Connect with us

Some help on bypass and filtering needed.

Discussion in 'General Electronics Discussion' started by AndreasC, Apr 1, 2014.

Scroll to continue with content
  1. AndreasC


    Apr 1, 2014
    I have a few concerns regarding bypass capacitors. For me, capacitors and how they can be used is a complete science and very hard to comprehend fully. I don't fully understand it and would therefor need some help on it.

    I'm designing a circuit board and I want to use bypass capacitors to filter out noise that can easily be induced in wires. If I've understood it right, their function is also to supply IC's with power when power sources like batteries can't keep up with the fast switching in logic gates.

    The circuit board consists of different circuits that is designed to be run independently from each other. There are some IC's like Ne555 and Ne556 etc and also some individual transistors. For instance, I have a transistor that is driven by an IC. When the output of the IC is high, the transistor turns on. This result in a relay connected to the transistor to be activated. The relay is in other words driven by the transistor.

    I essentially need help with three problems here. The first one being, how should I design the filter that filter out potential noise on the incoming +9VDC line? What should I use? Electrolytic? Ceramic disc type? A combination of different types? What values?

    My second concern revolves around the issue of having bypass caps for individual transistors or not. I'm using 100 nF ceramics for IC's to make them work better in a potentially noisy environment. My thoughts is if I can use those even for individual transistors? If so, how should they be implemented? My main concern is, in the case of having a relay powered through a transistor, that the relay activates even if the transistor is turned off? I figured this might happen as capacitors can leak. Is that concern legitimate?

    And my third concern involves inductive loads like DC motors, relay coils etc. I'm using flyback diodes on my small DIL reed relays to protect electronics nearby from back EMF. Can I combine them with bypass capacitors to filter out noise created by the relay coils? What could such a circuit look like?

    Hoping for some help and learning possibilities regarding this.
  2. Arouse1973

    Arouse1973 Adam

    Dec 18, 2013
    This is what I do. For every IC I use a 100nF capacitor X7R ceramic and a 10uF multilayer ceramic as close to the power supply pin as possible. Now most people use a large capacitor on the input also, this can give you issues if your power supply is being used with longish cables. You can get ringing which can produce spikes greater than the voltage rating of the capacitor.

    You really have to look at the circuit as a whole and not place large capacitors everywhere they are expensive and you don't need them. You also don't need one for every transistor in the circuit. I would only add extra capacitors if you are switching say relays or high power IR or you were using a push pull type of circuit that sort of thing. If you are using power planes then it is quite common to place a few extra small ceramic capacitors around just connected across the supply planes.

    If you have a large amount of noise coming in that is low impedance then you would need to use a power line choke and a capacitor. Don't expect the capacitor to do all the work as for high frequency low impedance noise the large capacitor usually won't be fast enough because of ESL to reduce the noise much at all.

  3. duke37


    Jan 9, 2011
    The 9VDC supply should have a filter in it to limit any noise. This is particularly important if it is a switched mode supply.

    The circuit that is being driven will produce transients from a very low frequency up to many MHz. A PCB will have tracks which are inductances, these can cause high voltages if switched quick enough.

    Bypass capacitors will need to be effective over the whole frequency range.
    Electrolytic capacitors can have a large capacity but have considerable inductance so that the capacitor has a lowish resonant frequency. The capacitor appears as an inductance above this frequency.
    Disc ceramic capacitors have lower inductance so can work at high frequencies if the leads are kept short. A capacitor is often placed across the input power of a digital circuit to stop interference escaping and affecting other parts of the circuit.
    Layout design of digital circuits is not easy.

    Your flywheel diodes are there to protect the switch (reed relay). Think where the transients are generated and place capacitors to bypass them with as short a lead as possible.
    Last edited: Apr 1, 2014
  4. KrisBlueNZ

    KrisBlueNZ Sadly passed away in 2015

    Nov 28, 2011
    Hi Andreas and welcome to the Electronics Point forums :)

    I'd like to add some minor points to those good answers.

    I think one 0.1 µF ceramic or multi-layer ceramic capacitor per digital IC is enough; I do not use a 10 µF capacitor on every digital IC as well. Exceptions are high-current high-density devices such as fast CPUs; in this case, the CPU manufacturer will give recommendations for decoupling.

    Yes, this is the main reason for having bypass capacitors. When digital ICs switch, they draw a very short pulse of current through their supply pins. The decoupling capacitor acts as a local reservoir that supplies this current. This current flows in a loop, from the positive terminal of the decoupling capacitor to the positive supply pin of the IC, through the IC, out the negative supply pin, and back to the decoupling capacitor. You need to (a) minimise the total resistance and inductance in the loop, (b) minimise the total loop area, and (c) keep the loop independent from the rest of the circuitry as much as possible.

    Items (a) and (b) are achieved by keeping the decoupling capacitor close to the IC and using thick tracks. Item (c) is achieved through the circuit topology. The power and ground feeds from the rest of the board should connect to the capacitor, then the capacitor should connect to the power and ground pins of the IC. This ensures that the current is contained within the small loop, and voltage spikes within the loop (due to track inductance) are not coupled back into the main supply rail. If the topology is changed so the power and ground rails feed to the IC pins, then the capacitor is tacked on across the pins with a separate loop, current will still flow in that loop, but, more noise will be fed back into the rest of the board.

    Not. I would have a larger decoupling capacitor, say 10 µF, across the supply to the relays. But relay coils are highly inductive, so the current through them cannot change quickly.
    No. Even if a decoupling capacitor was leaky, that wouldn't cause current flow in the relay coil. Only if the capacitor was connected across the driving tansistor would there be a possible issue there.
    Back EMF from an inductive load is clamped by the diode that's connected across it. When the transistor turns off, the current that was previously flowing through the transistor then flows through the diode, and this current should be isolated from the rest of the circuit using the same topology as a decoupling capacitor - that is, run tracks from the relay coil to the diode, then from the diode to the rest of the circuitry. Minimise the total area of the track from the relay through the diode to the transistor. One 10 µF decoupling capacitor for a bank of relays is fine, but ideally, use the same topology so that the relay coil currents are kept within a loop that's independent of the rest of the board.
    Last edited: Apr 2, 2014
  5. AndreasC


    Apr 1, 2014
    Thank you for all the great response.It cleared things up somewhat.

    So for the line filter part, I could essentially use an electrolytic and a multilayer ceramic in parallel, between + and - at the power input on the board? Mainly to cover as wide frequency as possible. 10 uF ceramic paralleled with a electrolytic of a couple of hundred uF? Is voltage rating important? The higher the better? The wires between the battery and the board will be about 30 cm long or so and are going to be twisted. Is an electrolytic going to cause problems in this case? Is a choke needed? I don't have a clue about the amount or level of any noise.

    About powerplanes. I use a ground plane on the bottom side. I don't use a power plane, I use thick traces for power and vias that connect components with the ground plane. Is this a good practise? Also, should caps be used in this case as you mentioned, Adam, or only if I use both ground and power planes?

    So, I don't need to use caps on transistors. I could try using one 10 uF ceramic per bank of relays as you said. If I needed one for only one relay, I could use a 100 nF?

    A thought about bypass caps. Is it important for them to be multilayer and/or class 2?
  6. Arouse1973

    Arouse1973 Adam

    Dec 18, 2013
    Ok lets give an example. If you have an IC that dissipates 1W and you have only a decoupling capacitor of 1nF with a supply voltage of 3.3V and you can only tolerate a 5% droop of the supply then this will only decouple for 0.5ns = C*0.05*V/P. That's not very long. We want something like 5us to give say the voltage regulator time to react and increase the current.

    That's one of the reasons we use them. This would require oh about 10uF. It is also assumed by some that the capacitance between power planes is sufficient, this is incorrect since the capacitance is C=Eo*Er*A/h. With a dielectric thickness of 10mils only give about 100pF per square inch. This is why we distribute capacitor around.You really need to two planes for this to work best.

    The capacitor for the relay supplies current for the small capacitances between the coil windings but more importantly if you are switching the same power through one of the contacts into a capacitive load then this capacitor provides the required current for this. It all depends on what you are doing with the relay I suppose. It's a lot easier to not fit it on the PCB than to bodge one on after if you need it.

    You will only need power line choke if the noise getting in becomes an issue. I mentioned them because you weren't sure on what noise you might get.

    Last edited: Apr 2, 2014
Ask a Question
Want to reply to this thread or ask your own question?
You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.
Electronics Point Logo
Continue to site
Quote of the day