have a device in a DC circuit that calls for decoupling capacitors on VIN and VOUT. The circuit is powered by two batteries in series. The above device is hooked up in parallel to the second battery with the device ground connected to the node shared between the two batteries like so:

Code:
-------------
|               | vin
_              00
.               00--------- +
_ vdc2       00
.                |             vout
|                |
---------------------- -
|                |
_                |
.                 X
_ vdc1         X
.                 X
|                 |
--------------

Should I connect my decoupling capacitors to the node shared by the the negative terminal of vdc2 or the node shared by the negative terminal vdc1? Does it matter? Topologically, with the real world circuit, it would be easier to connect to that bottom most node shared by the negative terminal of vdc1, but I'm not sure if having the potential of both batteries across the caps will cause bad things to happen...

Thank you!
 
hi there
Welcome to PF

what is the device?
what does that vertical XXX line represent ??

not really enough info and the ascii drawing doesn't really help ;)

Dave
 
Last edited:
is this more like what you are looking for ? ...

Circuit schematic of a dual-supply rail with a three-terminal block between +VDC and 0V, Vin and Vout labelled, electrolytic capacitors on the input and output, and a -VDC supply below 0V.



where the device is say a voltage regulator ?

Dave

edited
 
Attachments
  • Circuit schematic of a dual-supply power rail: +VDC and -VDC batteries with a 0V midpoint, a regulator block labelled Vin and Vout with electrolytic capacitors on input and output, all returning to the 0V line.
    dualrailsupply.GIF
    1.3 KB · Views: 691
Last edited:
Yes, exactly. It's just a linear regulator. Is the way you connected the decoupling caps correct? Or rather, would connecting the negative terminals of those caps to the node at the very bottom of the diagram be incorrect?
 
edegro said:
Yes, exactly. It's just a linear regulator. Is the way you connected the decoupling caps correct? Or rather, would connecting the negative terminals of those caps to the node at the very bottom of the diagram be incorrect?

the way I connected is correct. Note also that because you are supplying the regulator from a battery then the supply is going to be very smooth anyway. Any decoupling capacitors are going to have very small values that are mainly for inhibiting any internal oscillations of the regulator chip.
Practically speaking ~ 10uF on the input and ~ 100nF (0.1uF) on the output. And preferably ensure that the capacitors are rated to at least double the voltage at that point
eg. a 10V on input to the reg. use 25V rated caps etc You would get away with 16V rated 10uF cap, I personally like a good bit of headroom ;)

ok to take the circuit one step further ... you may want a positive and negative regulated supply, it would be done like this...

Circuit schematic of a dual-rail regulated supply: a positive regulator LM78xx and a negative regulator LM79xx, each with 10uF input capacitors and 100nF output capacitors, between +VDC, 0V and -VDC.


LM78xx could be a 7805, 7812, 7824 etc
LM79xx could be as above but the negative version

cheers
Dave
 
Attachments
  • Circuit schematic of a dual-rail regulated supply: +VDC and -VDC inputs from batteries, a positive regulator LM78xx and a negative regulator LM79xx, each with a 10uF input capacitor and 100nF output capacitor, sharing a 0V centre rail.
    dualsypply1a.GIF
    2.3 KB · Views: 732
Last edited:
Thanks a lot Dave. Very much appreciated.

The caps were intended for oscillations, but, just out of curiosity now, would it ever be valid to couple down to that bottom node? If the caps were just there for line noise or any other kind of signal that needed to be grounded out?
 
edegro said:
Thanks a lot Dave. Very much appreciated.

The caps were intended for oscillations, but, just out of curiosity now, would it ever be valid to couple down to that bottom node? If the caps were just there for line noise or any other kind of signal that needed to be grounded out?

you're welcome :)

no it would never be valid to decouple to the negative rail. You always decouple to the 0V rail in a single or 2 rail system.

Now if these regulators were fed from a transformer and a bridge rectifier then you would need additional large capacitors on the input to the regs. The general rule of thumb is 1000uF per 1A of current drawn. This is because the DC output from a bridge (or other) rectifier system is not smooth. It has a lot of AC ripple voltage present that needs to be taken care of.

cheers
Dave

NOTE just noticed a boo boo I did in that diagram
The 100nF caps on the output of the voltage regulator are NOT polarised ... will do an edit
 
Last edited: