Revisiting Musiland Monitor I2S Output
One can easily tap into the I2S lines of the Musiland Monitor devices. In my previous attempt, I soldered the wires with the DAC chip still in place. I tested it on the OPUS DAC (WM8741) and it sounded very good. The OPUS DAC, like most DACs, requires all 4 lines (Master Clock, Bit Clock, LR Clock, Data) plus GND.
In my new attempt I wanted to streamline the wiring and thus:
- Removed the DAC chip
- Removed the bypass caps for the DAC chip
- Hot glued the wires to the board
I am not tapping the MCK (Master clock) line because I plan to interface to the Buffalo II DAC. The Sabre32 DAC does not need MCK because it generates its own clock (because it has a built in ASRC). The 4 wires are: bit clock, LR clock, data and GND.
Often it is important to characterize the I2S lines, for example to:
- To determine which line is which if you don’t have the datasheet
- To determine if a source is compatible with a specific DAC
- To confirm the sample rate of the incoming signal
My $49 oscilloscope has an analog bandwidth of about 1 MHz. It can show the traces of the LRCK signals, but not much else. When the signals exceed the bandwidth of the scope, we can use the frequency counter feature which is good to about 8 MHz.
Measuring the LRCK line
LRCK signal when playing music at 44.1KHz sample rate and playing music at 192KHz sample rate:
If we measure the period we get the following results:
- 44.1KHz trace: T=22.5 usec; f=1/22.5=44.4KHz
- 192KHz trace: T=5.2 usec; f=1/5.2=192.3KHz
The maximum resolution of the scope is .5 usec. The following trace is for 44.1KHz. In reality, the rise time may be faster than this; the cheap scope just gives you a general idea on these waveforms.
Measuring LRCK of the I2S line allows you to confirm the sample rate of the source. Notice also that the level is about 3.2v which means TTL level.
Measuring the DATA line
Data signal when playing music at 44.1 KHz sample rate. The signal is not a regular signal and the scope is not fast enough. Notice though that the level is ~ 3v, which confirms that there is “good” data in this line.
Measuring the BCK (Bit clock) line
Because the bit clock runs at a large multiple of the sample rate, typically 32X, 48X or 64X, the scope does not have the sufficient bandwidth to look at the bit clock trace. However we can use the frequency measurement feature of the scope.
The two photos below show the bit clock frequency for a 44.1KHz and a 88.2KHz signal
We find that
- 44100 x 64=2822400
- 88200 x 64=5644800
It is then determined that the bit clock is running at 64 times the sample rate. This is GOOD because the Sabre32 DAC requires 64fs bit clock.
Measuring the MCK (Master clock, also known as System Clock) line
The master clock runs at 2X (or more) faster than the bit clock. We can use the frequency measuring feature of the scope
The above measurement corresponds to playing music at 44.1 KHz sample rate. We notice that
The Master clock is running at 128x the sample rate. It is important to determine the master clock fs because DACs are designed to work within a range of master clock fs. Operating outside of this region may impact the performance of the DAC.
Bellow I’ve summarized the master clock requirements for two high-end dacs
From the above table we find that the Musiland MINI, when playing 44.1KHz material is not supported by the Wolfson DAC but supported by the Burr Brown DAC. In theory, the Wolfson DAC works with the Musiland MINI though I2s but perhaps not its optimal operation region. When I tested it with the OPUS DAC I found that the high oversampling filter would not work (audibly though I could not hear any differences).
CONNECTING TO BUFFALO II DAC
I connected the Musiland to the Buffalo II DAC with the current version of the s/w (most of the registers are at default configuration). Result: no sound. Next: review the registers and change the s/w for I2S