Audio Circuit Help

As long as your GND is sufficiently low-Z there is no problem. This is achieved by wide, short traces or a ground-plane on the pcb. avoid creating a loop in the ground circuit.

When using ONE 9V power supply you have to create 1/2 supply as a reference. Make sure this one is also low-Z, at least for audio.

When using just 9V power, your choice of op-amps is limited and specially low-noise and low distortion don't match with low supply.

To low values => not enough drive current To high values => noise + limited bandwidth

2kOhm - 100kOhm is OK op-amp outputs driving an interconnection-cable should have at least some 100 Ohm series resistor to avoid oscillation.

RF rejection of inputs by using inductors, ferrite-cores, small caps. Op-amps with moderate to high gain or driving loads should have some 22pF compensation in their feedback. Avoid multiple filtering ( low-pass or high-pass ) at the same frequency If your project needs to pass official tests concerning EMC, also apply RF filtering of outputs and supply. There is a lot of info on the web about this.

Robert

Ditto with Robert. The filtering should be evaluated at the output stage of each amp. You do not want to feed unwanted signals into the next stage Unless they are so small you need to amplify them out. As Robert indicated Low Z. no antennas and sometimes using larger wattage resistors help with noise e.g. 1/2 watt instead of 1/4. As far as amplifying radio stations, I think your speakers will only go to 20k in frequency which rids that problem. Even if they get through you will not hear them!

Thanks to you and Robert both for your responses. It looks like adding a bandpass filter from stage to stage is the way to go, along with lower resistances for controlling gain in the opamps. It sounds like 10K on the input and a 100K pot on the feedback loop might be a good choice. Any opinions?

Thanks for all of your recommendations. So what is the reason for not filtering multiple times at the same frequency?

It will create uncontrolled filtering leading to to steep cut-off and / or phase shift.

Better is: Dominant filtering at first stage / input by using single pole low-pass and high-pass filtering ( one C - one R ) All other stages get only frequency compensation by a little C ( 22pF ) around the feedback.

Make sure that the in and outputs of a potmeter are AC-coupled by using large enough capacitors. These will most likely be electrolytics because of the value needed. In a single supply circuit this will be no problem. For a circuit with symmetrical supply there will be the discussion of the ( polarized ) coupling capacitor being wrong polarized every half cycle. But they are not because there is no voltage- drop over these capacitors. Use high temp / high rel. types and they will last.

The slightest DC into a potmeter will make it noisy within weeks.

Robert

Amen to that. Steepness is not usually an issue, since most of the time it's desireable. And in fact, cascading simple single-pole stages won't have much effect on steepness, but it might have a big effect on the cut-off frequency.

Consider that the cut-off is the -3dB frequency... let's say it's 10 kHz for a single stage. Cascade two stages together, and at 10 kHz the response is now -6dB. The 3 dB point will be effectively moved lower. For simple single-pole R-C stages, it would be about 70% of the original, or 7 kHz in this example.

Multi-stage filters are normally designed with two poles per gain stage. To get a specified overall cut-off frequency and response shape, the stages have to be designed to work together. Each has a slightly different cut-off and damping (Q) which when cascaded give the target response. However, note that the more stages you use, the tighter the tolerance of each stage must be if you want to get the desired response. 4 or 5 stages (8-10 poles) is about the max I'd want to mess with, and even there you need to hand-pick parts.

Best regards,

Best regards,

Bob Masta DAQARTA v3.50 Data AcQuisition And Real-Time Analysis

Scope, Spectrum, Spectrogram, FREE Signal Generator Science with your sound card!

Usually completely unnecessary.

A lot of BAD info it would seem if you came to that conclusion.

FYI, I have designed many audio mixing products that have been in volume manufacture and not ONE of them needed explicit *filters* to pass the required EMC tests for CE. Mostly needed just a few extra caps of small value in a few judiciously chosen places. May of those were merely 'belt and braces' too.

Graham

Errr... Hallo !

If it's audio, the bandwidth will be constrained to audio frequencies, so no filtering will be required.

What antennas ?

Complete nonsense.

Graham

NO.

You've been given very bad, over-the-top and completey erroneous advice.

I'm a pro-audio designer by the way so I damn well ought to know.

Graham

Lets all kneel down and bow our heads for Eeyores eternal wisdom and authocracy.

Now shut down this thread. It starts to smell .

Robert

--- Really???

What do you think the 'T' in

E = sqrt(4kTR df)

is about?

And which do you think will generate more noise: a 100 ohm 1/2 watt resistor dissipating 1/2 a watt or a 100 ohm 1 watt resistor dissipating 1/2 a watt?

-- JF

Modern audio design doesn't run resistors hot in sensitive gain stages.

Graham

--
Precisely because of the reason poogie stated, which you called
\'complete nonsense\', you utter hypocrite.

NO.

It is complete nonsense when you're talking about dissipating at maximum maybe a few milliwatts in resisitors.

Audio designers don't throw away significant fractions of a watt as heat in resistors. Stop pontificating about stuff you're not familiar with.

Graham

I have designed audio mixers before.

I always limit the gain at high frequencies using caps in the op-amp feedback cct.

On inputs its always a good idea to have a low value cap to ground to get rid of radio frequencies.

I always put a low value resistor in any outputs from the mixer in case of accidental shorts.

Try to keep any output wiring away from input wiring.

When you split the 9v to get 4v5 use a tantalum and a polyester cap to short out an AC fluctuations.

Not so that it encroaches on the audio band I hope.

Graham

--
Dumb ass, unless you know a trick, the slope of the filter\'s skirts
will either always allow out-of-band signals to encroach on the
audio band or squish the band to where it\'s no longer flat between
20Hz and 20kHz.

So what stage roll-off would you use so that say 10 stages in series is still < -0.3 dB @ 20 kHz ? A worst-case figure for my designs btw.

Graham