I am with edorr let’s see. Judging by Paul’s grinning attitude I think we will be surprised edorr. So when is a universal player coming ?
Al
Volume attenuation will not be made to the DSD data stream. You cannot perform math on a 1-bit stream; there is no room, you need multi-bits to perform any math.
The White Paper does appear to imply that volume is attenuated in DSD, but I think what it intends to indicate is that volume changes are made without any adverse impact on DSD data.
Volume changes may be done as part of the conversion. just prior to converting to DSD.
Elk said: Volume changes may be done as part of the conversion. just prior to converting to DSD.
Yes, my first thought that it's done in the PCM domain after upsampling. Another possibility is the LPF with a digitally switched resistor ladder (passive preamp with a low pass filter), but I doubt it...
In DSD direct (native) there is no volume control, but in DSD plus (DoP) there is, at least for the Wolfson DAC’s. There might be a difference between multibit and 1-bit DAC chips for all I know.
Presumably the volume control will work with DSD files, which don’t go through the PCM to DSD conversion.
frode said: but in DSD plus (DoP) there is, at least for the Wolfson DAC's.
Isn't it done in the PCM domain as well? When I checked the Wolfson chips I noticed that they have internal DSD->PCM conversion.
Yes, just checked again. http://www.wolfsonmicro.com/products/dacs The DSD stream is converted to PCM first.
OK, it’s in the WhitePaper that Paul sent. The attenuation is done before converting to 2xDSD. However, the DSD stream is in the same path. Then:
"10X DSD rate. Regardless of input format, whether PCM or DSD, all data are
upsampled to 30 bits running at 10 times the standard DSD rate and then back
down again to double rate DSD for noise shaping."
It’s not 10xDSD. It’s 10x DSD rate or 28.224MHz, 30bits @ 28.224MHz. I wonder if a true DSD stream is converted to PCM, then gets upsampled to 30/28224, then downsampled to 5.6448MHz, then gets attenuated, and only after the volume control the PCM stream is converted to 2xDSD.
Good reading.
I cannot tell from the documents PSA released how the volume control in the DirectStream product works. I just know it cannot be done on the 1-bit DSD datastream itself. 
If I recall correctly, Sabre has a chip that scales DSD by representing each data point with a 32-bit number. It sounds like a DSD/PCM hybrid in a sense.
It will also be fun to learn how PSA’s incorporation of FPGAs is the same/different from Chord’s.
Oh, just noticed another interesting thing:
"Transformer coupled output. The output of the DSD engine is fed directly into
the output stage which is based on high speed video amplifiers, passive filtering,
including a carefully crafted high bandwidth audio transformer at the output."
No coupling capacitor needed - good! OTOH, transformers are not linear.
Also, one of the “popular mods” is replacing the output stage with a transformer.
Howdy all
I thought I’d answer the technical question I saw in one big post. If I’m not clear or you have other questions please ask:
Volume control:
All inputs (whether PCM, DSD or DoP) are upsampled to 30 bits @ 10 x the normal DSD rate (28.224MHz). Then a 20 bit volume is applied (this is all the bits that are needed to represent the volume from 100 == full scale to 1 which is -49.5dB). All significant bits from the output of the volume multiply are used in the conversion to single bit double rate DSD. There isn’t any truncation, rounding or dither used or needed.
Any references to Wolfson (or any other) DAC chips are irrelevant, we don’t use them. This is a fully discrete implementation.
“…DSD direct (native) there is no volume control, but in DSD plus (DoP)…” There aren’t such modes in the DirectStream. The DirectStream accepts DSD directly over the I2S inputs and DoP (DSD over PCM) on all inputs (well if your TOSLink runs up to 176.4k). DoP is a lossless packaging of the raw DSD bits into a 24 bit 176.4kHz stream. Foobar2000, JRiver, etc. can play DSD via DoP.
Internal shielding:
By using balanced differential pairs for all of the signals downstream of the FPGA (including over the ribbon cable) there is little radiated interference from the power supply or the bridge. By ear (or the Audio Precision) there isn’t any haze from EMI.
"Will the PWT be firmware upgraded to handle DSD (SACD) or the drive, etc. replaced to do so?"
I can’t speak to PS Audio’s other products but part of a solution for the PWT is to use foobar2000 (or similar) to convert any DSD to DoP (which is lossless). That will produce a 24 bit 176.4kHz file which can be written to a CD or DVD. Then when the PWT sends the file to the DirectStream, the DirectStream will recognize it as DoP and treat it appropriately.
There are multiple DACs (and many other products) which use FPGAs out there. There is little similarity between the DirectStream and, say DcS, certainly the architecture, algorithms, upsampling and conversion to DSD are entirely different. Outside of a few Xilinx library/macros the code is all original. Perhaps the closest might be the Playback Designs/EMM Labs, in that they also convert to double rate DSD. Tho Andreas and I have talked I doubt that there is much similarity otherwise.
The DirectStream digital board is indeed different than the PWD digital board. There is little in common except the connectors. Too a first approximation, all inputs go to the FPGA and the FPGA does its work and passes the DSD signal to the analog card. The most complicated thing outside of the FPGA on the digital card is the XMOS USB chip which gets its non USB inputs and clocks from the FPGA and hands its data back to the FPGA over I2S.
USB, I2S, S/PDIF, TOSLink, AES/EBU are all gapless and in fact there’s no gap, click or pop when you change from one input to another, change sample rates or go from PCM to DSD or back.
Indeed a transformer is use for audio output: its main function is to isolate common mode noise. Tho I too was initially skeptical about using a transformer, it’s benefits far outweigh it’s downsides (most of which are ameliorated by using a much bigger transformer than some might use.)
-Ted
1 Like
Our Hero! =D>
I didn’t get to meet you earlier this month when I visited but I heard your work. Spectacular!
Congratulations on your achievement.
Oh yeah… Welcome to the forum, Ted.
Thank you, Ted!
Great information and very exciting to hear it directly from you.
I can only imagine how proud you are. =D>
@elk
Thanks. I started this just to show that DSD sounded good. Boy does it
Ted would you expect the USB input to be superior to the Bridge Ethernet connection now?
or is it the same?
@magicknow
"Ted would you expect the USB input to be superior to the Bridge Ethernet connection now?"
There are a few answers:
The raw bits from any input sound the same, my goal was to nuke input jitter and I’ve succeeded better than any DAC I know. The bridge, USB, I2S, TOSLink will all sound the same. (At least if TOSLink has the bandwidth you need).
USB and the bridge have two totally different user experiences. It’s your pick.
At one level lower, all non TOSLink inputs including USB and the bridge have the potential to create ground loops in your system. Tho we do our very best (including using audio output transformers and solid grounding) there’s always the potential (No Pun Intended) for there to be some system specific sound differences whenever a ground loop exists.
Thanks very much Ted. This was obviously a labor of love and we are all beneficiaries. Can’t wait. I guess there will be some redundant buttons on the PWD remote (sample rate, filters, etc.).