FredA
Headphoneus Supremus
- Joined
- Dec 15, 2013
- Posts
- 5,605
- Likes
- 6,952
Maybe Kingwa will read this. How is the isolator adding jitter? Augmented rise time? Noise added?There is a new R-28, NOS-only version called "R28 NOS". A price is $100 lower than R-28 2022 model. It would be a natural upgrade path from my R2R-11, it is why I took a closer look at this model, but read on:
TL;DR: Limitations are more severe than just removing oversampling feature. FPGA which was responsible for many functions is gone, now clock design resembles R2R-11mK2. Rather get a 2022 model while stock lasts!
Now it goes on:
Part A: Missing many functions serviced by FPGA:
1) Oversampling. Now NOS-only, no digital processing is available. Previously there was a selectable option 2x/4x/8x. Not a problem for me and all users are allerted accordingly. It is why is 100 bucks cheaper.
2) Clock management, synchronisation and jitter reduction with built-in PLL. In some cases it is not needed, by example with asynchronous USB transfers. However Kingwa has decided to add galvanic isolation on all ports, including the USB and I2S port. Read part B about consequences. In this case lack of clock management will degrade performance on USB and I2S port to the level off a DAC in $500 range. It is serious, not kidding...
3) S/PDIF receiver. Now a receiver use dedicated chips, perhaps the same parts as in R2R-11 Mk2. This is a chip with a good reputation, but in a standalone configuration do not have support for 176kHz sampling rate. As a consequence, there is also no playback of DSD-64 (or SACD) on the coax/optical ports. SACD playback will be still possible if a player has HDMI-I2S or USB port.
Im my opinion such limitation is unusual in the DAC of this price range. This particular chip can be reprogrammed on the fly which bring support for 176kHz, it requires adding a few $ worth control unit for autodetection of the incoming sampling rate. There is a small microprocessor inside a display panel, Kingwa could use. It hasn't been done, unfortunately. Lousy, opportunistic implementation.
Part B:
The new model adds galvanic isolation on all ports (not an optional feature like in R2R-11mk2), including USB port, 2022 model do not have isolator. However 2022 model use a pair of Accusilicon ultra-low jitter oscilators. A clock for USB transfers is delivered from these oscilators (Amanero module has no oscilators), giving support for asynchronous transfers based on a high precision internal clock, so there is no need for jitter reduction processing (PLL in the FPGA). I expected that R28 will be improved with a clock feed-back isolator like it is done in higher models (R-8 and up). There is zero added jitter in this solution. The 2022 model model made a step up in this direction, I expected it would be continued, is not!
It is actually step back, replicating design from the lower priced R2R-11mk2. Isolator is adding jitter, there is an urgent need to fix it with PLL, but there is no FPGA to do it anymore!
Look at the photos, there is no Accusilicon clocks on the main board anymore. Instead there are clocks on the Amanero board. Judging from a flat shape, Crystek CCHD-571 (low jitter) type, but it doesn't help much, as clock performance is lost on the isolator.
This is all from me, I always supported Audio GD. I understand a reason of dropping FPGA. I restrained myself from criticising galvanic option on the entry model R2R-11mk2, but in the R1/R28 range it is a major step back. With little more attention to details could be done better at no extra cost. In other words, not acceptable.
There is a CPLD however as a replacement. Not sure the spdif input go through isolation. And we do not see what clock is used for the spdif receiver. I would assume that using a ddc, the r28 nos will really be awesome. And still best the r2r 11 mkii though usb.
Last edited: