|
Moe
- Re: Noisy chip
|
(no rating)
2008-12-05 |
Apologies for the delay. I heard back from the product manager and applications support and this is their answer:
Thank you for your interest in the MAX5486. We would like to help you overcome the noise issue. The low end of the pot where you said noise was worst is the bias supply. The data sheet shows, on page 2, that the Bias output noise is typically 2.2 micro Volts RMS. If we have a 1Volt RMS signal, that is a 113dB SNR. So, we need to look elsewhere for the noise source.
Audio is a difficult discipline. In video 60dB Signal to Noise (SNR) makes a good picture. In audio, this SNR is considered very noisy. If we have a 5 Volt power supply and want 100dB SNR, noise from all sources has to be less than 50 micro volts (5 Volts divided by 100,000). It is not easy to make power supplies that quiet. The decoupling capacitors are critical. One normally thinks of the capacitor transferring noise from the power supply to ground. To do this, the ground must be a low impedance, quiet place. In digital systems, the digital ground may be as noisy as the supplies. This is why we have analog supplies and ground. In fact, connecting a decoupling capacitor from analog power to a digital ground actually adds noise to the analog power. The ground current return must be controlled to control noise. For example, if the current from a noisy brush motor flows under the audio circuits, that noise will hurt the audio signal.
Illustrating how clean the power supplies must be, the Power Supply Rejection Ratio (PSRR) for the Bias Voltage (page 2) is only 60dB. Reference the PSRR plat in the lower left on page 5. The rejection is 55dB at 10kHz. Thus, noise on the supplies is only attenuated by a ratio of 500 to 1000. The rest of the noise suppression must come from clean supplies and grounds.
|
|
|
