Chord Electronics Hugo M Scaler upsampling digital processor

June 30, 2023

The idea of using digital signal processing (DSP) to convert digital audio data sampled at 44.1kHz or 48kHz to a higher sample rate is not new. I first heard the beneficial effects of upsampling at Stereophile‘s 1998 hi-fi show in Los Angeles, where a pro-audio dCS 972 digital-to-digital processor was being used to convert 16-bit/44.1kHz CD data to a 24/192 datastream (footnote 1). So persuaded was I of the sonic improvement offered by upsampling that I bought a dCS 972 to upsample CDs to 24/88.2 to feed my then-reference Mark Levinson No.30.6 Reference DAC. As D/A processors improved, my need for upsampling faded away. (While my dCS 972 still sees some use, this is for the opposite purpose: downsampling my hi-rez recordings to 16/44.1 to produce CD masters.)

I subsequently wrote that I was convinced that the sonic improvement I heard with the dCS 972 was due to its using a different oversampling digital reconstruction filter with a different number of taps and, as a consequence, different passband ripple and stopband rejection (footnote 2). Twenty years after those words appeared in print, the Chord Electronics $4795 Hugo M Scaler arrived in my listening room, and I found myself returning to the subject of digital filters and upsampling—or upscaling, as the British company calls it.

220chordsc.bac

Description
The Hugo M Scaler is a small, almost square component powered from a supplied 15V wall wart and controlled with either a small remote or the front-panel buttons. It features five digital inputs: a galvanically isolated Type-B USB, two coaxial S/PDIF on BNC, and two TosLink optical. (DSD data is converted to PCM with a 6dB reduction in level.) The M Scaler doesn’t have analog outputs, but it has three digital outputs: one coaxial BNC S/PDIF, one optical, and a pair of galvanically isolated BNC jacks that enable upsampling to 705.6kHz or 768kHz—but only when used with compatible Chord Electronics DACs. The M Scaler will work with D/A processors from other manufacturers, but only, of course, up to the maximum frequency the DAC can accept. (My PS Audio DirectStream DAC indicated it was receiving data sampled at 384kHz when I set the M Scaler to output that rate via an S/PDIF connection, but the DirectStream is limited to 192kHs and there was no sound.)

220chordsc.2

The front panel offers six of Chord’s traditional glass-sphere buttons, which illuminate in different colors according to what the M Scaler has been asked to do. Only four of these are currently functional; the rightmost pair, marked “DX,” are intended for a future product design. From left to right, these indicate: whether video mode or automatic input detection is selected; which input is in use; the output sample rate; and the input sample rate.

The M Scaler offers a video mode, as the upsampling introduces a latency up to 600ms, too long for synchronizing audio with video. In video mode, the M Scaler uses an interpolation filter with lower latency. For audio use, a pass-through option, with the output sample rate the same as the input rate, is provided to allow comparison to upsampled output. However, as the upsampling can introduce digital “overs”—interpolated sample values exceeding 0dBFS—both the passthrough and upsampled signals are reduced in gain by about 2.8dB, which complicates such direct comparisons.

The core of the M Scaler is a Xilinx XC7A200T FPGA (field-programmable gate array) on which runs the code for the Watts Transient Alignment reconstruction filter—named for design consultant Rob Watts—first seen in Chord’s Blu Mk.II upscaling CD transport. The FPGA has 740 DSP cores; Watts’s filter uses 528 of them running in parallel at 16Fs and a bit depth of 56 to achieve a filter length of an extraordinary 1,015,808 taps. For comparison, the WTA filter in Chord’s DAVE D/A processor, which I reviewed in June 2017, used 164,000 taps implemented in 166 DSP cores.

In preparing the DAVE review, I asked Watts what is the advantage of using ever-longer digital filters. “If you have a conventional filter with 100 taps, you will recover some of the transient information,” he explained. “A 100-tap filter gives you sufficiently good frequency-domain performance, but not in the time domain. . . . Every time you increase the number of taps, you improve the perception of pitch, timbre gets better—bright instruments sound brighter, dark instruments sound darker—the starting and stopping of notes becomes easier to hear, the localization of sounds get better. There is less listening fatigue—the brain has to do less processing of the information presented to it to understand what’s going on.”

Digital filters and upsampling
In the promotional literature for the M Scaler, Chord writes, “The Hugo M Scaler . . . takes the digital file and repairs it, adding back the information lost between the samples, then it sends the repaired file to the DAC. . . . With 705,600 samples per second, a huge amount of important information that was lost when creating the 44.1 digital file is now recovered. The more samples, the closer you get to the original analog signal. . . . The Hugo M Scaler in essence places 15 additional new musical samples in between each original musical sample, resulting in an astounding improvement in the recreation of the original music signal.”

My eyebrows raised, I kept reading. Referring to the figure reprinted here, the text states that “The Hugo M Scaler takes a rough stairstep CD quality waveform and transforms it into a smooth analog-like waveform. That quantum leap in sampling brings a breathtaking leap in detail, accuracy and realism to your music.”

220Chord-Theoryfig1

Hmm. The measurements I performed to accompany our reviews of the dCS 972 and Purcell definitively showed that upsampling doesn’t add information above the Nyquist frequency—22.05kHz with CD data—of the lower sample rate. So what is the M Scaler doing?

In one of the first articles I wrote for Stereophile, “Zen & The Art of D/A Conversion,” which was published in September 1986, I discussed how the recovered analog signal is not directly described by the levels of the digital samples. Instead, the interaction between those samples and the impulse response of a digital low-pass reconstruction filter recreates the analog waveform—not just at the sampling intervals but between them (footnote 3). By processing the incoming data with a low-pass filter featuring an extremely long impulse response, the M Scaler makes it possible for the accompanying DAC to more accurately reconstruct the analog signal. In effect, it replaces the DAC’s digital filter with its own, as the DAC’s filter is now operating at the higher sample rate, and its cutoff is one or more octaves above the original data’s Nyquist frequency.

Listening with the Chord Electronics DAVE
I performed most of my auditioning of the M Scaler with a sample of Chord’s DAVE D/A processor, sourcing audio data from my Roon Nucleus+ server via USB and sending the upsampled data to the DAVE with a dual-BNC connection. I also used the M Scaler with my PS Audio DirectStream and Mark Levinson No.30.6 DACs, using a single S/PDIF connection. I tried using the M Scaler with my NAD M10 integrated amplifier, sending the latter data upsampled to 96kHz or 192kHz via an optical S/PDIF link. However, while the NAD would play music for a few seconds, it stuttered and then stopped. I was using the M10’s Dirac room correction and wondered if that was the problem, so switched off the Dirac filter, but there was no improvement.

Footnote 1: Jonathan Scull reviewed the dCS 972 in February 1999. dCS subsequently introduced a functionally identical but cosmetically improved consumer version, the Purcell.

Footnote 2: In a letter responding to this essay, Bob Katz conjectured that because the inevitable quantization distortion is spread over a wider frequency range with upsampled data, the audibility of this distortion is significantly reduced.

Footnote 3: For a detailed explanation of how a digital filter operates, see an article I wrote in September 2018.

NEXT: Page 2 »