TOPPING D90SE Updates to D90 III Discrete

[OhmsClaw]

That's just marketing talk from akm side and nothing more.
All these modern chips are multibit desta sigma designs.

False. There's DSD to SDM or PCM conversion going on in even suboptimal AKM/ESS implementations.

AKM actually supports... direct DSD allowing modulation and multibit bypass. This may measure worse when fed unadulterated DSD. If one is running HQP filters then this is the preferred DS chip family. Some OEM's may disable this feature for whatever reason while another with the same chip allows bypass.

Non bypass AKM flagship does 7 bit multibit DSD processing with filtering and lower noise due to the modulator being offboard.

ESS has been explained to run 64 parallel 1 bit modulators allowing for differential digital volume control. It's a "multibit" approach to native 1 bit sampling.

It's not marketing when their approaches are core to their technology and in house reconstruction filters. Whether anyone cares or it sounds better/worse is up to the consumer.

 

[lantian]

False. There's DSD to SDM or PCM conversion going on in even suboptimal AKM/ESS implementations.

AKM actually supports... direct DSD allowing modulation and multibit bypass. This may measure worse when fed unadulterated DSD. If one is running HQP filters then this is the preferred DS chip family. Some OEM's may disable this feature for whatever reason while another with the same chip allows bypass.

Non bypass AKM flagship does 7 bit multibit DSD processing with filtering and lower noise due to the modulator being offboard.

ESS has been explained to run 64 parallel 1 bit modulators allowing for differential digital volume control. It's a "multibit" approach to native 1 bit sampling.

It's not marketing when their approaches are core to their technology and in house reconstruction filters. Whether anyone cares or it sounds better/worse is up to the consumer.

So what your saying is they have a separate ds modulator in ther just for dsd with no additional processing and digital filters, like bur brown has?
Delta sigma modulation can not be bypassed for a dsd signal.

Ps. dsd is delta sigma, so what conversion are you implying there?

 

[lantian]

False. There's DSD to SDM or PCM conversion going on in even suboptimal AKM/ESS implementations.

AKM actually supports... direct DSD allowing modulation and multibit bypass. This may measure worse when fed unadulterated DSD. If one is running HQP filters then this is the preferred DS chip family. Some OEM's may disable this feature for whatever reason while another with the same chip allows bypass.

Non bypass AKM flagship does 7 bit multibit DSD processing with filtering and lower noise due to the modulator being offboard.

ESS has been explained to run 64 parallel 1 bit modulators allowing for differential digital volume control. It's a "multibit" approach to native 1 bit sampling.

It's not marketing when their approaches are core to their technology and in house reconstruction filters. Whether anyone cares or it sounds better/worse is up to the consumer.

Can you specify which akm chip it is that does this? Would love to read the datasheet.

 

[Edward Ng]

Can you specify which akm chip it is that does this? Would love to read the datasheet.

It is almost certainly the two-chip AK4499EX+AK4191EQ DAC-set, which allows this bypass by separating the signal conversion from the signal processing into separate modules/stages.

-Ed

 

[sendler2112]

Can you specify which akm chip it is that does this? Would love to read the datasheet.

The new AK4499EX (EX stands for EXternal modulator chip) uses a separate AK4491 chip as the modulator and they allow bypassing this to send SDM signals directly to the converter chip(s). The Gustard A26 allows SDM direct inputs from HQ Player for example and uses two AK4499EX for dual mono. Non Over Sampling R2R DACs like the Holo Audio May/ Spring/ Cyan for example can also receive direct SDM.

https://www.akm.com/us/en/products/audio/audio-dac/ak4499exeq/

Single bit resistor ladders, whether in a chip like the A26 or discrete, like the Holo Audio May etc., are easier to trim and sort to the exact value since the resistance value is all the same for each individual component. Whereas a PCM resistor ladder has all different values. This allows SDM convertors to measure with less distortion.

One caveat is that the processor demand is very high for an external computer to do a high quality conversion of Pulse Code Modulation (cd and streams) to Sigma Delta Modulation with HQ Player etc. This requires a gaming computer with a powerful nVidia graphics card doing the work in an offload to even make it to SDM 256 with a decent sounding advanced filter and modulator (from the dozens available in HQP) let alone SDM at 48x512.

After playing with external pc SDM conversion via HQ Player at the highest settings that my laptop can perform, I have still decided that I am a PCM person with my Gustard R26. I don't own any DSD recordings and probably never will and compared to my good R2R NOS PCM DAC receiving 23/ 768 from HQP, I find the SDM into the Cyan2 to sound robust and fleshed out, but subdued. The high rate PCM is wider, deeper, which makes it sound better separated and blacker, and more exciting and dynamic for Rock.

I sold my A26 and will be selling my Cyan2 to stick with the R26 for now until the new Gustard R30 comes out next year or I end up with a used May.

 

[Edward Ng]

The new AK4499EX (EX stands for EXternal modulator chip) uses a separate AK4491 chip as the modulator and they allow bypassing this to send SDM signals directly to the converter chip(s). The Gustard A26 allows SDM direct inputs from HQ Player for example and uses two AK4499EX for dual mono. Non Over Sampling R2R DACs like the Holo Audio May/ Spring/ Cyan for example can also receive direct SDM.

https://www.akm.com/us/en/products/audio/audio-dac/ak4499exeq/

Single bit resistor ladders, whether in a chip like the A26 or discrete, like the Holo Audio May etc., are easier to trim and sort to the exact value since the resistance value is all the same for each individual component. Whereas a PCM resistor ladder has all different values. This allows SDM convertors to measure with less distortion.

One caveat is that the processor demand is very high for an external computer to do a high quality conversion of Pulse Code Modulation (cd and streams) to Sigma Delta Modulation with HQ Player etc. This requires a gaming computer with a powerful nVidia graphics card doing the work in an offload to even make it to SDM 256 with a decent sounding advanced filter and modulator (from the dozens available in HQP) let alone SDM at 48x512.

After playing with external pc SDM conversion via HQ Player at the highest settings that my laptop can perform, I have still decided that I am a PCM person with my Gustard R26. I don't own any DSD recordings and probably never will and compared to my good R2R NOS PCM DAC receiving 23/ 768 from HQP, I find the SDM into the Cyan2 to sound robust and fleshed out, but subdued. The high rate PCM is wider, deeper, which makes it sound better separated and blacker, and more exciting and dynamic for Rock.

I sold my A26 and will be selling my Cyan2 to stick with the R26 for now until the new Gustard R30 comes out next year or I end up with a used May.

The pure DAC mode (when Preamp mode is disabled) in the E70 Velvet is considered among the most affordable ways to take advantage of this implementation of the AKM 4499/4191 DACset for DSD material.

-Ed

 

[lantian]

After having a look a the datasheet found what you all are talking about, thanks. Though not sure how it can work without ds modulation.

 

[Lucus]

Only the (old) 4499 allow pure/native DSD. The new combo 4499EX/4191 does allow native DSD but only until 256....

 

[lantian]

If anyone has more technical information on how the new akm chips work please do share, the datasheet is just as limited as ess datasheets.

 

[OhmsClaw]

So what your saying is they have a separate ds modulator in ther just for dsd with no additional processing and digital filters, like bur brown has?
Delta sigma modulation can not be bypassed for a dsd signal.

Ps. dsd is delta sigma, so what conversion are you implying there?

You're conflating the colloquial pcm modulation and SDM conversions with PDM i.e. 1 bit DSD vis a vis delta sigma on purpose.

Below responses cover the rest.

PS: all the DS companies have different filters as well which may or may not be perceived by humans (outside of obvious slow/psuedo-NOS playing redbook vs 192 content or fast filters; the early roll off is obvious while heavy bass attack/decay is more tricky to provoke) but are certainly measurable and food for thought when choosing one a academically.

 

[OhmsClaw]

The new AK4499EX (EX stands for EXternal modulator chip) uses a separate AK4491 chip as the modulator and they allow bypassing this to send SDM signals directly to the converter chip(s). The Gustard A26 allows SDM direct inputs from HQ Player for example and uses two AK4499EX for dual mono. Non Over Sampling R2R DACs like the Holo Audio May/ Spring/ Cyan for example can also receive direct SDM.

https://www.akm.com/us/en/products/audio/audio-dac/ak4499exeq/

Single bit resistor ladders, whether in a chip like the A26 or discrete, like the Holo Audio May etc., are easier to trim and sort to the exact value since the resistance value is all the same for each individual component. Whereas a PCM resistor ladder has all different values. This allows SDM convertors to measure with less distortion.

One caveat is that the processor demand is very high for an external computer to do a high quality conversion of Pulse Code Modulation (cd and streams) to Sigma Delta Modulation with HQ Player etc. This requires a gaming computer with a powerful nVidia graphics card doing the work in an offload to even make it to SDM 256 with a decent sounding advanced filter and modulator (from the dozens available in HQP) let alone SDM at 48x512.

After playing with external pc SDM conversion via HQ Player at the highest settings that my laptop can perform, I have still decided that I am a PCM person with my Gustard R26. I don't own any DSD recordings and probably never will and compared to my good R2R NOS PCM DAC receiving 23/ 768 from HQP, I find the SDM into the Cyan2 to sound robust and fleshed out, but subdued. The high rate PCM is wider, deeper, which makes it sound better separated and blacker, and more exciting and dynamic for Rock.

I sold my A26 and will be selling my Cyan2 to stick with the R26 for now until the new Gustard R30 comes out next year or I end up with a used May.

I will never understand why people obsessed with "detail/as the artist intended/high quality" are the same to pimp out a red book 44.1khz into HQP for 1024 DSD

Native DSD recordings are a whole nother story. Very well recorded, utmost care in mastering, expensive and data hogging. No need to go above 256. Very small library of material or niche audiophile test track bait.

Only the (old) 4499 allow pure/native DSD. The new combo 4499EX/4191 does allow native DSD but only until 256....

still does it anyone doing 512->1024 is running an R2R anyways and upsampling/converting the living daylights out of their pcm library.

If anyone has more technical information on how the new akm chips work please do share, the datasheet is just as limited as ess datasheets.

Data sheets will never explain proprietary processes. It's not the purpose anyhow. Many DAC manufacturers who work with the chips, DIY engineers and occasionally the company themselves will lift the hood on the theory of operations. The data sheet only provides hints with the pin-out.

 

[felix37]

About the peq : can the d90III discrete give the user the posibility to choose from various presets loaded from the tune application (working offline)
When using the eversolo dmp-a8 peq function the machine do downrates the incoming signal to 48k so degrading the audio , so my question is if the d90III discrete also deteriorates/degrades the SQ when it is active?

 

[skasundr]

About the peq : can the d90III discrete give the user the posibility to choose from various presets loaded from the tune application (working offline)
When using the eversolo dmp-a8 peq function the machine do downrates the incoming signal to 48k so degrading the audio , so my question is if the d90III discrete also deteriorates/degrades the SQ when it is active?

No. There were no presets in the Topping Tune app itself, but you can upload ones you got from someone or found online to the app. If your DAC is connected via USB to your PC that you are using as your audio source and also has Topping Tune, the app can run in the background and you can switch between your different PEQ profiles and it is instantly changed in the DAC. Or, you can upload some profiles to your DAC if it is not going to remain connected to the PC with Topping Tune. The only caveat is that you have to do the hard off/on with front button held to get to the DAC menu to change PEQ profile. It also just says profile 1, 2, etc. so you have to remember which PEQ profile was which when you uploaded from Topping Tune. Adjusting/creating a PEQ profile in Topping Tune is pretty standard just like most PEQ apps, so no complaints there. I don’t think it degrades sound quality when apply PEQ, but it only supports up to 192 khz and only on USB and maybe the SPDIFs (haven’t tried), but it does NOT apply PEQ to I2S, which sucks.

 

[skasundr]

There is something about this DAC that I can’t quite put my finger on, but it sounds different (in a positive way) than anything I’ve heard before, and it is definitely not like any other Topping DAC I have heard to date. I have it connected to Woo Audio WA8 via RCA out to 3.5 line in, and the sound is just sublime.

 

[sendler2112]

Still no comparisons with Cyan2, R26, D90III, SU-10, Denefrips? No Youtube reviews?