When making a pcb layout, does order of nodes have any importance?

[Herr Masel]

I'm making my first pcb layout for the bsiab, but I'm wondering what importance (if any) there is for placement order of several components connected to each other. For instance, if the schematic shows a cap, a resistor to ground and a transistor (drawn in that order), all sharing a node, do I need to keep that order in my board? I guess if they're connected to each other it's ok that I place the transistor "before" the capacitor or resistor, right?

[MartyMart]

That depends on the circuit, if it's say a cap to ground "before" being amplified by a tranny, then moving the
cap after the tranny would not be the same ... yes ?
If it all happens "just before the tranny" then the node will work either way .
You can move nodes around, but just be careful of parts placement and "flow" of the circuit.

MM.

[Gilles C]

I would say, the bigger the current flowing into the node, the bigger the difference it makes. Like relays and Leds for the worst examples.

Gilles

[Paul Perry (Frostwave)]

if the schematic shows a cap, a resistor to ground and a transistor (drawn in that order), all sharing a node, do I need to keep that order in my board? I guess if they're connected to each other it's ok that I place the transistor "before" the capacitor or resistor, right?

All that matters is that they are connected.
Shuffle the parts around all you want, just keep the same connections, the order doesn't matter.

Having said that, some arrangements may be more prone to pick up hum or noise, but that's a whole other thing. Fortunately, stomp box building is (usually) not as critical as eg. radio work.

[Herr Masel]

Thanks for the replies. They are a bit contradictory, so from now on I'll try to keep it in the same order drawn in the schematic.

[Paul Perry (Frostwave)]

Thanks for the replies. They are a bit contradictory, so from now on I'll try to keep it in the same order drawn in the schematic.

Fair enough, but note that no matter how you draw a schematic, it is the same circuit!!
The apparent discrepancy in hte replies is because I was ignoring the possibility of picking up noise from sudden current changes when diodes switch on and off. Schematics won't be drawn to take that into consideration, anyway.

[Torchy]

The physical order in stompboxes is only important in certain areas, mainly with power filtering.

Keep filter caps, reverse polarity protection and any resistors for RC filters close to the supply point on the board.
Reservior caps need to be kept as close to the chip as possible.
Avoid long signal traces and watch the proximity of signal traces to power rails.
Avoid ground traces going all the way round the pcb (acts as an aerial).
Try to dress out the pot/IO traces to the edge of the board to keep connecting wires short.
Try and keep digital and analog grounds separated.

Ive had problems with these in some of the vero layouts Ive done. Now Im into pcbs these were tips given to me by a "veteran" radio enthusiast.

[R.G.]

The order of connections on a copper trace has no importance at all to the degree that the currents in the trace cause no voltage drops along the trace. As Giles says:

I would say, the bigger the current flowing into the node, the bigger the difference it makes. Like relays and Leds for the worst examples.

Let me translate that. If copper were a perfect conductor, the order of nodes in a net (that is, all the connections which are connected by a copper trace) would make no difference at all, because the voltage would always be the same everywhere on the trace. However, copper is a resistor. The resistance causes the voltage to vary depending on how much current flows. Because the resistance is small, the copper trace is ALMOST a perfect conductor, but at high currents and especially with sensitive, high gain circuits, even the little voltages that build up can matter.

This is worst where there are momentary high current spikes - as Giles says, with relays, LEDs, and as BOM says, with rectifier power supplies. Even in effects pedals, these things can cause sudden current spikes that cause signals of a few millivolts to occur between points on the same copper trace. That doesn't sound like much, but a few millivolts times the hundreds or thousands gain of distortion pedals can make big, easily heard signals.

The Answer is to know where the big currents are flowing and to make those currents flow on copper traces that are NOT shared with signal circuits.

To the extent that the currents are low enough not to cause signals big enough to be heard, the order of nodes on a net does not matter at all.

You're now feeling frustrated, no doubt, because what I did was tell you that you simply have to know in detail how the circuit works and how big currents are that flow in every net.

I apologize, but that is the way it works. You have to know.

[bioroids]

Avoid ground traces going all the way round the pcb (acts as an aerial).

Oops, I sometimes do that, and also fill all unused board space with gnd copper. Why is that bad? Is it because it forms a loop, or is it because the large surface of ground?

Of course, digital and lfos and analog grounds, I keep them all separated. Also power supplies if posible

Luck

Miguel

[Herr Masel]

Thanks R.G., maybe frustrating but with a goal to work for!

Do you really keep LFO and regular grounds seperate? That might of just solved my noisy EA tremolo..

[Torchy]

... and as BOM says, with rectifier power supplies.

Who's BOM

[Gilles C]

One thing that I like to do when I really want to do a good pcb layout is I separate the circuit into modules.

I try to see each module having an input and an output, and connect every modules together after that. More work, and it can take more place on a pcb that way, but sometimes I find it easier to do big layouts that way, by splitting the problem in several pieces. And at the same time, it separates the nodes a bit more.

For small layouts, or when I feel lazy, or when I want a really compact layout, I just put everything together at the same time.

I think that it can help, for a beginner to look at schematics as being smaller modules as an input buffer, a tone section, an overdrive part, and an output module for example. Each module can even be tried separately on a breadboard. That's a good way to learn.

Gilles

[Ardric]

One thing that I like to do when I really want to do a good pcb layout is I separate the circuit into modules.

I try to see each module having an input and an output, and connect every modules together after that. More work, and it can take more place on a pcb that way, but sometimes I find it easier to do big layouts that way, by splitting the problem in several pieces. And at the same time, it separates the nodes a bit more.

Agreed!  It makes it easier to visualize where those big currents are within the module and plan how to get them in and out.  Then do the same with all the modules, like bringing all the module grounds together at the power supply.  I try to use two grounds per module: a "clean" one for signals, and a dirty one for everything else.  Extra traces for shielding can be dumped into the latter.  Then if I run two ground wires to the star gnd point at the power supply, I should be all set.

Schematics won't be drawn to take that into consideration, anyway.

Some schematics are drawn in a way that indicates how the power supply and grounds are treated, though it's not that common.  The worst use a single ubiquitious ground icon everywhere.  Better style is to use two icons, ie agnd and dgnd.  But best IMHO is when circuits to ground in the schematic are drawn as lines to a common ground local to that part of the circuit, so the schematic explicitly shows the same module scheme and grounding that Gilles speaks of.

Here's a random example, a Crate preamp schematic:  http://www.blueguitar.org/new/schem/misc_amp/crate_vc3112pre.gif

There's two grounds, one for signals and one for switching.  The first triode at the input is all grounded to one symbol.  This is a circuit module, and this is a ground node that deserves it's own wire to the star ground point.  The voltage/waveform table is a nice touch, too.  I like this style.

[Paul Perry (Frostwave)]

Well, I agree entirely with everything everone has said. And indeed some of the effects I build have separate analog & digital grounds (you won't get much more than 6 or 7 bits A to D conversion wihtout precautions like that).
but, for a beginner like the original poster..... the important thing is, the LOGIC and basic behavior of the circuit doesn't change as you change the order of connection to the nodes. Which I thnk is what he wanted to know.
As for picking up problems from the resistance of vero or other traces.... yeah I've been there. Even soldered fat copper 8 gauge wire on top of the vero to reduce the unwanted voltages (making V=IxR work in your favor).
Is it easy to get a 'working' layout? yes.
And with some effort and knowledge & experience, you can do an 'acceptable' one.
But a PERFECT one? Not in THIS universe

[brett]

Hi
For what it's worth, here's my rules.  They worked very well so far:
1.  Low gain circuit (about 10) - do what you like
2. Medium gain circuit (up to about 200) - filter cap (at least 1uF per mA) at the supply input.
3. High gain circuit - filter cap (2.2uF per mA or 100uF) plus a 0.1uf film cap at the supply input, and 0.1uF cap to de-couple any op-amps, plus star grounding.
cheers