What are different ways to provide external clock signal to an audio ADC

[matrixofdynamism]

I know that I can use an oscillator or perhaps a PLL. A PLL appears to be a more complex component and I am not sure how I shall use it right now.

Anyway, this question is related to Texas Instrument's PCM1808 Audio ADC. It needs a clock input. The table in its data shows the the relationship between sampling rate and the clock input. It says:



Now I want to be able to do an experiment with ADC working at different sampling rates to see effect on sound quality. For this reason I guess I shall need to change the clock input to the ADC as well. What method is best to change the clock input into the ADC with ease once I have a PCB with the ADC on it?

 

[barry]

A PLL might not be your best solution because of the jitter it will add to the sampling process. If you are just doing an experiment I'd suggest you use a signal generator for your clock source.

 

[matrixofdynamism]

barry, I agree with you about the signal generator. If we want to have a PCB with a system where we can choose the sampling rate for the audio, how do you think that can be done?

Actually, another confusion I have is that if the ADC is a slave device (this one can be a master as well, strange), the the microprocessor/microcontroller will continuously signal the ADC to provide data. In this way, the sampling rate should be controllable via the interrupt timer inside the microcontroller. However, here, the sampling rate of audio is dependent on the external clock input. I wonder why.

How will jitter effect the system performance? You mean the audio quality will be bad?

 

[ads-ee]

barry, I agree with you about the signal generator. If we want to have a PCB with a system where we can choose the sampling rate for the audio, how do you think that can be done?

Actually, another confusion I have is that if the ADC is a slave device (this one can be a master as well, strange), the the microprocessor/microcontroller will continuously signal the ADC to provide data. In this way, the sampling rate should be controllable via the interrupt timer inside the microcontroller. However, here, the sampling rate of audio is dependent on the external clock input. I wonder why.

How will jitter effect the system performance? You mean the audio quality will be bad?

I don't think you understand ADC intefacing that well.

Most ADCs you'll find have a control interface (usually SPI), which is a slave interface. The D in ADC the digitized analog output of very high speed ADCs may actually generate a clock output (derived form the sample clock, possibly from a built in PLL), data, and valid signal that are source synchronous. Other slower ADCs typically expect an external device to drive the sampling/output clock and capture the data from the output.

In either case the sampling clock should be clean, with minimal jitter if you want good system performance.

 

[KlausST]

Hi,

Just some additional information.

An audio ADC needs a countinous clock.
If the ADC is a slave device, then the clock must come from the master device. Both (master and skave) frequencies usually are 128x, 256x or 384x the sample frequency.
Both master and slave must run from the same clock. Otherwise you need a sample rate converter to meet both datastreams

Klaus

 

[matrixofdynamism]

My assumption was that the Microcontroller can signal the ADC to start conversion and once the ADC ready with data, it shall send a signal back to the microcontroller. This signal can be processed via an interrupt which shall read the ADC data output. In this way, the microcontroller shall control the sample rate of the ADC. However, now to me it seems that this is infact not the case.

There are basically two type of ADCs, one are flash converters and the other use some sort of mechanism to approximate the a digital value for the analogue signal. This is done by internally generating a digital value, change it to analogue and then use a comparator to see if it matches the actual external analogue input. If it does, the digital value is considered accurate and is output to the master device. Am I correct? In this way, the latter type of ADCs would actually need a clock input.

... right , so at the moment I feel .. flabbergasted

 

[KlausST]

Hii,

What you need is a SAR converter. Most audio ADCs are delta sigma, they need a continous clock..

I just looked into the datasheet. Definitely your ADC needs a fixed, stable, continous clock to operate.
It is not possible to start a "single conversion".

Sorry.

Klaus

 

[D.A.(Tony)Stewart]

If you just want to hear the difference get free audacity and load an MP3 then save as MP3 with different settings and play on different players that may have different codecs. Eg car, PC.

I prefer variable bit rate VBR that speeds up for impulse sounds and high f content but better compression to SNR to BW to File size ratios.
You can display spectrum of nyquist filtered audio in Audacity.

- - -