Audio Pcb Layout Techniques - Page 3 - diyAudio
Go Back   Home > Forums > Design & Build > Construction Tips

Construction Tips Construction techniques and tips

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools Search this Thread
Old 15th May 2013, 07:43 AM   #21
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Quote:
Originally Posted by aspringv View Post
Hi Gootee,
I've been etching pcbs as I go on various projects using the toner transfer method. I've mostly completed a small cnc router, with the intent that I can whip up a layout in eagle and then output code to cut it for me. Hopefully once thats working properly I'll have a workflow that allows time efficient experimentation.

I actually really enjoy the circuit layout side of the designing process. It intrigues me like a complex puzzle that actually has no perfect solution but can be tweaked to be 'best' in a number of different ways. I'm really happy putting hours into what I hope is a good design - although explaining why I think it's an enjoyable challenge is harder I guess I'm looking for the 'rules' that guide this part of the process to the best possible result given the constraints.

In a perfect world I'd just build multiple iterations of whatever I'm trying to build and compare empirically as to the improvements, but cost and time prevents this being a viable way forward for most projects. Hence me looking to develop a theoretical background on which to base design decisions on.

In terms of testing finished builds, I've a few older oscopes around - the best of is a 50MHz unit which appears to only have one working channel. Standard probes and ARTA on a pc for testing amp distortion into dummy loads. The sort of equipment you're talking about is beyond my needs, and sadly beyond my understanding as yet
You're living the dream, aspringv. Living the dream. I'm jealous of your CNC router. I was on that path once. But life got in the way and I never got back to it. Around that time, I was also doing a few PCB layouts, for kits I had designed. I too really enjoyed doing them. But that was back when I was just starting to try to get back into circuits and I actually wasn't very good at it, because I had no experience and no mentoring. I could make the circuits work fairly well, and even made them robust and stable in certain ways, and could come up with clever ways to fit them onto PCBs, and even greatly enjoyed the puzzle-working aspect of that.

My fairly-rusty EE degree was helpful, but would have helped a lot more if I had known what questions to ask myself, and what to analyze. LT-Spice was a Godsend. And I became extremely proficient with it, in many ways. But again, I didn't really know WHAT to simulate and analyze, in order to be able to generate good PCB layouts. I was trying to learn, but I was also in production mode most of the time.

The fundamental physics that is involved is not too conceptually difficult. And a well-done EE degree gives probably all of the needed tools to even apply it quantitatively. But I also have, for example, woodworking tools. And I can tell you, for sure, that possessing the tools does not make one a master craftsman of fine furniture. (They don't even tell me that such a thing as furniture could exist, if I don't already know.)

I could have tried to imagine all of the possible cases of what electromagnetic theories or equations might apply where, and how. But even if I thought of ALL of the possibilities and how they might apply to circuits on PCBs, although it's possibly a lot better than never having thought about them there would still be a long way to go, because, as often occurs when thinking about the behaviors of complex systems, I still wouldn't know, or even have a good "feel" for, which effects might be significant, and in which situations, and which ones would totally waste my time to even consider them. i.e. I probably still wouldn't have a good estimation of what was important to think about, or which approaches to those considerations would be useful and practical.

Basically, life is too short to make all of the possible mistakes, by yourself. So you need to try to learn from the mistakes of many others.

Even that takes some significant amount of time, before you can feel a little comfortable in the amount of knowledge and experience you have acquired by reading and examining and then applying what others have learned the hard way. "Experience" is also needed. But again, life is too short to go down too many wrong paths in the dark. So I think that you have to spend at least SOME time trying to reach a sort of "critical mass" just by absorbing the "distilled experience" of the many who have come before us.

There is a famous quote about "standing on the shoulders of giants". But if each one stands on the shoulders of all of those who came before them, altogether they will BE a giant. And it should be much quicker and easier to scramble up to the top of their shoulders than it would be to grow that tall, yourself.

Here is one post that I think touches on, in simple terms, examples of a couple of the types of basic areas that need to be considered, in audio circuit design and layout. This is just an example of how you will want to learn to think. But it also serves to point out that WHO you read and listen to is important.

The (high-cap.) unregulated PSU for chipamps

It took me a long time to even think about what was really going on and what was important, in the high-power portions of an audio power amplifier layout, and even longer to then think the right way about it. But in hindsight, it seems obvious and simple. And it helps me to know what layout questions to ask, and how to answer them:

High-power-capable output transistors are used as small-signal-controlled current valves. The small control signal (the music) varies their channel resistance in exact proportion to the music (not considering feedback, yet), allowing current from the power supply capacitors to whoosh through them, to the speakers, with exactly the right amount of current at exactly the right time. The main important action in the PSU is the current, which IS, directly, the music signal that we hear from the speakers. That current almost all comes from the capacitors, either the PSU reservoir caps or the decoupling caps near the output device. There needs to be enough capacitance to sustain the worst-case current amplitude for the worst-case time duration (the time between rectifier charging pulses), or else the current will "poop out" on strong bass tones. But we also require very agile current delivery, for accurate fast transients and high frequencies. Inductance delays the delivery of current so inductance might be significantly bad. Inductance will be caused by conductor length and geometry and layout. We can relatively-easily calculate the minimum required capacitance AND how far away from the power pins it can be (i.e. the maximum tolerable inductance), to still be able to get accurate-enough transient currents, and thus we can FIND OUT how significant that effect is, and what to do about it.

Drawing current causes the capacitor voltages to fall. And drawing current through conductor inductances causes voltage spikes and sags. Both might be significantly bad. Or they might not. But we can decide how much voltage variation is acceptable and then add capacitance and/or lower inductance (e.g. shorten and fatten conductors) until it's "good enough", or maybe even "as good as we want".

I was going to mention feedback and PSRR and how it's possible that the voltage sags don't makes the response non-linear (same small control signal makes same resistance but now maybe PSU voltage across device is different, giving different current; the definition of non-linear!). But there's more-basic stuff that's important, too, and this post is already very long.

Conductors have parasitic inductance and resistance. So currents through them induce voltages across them. Even small currents that are rapidly time-varying can make large voltages across an inductance. So not sending large or fast-varying current through the same conductor that ties your amp input's reference ground to the PSU ground, for example, is basic. It's "star grounding". Using star power AND ground is possibly even better. But much better still would be using full unbroken planes of copper for power rails and grounds.

LOOPS: Faraday's Law (part of Maxwell's Equations) reminds us that a time-varying magnetic field will induce a time-varying current in a conductive loop, and the amplitude of the current, all else being equal or held constant, will be proportional to the geometric area enclosed by the loop. The converse is also true. i.e. A current in a loop will induce a field in the surrounding space. So conductor loops can act as both receiving and/or transmitting antennas. And, they do. In fact ALL of them _ALWAYS_ DO. We can only control "by how much" they do.

(Often-)Unfortunately, for us, "circuit" MEANS "loop". An audio input, for example, with a source plugged in, and with two internal wires leading to an amplifier PCB, usually to both ends of a resistor between a device input and ground (where "device input" could be opamp input pin, triode grid, transistor base or gate pin, etc), MAKES A LOOP.

Leaving any space between the two conductors, anywhere along their entire lengths, will increase the possibility that an electromagnetic field in the surrounding space might induce a current in the conductors. That current would then induce a voltage across every impedance in the loop, such as, the resistor connected between the first active device's input and ground, which happens to be the input for a high-gain amplifier. Combine that with a power transformer, with space between the pair of AC Mains wires coming in, and space between a pair of secondary wires. HUM! Even separating the signal and ground traces on the PCB, between where they connect to the PCB and where they finally get to the input resistor, can allow significant hum.

Everything above still only barely touches on the MOST BASIC of the basic considerations. For example, what about the extemely-high-frequency stuff that the high-speed digital PCB gurus worry about? We are only dealing with audio frequencies, right? Unfortunately, no. RF and high-speed signals can and in fact ALWAYS WILL sneak in from the outside. It's only a question of how much, and what effects might occur to what degree. And since we can't predict the future environment in which our circuits might be operating, we have to assume that there could be significant RF present, at unknown frequencies. Trouble is, "everything's an input", for RF, including the outputs, the feedback loops, the power supply, and, of course, the inputs.

There are many sources of good information about thwarting RF incursion, which besides the obvious effects like hearing AM radio station programs from your speakers, can have subtle and insidious effects inside circuits and ICs, such as slightly shifting the DC operating points of transistor cicuits. Some basic ideas are mentioned in Chapter 7 of ADI - Analog Dialogue | Op Amp Applications Handbook .

Additionally, most of the active devices we use will be quite happy to oscillate or ring at high frequencies, even without any RF present. You have to pay attention to things like the effect of having open traces near IC pins. That's often bad because the tiny amount of parasitic inductance or capacitance could make the IC behave badly. So usually any component that connects to an IC pin should be connected right at the pin, not centimeters away, for example.

Here is another link I saved:

Analog Devices : Analog Dialogue : PCB Layout

Anyway, I'm certainly not an expert at any of this stuff, yet, but I hope that you get the idea.

Do some searches at some of the IC manufacturer sites (ti.com, analog.com, linear.com, maxim.com, and so on, for things like "PCB layout". And google it. Also, I've noticed that adding the word TUTORIAL to google searches can often be very helpful.

And learning LT-Spice well could also be extremely helpful. It's a free download from linear.com . Pay attention to parasitics, in both the components AND the conductors that connect them.

Cheers,

Tom

Last edited by gootee; 15th May 2013 at 07:49 AM.
  Reply With Quote
Old 15th May 2013, 08:57 AM   #22
marce is offline marce  United Kingdom
diyAudio Member
 
Join Date: Jun 2007
Location: Blackburn, Lancs
Before you have a meltdown, why dont you post your initial schematic and we can all work through it with you.
Doing layouts is the best way to learn, we train apprentices that way (generaly have a degree or other electronics qualification).
Also download this and have it handy when you are doing a design, it has most of the calculators you require:
Saturn PCB Design - PCB Via Current | PCB Trace Width | Differential Pair Calculator | PCB Impedance
  Reply With Quote
Old 16th May 2013, 12:36 AM   #23
diyAudio Member
 
Join Date: Sep 2010
Location: Perth, Western Australia
Default thank you for taking the time to reply in such depth

Gootee; Wow.

I'll take me a little bit of time to absorb your post (and do the amount of effort you've put into it justice), but I primarily wanted to say thanks for the time and thought you've put into that post. I now appear to have an astonishing amount of reading to do!

I'll reply in depth later today sometime...

Last edited by aspringv; 16th May 2013 at 12:59 AM. Reason: name fail
  Reply With Quote
Old 16th May 2013, 12:57 AM   #24
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Marc is actually a top professional in the PCB-design world. Whatever advice he gives us will be golden. You should post a schematic as he suggested, and work through doing a PCB for it. It's a really-wonderful opportunity for all of us to learn from one of the best.

Last edited by gootee; 16th May 2013 at 01:19 AM.
  Reply With Quote
Old 16th May 2013, 01:09 AM   #25
diyAudio Member
 
Join Date: Sep 2010
Location: Perth, Western Australia
Default working through an example

Marce - thanks for the offer! The last thing I tried to design was a preamp so If I could present that here for discussion would that suffice? If there's a better class of example circuit for working through then please let me know and I'll pull a schematic up from somewhere and work through that

I'd better cast another eye over that preamp pcb though - all of a sudden this is a slightly intimidating audience for my baby steps in pcb design

Last edited by aspringv; 16th May 2013 at 01:14 AM. Reason: I should stick to one thing at a time...
  Reply With Quote
Old 16th May 2013, 06:38 AM   #26
marce is offline marce  United Kingdom
diyAudio Member
 
Join Date: Jun 2007
Location: Blackburn, Lancs
Thanks for that Gootee was a Lumberjack We are so busy in the bureau, its getting crazy. This is a sad reflection on the UK electronics industry, our gain is other people loss as numerous PCB designers have been made redundant in the last several years as firms tighten their belts.
Interestingly, over the last couple of years, I am seeing more designs where the GND (0V, return) is split into definite power, analogue and digital planes joined either by ferrite beads or a star point. This is becoming more prevalent with our automotive customers, a shift brought on by the recent Toyota problems in the USA.
I look forward to us all playing with a design, it will be fun to do a board for fun and my own enjoyment.
  Reply With Quote
Old 16th May 2013, 12:41 PM   #27
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
That IS interesting! If you were referring to the "sudden acceleration" problem, then it's even more interesting! I have to admit that I pretty-much quit following all news (and all media, actually), well before that, but I remember that many people here assumed that whole affair was probably bogus. (Or maybe it was just me. ). To learn that it might have been caused by grounding and power circuit problems, and what has been changed to attempt to fix it, and why, is very gratifying. Can you please say a bit more about the differences between the behaviors of the old and new topologies?

Last edited by gootee; 16th May 2013 at 12:45 PM.
  Reply With Quote
Old 16th May 2013, 03:51 PM   #28
marce is offline marce  United Kingdom
diyAudio Member
 
Join Date: Jun 2007
Location: Blackburn, Lancs
Yep, the associated braking problem was not electrical or mechanical, but driver error, saving a lot of money for Toyota.
The actual problem was finally blamed on tin whiskers, I have the 700 page report from NASA (I was working on a project for Cobham/NG and part of the project was an investigation into moving to lead free solder and the implications for Military electronics, so had to read the report!!!). As part of the investigation Keith Armstrong (Cherry Clough Consultants) was also involved looking at the possibility of EMC being the cause, another few hundred pages of exciting reading.
What it did do is make the automotive industry nervous about future problems and the possibility of major law suits, so a lot of the information from both reports was examined and design practices for car electronics were updated and changed (to the extreme) to cover all possibilities for failure. A lot of automotive layout we now do is more stringent than mil/aero and medical, including such practices as all SMD MLCC capacitors doubled up and laid out at 90 degrees to each so that if one cracks under PCB flexing you minimise the chance of a short circuit (these being used for decoupling a short would take out the power, not what you want on say an ABS braking system.
As a result of the Keith Armstrong report a lot of transport based designs are now going back to segregated grounds (not only automotive but also railway systems), this is becoming more common, especially isolating the boards 0V from the main system 0V which as you can imagine can be quite noisy in a car.
Interestingly instrumentation and other areas of design are pretty much one ground for all, though obviously high current areas are separated from low current to avoid the obvious problems that mixing these two can cause.
Over the forthcoming years as more and more devices use RF, power line communication, LCD display etc there will be (there is) a move towards isolating more and more the core circuitry on a board from the rest of the world, ie its IO. This already happens a lot on commercial designs with both EMC and ESD protection on most inputs and outputs, for both incoming problems and to minimise any chances of the equipment transmitting crud to the outside world.
What segregation does mean that we have to be more pedantic with component positions and routing as crossing any barriers in ground planes becomes even more of a no no.
Have Fun
Marc
  Reply With Quote
Old 17th May 2013, 11:01 AM   #29
gootee is offline gootee  United States
diyAudio Member
 
Join Date: Nov 2006
Location: Indiana
Blog Entries: 1
Wow. Thanks for that!
  Reply With Quote
Old 17th May 2013, 07:28 PM   #30
diyAudio Member
 
Join Date: Oct 2007
Location: Aarhus, Denmark
Hi all,

Quote:
why dont you post your initial schematic and we can all work through it with you.
Doing layouts is the best way to learn, we train apprentices that way (generaly have a degree or other electronics qualification).
... sounds like a great offer and chance to learn (for me as well)! BTW I've been reading through parts of Henry Ott's book and for a newbie in higher speed PCB layout (me) I find it to be surprisingly straightforward and illustrative. Currently I'm looking into Howard Johnson's High speed Digital design (a handbook of black magic) which to me is also a superb read.

It seems to me that one of the main challenges in PCB layout is the overall signal path lenght and this also relates to the inductance of the capacitors used. To that end different capacitor types (SMDs or leaded) seem have quite different inductances associated with their terminations. Might one of you know of a source for impedance/inductance vs. frequency measurements on high-quality capacitors (e.g. like rubycon ZLH (radial leads) http://www.rubycon.co.jp/en/catalog/...inum/e_zlh.pdf

or rubycon SWZ (SMD) http://www.rubycon.co.jp/de/catalog/..._NSX_SWZ.pdf)?

Hmmm... Let me know if I'm bumping the thread too much with this question ;-) Otherwise I look forward to following it!

Jesper
__________________
... Being mindful about sustainability also in my audio endeavors matters to me ... paying CO2 - and a quite fair pollution compensation helps create a good listening "energy" ...

Last edited by gentlevoice; 17th May 2013 at 07:32 PM.
  Reply With Quote

Reply


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off


Similar Threads
Thread Thread Starter Forum Replies Last Post
Eagle vs. Sprint-Layout for PCB design/layout hollowman Parts 11 12th January 2014 09:01 PM
Cadstar PCB layout package for Audio treez Everything Else 3 24th September 2012 02:28 PM
Can any one give e a good audio equiliser schematic and pcb layout?? diptangshu Chip Amps 0 11th May 2011 06:49 AM
PCB layout for audio panson_hk Solid State 34 3rd September 2007 10:08 PM
PCB layout of audio controller with PGA2310 louischia Pass Labs 20 24th April 2003 11:11 PM


New To Site? Need Help?

All times are GMT. The time now is 03:15 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio

Content Relevant URLs by vBSEO 3.3.2