Expert review and test of the Questyle Reference System Golden Edition by Australian Hi-Fi Magazine. Free download.

Questyle is a newcomer to high-end audio, having been established only five years ago in Shenzhen, China, to build products based around a current mode circuit its founder Jason Wang (Wang Fengshuo) started developing while he was an electrical engineering student at university, and completed after graduation. The first product to use this concept was the CMA800 headphone amplifier, the forerunner of the CMA800Rs used in this Questyle Reference Golden Edition System. Why two headphone amplifiers? According to Wang, his thinking process was: ‘top power amplifiers are often designed as monoblocs to improve performance, why not do the same for a headphone amplifier?’ Why not indeed?

The Equipment

So, in addition to featuring two Questyle CMA800R headphone amplifiers (each of which has a retail price of $4,299), enabling ‘dual mono’ operation, with the signal from one headphone amplifier going to the left earpiece of a pair of headphones and the signal from the other headphone amplifier going to the right earpiece, the Questyle Reference Golden Edition System also has a ‘Current Mode’ preamplifier (CMA800P, retailing for $5,199) and a DAC (CAS192D, which retails for $4,299).

As you have no doubt already realised, Questyle’s model numbers are descriptive, so that the DAC is capable of 192kHz operation and also of processing DSD signals. The ‘CMA’ of course stands for ‘Current Mode Amplification’ and the P is so obvious as not to need explanation at all. However, the ‘800R’ in the headphone amplifier stems from the fact that Sennheiser’s HD800s were Wang’s favourite headphones, and he designed the CMA800R specifically for them, so he ‘borrowed’ the ‘800’ descriptor, as a ‘homage’ to the HD800. (That said, the CMA800R can drive dynamic headphones from any manufacturer… it’s not just restricted to driving HD800s!)

Despite the emphasis on the words ‘Current Mode Amplification’ in all Questyle’s literature and model numbers, Questyle’s products are not ‘current source’ (a.k.a. transconductance) amplifiers—they are conventional voltage source amplifiers, but internally use a circuit concept similar to that used by Krell in its ‘CAST’ (Current Audio Signal Transmission) amplifiers, where after a voltage has been applied to the input terminals, it’s then amplified via current mode, then converted back to voltage source before being delivered to the output terminals. Or, to use Questyle’s own terminology exactly: ‘voltage input and output, while the core amplification takes place in the current domain operating in a pure Class-A state, and allowing the output stage to be either Class-A or Class-AB.’ According to Wang, the result is ‘extremely wide bandwidth, ultra-low distortion and a total lack of transient intermodulation distortion.’

Also, despite the fact that Questyle makes much of the fact it’s using balanced amplifiers to drive the headphone outputs, and claims this is superior to using unbalanced (single-ended) amplifiers to drive headphones, all headphones are electrically ‘balanced’ irrespective of whether they’re being driven by a balanced amplifier or an unbalanced amplifier. Looking at it at a truly basic level, for any true balanced circuit to operate, three wires are required: positive, negative and ground. Headphone drivers have only two wires and since the current into and out of any two-wire network is always equal and opposite, this means headphones cannot ‘tell’ whether the audio signal comes from a single-ended or a balanced voltage source.

Notable on the ‘Golden Edition’ series is Questyle’s use of Rogers 4350 ceramic printed circuit boards (PCBs), which are only 0.8mm thick, yet have a dielectric constant of 3.48 compared to just 2.55 for, say, Teflon PCBs. Ceramic PCBs are used in high-insulation, high-frequency, high-temperature applications and are particularly suited for low-volume electronic products. However it’s important to note that these ceramic PCBs are used only in the Reference ‘Golden Edition’ Series components, not the ‘Silver Edition’ components. Another point of difference is that whereas the PCBs of the ‘Silver Edition’ components are populated by high-quality electrical components, Questyle uses even-higher-quality hand-selected components in its Golden Edition components.

Questyle’s CAS192D DAC is unusual because whereas most high-end DACs that offer DSD use a digital-to-analogue conversion technique known as DoP (DSD over PCM), the CAS192D processes the DSD bit-stream directly—it even bypasses the CAS192D’s own onboard (and switchable) digital filters to provide true DSD. (Note that computer playback via Windows—everything from XP up—requires installation of driver software that Questyle provides on an optical disc provided with the CAS192D.)

There is certainly no shortage of digital filters selectable on the CAS192D, because Questyle has activated all the optional filters that are already resident inside the Cirrus Logic Wolfson Microelectronics WM8741 DAC that it’s using as its digital ‘engine’. (The WM8741 natively also provides the options of direct DSD, or DSD via PCM.) The filters available for use vary depending on the conversion mode you choose (that is, whether you choose to use oversampling or no oversampling) and the bit-rate of the digital input signal, but altogether you can select between FIR Brickwall, FIR Apodising, FIR Soft-knee, FIR Half-band, IIR Apodising, IIR Soft-knee and IIR Half-band.

A full explanation of the differences between these filters is beyond the scope of this review, but there is a fundamental difference between an FIR (Finite Impulse Response) and IIR (Infinite Impulse Response) which is that an FIR filter has time-reversed ringing, whereas an IIR filter doesn’t (a result of having to have a higher Q than an IIR filter to meet the same specification regarding stop band, ripple, and roll-off.) This is best illustrated visually, as per the three shots from an oscilloscope. The first image shows an analogue pulse, the second is what that pulse looks like after passing through one of the Questyle’s FIR filters, and the third after it’s been passed through one of the Questyle’s IIR filters. Although the IIR filter has more ringing after the pulse, there is no pre-ringing. (And, just in case you were wondering, the reasons for the names are that a pulse processed by an IIR filter response never falls to zero, but continues indefinitely, whereas one processed via an FIR filter eventually does.) Questyle correctly points out in its manual that: ‘Vibrating objects in nature produce a soundwave which has similar features to the IIR digital filter. There is no pre-ring before the vibration and the vibration amplitude decreases gradually after the vibration’ … but fails to add that in nature, eventually to zero!

Questyle’s oversampling circuitry is very clever. To avoid repeating decimals and associated rounding errors, as well as jitter, it uses two clocks, rather than just a single clock, so that 44.1kHz and 48kHz signals are only upsampled four times (from 44.1kHz to 176.4kHz  and from 48kHz to 192kHz), while 96kHz signals will be upsampled only twice… assuming you choose to upsample at all). For the sake of completeness, I should point out that the Questyle has a third clock, used for the USB input. The power supply uses a custom Plitron toroidal transformer, Schottky rectifier, Nichicon 2200μF (×22) FG capacitors, and 22 power regulators.

The Questyle CAS192D DAC has three digital inputs: optical (Toslink), SPDIF (RCA) and USB (Type B). The omission of an AES/EBU input is surprising, particularly since there’s plenty of room on the rear panel, but not without precedent on a DAC intended for home use. There are two sets of analogue outputs: balanced via XLR or unbalanced via RCA.

But whereas the DAC has ample inputs and outputs, the Questyle CMA800P preamplifier is almost bereft of them, with only two inputs (one balanced, the other unbalanced) and only two outputs, balanced (via XLR) and unbalanced (via RCA).

The CMA800R is also shy on inputs, but this time there are three: balanced and unbalanced stereo inputs, plus a single ‘full balanced mono’ input (used when using the CMA800R as a monobloc) and two unbalanced outputs (via RCA). On the front panel are two unbalanced 6.35mm ‘phone jacks, so you can drive two pairs of headphones, plus a single balanced output for driving one side of a pair of headphones wired in dual-mono mode.

If you find yourself confused by the wiring instructions in the CMA800R manual, it’s because Questyle has used the wrong rear panel diagram. Instead of being a diagram of the CMA800R, it’s one of the rear panel of the CAS192D. Questyle has also made an error with the test report that accompanies each CMA800R. Whereas the company says it tests ‘over 30 factors of specifications before being shipped out’, with a test report being ‘put into carton box’, the only thing in the carton is a generic, mass-produced, card that has two generic pre-printed graphs (THD vs. Voltage and a Frequency Response) plus two QC stamps. That is, the test report enclosed in the box is not that of the specific model you’ve purchased.

Although the components look ‘full-size’ in the illustrations accompanying this review, each one is only 330mm wide, 300mm deep and 55mm high. This sizing is because Questyle regards these as ‘desktop’ components, and the system as a ‘Desktop’ Reference System. Each component is available in either silver or gold finish, but as noted earlier, only components with the gold finish have the Rogers ceramic PCBs and the upgraded, hand-selected internal components. Due to the quality of these components, and the price of each one, I was surprised at the brevity of the warranty period.

In Use and Listening Sessions

When I first unpacked the Questyle Reference System Golden Edition system, it appeared to me that the CMA800P was ‘superfluous to requirements’, because since there is only one ‘source’ component, it would make more sense to connect the CAS192D directly to the two CMA800R headphone amplifiers, thus improving the performance (eliminating superfluous circuitry, no matter how good, will always reduce noise and distortion) and simultaneously reducing the price of the combo and lowering the height of the stack. This made even more sense to me because whereas in other systems removing a pre-amplifier would also prevent you switching between multiple components, using processor loops, and maybe even allowing control over bass, treble and balance, none of these niceties are available on the CMA800R, so you’re essentially missing out on nothing.

Alas, when I tried eliminating the CMA800P, I quickly discovered why it’s an essential part of Questyle’s Reference System. That reason is simple: It’s the only component that allows you to control volume level when you’re using the CMA800Rs’ ‘Full Balance Outputs’ (which is the configuration designer Jason Wang recommends). If you don’t use it, your headphones will be running ‘full bore’ all the time. This is because although the CMA800R has a volume control, it only controls the volume of the unbalanced outputs, not that of the ‘Full Balance Output’. And, of course, there is no volume control at all on the CAS192D!

My statement that the CAS192D does not have a volume control may seem confusing for anyone who’s looked at the advertisements for it, many of which show a small plastic remote control with two volume buttons. Sadly, these buttons do not work at all (i.e., they’re inoperative) and this despite the fact that one of the features built into the WM8741 inside the CAS192D is a digital volume control. The mute button on the remote does work though. While it’s not exactly dowdy, the build quality of the remote is no-where near that of the main components.

I first tried to determine which of the many filter options I preferred, which was complicated slightly by the fact that if I used 44.1kHz or 48kHz source material, I had three IIR and two FIR options from which to choose, whereas at higher bit-rates, I had three FIR and only two IIR options. (And when playing back DSD there are no filter options at all.) It didn’t take too long at all to find that I preferred the IIR apodising filter, except that when doing the comparisons, I’d only been using the same small section of classical music, played over and over. Later, when I accidentally selected a different filter while playing some rock music, I found I preferred the FIR apodising filter for rock. Review time constraints meant I could not establish if I preferred different filters for other musical genres, but I really wouldn’t be surprised if I did. When switching between filters, the volume fades-out then fades-in, which I found annoying: I would have preferred an instant switch, though I guess Questyle is doing the right thing by making sure that no switching transients can ever get near your loudspeakers.

When working out my filter options I was using my trusty Sennheiser HD800S headphones, all set up for dual-mono, dual-balanced operation (via a four-pin to dual XLR adaptor lead)… at which point I should mention that one huge advantage of using Questyle’s approach is that you get twice the voltage swing and thus four times the power compared to using just a single unbalanced amplifier. And boy did the Questyles make the HD800Ss sing! As, I think, every reviewer has noted of the HD800Ss, their exquisite clarity and the way they render the finest details of the music they reproduce is beyond exceptional. But with the Questyles driving the HD800Ss, their already state-of-the-art sound quality in these aspects of performance was ratcheted up a large notch—taking them from beyond exceptional and all the way out to fantastic. Listening to the intro to Never Give All the Heart (Brenda Fricker and Anuna, backed by The Chieftains) was like a holy experience… an auditory version of the visual catharsis one experiences when first looking up into the naves of Antoni Gaudi’s Sagrada Família. The crisp delineation of the vocal lines… the incredible feeling of harmonic unison from the merging of the voices… there are few words that can describe the experience.

Also taken to yet another level was the bass delivery, which accelerated from being just ‘fast’ to ‘lightning fast’, such that for the very first time I completely appreciated the complexities of the drumming on Dream Theatre’s Dance of Eternity. And not just the complexities of the drumming … I also finally totally appreciated the rhythmic interplay between the musicians, not simply on a sonic level, but also on an intellectual level. The combo of the Questyles and the HD800Ss just peeled back the layers of sound, revealing the whiteness of the bones below.

My only previous criticism of the Sennheiser HD800S’s might have been their midrange, which I’d have described as being translucent, rather than totally transparent, but it appears that I was obviously not using the correct associated electronics to drive them when I made that assessment, because when they were driven by the Questyles, the midrange of the HD800Ss went beyond transparent to being totally invisible. When listening to female vocalists, the closeness and realism—indeed the humanity—of their voices was so uncanny it was almost embarrassing… as if I were intruding on their personal space just by listening. I’ve heard nothing like it.

In order to insert a modicum of fiscal reality into this review, I re-wired the HD800Ss back to a single 6.5mm phone jack, and reconnected the CAS192D directly to a single CMA800R switched for normal stereo operation, which rendered the volume control operational. A retrograde step? Well, yes… but whereas I was expecting there to be a huge difference, I was more than a tad surprised to hear the HD800S’s were still operating on a whole higher plane than I expected. Perhaps not quite at the level I’d been experiencing with the dual CMA800Rs, but close… very, very close. So close that if I had to pull $4,299 out of my pocket in order to return to the sonic nirvana I’d been enjoying only a few moments earlier, I think I would hesitate. But for how long? That I don’t know.


Throughout the time I was evaluating this system I could not help but wonder why Questyle left out so many facilities that many audiophiles might expect, such as a volume control and a headphone output on the DAC, additional inputs on the pre-amplifier, and a certainly a remote control that could actually be used to adjust volume level! At least these omissions means that there’s a clear upgrade path should you wish to work your way up to owning a complete Questyle Reference System Golden Edition System by buying one component at a time. But whether you buy this system incrementally, or take the plunge and do it all at once, you’ll end up the proud owner of what just has to be the ne plus ultra of desktop headphone audio. # Ernest Denman

Glorious sound
Switchable filters
Selectable outputs

Volume control
Warranty period





For more information, contact Audio Dynamics


Readers interested in a full technical appraisal of the performance of the Questyle Reference System Golden Edition System should continue on and read the LABORATORY REPORT published on the following pages. Readers should note that the results mentioned in the report, tabulated in performance charts and/or displayed using graphs and/or photographs should be construed as applying only to the specific sample tested.

Laboratory Test Report

Newport Test Labs measured the frequency response of the Questyle CMA800P as extending from less than 1Hz to 69kHz ±0.5dB and from less than 1Hz to 110kHz –3dB. Across the audio band, the response was 20Hz to 20kHz ±0.1dB. Channel separation was more than 100dB right across the audio band, measuring 116dB at 20Hz and 1kHz and still an excellent 106dB at 20kHz. Channel balance was equally good, with the lab measuring 0.054dB at 1kHz.

Distortion was extremely low, as you can see from Graph 2, with a second harmonic at –110dB (0.00031%), and second and third harmonics at around –115dB (0.00017%). A few other harmonics are visible above the noise floor, but they’re all more than 120dB (0.0001%) down. Overall THD+N, as shown in the tabulated results, was measured at 0.0009%.

The noise floor across the audio band was extremely low, as you can see on Graph 2, where it’s more than 140dB down. There was some low-frequency noise (the spike at the extreme left of the graph), but the overall signal-to-noise ratios measured by Newport Test Labs were exceptionally good: 101dB unweighted and 107dB A-weighted.

The frequency response of the CMA800R was almost as extended as that of the CMA800P, but within the audio band rolled off slightly earlier, to be 0.4dB down at 20kHz, then 1dB down at 35kHz as you can see in Graph 3. The results of Newport Test Labs’ measurements of the CMA800R’s distortion are shown in Graph 4.

You can see a second harmonic at –108dB (0.00039%), a fourth harmonic at –116dB (0.00015%) and a sixth harmonic at –120dB (0.0001%), but that’s it. All are so low as to be totally inaudible, but even if they were, the fact that they’re even-order harmonics (as opposed to odd-order) would tend to make the sound seem ‘richer’ than if there were no distortion at all. The noise floor of the CMA800R across the audio band is shown on Graph 4 to be tracking at –130dB, but some low-frequency noise (visible at the extreme left of the graph) resulted in the overall signal-to-noise ratios being measured at 91dB unweighted and 97dB A-weighted.

The frequency response of the Questyle CAS192D is shown in Graph 5 for four different settings of the filters and you can see that with the exception of the FIR brick wall filter, which cuts off steeply just before 20kHz (you can see why it’s called a ‘brick wall’ filter!), all other filters roll off very gently, to be around 0.2dB down at 20kHz, for a normalised response of 20Hz to 20kHz ±0.1dB.

The frequency response of the Questyle CAS192D is even-further extended when oversampling is switched on, as you can see in Graphs 6 and 7 which show the impulse response, but also serve to show high-frequency response, with Graph 6 showing the response without oversampling and Graph 7 with oversampling. You can see the extension of the response when using oversampling.

Newport Test Labs tested the Questyle CAS192D using both CD-standard digital signals (44.1kHz/16-bit) and AES-17-standard digital signals (48kHz/24-bit), the results of which are shown in the respective tabulated results, and in the graphs.

Graph 8 shows THD for a 0dB CD-standard signal and you can see that with the exception of the third harmonic, at –95dB (0.00177%), all harmonic distortion components are more than 100dB down (0.001%) and most are more than 120dB down (0.0001%).

Reducing the level of the test signal by 10dB, which gives the results that you would expect when playing music (which is never recorded at 0dB) the Questyle CAS192D returned the results shown in Graph 9, with a third harmonic at –108dB (0.00039%) and all other harmonics (and note there are fewer than there were at 0dB) are more than 120dB down (0.0001%).

Measured with extremely low-level test signals, down around –90dB, the Questyle CAS192D delivered excellent performance, no matter whether the test signals were dithered (Graph 10) or undithered (Graph 11).

The undithered signal results in the CAS192D delivering some odd-order harmonic distortion, but all components are more than 109dB down (0.00035%). Dithering the test signal results in a rise in the noise floor, but as you can see, that noise floor is still down around –140dB, and dithering also removes almost all the distortion components, with the two that are left sitting down at –122dB (0.00007%) where they’d be completely inaudible. The THD+N results with AES-17 signals were similar, but slightly better than the results for CD-Standard signals as you can see in Graph 12. The THD vs. Level results for AES-17 are shown in Graph 14.

Intermodulation distortion (CCIR-IMD) using 44.1kHz/16-bit test signals was extremely low, as you can see in Graph 13, with all high-frequency sidebands more than 110dB (0.00031%) down, and the unwanted regenerated 1kHz signal down at –118dB (0.00012%).

The very few high-frequency sampling-related products that are visible on the graph (far right) are nearly 120dB down. Using AES-17 IMD signals the overall results were –99.851dB for 18kHz/20kHz 1:1 ratio IMD and –100.304dB for 41Hz/7993Hz 4:1 ratio signals, which are also excellent results.

Channel separation was excellent for both digital standards tested, with the results tabulated for 44.1kHz/16-bit and shown in Graph 15 for 48kHz/24-bit signals. The results achieved using the CD standard are absolutely outstanding, with Newport Test Labs measuring 142dB at 16Hz, 132dB at 1kHz and 110dB at 20kHz.

Channel balance was equally outstanding, at 0.009dB, while interchannel phase was 0.01 degrees at 16Hz, 0.02 degrees at 1kHz and 0.036 degrees at 20kHz. The phase for the AES-17 signals was graphed, but since the trace was just a straight line across the 0 degree calibration, I thought I’d omit it to save space.

Signal-to-noise ratios were exceptionally good, at 105dB without weighting, improving to 109dB with A-weighting for CD-standard signals and 111.744dB with CCIR-RMS weighting for AES-17 test signals.

Although a de-emphasis option is fitted to the DAC Questyle is using in the CAS192D, it has not been implemented, so if you’re playing CDs that were manufactured prior to 1990 that have been emphasised, they’ll sound a little bright in the high frequencies, but that will be the only consequence.

The effect on the time domain of the different digital filter options fitted to the Questyle CAS192D varies from subtle to dramatic depending on the filter option you choose and the digital signal you’re using, so there too many variables to be able to illustrate them all in the space I have available. However I have included two of the oscillograms taken by Newport Test Labs that show the affect on an square wave of using an FIR apodising filter and a IIR filter.

The Questyle CAS192D puts out nearly 5 volts at its XLR outputs, which will be more than sufficient to correctly drive any ancillary equipment, and obviously perfectly suited to other Questyle components. The power consumption in standby is quite high (3.66-watts), so it’s not really what I’d call a ‘Standby’ power mode at all, but it consumes so little power that leaving it in Standby mode is not going to impact on your power utility bill.

All three of these Questyle components delivered superb performance on the test bench. I was seriously impressed by all three, but probably most of all by the CAS192D.# Steve Holding