0
\$\begingroup\$

I have two almost-identical circuits:

circuit

Before continuing, I want to point out that I used three sets of jumpers for making logic selections to some of the pins. I know the better approach would be to use a pull-down or pull up resistor with a key, but my question isn't focused on those inputs.

I'm sharing the same power source with other circuits and if I don't use any EMI protection, then one IC in one connected circuit could screw up all other circuits relying on the same power source.

What I want to know is:

Which of these two circuits is recommended for best protection against EMI? and if the answer is the first one, will I need to replace the wire located at the bottom-right of the big X with another RC filter?

Also, do the resistors that form the EMI protection network (resistors connected to +ve) need to be higher than 1/4 watt? I intend to make the values low enough (in ohms) so current can flow through but I don't want too low of values or I'll feel I'll make the EMI protection circuit useless.

Additional question

The reason why I considered resistors is because then each subcircuit can draw less power from the main battery, that way, there won't be any chance of components fighting for power.

Based on the answers, it is claimed that using any resistor for R1 and R3 increases EMI. I think there are two issues with shorting out those resistors. For one, I still want some sort of noise filtering, and also, It makes PCB routing more difficult.

Perhaps I should just hit ebay and buy axial ferrite beads. But wouldn't inductors in the uH be better to filter out bad frequencies? After all, they are very low resistance as well.

\$\endgroup\$
2
  • \$\begingroup\$ Why do you have resistors in the power line to the ICs. This will cause significant switching noise on the power line and everything attached to it. \$\endgroup\$
    – Trevor_G
    Mar 25, 2017 at 5:42
  • \$\begingroup\$ When I look at passive filter circuits online, I see that they require a capacitor and a resistor/inductor. Maybe I should opt for an inductor instead of the resistor? I want to filter out the noise. \$\endgroup\$
    – user143136
    Mar 25, 2017 at 17:05

4 Answers 4

1
\$\begingroup\$

Neither...

Do not confuse signal filtering with power filtering.

Signal filtering is usually done on high impedance low current lines. You should not filter noise on a power line with an RC circuit when driving anything digital unless you TOTALLY understand your circuits current demand in the time domain. Even then, keep that resistor small.

Usually, for digital systems, a sufficient amount of carefully placed capacitance attached between Vcc and Ground is sufficient to reduce voltage spikes on to an acceptable level.

Additionally an in-line inductor can be added to the power line to limit current spikes. However, use the latter with caution. An inline inductor can in fact make things worse if your digital circuit is repetitively switching demand current levels current in a homogeneous manner rather than all the signals randomly. (Think Army Marching across a bridge.)

Further, if your Vcc supplies a high current load, like a relay, motor, bank of 1000 LEDs, that inductor will give you problems and may blow up some ICs.

As for emissions.

First, it depends on what "clocks" are running in your circuit. If your circuit is a simple logic board that gives an output based on some input condition or has a slow clock, emissions will not usually be a problem if you are not switching LARGE currents and voltages.

Obviously high frequency clocked boards will radiate that frequency as radio waves. That can be reduced by careful layout and ultimately shielding. However, even a relatively low frequency board will generate EMI.

The source of MOST digital noise is the rise time of the signals themselves. Digital circuits are designed, for obvious reasons, to switch their outputs as fast as their technology will allow.

However, in most applications, that switching speed will be overkill.

It is therefore common to see added resistors at the source of some of the busier digital signal lines, like clocks, address and data lines. The combination of this resistor and the capacitance of the PCB trace and input gates, forms a low-pass filter. This effectively reduces the rise time of the signal.. and with it.. the noise generated by that signal.

Unfortunately, this can leave the trace a little "floaty". Meaning it can be more susceptible to noise reception. If that's an issue you may need to filter the signal at both ends.

Any digital signals that leave the board via a connector and or wires should also be treated as above, but close to the connector. Even low frequency signals. Wires make wonderful antennas. That includes wires to motors, relays, sensors etc. too.

Any signal which enters a circuit from an external source through a wire or connector should be filtered for noise close to the connection point using a suitable RC filter. Additionally, when appropriate, other protection circuitry may be required like transorbs.

And the above is just the start. I did not even mention grounding, power planes, signal planes, antennas, trace resonance, shielding and the myriad of other things that affect a boards EM characteristics.

\$\endgroup\$
2
  • \$\begingroup\$ I do use ground planes as much as I can. So I'm better off then to filter clock inputs? and the highest speed clock for me (except for the attached radio) would be 20Mhz (a crystal connected to a micro) and the second highest speed clock is about 1.6Mhz, the speed of the ALE pin on the micro. The rest of the clocks are controllable via the micro and their speeds are less than 1.6Mhz. \$\endgroup\$
    – user143136
    Mar 26, 2017 at 2:47
  • 1
    \$\begingroup\$ @Mike The crystal is usually an ac waveform so you can not really filter that one. \$\endgroup\$
    – Trevor_G
    Mar 26, 2017 at 11:02
1
\$\begingroup\$

I'm sharing the same power source with other circuits and if I don't use any EMI protection, then one IC in one connected circuit could screw up all other circuits relying on the same power source.

EMI (ElectroMagnetic Interference) comes in two flavors - conducted and radiated - which require different techniques for effective suppression. Your concern about the IC 'screwing up' other circuits that use the same power source indicates that you are talking about conducted EMI. This concern is justified because your power source (a 3V Lithium primary cell) has high internal resistance, so even small currents will cause significant voltage drop.

An RC filter is effective at reducing voltage dips that are caused by current surges when CMOS gates change state. However the resistance also causes a DC voltage drop, so any circuit that draws significant continuous current will get lower voltage than it would otherwise.

In the first circuit, when you turn on LED2 the supply voltage to IC2 will drop more due to LED current flowing through R3. However the capacitor takes time to discharge, so the voltage drops slowly. If the LED is only flashed briefly then the RC time constant may be long enough to hold the voltage up during flashes, otherwise the resistance must be made small enough to not drop the DC voltage below the minimum required for correct operation.

The second circuit doesn't suffer from this problem, but it doesn't suppress voltage variations in the power source due to varying LED current. However if the other circuits use the same technique then they will also be protected. As the RC filter doesn't have to handle LED current, C1 can be made much smaller and R1 much larger. The IC switches quickly producing short duration spikes, so the filter can have a high cutoff frequency and still be effective. However it will not filter out lower frequency (perhaps down to DC) voltage variations at the battery caused by high current loads.

If battery voltage drop is a problem then you can either increase filter capacitance and resistance enough to keep the voltage up while the LED is on, and/or reduce LED current. If that still isn't enough then insert a Schottky diode in series with R1. This will prevent C1 from discharging when battery voltage dips, allowing it to hold Vcc up for longer (CMOS gates need almost no current to hold their state, so this could be a very long time).

\$\endgroup\$
0
\$\begingroup\$

The first circuit is better, because all noise on the power supply lines can be reduced by the decoupling capacitor C2. In the second circuit, R1 hides the LED1 switching noise (and anything happening on the jumpers) from C1.

But "better" is relative. As mentioned by Trevor, the resistors R1 and R3 actually increase EMI, because any change in IC1's or IC2' power usage will result in a change in the voltage drop over these resistors. Remove them.

To reduce EMI, you need components that remove high-frequency noise. The decoupling capacitor C2 is such a component: at high frequencies, it has low impedance, i.e., high-frequency noise gets shorted to ground.

To improve the circuit further, you could put a ferrite bead where R3 was previously: it has low impedance only at low frequencies, which means that the steady power supply current gets through, while high-frequency noise gets blocked.

\$\endgroup\$
1
  • \$\begingroup\$ I'd give you a +1 for this answer but I do believe the OP thinks he can get 5V from his battery by simply using a resistor. \$\endgroup\$
    – Trevor_G
    Mar 25, 2017 at 8:48
0
\$\begingroup\$

None is good!

if you insist to use the RC filtering then every current consuming connection to +supply should have own RC pair, even the led. The C's should cover the current spikes that tha parts draw when they change their states. The resistors are difficult to select if the current consumptions are not known. They depend on the temperature and the switching freqyency. As others are already commented, you should use RF choking by using ferrites in place of the filtering resistors.

The decoupling capacitors are a common practice even without the resistors. Without the caps the inductance of the wires and also the resistance of extremely thin wires can cause substantial voltage drops at the state changes.

RC filtering has a point: It can effective at very low frequencies if wanted. This has a drawback: The circuits can behave totally unexpectedly if the supply voltage comes on slowly.

You have totally forgotten the ground. The ground side should also checked. It spreads RF interferece as well as the + supply side. No sensitive circuit should be exposed to the current spikes drawn by another circuit through the common ground wire.

Noise making and noise sensitive circuits really should be separated. Read at first this to find the right terms for more searches:

https://www.digikey.com/en/articles/techzone/2013/jan/rf-shielding-the-art-and-science-of-eliminating-interference

\$\endgroup\$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.