I have discovered something recently, which may be connected with your problems. I use CM6631 USB-I2S converter and it's clocks. I had problems, because I had to switch x45 and x49 clocks or resample everything to 44,1 or 96kHz depending on the clock connected, otherwise I heard "frying" sound. I keep the CM6631 board close to the DAC in order to shorten the wire connecting clock. Since I have shielded the top of the ES9018 chip with a piece of grounded metal, I do not have to switch the clocks or resample anymore. It plays flawlessly in every clock/sample rate configuration. The idea comes from glt's blog and it works!
I have not cut the traces myself, but based on the pin assignments, you will need to cut the three bridges on the right side to separate the right avcc from left avcc.
many people seems like L R seperated power connector, and I also found that the left and right seperation can be more better. We will have a new 9018 version with AVCC left and right power connector
It's better to assemble the current version and enjoy it now. send it to your friend one month later and build the new version to compare the improvement is funanyway, I am always listen the AK4399 with OPA627 in this few days, female vocal is lovely and touch
Hmmm... Cutting the right one you circled "should have" disconnected one of the corner AVCC pins. There are 8 AVCC pins: 4 in the corners and 4 in the sides/middle of the chip.
When you cut the trace, do you still have continuity?
Another "discovery": it turned out that the C-Media Win XP DirectSound driver for CM6631 USB to I2S converter always converts signal bit depth to 16. It simply ignores any application settings. There is no control panel (I could not find one) to change this setting. All my previous measurements were made for 16 bit ESS playback and 24 bit EMU recording. But I have found workaround: I play RMAA test signal wav file using the foobar2000 and the C-Media ASIO driver. ASIO driver has small control panel for bit depth setting (default is 16 bit) which is accesible from foobar. Then I use "recording only" option of RMAA. This way I have full 24 bit measurement. It now shows that the noise level of ESS board is really lower than EMU ADC recording capabilities. Of course my previous results are still valid. Only noise level shown was too high. Power supply bybpass capacitor influence on harmonic distortion can be observed with this new measuring method too. My test setup looks like this:
and my power supply:
th c-media control panel:
and the results: noise level
THD+N
An externally hosted image should be here but it was not working when we last tested it.
and my power supply:
An externally hosted image should be here but it was not working when we last tested it.
th c-media control panel:
An externally hosted image should be here but it was not working when we last tested it.
and the results: noise level
An externally hosted image should be here but it was not working when we last tested it.
THD+N
An externally hosted image should be here but it was not working when we last tested it.
Impressive ! Getting closer to the theoretical limit for 24 bit.
A few years ago, I borrowed an spl meter - my hifi creates a maximum 119dB in my living room in the listening position, which is about 20bit, and when ambient noise is factored in, I don't need much over 16bit. And yet, I can hear differences that maths and measurements say I should not be able to, at volume levels significantly less than maximum.
I can hear things I can't measure, and I can measure things I can't hear. But that's just me.
A few years ago, I borrowed an spl meter - my hifi creates a maximum 119dB in my living room in the listening position, which is about 20bit, and when ambient noise is factored in, I don't need much over 16bit. And yet, I can hear differences that maths and measurements say I should not be able to, at volume levels significantly less than maximum.
I can hear things I can't measure, and I can measure things I can't hear. But that's just me.
Unfortunately, what we can see on my graphs, makes only -111dB total THD+N. It is around 18-19bit performance. 24bit in theory gives 144dB dynamic range. But now we are looking at my EMU noise floor rather than ESS performance. This C-Media driver bug was really a failure.
My power amplifier is capable of 90dB dynamic, so ESS board as a source is more than perfect for it.
RMS figures right ? My soundcard can't even do that - one of the many reasons I gave up RMAA testing. You need to get a 32-bit card these days or it's pointless, and who can hear the full 32 bit range on a domestic hifi ? Most speakers can't do it and it'd be permanent hearing damage anyway.
RightMark Audio Analyzer test : [MME] SE-200PCI LineOut
RightMark Audio Analyzer test : [MME] SE-200PCI LineOut
Last edited:
Hello folks,
If you set register 17 to all mono. And use I2S input on this DIYINHK board. Does this then mean you have a mono balanced board, with +/- output on DAC2,4,6,8, and DACB2,4,6,8. ? And dont need any more modifications then f.ex sharing the master clock between two boards ? Or is there more settings to be made, and more mods to be done ?
If you set register 17 to all mono. And use I2S input on this DIYINHK board. Does this then mean you have a mono balanced board, with +/- output on DAC2,4,6,8, and DACB2,4,6,8. ? And dont need any more modifications then f.ex sharing the master clock between two boards ? Or is there more settings to be made, and more mods to be done ?
- Status
- Not open for further replies.