I tried Andrea's RTZ dac with HQ Player, but not with all of the extensive combinations of settings.
Also, I believe that is possible to make a better FPGA-based PCM->DSD converter than Simple DSD Converter. Its very good in its present version, but probably not the best possible if a more powerful FPGA were to be used.
What sounds best in terms of DSD conversion does seem to change sometimes with the state of a particular dac design and or the type of dac architecture being used. Its sort of like the DSD conversion algorithms chosen can also be doing something to compensate for the sound of the dac itself.
Also, I believe that is possible to make a better FPGA-based PCM->DSD converter than Simple DSD Converter. Its very good in its present version, but probably not the best possible if a more powerful FPGA were to be used.
What sounds best in terms of DSD conversion does seem to change sometimes with the state of a particular dac design and or the type of dac architecture being used. Its sort of like the DSD conversion algorithms chosen can also be doing something to compensate for the sound of the dac itself.
Mark,
Fully understanding and appreciating your aiming for the best possible sound, but I can’t remember you ever tried Fifo PLUS buffer/reclocker but WITHOUT Se/Diff in NON RTZ mode, and letting Marcel’s board do the clock doubling and RTZ logic, am I correct ?
The way the shift registers are being fed now have too little ressemblance with the original design to make any possible SQ judgement on Marcel’s clever digital design.
Hans
Fully understanding and appreciating your aiming for the best possible sound, but I can’t remember you ever tried Fifo PLUS buffer/reclocker but WITHOUT Se/Diff in NON RTZ mode, and letting Marcel’s board do the clock doubling and RTZ logic, am I correct ?
The way the shift registers are being fed now have too little ressemblance with the original design to make any possible SQ judgement on Marcel’s clever digital design.
Hans
For my own use, PCM2DSD is always followed by isolation and reclocking before going into a dac. FPGAs are known to be jittery and to produce EMI/RFI type noise.
Also, for some listening instead of using the speakers there are a few different HPAs available. Best is NDA. Headphones are usually Audeze LCD-X.
That was tried. It sounded better with the better clocking and better isolation than it did with I2SoverUSB.
Also tried non-RTZ with Marcel's clocking circuitry bypassed, and letting the Andrea's boards drive the output shift registers.
The hardware hookup only became a problem when I set the FIFO board to output RTZ. Then there was a problem with the inverting shift register input signals. However I was not using the inverting outputs most of the time anyway since the dac sounded better without the filter board and in SE mode, and with external passive filtering.
Please note that the dac of this thread should have a proper output filter at some point, just not one that contributes to blurred sound.
Last edited:
Mark,
You did not answer the question I asked, that could have been a simple YES or No.
I didn’t argue the FiFo and the Reclocker with better clocking and Isolation.
I just mention that you seem to have missed a combination.
When feeding Marcel’s board with the superior and buffered clock and non RTZ data, you won’t worsen Vref on the shiftregisters and keep the design as it was meant to be.
When this turns out to be sounding worse, then you have a point, but I would be surprised.
Hans
You did not answer the question I asked, that could have been a simple YES or No.
I didn’t argue the FiFo and the Reclocker with better clocking and Isolation.
I just mention that you seem to have missed a combination.
When feeding Marcel’s board with the superior and buffered clock and non RTZ data, you won’t worsen Vref on the shiftregisters and keep the design as it was meant to be.
When this turns out to be sounding worse, then you have a point, but I would be surprised.
Hans
Acko and Marcel,
I see a couple of things I think it would probably make sense to try:
1. Fix the RTZ for the inverted shift register input signals, which will require a wait to get parts. Then try dummy loads on unused outputs to even out current draw.
2. Consider trying metal foil output resistors. They have less excess noise, and are available in .01% tolerance. Should help both noise and element matching between phases. Downside would be cost of the parts.
Something we know about but I am probably not going to do at this time:
1. Make a better output filter. Can't get inductors I would want to use for passive, and no time nor interest to design and develop a discrete active filter.
I see a couple of things I think it would probably make sense to try:
1. Fix the RTZ for the inverted shift register input signals, which will require a wait to get parts. Then try dummy loads on unused outputs to even out current draw.
2. Consider trying metal foil output resistors. They have less excess noise, and are available in .01% tolerance. Should help both noise and element matching between phases. Downside would be cost of the parts.
Something we know about but I am probably not going to do at this time:
1. Make a better output filter. Can't get inductors I would want to use for passive, and no time nor interest to design and develop a discrete active filter.
Just to reiterate the contents of my earlier posts regarding the comparison of HQPlayer and PCM2DSD - it was a quick workbench exercise! I kept the components similar, as best I could in the context and of course with some time I could make improvements but my intention was just a quick initial comparison - my initial objective was simply to compare the newly reflashed PCM2DSD against a known baseline.
I am certain that an improved PCM2DSD is possible at extra cost and providing someone with the skills is prepared to work up the design and code - that person certainly isn't me.
So it seems I am the only one to have experience of using HQ Player to deliver DSD to the RTZ decoder and perhaps that's a factor in my not hearing the bluriness that has been reported? Anyway, at some point I'll do a quick explore of the various modulators/filters again - my glass is generally half-full so I prefer to think in terms of potential optimisation rather than compensation.
I am certain that an improved PCM2DSD is possible at extra cost and providing someone with the skills is prepared to work up the design and code - that person certainly isn't me.
So it seems I am the only one to have experience of using HQ Player to deliver DSD to the RTZ decoder and perhaps that's a factor in my not hearing the bluriness that has been reported? Anyway, at some point I'll do a quick explore of the various modulators/filters again - my glass is generally half-full so I prefer to think in terms of potential optimisation rather than compensation.
Once again the mysterious "we". Nothing wrong with the original filter. Using inductors or discrete active does not make it better.
Acko, my recommendation would be to get the RTZ back while it's still functioning so you have a chance to listen to it and form your own opinion.
All the changes I made are easy to revert back. Some of them can be done with jumpers such as filter common mode rejection bandwidth. Some would require removing jumpers and soldering over cut traces, such as bypassing of the last output stage opamp as others have already done. Easy for me to remove the film caps on Vref and replace them with X5R. For Marcel's RTZ circuitry, it only requires replacing the series termination resistors that allow drive signals to reach shift register inputs. If Acko wants I can restore it to original and remove all the u.fl connectors. Also can hook up the output stage power to the dac board power if someone wants dirtied up output filter power. Just say what you want.
However if I do all the above, Acko will learn nothing about what changes can do. His choice.
EDIT: BTW, just hooked up Andrea's dac again. Its the one I would choose at this stage of development. A+ (but not at the very top of that league)
However if I do all the above, Acko will learn nothing about what changes can do. His choice.
EDIT: BTW, just hooked up Andrea's dac again. Its the one I would choose at this stage of development. A+ (but not at the very top of that league)
Last edited:
Mark,
Looking back, I can see you did.
And yes the S and Sh problems you got when inserting a simple inverter board violating RTZ, that may or may not have been solved with the Rubicon caps.
The idea was to get a 2*BCLK crystal clock directly on the shiftregisters instead of an XOR generated 2*BCLK.
But DSD128 is now the maximum rate, a far cry from Marcel’s DSD512 having proper RTZ signals
I wonder how much jitter can be produced by his XOR that is driven by your super stable OCXO Fifo’s external crystal clock.
Maybe Marcel can comment on that.
So I have the idea that you didn’t invest enough time in trying to improve the sound using Marcel’s clock doubler and his well thought RTZ cicuitry, before reverting to skip 90%.
Maybe the Rubicons could already have done the job, but at least the digital design would have stayed intact plus at least a higher DSD256 rate.
Hans
Its because I call balls and strikes as I see them. Its not about favoritism. I admire Marcel's expertise. I also admire Andrea's hard work. Both have earned my respect. However, I will tell the truth as I see it.
This DAC does not match my Amp going direct. Output is too low 2V rms. Amp needs 9V rms. This may appear to sound less favourable compared to the other DACs that I have. So one of the reasons why I thought getting feedback from others with compatible systems would be better.
I have already formed an opinion from what you and Mark have experienced plus all the explanations from Marcel. There is so much useful information from all the posts here.
This is a DAC that is worth pursuing further. In fact if you wish to I am happy to cover the build costs for the units that you were planning to do by yourself. If it is too tedious as you explained why not get factory to do it for you. How much will it cost and let me know? I have spent heaps for EES and AKM so why not do the same for something more deserving here.
Please let me know.
Mark, I trust the mods that you have made and the skilful way it was executed. Others seems to be horrified but I hope you are not deterred. If the board fails, not a problem for me as it can be repaired or replaced
How on earth you can extract anything useful of the impression stated here is beyond me - acko - you haven't heard any of the rest of the systems components - you as an individual "commercial" developer need to form you own idea about sound and prove that you continuously master to create superior sounding gear rather than state that you trust "op amp rollers" to make your design decisions - this will not up your brand value... I'd be careful...
//
//
As I have made and listened to both I'm sorry to say that ES9039Q2M is not better sounding.