11
\$\begingroup\$

I am going to design a stereo audio amplifier with TDA2030A. This is going to be an amplifier that is going to be powered from 27.6V DC and 12W per channel.

Is there any sources that I can learn PCB design guidelines for audio amplifiers from? Or can you give some tips?

\$\endgroup\$
  • 4
    \$\begingroup\$ The data sheet includes a PCB layout. Why not use that? I've used that amplifier in a design and didn't have any problems with the PCB. \$\endgroup\$ – Leon Heller Oct 9 '11 at 12:33
  • 1
    \$\begingroup\$ I don't just want to use it directly, I want to learn so that I can be able to design with other chips, too. \$\endgroup\$ – abdullah kahraman Oct 9 '11 at 13:04
  • 4
    \$\begingroup\$ @abdullahkahraman, I would say that 98% of practicing engineers I know use the reference design. You must pick your specialty and master it, no single engineer designs the board, layout, firmware, software and performs all testing himself on a regular basis, it takes forever to get products through the pipe. However, good you for you to attempt to learn more. \$\endgroup\$ – Kortuk Oct 9 '11 at 13:45
  • 1
    \$\begingroup\$ Well, I don't know what to say, I am just a fresh engineer so I want to learn something on everything, I don't think that in this level I need to specialize on something, because I am not too deep, in my opinion. However, later on, I will of course specialize on something I like most to work. \$\endgroup\$ – abdullah kahraman Oct 9 '11 at 14:03
  • 2
    \$\begingroup\$ 111014 - TI Circuit board layout techniques ~= 30page chapter from their book "Opamps for everyone". \$\endgroup\$ – Russell McMahon Oct 13 '11 at 13:21
17
\$\begingroup\$

I applaud your wanting to learn instead of blindly copying the reference design. I guess I'm the 2% of engineers Kortuk refers to. I may look at a reference design, but I'm not just going to follow it. I try to understand what the important points are and why they did what they did, then make sure to incorporate those if I deem them applicable to my design. Datasheets are often written in isolation, and any real design has other tradeoffs and issues that must be considered. I'd say that 98% of good engineers don't just copy whatever the datasheet examples are. Of course you have to read the datasheet carefully and make sure you understand what they are saying about particular needs of the part.

So to answer your question. The characteristics of audio are that it is low frequency but high signal to noise ratio. That means you don't have to worry about transmission line effects and the like. However, you do have to think about every little place noise can come in and try to prevent it in as many ways as possible. Separate filters on the power to any IC is a good idea, something like a small ferrite in series and a cap to ground right at the chip. This is for anything that doesn't handle the final output power. That needs a low impedance connection to the power supply.

Capacitive coupling onto signal traces has to be considered. This can be handled in routing, and sometimes you route extra ground traces around a signal path just for this reason. Keeping signal nets low impedance helps, but that's not always possible. Keep the sensitive traces away from traces with large voltage swings, like the final power output. Keep the power supply away from signal traces as best you can. Eventually the power supply needs to power the circuitry, but make sure it is well filtered before it gets there to that it it not a source of capacitive noise. In some cases you have to consider inductive coupling between traces, but that's usually not as big a deal as capacitive coupling, especially if you keep the high current final output traces away from the sensitive input traces.

Another source of noise is external coupling from the power line frequency or radio stations. Keeping out power line noise is one of the few places a shield can actually be a good idea. Putting the circuit in a metal box that is tied to the signal ground in one place is a good place to start. Simple R-C low pass filters well above the audio frequency but still well below radio helps keep down radio pickup. For example, one R-C pole in the 50-100kHz region won't effect audio, but will attenuate even AM radio.

There are a lot more details, and there are probably whole books written on this, but this should give you a place to start. A good way to learn is to try these things, then play around with them and see how they effect the output.

| improve this answer | |
\$\endgroup\$
  • \$\begingroup\$ I know many that do. I have spent a lot of time teaching myself PCB design techniques, so I would not include myself, but I find myself to be the exception, not the rule. I assumed a number of our top users are the same way. \$\endgroup\$ – Kortuk Oct 13 '11 at 16:22
  • 2
    \$\begingroup\$ 98% of bad engineers don't just copy whatever the datasheet examples are, either. I seriously wish they would. It would save me a lot of time and headaches cleaning up their terrible layouts. If you don't know what you're doing, copy the damn example, as closely as possible. When you learn why it's done the way it is, then you can modify it. \$\endgroup\$ – endolith Oct 18 '11 at 1:10
3
\$\begingroup\$

Starter for 10:

Moderate

Maxim - goodish - 4pp

Maxim - keeping RF out of your audio

TI - Achieve good audio quality in portables - goodish
PDF version of ame

IBEX - useful

Audio PCB layout guidelines. Targeted at a specific IC but useful. Downloads a useful PDF via google search link - absolute address unknown. Useful


How to get many many many more ... => where all the above came from


111014 - TI Circuit board layout techniques
~= 30page chapter from their book "Opamps for everyone".

Wow!
464 page TI book Op Amps for everyone - includes above PCB designm chapter.
This may be version one. You can but version 2 on web for about $US60.

Or you can buy version three from the just maybe upstanding and moral gentlepersons here a chapter at a time online for about $30/chapter or around $5600 for the book. I'm not at all sure why anyone would want to give them their business.

| improve this answer | |
\$\endgroup\$
  • 1
    \$\begingroup\$ Thanks Russell! Did that Google search and already checked all of these sources before. I was looking for AP notes or books, or maybe something more like first maxim link. However, Google didn't give me much satisfaction. \$\endgroup\$ – abdullah kahraman Oct 9 '11 at 15:00
  • \$\begingroup\$ Aha - the phantom anonymous downvoter is following me around again :-). A certain pattern emergeth. How about saying what's wrong with something that others are upvoting so we can all learn something? \$\endgroup\$ – Russell McMahon Oct 13 '11 at 18:06
  • \$\begingroup\$ I can't see any legitimate reason for a downvote either, so I added a upvote to cancel it. I also think that those who downvote should explain what their objection is, except in rather blatant cases. \$\endgroup\$ – Olin Lathrop Oct 13 '11 at 20:55
  • \$\begingroup\$ Don't blame me, I am an upvoter. \$\endgroup\$ – abdullah kahraman Oct 14 '11 at 13:16
2
\$\begingroup\$

The most important thing to understand about audio PCB layouts is that copper is not a perfect conductor. It has a small amount of resistance, and a current flowing through a trace will induce small voltage differences at different points along that trace. If you have a power supply draining heavy currents through a ground trace, and then you amplify a signal using another point on that trace as your ground reference, any noise in the power supply will be added to your signal.

http://www.aikenamps.com/StarGround.html

If you're working with an amplifier that has a differential input, run the ground trace back to wherever the signal is coming from, without touching anything else. Don't just ground it locally to the power amp's ground. It's the reference for the signal, and it should be connected only to the source's ground so that any difference in voltage between the source and amplifier grounds will be cancelled out by the diff amp.

Traces with nothing but high impedances connecting them to other circuitry will pick up interference easily. In general, if you have a choice, this op-amp layout:

▷—————————[MΩ]—▷

is better than this:

▷—[MΩ]—————————▷

Because the latter has a long trace with >MΩ impedance on both ends, opening it up to capacitive coupling, while the former has a long trace held "stiff" by the op-amp's low-impedance output, so that a coupling from a higher-impedance source into the trace will not have much effect.

| improve this answer | |
\$\endgroup\$

Not the answer you're looking for? Browse other questions tagged or ask your own question.