Hi All,

I was laying out a pcb tonight for a line amplifier using a tool called PCB
Wizard (a good little tool) and reminiscing to myself how I used to do it in
the past using black tape and a scalpel onto clear film.

Despite the ability of PCB Wizard to autoroute etc., I still prefer to lay
the track out myself. As well as the obvious topological challenge, I
was trying to wonder why and came to the conclusion that, instinctively, I
avoid the use of sharp edges such as right angle turns - I always use
rounded bends and avoid the obvious stuff of high-level and low level
signals being non parallel runs, star earthing etc. that I was trained to do
and that autorouting has no general concept of.

The line amp I am laying out is based on one from an EMI TG mastering
console where the boards were originally laid out at twice full size using
the tape and scalpel method as was the norm during the late 60s and 70s and,
as well as the low number of active devices in the signal path and good
electronic design, modern equipment seems incapable of equating in terms of
sound quality. The evidence of this is that these consoles are still
preferred when it comes to the audio mastering process at the likes of Abbey
Road studios for their 'sound' as well as their flexibility.

I suppose the ultimate in this is point to point wiring, but that is not
what I want to do in wiring up a line amp.

So my question is this: is there any evidence that the layout of pcbs does
affect audio quality? If so why? The type of evidence is things like the
use of rounded instead of right-angle corners. (For example, if I remember
my EM theory correctly, sharp points are the most effective for radiating an
electric field so could right angles could radiate and produce cross-talk)

Perhaps somewhat esoteric, but worth a few lines of discussion.

I would be interested in the references for hard evidence and not hearsay.

Best regards

Mike

On Thu, 24 Dec 2009 00:28:14 -0000, "fredbloggstwo"
wrote:

Hi All,

I was laying out a pcb tonight for a line amplifier using a tool called PCB
Wizard (a good little tool) and reminiscing to myself how I used to do it in
the past using black tape and a scalpel onto clear film.

Despite the ability of PCB Wizard to autoroute etc., I still prefer to lay
the track out myself. As well as the obvious topological challenge, I
was trying to wonder why and came to the conclusion that, instinctively, I
avoid the use of sharp edges such as right angle turns - I always use
rounded bends and avoid the obvious stuff of high-level and low level
signals being non parallel runs, star earthing etc. that I was trained to do
and that autorouting has no general concept of.

The line amp I am laying out is based on one from an EMI TG mastering
console where the boards were originally laid out at twice full size using
the tape and scalpel method as was the norm during the late 60s and 70s and,
as well as the low number of active devices in the signal path and good
electronic design, modern equipment seems incapable of equating in terms of
sound quality. The evidence of this is that these consoles are still
preferred when it comes to the audio mastering process at the likes of Abbey
Road studios for their 'sound' as well as their flexibility.

I suppose the ultimate in this is point to point wiring, but that is not
what I want to do in wiring up a line amp.

So my question is this: is there any evidence that the layout of pcbs does
affect audio quality? If so why? The type of evidence is things like the
use of rounded instead of right-angle corners. (For example, if I remember
my EM theory correctly, sharp points are the most effective for radiating an
electric field so could right angles could radiate and produce cross-talk)

Perhaps somewhat esoteric, but worth a few lines of discussion.

I would be interested in the references for hard evidence and not hearsay.

Best regards

Mike



The sharpness of corners does not matter for audio, though it does for
RF (for microwaves corners are carefully calculated).

For audio the most important thing is the star ground point. That will
have a real effect on pickup from power supply lines, and confer
invulnerability from external pickup.

Routing of tracks is also important. Power lines will be dirty and
should not go near low level signal lines. Tracks on opposite sides of
the boards should cross each other at right angles to minimize
capacitive coupling, but this is a secondary worry, and not likely
ever to be a problem..

d

"fredbloggstwo" wrote

I would be interested in the references for hard evidence and not hearsay.


It's my personal experience that good layout is vital. Poor layout can
result in instability and significantly increased noise levels, but I'm
afraid that I cannot offer "references for hard evidence".

With something as simple as a line amp I'm suprised that you need
auto-routing. Most PCB layout software offers a manual routing option, why
not use that?

David.

In article ,
David Looser wrote:
It's my personal experience that good layout is vital. Poor layout can
result in instability and significantly increased noise levels, but I'm
afraid that I cannot offer "references for hard evidence".

With something as simple as a line amp I'm suprised that you need
auto-routing. Most PCB layout software offers a manual routing option,
why not use that?

I don't make anything very complicated, but working out the routing is
part of the fun. Might be different if it were a job, though.

--
*Money isn't everything, but it sure keeps the kids in touch *

Dave Plowman London SW
To e-mail, change noise into sound.

Well, I cannot do the hard evidence stuff, but I've encountered some pigs in
my time.
Who recalls the Sinclair IC12, in which it was all too obvious that inside
the chip small and large currents came too close to each other making the
sound distorted by intermodulation, one assumes. The pcb was not much
better in this respect despite the advice on keeping input and output earths
apart, it seemed the pcb design had been made before this was written.
They also seemed to have some nicely tuned wiring designed to pick up radar
pulses as well.

On a similar note, one amplifier I encountered in the 70s, had a unique
feature on its balance control. as you got near the end of travel, the
opposite channel came up again with a very distorted version of the audio.
This seems to be once more, inappropriate earthing of the pcb mounted pot.


I'd have thought the sharp point problem you mention might matter more in rf
or in high impedance audio more myself. However long runs of audio near
other signals, particularly if they are not audio is to be avoided. Ref some
Cassette machines with inverters for the displays getting into the audio
chain. Did I say Techniques? grin.

Brian

--
Brian Gaff....Note, this account does not accept Bcc: email.
graphics are great, but the blind can't hear them
Email:
__________________________________________________ __________________________________________________ __________


"fredbloggstwo" wrote in message
...
Hi All,

I was laying out a pcb tonight for a line amplifier using a tool called
PCB Wizard (a good little tool) and reminiscing to myself how I used to do
it in the past using black tape and a scalpel onto clear film.

Despite the ability of PCB Wizard to autoroute etc., I still prefer to lay
the track out myself. As well as the obvious topological challenge, I
was trying to wonder why and came to the conclusion that, instinctively, I
avoid the use of sharp edges such as right angle turns - I always use
rounded bends and avoid the obvious stuff of high-level and low level
signals being non parallel runs, star earthing etc. that I was trained to
do and that autorouting has no general concept of.

The line amp I am laying out is based on one from an EMI TG mastering
console where the boards were originally laid out at twice full size using
the tape and scalpel method as was the norm during the late 60s and 70s
and, as well as the low number of active devices in the signal path and
good electronic design, modern equipment seems incapable of equating in
terms of sound quality. The evidence of this is that these consoles are
still preferred when it comes to the audio mastering process at the likes
of Abbey Road studios for their 'sound' as well as their flexibility.

I suppose the ultimate in this is point to point wiring, but that is not
what I want to do in wiring up a line amp.

So my question is this: is there any evidence that the layout of pcbs does
affect audio quality? If so why? The type of evidence is things like
the use of rounded instead of right-angle corners. (For example, if I
remember my EM theory correctly, sharp points are the most effective for
radiating an electric field so could right angles could radiate and
produce cross-talk)

Perhaps somewhat esoteric, but worth a few lines of discussion.

I would be interested in the references for hard evidence and not hearsay.

Best regards

Mike

In article , David Looser
wrote:
"fredbloggstwo" wrote

I would be interested in the references for hard evidence and not
hearsay.


It's my personal experience that good layout is vital. Poor layout can
result in instability and significantly increased noise levels, but I'm
afraid that I cannot offer "references for hard evidence".

Instead I can offer a simple experiment.

Take a power amp and set it delivering something like a few watts of a
20kHz sinewave. Monitor the output with a distortion or spectrum analyser.
Try wiggling/moving the wires that convey the dc lines/output so that their
relative locations and spacings and orientations change. If any components
can be moved, try also moving them. (Use some plastic pliers for safety.)
Watch how the distortion changes. Chances are this will be noticable unless
the level of distortion is high in the first place.

Note also any changes in ripple at the output if using the amps own PSU
rather than an excellent bench supply.

I've certainly in the past found that changes in wire looming and PCB
layout and tracking measurebly change performance in terms of
ripple/hum/distortion/stability/rf sensitivity. Reasons for this should be
obvious.

Also sometime the orientation of capacitors can matter. If they are 'swiss
roll' types for example, turning them around changes which pad is linked to
the 'outside' of the cap and directly exposed to coupling externally.

With something as simple as a line amp I'm suprised that you need
auto-routing. Most PCB layout software offers a manual routing option,
why not use that?

Perhaps worth adding one point.

People have sometimes chosen to make PCB 'tracks' by just etching away gaps
between areas of conductor. This saves money on the chemicals as it reduces
the area to be etched. The snag is that you get increased coupling between
'tracks' so may have more parasitic effects. This might help performance,
but can also cause any instability, etc, to be worse than if using the
traditional approach of 'lines of conductor of defined widths'.

No idea about references. I learned these things by experience. Have fun.
:-)

Slainte,

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

Damned spellchecker, I must give it some education..
Brian

--
Brian Gaff....Note, this account does not accept Bcc: email.
graphics are great, but the blind can't hear them
Email:
__________________________________________________ __________________________________________________ __________


"Brian Gaff" wrote in message
om...
Well, I cannot do the hard evidence stuff, but I've encountered some pigs
in my time.
Who recalls the Sinclair IC12, in which it was all too obvious that inside
the chip small and large currents came too close to each other making the
sound distorted by intermodulation, one assumes. The pcb was not much
better in this respect despite the advice on keeping input and output
earths apart, it seemed the pcb design had been made before this was
written.
They also seemed to have some nicely tuned wiring designed to pick up
radar pulses as well.

On a similar note, one amplifier I encountered in the 70s, had a unique
feature on its balance control. as you got near the end of travel, the
opposite channel came up again with a very distorted version of the audio.
This seems to be once more, inappropriate earthing of the pcb mounted pot.


I'd have thought the sharp point problem you mention might matter more in
rf or in high impedance audio more myself. However long runs of audio
near other signals, particularly if they are not audio is to be avoided.
Ref some Cassette machines with inverters for the displays getting into
the audio chain. Did I say Techniques? grin.

Brian

--
Brian Gaff....Note, this account does not accept Bcc: email.
graphics are great, but the blind can't hear them
Email:
__________________________________________________ __________________________________________________ __________


"fredbloggstwo" wrote in message
...
Hi All,

I was laying out a pcb tonight for a line amplifier using a tool called
PCB Wizard (a good little tool) and reminiscing to myself how I used to
do it in the past using black tape and a scalpel onto clear film.

Despite the ability of PCB Wizard to autoroute etc., I still prefer to
lay the track out myself. As well as the obvious topological
challenge, I was trying to wonder why and came to the conclusion that,
instinctively, I avoid the use of sharp edges such as right angle turns -
I always use rounded bends and avoid the obvious stuff of high-level and
low level signals being non parallel runs, star earthing etc. that I was
trained to do and that autorouting has no general concept of.

The line amp I am laying out is based on one from an EMI TG mastering
console where the boards were originally laid out at twice full size
using the tape and scalpel method as was the norm during the late 60s and
70s and, as well as the low number of active devices in the signal path
and good electronic design, modern equipment seems incapable of equating
in terms of sound quality. The evidence of this is that these consoles
are still preferred when it comes to the audio mastering process at the
likes of Abbey Road studios for their 'sound' as well as their
flexibility.

I suppose the ultimate in this is point to point wiring, but that is not
what I want to do in wiring up a line amp.

So my question is this: is there any evidence that the layout of pcbs
does affect audio quality? If so why? The type of evidence is things
like the use of rounded instead of right-angle corners. (For example,
if I remember my EM theory correctly, sharp points are the most effective
for radiating an electric field so could right angles could radiate and
produce cross-talk)

Perhaps somewhat esoteric, but worth a few lines of discussion.

I would be interested in the references for hard evidence and not
hearsay.

Best regards

Mike

"David Looser" wrote in message
...
"fredbloggstwo" wrote

snip

With something as simple as a line amp I'm suprised that you need
auto-routing. Most PCB layout software offers a manual routing option, why
not use that?

David.

Hi David

I am doing the manual routing - I thought that I had mentioned that in my
original post as I like the challenge of doing so even more so on a single
sided board :-).

As regards the RF considerations that Don mentioned. I also seemed to
remember doing university experiments where it was clear that low frequency
signals do radiate at sharp corners and plotting them on conductive paper as
so many volts per metre.

The other points raised such as grounding and power lines are taken for
granted. I was just wondering if there is any evidence on the more esoteric
aspects as a point of interest.

On the point of 'a complicated line amp', its a bit of an unusual one as it
is a Class A design with a constant current output stage, quite elegant,
that drives a chunky transformer, so having 4 transistors and a plethora of
resistors and diodes, its not quite like a straight forward your average
5534 with a beefed up output stage.

Mike

In article , fredbloggstwo
wrote:


As regards the RF considerations that Don mentioned. I also seemed to
remember doing university experiments where it was clear that low
frequency signals do radiate at sharp corners and plotting them on
conductive paper as so many volts per metre.

Its true that 'sharp' features (local curvatures) do increase the local E
field levels around the point. And that the E-field lines tend to 'radiate'
in the geometric sense from them. However it will then depend on other
nearby conductor shapes how much E-field coupling there will be. Also how
long the straight parts are either side of the corner/point.

For low frequencies you can work these things out using quasi-static
approximates, etc. If you want some good books on the theory then my
personal recommendations are "Electromagnetics" and "Antennas" both by
Kraus. His book "Radio Astronomy" is also excellent if you have an interest
in things like RX design, enviromental factors, etc.

Kraus worked for many years designing and developing high-performance
systems. His books explain *clearly* a number of details that many other
books on EM, etc, either gloss over or make a dog's dinner of explaining.

Ramo, Whinnery and van Duser is also pretty good, but IMHO much less
clearly written than Kraus.

The other points raised such as grounding and power lines are taken for
granted. I was just wondering if there is any evidence on the more
esoteric aspects as a point of interest.

On the point of 'a complicated line amp', its a bit of an unusual one
as it is a Class A design with a constant current output stage, quite
elegant, that drives a chunky transformer, so having 4 transistors and
a plethora of resistors and diodes, its not quite like a straight
forward your average 5534 with a beefed up output stage.

One advantage of steady rail current Class A designs is that you don't have
to worry so much about coupling with the rails since they aren't radiating
any ac. Although with more common designs they will of course be showing
variations like those of the signal - but that isn't likely to be a source
of distortion. :-) The downside is that ripple needs more attention in
terms of having an adequate PSU for the high demands.

Slainte,

Jim

--
Please use the address on the audiomisc page if you wish to email me.
Electronics http://www.st-and.ac.uk/~www_pa/Scot...o/electron.htm
Armstrong Audio http://www.audiomisc.co.uk/Armstrong/armstrong.html
Audio Misc http://www.audiomisc.co.uk/index.html

"fredbloggstwo"
wrote in message

So my question is this: is there any evidence that the
layout of pcbs does affect audio quality?

Depends, but often there is.

IME power amps are the trickiest, as are regulated power supplies. It's
about high currents. One of the worst power supply designs I've ever found
was in a Heath THD analyzer. I knocked the out-of-the box residual down by
about 1/3 just by changing how the filter cap land patterns were laid out. A
strategically placed piece of 14 guage copper wire in one place, and a cut
land pattern someplace else, if memory serves.

People talk about star grounding, but IME stars vary, depending on what the
points connect to, and how.

Mixed signal circuits (e.g. DACs and ADCs) can be very sensitive to layout.

If so why?

In audio circuits, it seems like grounding can be the big issue. Of course
at very high RF frequencies, things get very tricky.

I've experiments with various grounding schemes while the circuit was
amplifying say a 20 KHz sine wave, with the output of the amp hooked to an
analyzer. The differences often show up as spiky distortion residuals.

The type of evidence is things like the use of rounded instead of
right-angle corners. (For example, if I
remember my EM theory correctly, sharp points are the
most effective for radiating an electric field so could
right angles could radiate and produce cross-talk)

That's RF, for you. Probably more of an issue with mixed signal circuits,
because of the clocks and other digital lines.