– Audio Precision APx555 B-Series Analyzer with 200kOhm input impedance
– WyWires Platinum XLR interconnects
– USB Source: AMD PC via ifi iGalvanic 3.0 and iUSB 3.0
– Audioquest Carbon USB and SPDIF cables
– All measurements shown in this post are taken with USB input, XLR output, 44.1khz sample rate, Linear filter, Dither On, and Clock in ‘Local’ mode unless otherwise specified
– DAC is on latest firmware at the time of writing (30/08/2021)
– Full reports containing additional data and test configurations are attached
– Exact analyzer/filter configurations for each measurement are detailed in the full reports
The Rockna Wavedream holds an exceptionally high reputation amongst many audiophiles, often considered one of the best dacs money can buy. It is a dual-mono R2R dac and comes in four versions
- Wavedream Edition (£4,895)
- Wavedream Edition Balanced (£6,650)
- Wavedream Signature (£8,995)
- Wavedream Signature Balanced (£11,995)
The single ended vs balanced options are self-explanatory, and the ‘signature’ gets you a claimed increase of 12dB THD+N and 10dB dynamic range. Though also costs nearly twice the cost of the standard model.
The DAC offers proprietary, high tap-count digital filters, and can accept sample rates up to 384khz. DSD is converted to PCM internally and played on the PCM ladder.
It also has quite a hot output at 7Vrms, so make sure your preamp/amp can handle it! Otherwise you’ll need to add a few dB of headroom to ensure you don’t clip your amp input.
The Wavedream NET and Wavedream Signature Balanced DAC were kindly loaned to me by m17xr2b on SuperBestAudioFriends.
Full Measurement Reports:
Reports available here:
Dynamic Range (AES17): 127dB
Noise Level RMS (20-20khz): 2.930uVrms
Noise Level RMS (20-90khz): 6.710uVrms
DC Offset: 0.92mV Active, 1.06mV idle
1khz 0dBFS (Linear Filter, Dither On):
1khz 0dBfs (Filter Off, Dither Off):
THD+N / Frequency:
Low signal level performance is exceptional!
Definitely worth keeping dither set to ‘on’, as turning it off does hurt linearity (clever digital processing can be a great way to improve accuracy!):
SINAD/THD+N vs output level:
These graphs show some interesting behaviour. For those who don’t know, most R2R dacs nowadays aren’t made with a single, standard R2R ladder, they’re often made with two (sometimes more) ladders using different reference voltages, and switch between them depending on the signal level.
This allows for accuracy at low-levels that would otherwise be unachievable with a ‘traditional’ R2R design.
This is what causes the ‘sawtooth’ shaped behaviour in these graphs, which is not normally seen on Delta Sigma dacs.
The Wavedream seems to differ slightly from other R2R DACs though as this transition area seems to be larger, with more steps, and with a larger overall change in THD+N than DACs like the Denafrips Terminator Plus as an example.
The Wavedream features inbuilt DSP volume control. This is nice to have, but as with any DSP volume control, does result in a loss of dynamic range when you use it. A proper preamp would be preferred, but certainly nice to have this feature regardless, especially if you’re only wanting to adjust things slightly and not add an extra component to the chain.
It’s also useful to have given as the Wavedream’s hot output of over 7Vrms might be too high for some devices to accept. You can bring it down to a ‘standard’ 4v by setting the volume to -5dB.
My test for dacs with inbuilt volume control going forward will be to test dynamic range and THD+N with 20dB of volume control applied.
IMD (SMPTE) vs Output Level:
-90.31dBfs 1khz Sine (Recorded with 192khz ADC Sample Rate):
Filter Ultrasonic Attenuation:
It looks like ‘filter off’ is not actually true NOS, though it was never directly advertised as NOS anyway, in fact Rockna proudly advertises their high-performance oversampling filters and recommends using them.
The above impulse response shows the ‘Filter Off’ impulse response which gives a result close to ‘sample and hold’, like what you’d see from a NOS dac, though with some pre and post ringing, indicating that with the digital filter off there is still some DSP going on.
We can also see some stepping behaviour if we look in very close at one of the other impulse responses.
Note: This is NOT a bad thing, I’m just showing it because I think it’s interesting.
The ladders can’t run at an infinite speed after all.
Idle Noise Spectrum upto 1.2Mhz:
Noise floor overall is very very low, though we can see some content around 705.6khz. This is due to the oversampling. If we turn the digital filter and dithering off, then the idle noise looks different in the area where 16x the incoming sample rate falls.
Now the shape of the 705.6khz content is a fair bit different, but we still see a lot of activity.
Lastly here is the idle noise with the linear filter back on, but dithering off:
44.1khz Jitter (USB):
48khz Jitter (USB):
Jitter is exceptionally clean, not only are all artefacts under -160dB, the stem of the fundamental is razor thin all the way down past -160dB.
Interestingly, even though we’re on USB, when changing to ‘stream’ clock mode, the jitter pattern does change:
Whilst at first glance this might look like it’s overall better, the key thing to note is that the fundamental has a thicker/more spread-out stem, with some very close in sidebands that are hard to see in the image.
Regardless, both modes offer exceptional jitter performance, and I’d recommend trying both and seeing if you prefer one or the other, you’re getting a fantastic result either way.
You can also look at my measurements of the WD with the NET here to see how it changes jitter:
NOTE: I am using a new, higher FFT count method for this test.