HB has tested the first version of Katana, not V1.2. He might test the new version
Chord Mojo has independent measuring of THD+N -101.
In my biased opinion , Katana with good PSUs its on a different league.
Agree. In my experience with my music choice, My Chord Hugo was outperfromed by the Katana v1.2 w/ isolator, broken in for a week. I used LPS in the microcontroller.
Please can you elaborate further?
How many LPS?
Output stage separately +-15V?
Thanks
Matt
Last edited:
He told about Hugo, but it was corrected (written) in the vid that he meant Mojo.
Matt
Hi Mark,
Just to ask fo morons like me, is it just film caps soldered in paralell between PSU and +-15 connectors on microcontroller, or is it more complex?
Thank you
It's just like in the picture I attached to a few posts up from this one and in the 'ES9038Q2M board' thread. About 100uf sum total of parallel film caps from +15v to ground, and another 100uf from -15v to ground. The 100uf or so in my case was made up from some 22uf, 33uf, and 10uf film caps soldered in parallel. However, its not too expensive to order 10 each Wima 33uf, 63v film caps from Mouser and solder half of them in parallel with each other and use them on one rail, and do likewise with the other half of the film caps on the other rail. That would give 5 x 33uf = 165uf per rail, but that might be even better, don't know. Only suggest buying 10 because then you get a quantity discount.
Thank you Mark.
I saw you picture but because of poor light, it's not very detailed.
I just wanted to be sure it was like that.
I will try to add it to my katana.
Hi guys,
in his video, HB said that he prefers the "sound quality" output stage to the thd one.
I didn't already received the "sound quality" one, as Allo was short on stock, so I can tell for myself.
And you? Which output stage do you like more and why.
With chinese 9038q2m, I upsample to dsd256 like this:
RPI->Kali reclocker->AK4137->ES9038q2m
So I think you don't need usb.
Am I wrong to do it like that?
I think the weakest part is the AK4137.
According to reports from CA forum a solution with software upsampling to DSD by HQPlayer or Audirvana+ is much superior.
So I would prefer to go this path.
But going this path I see no way outside of USB.
Matt
I know that HQPlayer and Audirvana+ produce great results at upsampling but as I use rpi for now, I guess AK4137 is the only way to upsample PCM or DoP to DSD.
Whether upsampling to DSD helps at all and what the best sounding method is depends on a lot of particulars. So long as people will be using the exact same hardware and software, then results are likely to be repeatable. That includes the USB to I2S board, the AK4137 board and its particular clocks, and the particular dac and its clocks.
For example, one thing upsampling does is to help reduce jitter, but once jitter is reduced there is no guarantee it will stay that way. Every device and wire it passes through before being converted back to analog audio may have some effect of making jitter worse again.
For the best dacs, DSD and or 16/44 will sound as good as good as possible without any conversion making it sound better. Of course, the best dacs may be using whatever tricks inside that happen to work best with them.
Last edited:
Matt,
I think upsampling helps with jitter reduction, but since we not using master mode, in order to get the most benefit from upsampling we have to adjust DPLL bandwidth in the dac registers. Also, I use any old USB to I2S boards and clocks, since I aim to reduce jitter after that. I find it easy for jitter to get worse again after upsampling if care isn't taken in the wiring and connectors between AK4137 and ES9038Q2M. Conversion to DSD seems to help sound quality for other reasons. ES9038Q2M uses different interpolation filtering for DSD as opposed to PCM. I prefer the DSD filter sound. External interpolation filtering for PCM might be another option. Also, the oversampling ratio in the sigma-delta modulator might be more favorable for reducing THD when DSD is used, although don't know for sure.
I also did some experiments with AK4137 using different clocks. I found the clocks that came with the boards sounded better with my dac than supposedly very low jitter clocks I tried. That may go to illustrate that jitter is not just one type of timing variation. It can be random and or deterministic. Could be the lower jitter clocks were also more deterministic. Don't know.
I think upsampling helps with jitter reduction, but since we not using master mode, in order to get the most benefit from upsampling we have to adjust DPLL bandwidth in the dac registers. Also, I use any old USB to I2S boards and clocks, since I aim to reduce jitter after that. I find it easy for jitter to get worse again after upsampling if care isn't taken in the wiring and connectors between AK4137 and ES9038Q2M. Conversion to DSD seems to help sound quality for other reasons. ES9038Q2M uses different interpolation filtering for DSD as opposed to PCM. I prefer the DSD filter sound. External interpolation filtering for PCM might be another option. Also, the oversampling ratio in the sigma-delta modulator might be more favorable for reducing THD when DSD is used, although don't know for sure.
I also did some experiments with AK4137 using different clocks. I found the clocks that came with the boards sounded better with my dac than supposedly very low jitter clocks I tried. That may go to illustrate that jitter is not just one type of timing variation. It can be random and or deterministic. Could be the lower jitter clocks were also more deterministic. Don't know.
Last edited:
