Another new guy question / clarification. Is my description here correct, I appreciate any additions, clarifications, or corrections. I’m trying to think clearly about how to best use HQPlayer with my various DACs.
In general it seems like the right strategy with HQPlayer is to find a way to bypass most (if not all) of the internal DSP in your DAC. This looks different for different DAC architectures.
For Sigma-Delta based DACs this is done by using HQPlayer to convert to SDM and setting the DAC to “DSD-bypass” (or the equivalent for various DAC chips). This avoids conversion to PCM, filtering, volume adjustment, etc within the DAC, and just passes the SDM stream straight to the output stage. I do this with my Topping E30 (AKM chip). But on my Allo Revolution and Soncoz LA-QXD1, both of which are ESS DAC chips, I don’t see a way to fully bypass the DSP. And to my ear both of these sound better with PCM data than SDM.
For R-2R based DACs the strategy is reversed. Use HQPlayer to up-sample to the maximum PCM rate the DAC can accept, and set the DAC at maximum volume output. If possible turn off any filtering, but if there’s still some digital filter happening it should be so far away from the audible range it won’t matter. I do this with my Soekris dac1321 and it sounds great. My understanding is if I feed this DAC DSD it will just convert it straight to PCM, so really very little benefit from DSD.
Is this description accurate? How have others configured HQPlayer and DAC settings to bypass the internal DSP as much as possible?
I think it’s hard to generalise. With my RME ADI-2 I know that running it to ‘DSD Direct’ mode will give best objective performance. I know this because Jussi himself has one (two!) and has shared measurements, including vs high rate PCM.
I had a Pro-Ject Pre Box S2 (ESS chip) and this measured best when fed DSD512. I know this because Jussi himself has one and showed the same as above.
It can differ with different products. Everything after the DAC chip is important (analogue stage).
The safest best is to get what Jussi has recommended (those are shared on his website).
I don’t really expect him to know how every ESS and AKM and R2R based DAC on the planet does for different inputs, without him actually having one in hand to test.
Note I’m only covering objective performance here. Subjective can vary wildly so I prefer to leave that to each individual what they like best.
On ESS it depends, but so far DSD inputs seem to work well, because the DSP section becomes simpler and allows higher digital filter oversampling factors than when feeding PCM.
Note that there are some dual-DACs on the market with a separate discrete DSD D/A converter. Like T+A and Holo Audio.
I wonder if the iFi Neo iDSD with a Burr Brown dac would be a good fit.
They claim:
“Not only is the NEO iDSD a purist design – no DSP, no digital filter selection and no change in analogue playback – it is also all designed around a balanced circuit design named ‘PureWave”
Seems like a match for HQPlayer to show off its filters? Any comments?
With PCM inputs the TI/BB DAC chip always has it’s delta-sigma modulator etc. in play. But if you feed it DSD, then it is able to perform bit-perfect D/A conversion (the chip has four conversion structure / analog filter options).
Yes, the TI/BB chip has four analog filters (conversion stage configurations) you can choose from. But it doesn’t do any DSP for DSD sources. Note that all these configurations have a different output level, so comparing sonics of those configurations without compensating for the levels accurately will be problematic.
I don’t know if iFi exposes selection for these filters. Earlier on iDSD Micro they had three choices with a switch. I wish they would expose it.
But in any case, you certainly get best performance sticking to DSD inputs with BB/TI DAC chips.
Not much in 0-100 kHz region, but the DAC structure and roll-off slopes are different. So it would be great to at least know which one they have selected. It mostly affects how much left-over ultrasonic noise there is in the output and relative output level.
The specifications in the “tech lowdown” on the iFi website say this:
“DAC: Bit-Perfect DSD & DXD DAC by Burr Brown”
I can’t tell if they mean the DAC plays bit perfect at DSD and DXD sample rates, or if they’re only describing the capabilities of the DAC. Based on what you said, it would be the latter. Is that correct?
Sidebar: you mentioned iFi’s GTO filter in the post I quoted above. Which HQPlayer filter/SDM combination would you say most closely resembles that GTO filter, please? I have the xDSD Gryphon DAC which has the GTO filter as a selectable option. It’s not nearly as “good” as some of the filters in HQP, but it’s not unpleasant, either.
One of the shortest minimum-phase filters, like minringFIR-mp for example is most similar.
Note that GTO works only up to 352.8/384k rate, while with external upsampling you can do double that for PCM outputs and of course much higher rates with DSD, up to DSD512.
@jussi_laako somewhere I came across that you recommended 5EC instead of 7EC for DACs using ESS chips. Is this correct? My Ayre QX-5 has an ES9038PRO DAC chip.
Yes, that’s the case. You can use 7EC as well, especially at DSD512. But my recommendation is 5EC for most implementations. This is because ESS has a gentle DSD filter and most DACs implement at most 2nd order analog low-pass.
If one knows that their DAC has 4th order or higher analog low-pass, then 7EC becomes recommended choice.
Do you know if this diagram refers to iFi “BP - Bit Perfect” filter ?
In xDSD Gryphone “Tech Notes” I found the following slide, I assume this may be answers my question… “no chip DSP” with BP setting. Encouranging for passionate HQP user…
