How can I keep interference to a minimum on a board for four-way audio?

[SethGI]

Hi,

I'm designing a board to input four audio signals, amplify, filter, and then run through an adc. I'm wondering if having four signals in one four-way opamp or one adc with four inputs would cause substantial interference. I'm using the board to process data from an acoustic pinger, and I'm comparing the wave forms of the signals to figure out the TDOA (time difference of arrival) of the signals. Basic summary of this is that I cross-correlate the signals to see when the are most similar, which tells me how far apart the signals coming from the same source are. Sorry for that shitty explanation, but it doesn't really matter for the purpose of the question. The main point is I need as little interference as humanly possible. So, would I be better off using four separate opamps and adcs, or is this overkill and i should just use four-ways for both?

Thanks!

[bobaruni]

If you need "little interference as humanly possible" then you should use separate single op amps and shielding traces between each channel on the PCB as a minimum.
You might get away with a single ADC and mux.
Have a look at the cross talk specs for the op amp you intend to use and see if that's good enough for your purpose (keeping in mind that these figures will be ideal based on a very well designed PCB).
Cross talk can also be optimised by choosing an ideal impedance and also depends on the frequency of interest.

[SethGI]

Great, thanks. That was my assumption, 4 opamps with 1 adc. That also keeps com with the main computer WAY simpler to only have one I2S interface instead of doing something else way more complicated.

[Cerebus]

Hi,

I'm designing a board to input four audio signals, amplify, filter, and then run through an adc. I'm wondering if having four signals in one four-way opamp or one adc with four inputs would cause substantial interference. I'm using the board to process data from an acoustic pinger, and I'm comparing the wave forms of the signals to figure out the TDOA (time difference of arrival) of the signals. Basic summary of this is that I cross-correlate the signals to see when the are most similar, which tells me how far apart the signals coming from the same source are. Sorry for that shitty explanation, but it doesn't really matter for the purpose of the question. The main point is I need as little interference as humanly possible. So, would I be better off using four separate opamps and adcs, or is this overkill and i should just use four-ways for both?

Thanks!

The datasheet for multiple op amps in one package ought to contain some characterisation of the cross talk between the amplifiers. To pick a random example, for the LT1013D the figure is given as "channel separation" with a minimum of 120 dB and a 'typical' figure of 137 dB (for the 'A' grade it's 123 dB and 140 dB respectively). Those figures are fairly typical for a well specified package. I would expect to find similar characterisation for ADCs. I suspect that figures like that would be down in the noise in any practical audio signal - my trusty old AKG C451 microphone is specified as having a signal to noise ratio of 76 db A weighted. If the C451 is unfamiliar, it's a professional grade capacitor microphone that back in it's heyday would set you back around £500 GBP, so not cheap tat. Real world transducer signals are quite noisy compared to the level of cross talk we're talking about here.

[SethGI]

Hi,

I'm designing a board to input four audio signals, amplify, filter, and then run through an adc. I'm wondering if having four signals in one four-way opamp or one adc with four inputs would cause substantial interference. I'm using the board to process data from an acoustic pinger, and I'm comparing the wave forms of the signals to figure out the TDOA (time difference of arrival) of the signals. Basic summary of this is that I cross-correlate the signals to see when the are most similar, which tells me how far apart the signals coming from the same source are. Sorry for that shitty explanation, but it doesn't really matter for the purpose of the question. The main point is I need as little interference as humanly possible. So, would I be better off using four separate opamps and adcs, or is this overkill and i should just use four-ways for both?

Thanks!

The datasheet for multiple op amps in one package ought to contain some characterisation of the cross talk between the amplifiers. To pick a random example, for the LT1013D the figure is given as "channel separation" with a minimum of 120 dB and a 'typical' figure of 137 dB (for the 'A' grade it's 123 dB and 140 dB respectively). Those figures are fairly typical for a well specified package. I would expect to find similar characterisation for ADCs. I suspect that figures like that would be down in the noise in any practical audio signal - my trusty old AKG C451 microphone is specified as having a signal to noise ratio of 76 db A weighted. If the C451 is unfamiliar, it's a professional grade capacitor microphone that back in it's heyday would set you back around £500 GBP, so not cheap tat. Real world transducer signals are quite noisy compared to the level of cross talk we're talking about here.

Great point... thanks. I'll look into those cross talk numbers in a bit more depth. The only reason I'm even considering separating everything is that I will have four inputs with almost exactly identical waveforms. Aside from any distortion/noise which comes from each mic, they will be the exact same input, only offset by about a milisecond (or less). I need to be able to compare these four signals to figure out the exact time delay. You could very well be right that the amount of interference I'm talking about is negligible. I just want to make sure! Thanks for the help!