How to calculate fs bit clock knowing crystal frequency?

Status
Not open for further replies.
Jump to Latest
#2
According to the data sheet for the WM8741, the crystal frequency is fs.

fs/divide ratio (32, 64, 128, 256, etc) gives the data rate.

eg. 24.576ZMHz clock /256(divide ratio) = 192KHz data rate.

I suspect it is dependent on the DAC.

What device are you using?
 
#4
The bitclock (BCLK) frequency can be calculated as follows below.

1. If the left and right channel data are in multiplexed serial format (such as they typically are before entering an oversampling interpolator) then: BLCK = original input sample rate (Fs) * the oversampling ratio * 64 bits per frame.

2. If the left and right channel data are in de-multiplexed parallel format (such as they typically are after exiting an oversampling interpolator), then: BCLK = the original input sample rate (Fs) * the oversampling ratio * 32 bits per frame.

For points in the DAC where oversampling has not been applied, the oversampling ratio in the above formulas = 1.

Note: There also is a rarely utilized de-multiplexed parallel 16 bits per frame I2S data format.
 
Last edited:
#11
#12
Oh sorry,

That was the calculation for the "master clock"

The bit clock is also implementation dependent, and irrelevant of the source clock. Most CD drives will output at 32*fs (16 bit for the left channel and 16 bit for the right channel). Most I2S compatible devices will output 64*fs (32 bit for L and 32 bit for R, padded with zeros if there is no data beyond 16 bit or 24 bit)

Some DACs such as the wm8741 supports both 32*fs and 64*fs. The ESS as you said, only 64*fs

If you look at p11 of the datasheet, the serial output seems either 32*fs or 48*fs, therefore it will not work with the ESS DAC
 
Last edited:
#17
merlin el mago said:
When you said "not work" do you refer "no sound" or do you refer can sound with dropouts?
Click to expand...

Typically the data clocked into a device is done at the rate of the bit clock. The LR clock goes high or low to determine whether or not the data passing to the device is for the left or right channel and during that time the data for that channel is clocked into the device using the bit clock and the data stream. Even if the device is expecting a certain number of bits per LR period it will still recognise what data is meant for which channel. The question is whether or not the device will cope with receiving the incorrect number of bits. My guess is that if it works it wont do so gracefully. If the device was capable of filling in the missing bits with zeros then surely it'd be listed as being capable of working with a lower rate bit clock. You could email the people at ESS and ask them.

merlin el mago said:
If use S/PDIF I will have the same problem: "not work"?
Click to expand...

I thought the ESS chip had an SPDIF receiver built into it? If it does then that will work fine as the receiver will output data that the DAC is capable of working with.
 
#18
glt said:
I don't know. Try it...
Click to expand...

kevinkr said:
Unlike I2S SPDIF is quite standardized and the ESS dac should work fine with a SPDIF input..
Click to expand...

5th element said:
Typically the data clocked into a device is done at the rate of the bit clock. The LR clock goes high or low to determine whether or not the data passing to the device is for the left or right channel and during that time the data for that channel is clocked into the device using the bit clock and the data stream. Even if the device is expecting a certain number of bits per LR period it will still recognise what data is meant for which channel. The question is whether or not the device will cope with receiving the incorrect number of bits. My guess is that if it works it wont do so gracefully. If the device was capable of filling in the missing bits with zeros then surely it'd be listed as being capable of working with a lower rate bit clock. You could email the people at ESS and ask them.



I thought the ESS chip had an SPDIF receiver built into it? If it does then that will work fine as the receiver will output data that the DAC is capable of working with.
Click to expand...

Thanks for support guys,

@glt I will try

@kevinkr Works with dropouts

@5th element If not work the solution will be to buy a TP Metronome to upsample to 96fs
 
#19
I´ll resurrect this thread:
I have some interesting example of a Sony CDP-103 from 1985 where the bit clock is used to drive the control circuitry/logic.
According to the service manual, pulse width of the bitclock is 240µS, the bitclock is derived from a 8.4672 master clock (which is made from 16.9344 using a divider).
Which frequency has the bitclock ?
To find the trace of the bitclock, look at IC 304 in the middle of the schematics, it buffers the bitclock.
All the best,
Salar
 

Attachments

  • Bitclock.jpg
    Bitclock.jpg
    202.1 KB · Views: 116
#20
Salar said:
According to the service manual, pulse width of the bitclock is 240µS, the bitclock is derived from a 8.4672 master clock (which is made from 16.9344 using a divider).
Which frequency has the bitclock ?
To find the trace of the bitclock, look at IC 304 in the middle of the schematics, it buffers the bitclock.
Click to expand...

The 240us. period shown on the schematic cannot be an audio bitclock rate because it simply is too low, equaling a frequency of, 1/240us. = 4,166.7kHz. The units of the figure shown in red may be a typographical error, more likely being 240 nanoseconds. This would result in a bitclock frequency of 4.1667MHz, which is very close to being exactly half the 8.4672MHz master clock frequency you cite. Perhaps, that 240us. figure shown in red on the schematic was measured by eye from an o-scope graticule, and not only should be in nanoseconds, but was actually 236.2ns., which would then exactly equal half the master clock frequency.
 
Last edited:
Status
Not open for further replies.
Top Bottom