When not to use a ground plane?

[cwalex]

Hi, I have been lurking for a while, love the forum and Dave's videos. I have been trying to learn to make PCB's and I have a question about ground planes. I am mainly interested in audio circuits, specifically guitar effects circuits. I have been practicing by trying to make some PCB designs for some classic guitar effects like big muff etc, is there anything wrong with making the bottom layer a ground plane and running the rest of the traces on the top layer? is there any time when I shouldn't use a ground plane when cost of PCB manufacture isn't an issue?

I have looked at examples of PCB's for the big muff from places like general guitar gadgets but they are all single sided. Their designs never seem to use copper pours and I was wondering if there is a reason for it because a lot of the PCB designs I have looked at that were for digital stuff like microcontrollers seem to use copper pours a lot even on single sided designs and I wonder why.

Thanks for reading

[Rerouter]

ground planes and large copper planes are for failry higher frequencies, audible 22-22KHz doesnt really need any high frequency consideration, most micro's run at 16Mhz or higher, thats where you start to worry about RF signal integrity and copper pours,

not to mention single sided can be done at home with tone transfers,

if cost wasnt a concern, i would say a full ground plane wouldnt be that bad, if only just to mop up half of the EMI the amplifiers and other audio gear would likely be spitting out,

[free_electron]

Its got Nothing to do with frequency. Frequency is irellevant. Its all about edge rate (dv/dt) ( edge rate is sometimes called rise time and fall time, both of which are edge rates )

A triangle wave of 1MHz has totally different edge rate than a sqquare wave of 1MHz.

[cwalex]

Thanks for your replies, that was really fast.

If I understand correctly the digital circuits use ground planes because the signals are higher freq and more square so the ground plane will absorb more of it.

So there would be nothing wrong with using ground plane and copper pours in the audio circuits but it would be unnecessary.

Am I understanding this correctly?

[gxti]

Correct, there's no reason not to use one if it doesn't cost you anything.

[jahonen]

Having a ground plane has basically an effect that it lowers the impedance of the traces above it, if trace width is about same or wider than dielectric thickness. Although DC-like applications like audio won't benefit from that, there is a benefit that traces become more immune to external interference than traces "in the air".

Regards,
Janne

[free_electron]

Actually ground planes can cause unwanted oscillations due to stray capacitance in audio circuits....

First audio systems were built dead bug style by stringin parts between vacuum tube pins. No shield in sight... Works perfectly fine.

Once again, it is NOT frequency ! The determining parameter is edge rate. Edge rate is not frequency.
A triangle wave of 1 MHz at 1 volt has an edge rate of 2 volts per second.
Take a 1 MHz 1 volt square wave with picosecpnd risetime and you have an edge rate of 1 volt per picosecond.. And that spells trouble !

[Baliszoft]

Once again, it is NOT frequency!

Totally out of topic,  but if you got a high rise time square and a low rise time square, which one will contain more and louder harmonics in the hf range?!

[w2aew]

Once again, it is NOT frequency!

Totally out of topic,  but if you got a high rise time square and a low rise time square, which one will contain more and louder harmonics in the hf range?!

The faster the risetime, the more high frequency harmonic content.

[Baliszoft]

So at the end it is something about frequency ;-) ?

[ejeffrey]

Ground planes increase parasitic capacitance but lowers inductance.  They also reduce capacitive cross-talk between nearby signals since the capacitance to the ground plane overwhelms and screens the trace-to-trace capacitance.  For instance, the data sheets /application notes for most high speed CFB opamps recommend relieving the ground plane under the inverting input, as even a small capacitance can compromise stability.

When constructing a two layer board it is fine to use the bottom side as a ground pour, but keep in mind that it is 1.6 mm away from the signal traces.  This means it will have limited effectiveness as a ground plane if your traces are much narrower or closer than 1.6 mm.  It (probably) won't hurt, but if you are expecting it to have magical properties you will probably be disappointed.  If you are only doing it because you can, and it is convenient to have 0VDC available everywhere, knock yourself out.

Finally, ground planes on 2-layer boards are somewhat uncommon because they are so hard to implement.  Doing so restricts your layout to a single side.  As your circuit becomes more complicated it gets rapidly impractical to do that successfully.  As soon as your start cutting up the ground plane you have defeated the purpose.  At that point it may be better to ditch the ground plane and focus on doing a good 2 layer layout.

[free_electron]

no . has nothing to do with frequency. a 1KHz 1 picosecond rise time wil produce more crap in the spectrum than a 1 Mhz 500nS rise time ...

and ejeffry has a point too : capacitive loadig everywhere AND since your pcb is 1.6mm thick it doesnt do snot for RF ...
ground planes work only in multilayers and you need to create the stack very carefully. simply using equidistant thickness for the pre-preg gets you nowhere...

[Baliszoft]

no . has nothing to do with frequency. a 1KHz 1 picosecond rise time wil produce more crap in the spectrum than a 1 Mhz 500nS rise time ...

and ejeffry has a point too : capacitive loadig everywhere AND since your pcb is 1.6mm thick it doesnt do snot for RF ...
ground planes work only in multilayers and you need to create the stack very carefully. simply using equidistant thickness for the pre-preg gets you nowhere...

You did not get the point at all, but thats OK. Just dont do any signal analysis and you will be fine :-P

[jahonen]

Once again, it is NOT frequency!

Totally out of topic,  but if you got a high rise time square and a low rise time square, which one will contain more and louder harmonics in the hf range?!

The faster the risetime, the more high frequency harmonic content.

How about single pulses without repetition frequency then? I think it is difficult to think those in the frequency domain. Better approach is to think how long the edge is physically on the PCB.

Regards,
Janne

[Mechatrommer]

no . has nothing to do with frequency. a 1KHz 1 picosecond rise time wil produce more crap in the spectrum than a 1 Mhz 500nS rise time ...

and ejeffry has a point too : capacitive loadig everywhere AND since your pcb is 1.6mm thick it doesnt do snot for RF ...
ground planes work only in multilayers and you need to create the stack very carefully. simply using equidistant thickness for the pre-preg gets you nowhere...

You did not get the point at all, but thats OK. Just dont do any signal analysis and you will be fine :-P

sure you can somehow convert rise time to "equivalent effective frequency" (highest harmonics), but stating it in term of frequency alone can be misleading esp for noobs. take a single pulse (LO-HI or vice versa) for eg, too short rise time will create visible reflection and SI/RF issue. and its hard to tell in term freq for a pulse, unless, you know what you talking about

[Baliszoft]

Oh no, im not an rf guy. Also wanted to be strictly off topic, BUT stating that rise and fall times have nothing to do with the spectrum (therefore frequencies) of the signal is just not true. If i got it wrong, let me apologize.
For me a signal is not described solely by its fundamental frequency, though i agree that you dont have to care about it unless you are in the hf/rf range. F.eg why do they limit (slow down) rise and fall times to reduce emi? Not for dampening the radiation of the fundamental freq. for sure.

[w2aew]

Oh no, im not an rf guy. Also wanted to be strictly off topic, BUT stating that rise and fall times have nothing to do with the spectrum (therefore frequencies) of the signal is just not true. If i got it wrong, let me apologize.
For me a signal is not described solely by its fundamental frequency, though i agree that you dont have to care about it unless you are in the hf/rf range. F.eg why do they limit (slow down) rise and fall times to reduce emi? Not for dampening the radiation of the fundamental freq. for sure.

The only signals that are described solely by their fundamental frequency are sinusoids.  Faster risetimes lead to higher frequency components, regardless of the frequency of the fundamental.  It doesn't matter if the signal is periodic or not.  Consider the following complementary relationships:

A pure sinusoidal signal is represented by a single "line" in the frequency domain (contains only one frequency component),
and
A single "impulse" in the time domain is represented by a broad, flat horizontal line in the frequency domain (very broad frequency content), just not for very long .

A simple digital clock (square wave), is ideally composed of frequency components at the fundamental, and its higher order odd harmonics.  The more odd harmonic components added, the faster the risetime.  Picture it this way, if you pass a squarewave through a lowpass filter, you remove the higher order harmonics, and can reduce the signal back to a sinusoid.

[cwalex]

Hey, thanks for all the great replies. Looks like I'm wasting my time doing a ground plane other than it is convenient for gnd connections. I'll have a better read through the replies when I get home from work. Cheers

[free_electron]

Oh no, im not an rf guy. Also wanted to be strictly off topic, BUT stating that rise and fall times have nothing to do with the spectrum (therefore frequencies) of the signal is just not true. If i got it wrong, let me apologize.
For me a signal is not described solely by its fundamental frequency, though i agree that you dont have to care about it unless you are in the hf/rf range. F.eg why do they limit (slow down) rise and fall times to reduce emi? Not for dampening the radiation of the fundamental freq. for sure.

well ,you still got it wrong !

a pure sinewave has no harmonics. so it does not radiate in the spectrum. you will only find it at its fundamental frequency.
any other signal is a compound of base frequecies + harmonics.

the sharper a signal rises or falls ( dv/dt) , which is called edge rate , the more harmonics there are present in the signal. the more harmonics the more energy isdispersed throughout the spectrum.

the base frequency of the signal doesnt matter for this.
a 1 hz signal with 1 ps risetime produce an equal amount harmonics  as a 1Mhz ignal with 1pS risteime. the base freuqency isshited, thatis all. the number of harmonics does not change.

therefore : it is purely edge rate that determines how much energy is blasted in the spectrum. the frequency if the edge determines where the harmonics begin.

a 1Mhz 1vpp squre wave with an edge of 100nS does not produce as many harmonics as a 1MHz 1vpp with a 1pS edge rate.
simply becasue you don;t need that many harmonics to get that edge to be so fast...

[Baliszoft]

I think we are just talking about the same thing. I am doing signal processing for more than fifteen years (worked for several big companies in this topic in the past), so i do not really need a lesson how a sine, square, pulse, ect looks like in the frequency domain. You say that rise/fall time is what counts, but not the frequency (spectrum) and it is completely wrong! The two is the same thing, one comes from other, it's just the point of view. If it is hard to admit this for you (in your current position/reputation on this forum), it still does not change anything on the facts.

Just a last thoguht; You have a low frequency square and a high frequency square with the same rise/fall times (as you state). If the frequency is too high, you wont be able to talk about a square anymore, but only about a triangle instead. When i say my signal is a square, it is a square, no matter how high its freq. is.

[beaker353]

Ground planes and grounded metal enclosures in pro audio gear have almost nothing to do with signals playing with each other inside the case. It's all about keeping stray electric fields in the environment out of the case and signal paths. The inherent nature of the tiny output from a guitar and the magnitude of gain to get that signal to a usable level makes it super sensitive to fields in the environment. Try running a guitar with a unshelled cable within a few feet of a fluorescent light fixture transformer and you will see what I mean. There are very good reasons why most true pro audio gear has thick metal chassis with high gauge wire connecting everything together...

- EM

[free_electron]

I think we are just talking about the same thing. I am doing signal processing for more than fifteen years (worked for several big companies in this topic in the past), so i do not really need a lesson how a sine, square, pulse, ect looks like in the frequency domain. You say that rise/fall time is what counts, but not the frequency (spectrum) and it is completely wrong! The two is the same thing

No it is NOT !  This is a common mistake that is made over and over amd is very hard to stomp out.

The 'frequency' is not the 'spectrum'. The spectrum is the sum of all frequencies.
Fouriers theorem : a square wave is made by adding a fundamental sinewave with its harmonics. Every harmonic increases the edge rate.

The frequency of the signal is still only the base frequency ! Adding harmonics does not change the time required to pass through one full period of the signal. 'frequency' means 'base frequency'.

If you make an assymetric edge controlled square wave , lets say  a rise time of 1us, 3 us high, 1us fall time, 95 us low.
The period is 100us or 10khz.... The crap in the spectrum is still generated by the edges. In case of an assymetrical signal the spectrum distribution is going to be very weird since the superposition theorem doesnt hold.... That only works with symmetric 50% duty cycle signals...

If i take the same period for this pattern but i pump up the edge rates to picoseconds. 1 picosecond up , 5 nanosecond (-2 ps) high, 1 ps down.. The period of this signal has not changed, so its frequency has not changed! Yet it radiates waaaay further in the spectrum. So , edge rate is not the same as frequency ! The only relation is that your edge rate can never be slower than 1/2 period. If you edge rate is 1/2 period you have a triangle wave.

Its got nothing to do with position on the forum. I'm the first to admit that there is at least 90% of electronics i don't know anything about. The above is how it is defined in all literature about emc and emi.
I followed the course given by Montrose ( one of the gurus in this field) at UCSC a couple of years ago (uni of California - Santa Cruz) Very interesting stuff. Final exam was a pcb stack and layout that had to be optimized for emc. The damn thing was full of traps.. Aced it  walked away with 100%

[Baliszoft]

Im not arguing with you, it is an odd conversation. You are just proving your own truth.

[Write_to_Smokegenerator]

Well I'm defently not an Expert, but you I think it's a really bad idea to use a ground plane under a switch mode power supply (at least I heard that some months ago)

[free_electron]

It can be bad...

There is only one answer in anything related to signal integrity and EMC , and that answer is : it all depends.

If you have a coil that leaks magnetic field you may want to remove the plane underneath it so it induces no current. ( here is another common mistake : lets put copper to shield the inductor..... Copper does not shield magnetics ! )