What makes a layout "sound" good?

[Solidhex]

Any thoughts? There's keeping the traces short. Making sure the signal path doesn't snake around and double back. What other things should someone keep an eye on when laying out circuits?
  I've built various Superfuzzes using different layouts. I've used the Ustomp layout once and other ones I've done myself in Eagle, I've found that only when I use a layout aping the original pattern from the Univox Superfuzz does it sound "best" and most like the original Superfuzzes which I've been able to compare to. That layout is pretty spread apart in that one. The other ones I've made while sounding pretty decent have a certain overblown sound to them, overcompressed, farty. All using same components, same transistors and operating voltages etc...Using the original layout gets a way more dynamic sound with a lot more depth. I've built enough to really be able compare the difference. Would the spread out layout add  impedances to the signal path? Is cramming everything into a smaller spot creating parasitic capacitances or interferences?

--Brad

[brett]

I don't know why you posted this amazing secret in the forum, but it's safe with me .

[JHS]

The sercret of "good" sound is a good board layout, choosing/selecting the "right" parts, a "correct" wiring and a finetuning of values via ear. IMHO choosing/selecting the "right" parts and a good wiring with a central ground is way more important than the board layout.

JHS

[jpm83]

I think that giving enough room for the parts and traces will reduce noise. I have a bad habit to sqeeze things to fit in small area, because I used to do my stuff on vero.

Janne

[darron]

i don't know if keeping the traces short makes all that much difference, but at the same time there's no need to make them long. i'd be more concerned with which traces run beside what. considering what goes on inside ICs, thick traces may not be all that important for stomp boxes also. resistance of the copper is in fractions of an ohm for a whole board, and consider that in many effects the signal will hit a one-THOUSAND ohm resistor (or many more times that) before it even gets to any buffers. but then there's no reason to keep them thin if you can afford not to. thicker traces will hold better.


prettier boards sound better i've found...

[John Lyons]

Brad
Were the parts you used for the sub par unit identical to the "good" superfuzz build using the original layout.
I would think that the Transistors hfe would play a large in getting a more open and deep sound.
If the transistors were too high in gain that would seem to lead to compression and being "overblown".
I would doubt highly that the original units had matched transistors but they most likely used a lower Hfe device.

John

[R.G.]

Two EE points to John. Your suspicions are correct. It is incredibly easy to assign audible differences to anything when in fact it's part tolerance.

The superfuzz was the first PCB I "laid out". That was in 1970. Etching that board was where I learned not to put FeCl in aluminum pie pans. 

In a layout -> everything matters, unless it doesn't <- .

Well, duh, R.G. - what a stupid platitude, right?

No. It's your job to *know* what matters. This involves knowing what signal, how big, driven by what impedance and received by what impedance is on every single trace. Only then can you make judgements about whether long traces, short traces, close traces, etc. make any difference. Or whether things like the variation in the characteristic impedance of the trace itself make a difference. On hypercritical high speed logic boards, traces must frequently have matched characteristic impedances and *constant* characteristic impedance as well as matched lengths because the speed of light (EM field actually) in copper matters to the signals.

This kind of background knowledge is not something that gets answered by a few posts, or even a few hundred posts in a forum. I've done layout for pay at times and because I wanted to and was interested in it since 1970. I'm still learning. I don't know nearly enough about it because I don't do circuit layout all day every day.

But you can arm yourself with some weapons. Do you know what the capacitance is between two copper traces 0.0014" thick (that's standard one-ounce copper) 0.025" apart is? How about the same 0.050" apart? Do you know if the width of the traces matters, or only the edge separation and copper thickness? What's the resistance of copper trace? (hint: rho*L/A) What's the inductance of 0.025" trace per inch? This is all stuff google gives you instantly. Notice how hard it is to say whether any of those translates into sounding good? That leap has to be made either inside your head as you do the layout or inside your ears when you complete the board.

The original question is also ill-formed. Layouts do not sound good or bad. They also arguably don't sound like the original or not, because in most cases, and certainly in the case of the superfuzz, the original varies more than most layouts have an effect. It is always possible to get a layout so bad that it oscillates, or is close to oscillating so has huge humps in frequency response. But there is no list of seven tips to avoid that. "PCB Layout for Musical Effects" runs to well over 100 pages and can only barely touch on the subtleties. And then there's the issue of where the wires go from the PCB to the jacks and controls...

So sorry - no easy list to check off. Like the English castle/manor house lawns, you gotta roll it for 200 years.

[DougH]

Just to expand on what John said-

If you want to build two units and have them sound the same, make sure you measure the actual resistor and cap values and match them between the 2 units too. All of these values (in addition to HFE for transistors) can vary quite a bit, depending on tolerance and etc.

Layouts are more of a "noise" issue than a "tone" issue, unless the layout is inherently unstable and the oscillation is affecting the tone. Differences between the "tone" of 2 apparently duplicate units will be accounted for by parts tolerances way before you even start considering the layout. Really poor layouts that are noisy and oscillate won't be perceived as sounding "good" or "not as good". Rather, you will wonder what on earth is wrong with it and why it doesn't sound right.

[Solidhex]

So far I've probably made about 7 superfuzzes. First one I socketed the transistors and tried a bunch of different ones. All the rest I've used 2sc828's in. Matsushita, panasonic. They generally fall in the 200-250 hfe range. Last three ones I've built are from the same batch of transistors. Normally I would just chalk all this up to parts tolerances etc but its just undeniable  that the ones I've built using the original layout just sound better. Not just to my ears either. They all seem to have the same noise level as each other. I've done 3 using the original style and 3 with my layouts and the first one was the Ustomp. I prefer using my own layouts since they are more compact and suited for the enclosures I like...
  Think I'll spread my layout out a bit and see what happens. Let you know how it goes...

--Brad

[earthtonesaudio]

Statistics, man... If you had built 100 or more, I would seriously suspect layout, but 7 could just be coincidence with part tolerances.  Especially when you consider that a 100pF cap with 20% tolerance will have most of the caps at either 120p or 80p... and relatively few in between.  Tolerances are funny like that.

However, you are correct that layout can have an impact on the sound.  But, it won't be as big an impact at audio frequencies as it is at radio and above, but it's still there.  Maybe a high gain circuit is worse, but who knows. 



I like that you brought it up though... anyone have any generalities about designing layouts like this?  I suppose one thing you could do is design layouts with these things in mind.  For example you could take into account a few picofarads going into a high-gain stage, which would roll off some highs, which affects the next stage in a certain way, etc.

[Paul Perry (Frostwave)]

The superfuzz was the first PCB I "laid out". That was in 1970. Etching that board was where I learned not to put FeCl in aluminum pie pans. 

The fact that RG was a newbie once, should inspire us all.

But I think he is too modest when he says his layout book only touches the edges. It certainly covers EVERYTHING that is likely to be a problem, in laying out a mere guitar effect.

[R.G.]

I prefer using my own layouts since they are more compact and suited for the enclosures I like...

My grandmother once told me that cakes that she decorated tasted better, even if someone else had originally baked them. 

Before someone throws a shoe at me: it is very, very difficult for human beings to be unbiased about anything they are involved in. The biases can be so deeply unconsious that they seem completely absent.

I also prefer my own layouts - but I have turned out some real turkeys, too, so I probably bend over backwards not to think it's good because it's mine.

[Solidhex]

Yo

R.G. What I've been experiencing with my layouts is the opposite of what you think I think I'm experiencing. I would hope my layouts would sound fine but what I'm hearing (basically my best set of tools is my ears) is that my layouts don't sound as good as the layouts I've made imitating the original Superfuzz circuit. I want to have a bias towards mine of course hehe...
  Most of what I have learned about layouts is from you, your book, and all the other great resources everyone here has generously given. I know they're many variables involved but I'm really noticing a trend here. I'm really just curious about it.

--Brad

[brett]

Hi
IMO RG was almost right when he said

it is very, very difficult for human beings to be unbiased about anything they are involved in

.

It is generally regarded as impossible for people to not display biases, even if they only have a mild involvement in something or expectations of a particular outcome (my PhD concerned the psychology of judgement and decision-making).  As noted in other threads like this, one way to "break" the involvement is to do blind testing.

Almost in the same sentence that RG's comments about bias are rejected, I note this comment :

I would hope my layouts would sound fine but what I'm hearing...

"hope" and "my layouts" are an emotive combination, as are "would sound fine, but..."

If you want to test this perceived difference, it's easy enough, and would be very interesting.  Simply make up two old layouts and two new ones.  Then use one set of old-style components on one of each pair and newer components on the other. So there's four combinations (board-components): old-old, old-new, new-old, new-new.  Now, get someone else to set them to about the same volume and distortion levels.  Then, while you give scores (say, out of 100) get your co-researcher to randomly switch between the 4 different circuits.  Do this until every circuit has been used and rated at least 10 times.  (You don't need to test every one the same amount of times)

When you've got the results, post them here and we can work out the statistics concerning the size of differences in the ratings and certainty that they aren't just part of background variation.
cheers

[mahali]

Flux can cause troubles when the traces are close.

[DougH]

Normally I would just chalk all this up to parts tolerances etc but its just undeniable  that the ones I've built using the original layout just sound better.

If you haven't actually measured the values of the parts you stuck in each one, you can't say for sure what it is causing the difference in sound.