Are you suggesting using a PCM->PWM chip (plus OPS if required) as a DAC as front-end to a classD (analog) amp? I've been having great results with such a set up - one of the advantages could be that with such high voltage drive straight out of the DAC an nCore module would not need any front-end gain stage.
Any closed loop class d have analog feedback (I think infinion might have a pure digital feedback, but not sure). I'm not a developer, but is it absolutely impossible to "hack" into any class d amp, and make a closed loop PCM to PWM system?
I'm running a pure DSD DAC through some Weiss OP2-BP opamps, followed by autoformer volume control, and straight into the NC-500's. That's as pure as I have been able to pull off so far with the Ncore's.
The SA-XR50 and 55 is very very good with digital input. It will not drive heavy loads, but anything above 6 ohm and not too dificult phase angles, the it's quite good full range.
It would be a fundamental redesign - you are better off starting from scratch (and there are good reasons people don't do feedback in the digital domain, so it won't be easy).
I just pick up what I read - and something I have picked up is that a class d amplifier with no buffer etc have a low Zin when you also need low noise. This requires a strong front end - but if you convert to PCM - PWM system then it's quite another situation with 75 or 110 ohm transmission line.
I f you can read danish there is an interesting thread her
Klasse D - HiFi2You
Last edited:
I do believe Hypex has had a "digital" feedback patent for some years, they do not seem to use in production, maybe just blocking advanced digital feedback for others ?
Next step up would be to bypass the discrete opamps altogether by designing the DSD DAC to run natively at 10VRMS output and varying its power supply for some of the volume control steps. TVC coarse steps to make up the rest of the volume range.
I'm working on a software based modulator that will allow a direct DSD input into a class D amp. All audio is upsampled to a fixed DSD rate in software. Streamed via Ravenna to the modules, and converted to analog in the amp. Works with Roon GUI controlling the system.
Well that's the tricky part. But I have some smart guys on the project who know what to do. It will be a 4 channel system, and all DSP for xovers/room correction runs in software using 64 bit floating point DSP. all native DSD is processed using multibit DSD at it's native rate. No downsampling to a standard PCM rate.
Similar concept to the Mola Mola DAC where everything is upsampled to a fixed frequency, then runs through 1 bit DAC's in the end. On the module end I use 1 ultra low phase noise clock direct off the pins with no PLL generation or divider needed. Because I only need to ever clock 1 frequency. There's no place for jitter to even get into the system as nothing is clocked until the endpoint. Fiber inputs ensure nothing external can possibly influence anything inside the modules. IEEE1588v2 ensures 2 nanosecond clock skew between modules, and scale-able to 100's of modules if needed.
Even though Mola Mola is 1 bit in the analog domain, its not 1 bit digitally so doesn't have the technical flaw of DSD.
It's a similar concept anyways. But all DSP done with powerful Intel processors rather than Sharc chips. And multichannel Ravenna over fiber to the endpoints. Then the purest possible path to the amps. I'm sure Bruno could build something similar if he wanted. But might cost a bit too much for these Fusion plates 🙂
Considering how sceptical Bruno has been about digital feedback, I'd be surprised, but I guess some of his concepts do translate into the digital domain.
This whitepaper covers some of the issues faced back in 2002 making a 1 bit DAC powerful enough to drive speakers direct:
https://www.hypex.nl/img/upload/doc/an_wp/WP_AES112BP_A_true_one-bit_power_DA_converter.pdf
https://www.hypex.nl/img/upload/doc/an_wp/WP_AES112BP_A_true_one-bit_power_DA_converter.pdf
Yes it's referred to as "DSD wide"
"A new studio format has been developed addressing the issue of DSD mixing, usually referred to as “DSD-wide”, which retains standard DSD’s high sample rate but uses an 8-bit, rather than single-bit digital word length, yet still relies heavily on the noise shaping principle. It becomes almost the same as PCM (it’s sometimes disparagingly referred to as “PCM-narrow”) but has the added benefit of making DSP operations in the studio a great deal more practical. The main difference is that “DSD-wide” still retains the native DSD sampling frequency while the highest frequency in which PCM is edited, is 384kHz."
It's the highest quality possible way to apply DSP to a DSD track. I'm unaware of any 24/11.289Mhz PCM material available.