[Mahdi8]

You can't go back to M7 since the firmware is different and hardware connection for master clock has been severed. Looking at the nos7 jumper setting here http://www.audio-gd.com/Pro/NOS/NOS7/NOS7EN_Use.htm bypass is not used for anything 

 

[bButcher]

  If it's set to "BYPASS" on NOS7, is the sound still considered Non-Oversampling (NOS)? Or will it revert back to M7?

 
In a traditional NOS implementation, the signal should be "bit perfect" transferred to the DAC chips, no oversampling involved, no alternation of any kind. AGD's implementation is more unconventional, though not oversampled but the input signal will be altered by low pass filter no matter in nos mode 1 or 2. In "BYPASS" mode, the low pass filter will be skipped, so NOS7 will actually work in "real" NOS mode. To see how much actually been filtered you can play a test CD, ex. Bink's Audio test CD @ around 1k Hz.
 
Although "BYPASS" setting will put the whole DAC back to synchronous operation - same as M7, but it won't revert back to M7.
 
 

Also, do you know if there is a difference between using the 7pin vs 5pin when connecting the F1 internally? Aren't both i2s? Thanks!

7 pin connector include all the signals in 5 pin connector plus 3.3V power.

 

[Jhya]

Thank you dButcher for the detailed description. So, with the BYPASS mode (jumper 7 removed) NOS7 and M7 will be in "real" NOS mode? Will both sound exactly the same or is there some difference?

So both pin location will be the same? That's great news! As I just want to upgrade the internal Amanero USB, but still keep the HDMI i2s port functioning.

I'm trying to install the Singxer F1 internally. If I just remove the DC/DC module will that bring up to at least the level of SU1? I'm having a hard time desoldering that little bugger with the 4 pin. I pumped all the excess solder and used a braid to suck up the rest, but the pins are still stuck to the side! Do you have any pointers to help remove that? Did you solder a motherboard standoff nut to the F1 to help you mount to the case?

Also, I will be reusing the ribbon cable from the Amanero USB. Can I power the output portion of the F1 if I connect the 3.3v pin to one of the 20 hole in the F1?

Sorry for so many questions, thank you!

 

[bButcher]

So, with the BYPASS mode (jumper 7 removed) NOS7 and M7 will be in "real" NOS mode? Will both sound exactly the same or is there some difference?

My is a NOS11, never has a M11 for comparison so can't say for sure. My guess is yes, both will work and sound similarly in BYPASS mode with one difference: M7 can only handle input up to 192k and NOS7 with new improved implementation can handle up to 384k. That could translate to less gates delay so a better jitter performance. But with the new M7 firmware due to be released next month which like NOS7 can handle 384k input, that should make the two sound identical in BYPASS mode.

So both pin location will be the same?

Pin order is different. It's printed on the circuit board.
 
The pin out for the 7 pin connector of NOS11 is:
(top to bottom)
LRCK (word clock)
DA (data)
BCK (bit clock)
MCK (master clock)
NC (not connected)
GND
V3.3
 
For the 5 pin connector:
DATA (data)
BCK (bit clock)
WCK (word clock)
MCK (master clock)
GND
 

If I just remove the DC/DC module will that bring up to at least the level of SU1?

Should be.
 

I'm having a hard time desoldering that little bugger with the 4 pin. I pumped all the excess solder and used a braid to suck up the rest, but the pins are still stuck to the side! Do you have any pointers to help remove that?

The key point is to keep the 4 pin heated almost simultaneously. For a small SMT device a big glob of solder tin enough to cover all the pin will do, but this is a big 4 pin dip device. You can try this: apply some fresh solder tin to all four pins, use one hand to move the solder tip quick among the four pins to keep the solder tine flow, with the other hand slowly pull and wiggle the device out.
 

Did you solder a motherboard standoff nut to the F1 to help you mount to the case?

I use two cable ties to fix F1: with one tie's tail pass through the screw hole left by Amanero board and wrap around the USB socket then go back outside through the other screw hole, use another cable tie's head to fix the tail then cut the extra tail from both cable ties.
 

Can I power the output portion of the F1 if I connect the 3.3v pin to one of the 20 hole in the F1?

Absolutely no. 3.3V on the 20 pin connector holes is for output, not input. But you can connect 3.3V from 7 pin connector to F-1's +5V input after B0505S removed. ADP150 will work fine as long as input power > 3.35V, the LDO for cpld should work for input > 3.4V. This can be a makeshift arrangement before you find a better +5V input for F-1.

 

[gk2013]

So, with the BYPASS mode (jumper 7 removed) NOS7 and M7 will be in "real" NOS mode? Will both sound exactly the same or is there some difference?

My is a NOS11, never has a M11 for comparison so can't say for sure. My guess is yes, both will work and sound similarly in BYPASS mode with one difference: M7 can only handle input up to 192k and NOS7 with new improved implementation can handle up to 384k. That could translate to less gates delay so a better jitter performance. But with the new M7 firmware due to be released next month which like NOS7 can handle 384k input, that should make the two sound identical in BYPASS mode
 
Bypass NOS DSP product will disable of the 98Mhz TCXO, hence NOS DSP has no asynchronous clock to deal with.
Bypass OS DSP for me is unclear what it does exactly. (Does it disable involvement of the master clock input?)
New NOS DSP supports up to 768kHz (first release already)
New OS DSP supports up to 384kHz. (old release up to 192kHz)
Different DSP software - different specs - different sound is what I expect but I did not test it, can't confirm.
 
Be careful testing the Bypass on NOS or OS - it is not something Audio-GD publishes on their website

 

[Mahdi8]

@bballas where did you get the oxco clock from? I'm thinking to upgrade my nos clock have you consider crystek femto clock?

 

[bButcher]

In I2S spec. there are three mandatory signals:
 
WCLK (word clock, or left right clock (LRCLK)) - 44.1k, 48k, 88.2k, etc.
BCLK (bit clock) - usually WCLK x 2 x 32 in most implementation
DATA (data) - include left and right channel data
 
MCLK (master clock) is optional, it is usually used to drive digital filter or to up-sampling the input signal.
 
What PCM1704 need is WCLK, BCLK, and DATA (only one channel). ["http://www.qlshifi.com/jszl/pcm1704.pdf"]
 
By setting NOS dsp in BYPASS mode, all dsp functions will be disabled (low pass filter, asynchronous clock, fifo) --- verified by Kingwa through email. WCLK and BCLK will pass through, DATA will be extracted and distributed to the corresponding PCM1704s. So the dac will operate in synchronous mode.
 
Although it is specified 768k in AGD's website. The actual measured PCM1704 input BCLK is (44.1k - 2.82MHz), (96k - 6.1MHz), (192k - 12.29MHz), (352k - 22.57HHz). So that means for 768k input the BCLK will be 49.152Mhz which PCM1704 can only support up to maximum 25Mhz. So I have strong reservation for the 768k spec.

 

[gk2013]

  In I2S spec. there are three mandatory signals:
 
WCLK (word clock, or left right clock (LRCLK)) - 44.1k, 48k, 88.2k, etc.
BCLK (bit clock) - usually WCLK x 2 x 32 in most implementation
DATA (data) - include left and right channel data
 
MCLK (master clock) is optional, it is usually used to drive digital filter or to up-sampling the input signal.
 
What PCM1704 need is WCLK, BCLK, and DATA (only one channel). ["http://www.qlshifi.com/jszl/pcm1704.pdf"]
 
By setting NOS dsp in BYPASS mode, all dsp functions will be disabled (low pass filter, asynchronous clock, fifo) --- verified by Kingwa through email. WCLK and BCLK will pass through, DATA will be extracted and distributed to the corresponding PCM1704s. So the dac will operate in synchronous mode.
 
Although it is specified 768k in AGD's website. The actual measured PCM1704 input BCLK is (44.1k - 2.82MHz), (96k - 6.1MHz), (192k - 12.29MHz), (352k - 22.57HHz). So that means for 768k input the BCLK will be 49.152Mhz which PCM1704 can only support up to maximum 25Mhz. So I have strong reservation for the 768k spec.

 
Correct and well explained. 
One remark: The Bypass option is not something "supported" by Audio-GD. The 768 input is - I presume - measured in asynchronous mode and is an input specification, hence the DSP might output something different to the PCM1704. But who cares about 768kHz anyway....

 

[bButcher]

  One remark: The Bypass option is not something "supported" by Audio-GD.

Yes, you're right.
 
So next time I mention BYPASS option, I should add "Try at your own risk!!". Since Kingwa never warn me of any adverse effect, figure at least it won't burn your DAC. --- I've used this setting for couple of  months and also have the 98.3Mhz TCXO removed from my DAC.

 

[Jhya]

Thank you for the excellent explanations! I just finished installing the F-1 internally. The 3.3v pin to the 5v input works just as you described.I noticed that the LDO chips on my F-1 reads a little differently: the LDO that powers the 575 clocks reads "LEJ#0Z" and the other 2 LDOs that power the cpld and XMOS reads "L337". Is that the same chips that came with your F-1? This is the finished photo:
 

 
Also, Kingwa just released NOS version 2 firmware! Very excited to try it! Especially the 3rd NOS mode. This is from their website:
"NOS 7 , NOS 11 and NOS 19 had update the Ver 2 DSP software since Oct. 12th.2016
The Ver 2 DSP software had improve on the timing performance, hence can support high res music better and working better on high oversampling input, users can hear more better detail in sound quality . And the new software had addition the 3rd Nos mode for users choice for their favorite taste."

 

[Mahdi8]

Just send KingWa an email for the new firmware this morning. Has anyone got the firmware?

 

[Mahdi8]

Just got the firmware from KingWa. Weekend just get more interesting

 

[Kingwa]

In my test the PCM1704's bck support least 98.304MHz, had not test higher yet.

 

[Mahdi8]

here is the new output file http://www.filedropper.com/outputfilenos for anyone that needs them

 

[bButcher]

  In my test the PCM1704's bck support least 98.304MHz, had not test higher yet.

Thanks for this information.