Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Placement of a Bypass Capacitor

Status
Not open for further replies.

Electrix

Member
I seem to have forgotten the capacitor fundamentals learnt in college...could someone help me clear this out:

I have a micro-controller that gets its supply voltage from a voltage regulator IC. I also have a bypass capacitor to filter the high frequency components of the power supply to gnd. So on the PCB, where should this bypass capacitor be placed, near the Vcc & Gnd pins of the controller or near the Vout pin of the voltage regulator ? What are the trade-offs of one configuration over the other ?
Also please assume that the circuit above has other ICs too (perhaps an EEPROM chip, H Bridge..)

Thanks,
Electrix
 
Last edited:
Nearest to the device it is intended to bypass. If you think of the traces as being common in both directions, the direction they're shortest to are going to be lowest impedance, for a supply bypass you generally want high impedance from the supply and low impedance to the destination.
 
Last edited:
You should place the bypass caps before AND after the regulator, as close as you can. download the datasheet for the reg you're using. It should give you suggested cap values to use. I use 7805 Regs with 330nF on the input side and 100nF on the output. They work well for what most of my applications and are cheap and readily available.
 
Nearest to the device it is intended to bypass. If you think of the traces as being common in both directions, the direction they're shortest to are going to be lowest impedance, for a supply bypass you generally want high impedance from the supply and low impedance to the destination.

Sceadwian, Thanks for replying.

If I place the bypass capacitor near the pins of the device (that means away from the regulator) I am increasing the trace length, which means increasing the series resistance & the series inductance, right ? So that could cause additional voltage drops along the filter line, which I do not necessarily desire. So would you consider this point while deciding ?

-Electrix
 
Do both. As said above for 7805 reg's I use 330nF on input and 100nF on output side (datasheet suggested), I then use another 100nF cap across the microcontroller (PIC specifications).
 
@ELECTRIX
Generally in most of the voltage regulators it is recommended to use bypass capacitor in between output of regulator & ground.Since it is intended to bypass ripple & noise it will be better if high capacitance capacitors are used I had been using 1uF or 10uF.
 
Do both. As said above for 7805 reg's I use 330nF on input and 100nF on output side (datasheet suggested), I then use another 100nF cap across the microcontroller (PIC specifications).
I agree. You should have bypass caps at both the regulator input and output, and at all ICs as close as possible to the ICs power pins between V+ and ground. If you don't bypass each IC, then the internal current switching spikes flowing across the trace inductance can sometimes cause errors in the chip's operation. Even if you are operating the chips at a low frequency, the internal switching is still occurring at a very fast rate.
 
I agree. You should have bypass caps at both the regulator input and output, and at all ICs as close as possible to the ICs power pins between V+ and ground. If you don't bypass each IC, then the internal current switching spikes flowing across the trace inductance can sometimes cause errors in the chip's operation. Even if you are operating the chips at a low frequency, the internal switching is still occurring at a very fast rate.

I've always used the bypass caps at the input and output of the voltage regulator, but i must admit I don't often bother with the 100nF accross vss and vdd. Seems to work well enough without it.
 
I didn't say anything about regulator bypass caps because those are assumed, the datasheet for all linear regulators I've ever seen either explicitly require one or recommend it's use and recommend additional bypass capacitors if trace lengths are more than a couple inches from the regulator output to the load. The actually bypass requirements of a given IC may be completely unrelated to those requirements/recommendations and is IC dependent and should be added in addition to the ones the power regulator recommends, which are commonly electrolytics where bypass caps for MCU's or digital logic may be more useful if they're ceramic. The main purpose of the smaller ceramic capacitors for logic chips is to avoid switching transients in the power, so if you only ever lightly load your chips or don't switch at high frequencies you may never need them.
 
The main purpose of the smaller ceramic capacitors for logic chips is to avoid switching transients in the power, so if you only ever lightly load your chips or don't switch at high frequencies you may never need them.
The reason for the caps to suppress these transients is to prevent erratic chip operation, in either the same chip generating the transient, or adjacent chips. But those transients are independent of the operating frequency. They are due to the fast current transients generated when an internal logic gate changes states. The rate of this change is determined by the chips internal switching speed, not the external signal or clock rate.

The chips may operate ok without the decoupling caps, but it's not good practice, and it's a crap shoot as to when it may cause problems. If you like troubleshooting circuits for erratic behavior then go for it.
 
But those transients are independent of the operating frequency.
Absolutely not true. Just look at the current usage of a modern micro controller vs it's operating frequency and you'll see what I mean. If that logic is switching external I/O ports which are in turn delivering power that's what I'm talking about. They want you to use a high capacitance (which generally means high ESR) cap along with a low capacitance low ESR caps in order to fully buffer the chip. Modern micro controllers and high speed logic circuits fall pray to these stray AC effects so much so that they are actually as if not more important than their DC properties.

Generally speaking the stray capacitance for high speed low current switching from a micro controller is inbuilt in a circuit which is why they aren't often needed.
 
Absolutely not true. Just look at the current usage of a modern micro controller vs it's operating frequency and you'll see what I mean. If that logic is switching external I/O ports which are in turn delivering power that's what I'm talking about
Absolute not, not true. Those switching transients may be proportional to the external load, but not the frequency for a given microprocessor. It's the logic transition time (which is independent of frequency) that's the primary cause of the power supply transients. Of course microprocessors designed for higher speed may have faster logic transition times, so there is a secondary effect due to that.
 
I knew something's amiss when my PIC is behaving funnily.. it's got no caps for the supply!!

Hi Vizier87,

here's how to bypass your PIC. (Solder the cap on the bottom directly to the supply pins.)

It's almost as good as an IC socket with built in bypass cap.

Boncuk
 

Attachments

  • bypass.gif
    bypass.gif
    24.6 KB · Views: 231
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top