Srajan
Ich bin ein Berliner. Famous but not last words. These days John Darko is a Berliner and Portugeezer. Summers in Germany, winters in Lisbon. As such he knows the German words for convert as ändern, umformen, wandeln or wechseln. If you change money in the temple, it’s Geld wechseln. If you change from man to werewolf, it’s wandeln as in, being a changeling. Changing up 2023’s D/A converter scene is Poland’s Ferrum Audio with its flagship Wandla. If it spelled Wandler, it’d be how all Germans call all DACs. Wanting to be special not all, it’s Wandla instead. Before ‘flagship’ mentions cloud your vision with too many zeros—this, after all, is the realm of ones and zeros—the sell price is €2’795 with Globtek 24V/2.5A switching power adapter. Upgrading to the companion Hypsos power supply adds €995. This 80-watt power supply with otherwise infinitely adjustable output voltage can run Ferrum’s Oor and Erco which combined burn up 31.5W at idle. “Off Hypsos you can easily use 3 splitters and connect 2 x Oor and 2 x Erco/Wandla and there should still be some spare watts left.” Wandla anticipates being hypnotized by the charms of Hypsos with that special power-sensing 4-pin 22-30V/2.5A socket.
Ferrum Audio the name reflects ancient iron ore deposits around Warsaw where the company is based; and the Corten steel accent on the half-size casing recalls it with its sealed red rust coloration. The beating heart of Wandla is Serce, a 5 x 7cm PCB whose ARM Cortex M7 processor combines what usually takes five chips into one. Optimization from simplification. It routes/transcodes USB, S/PDIF, AES/EBU, differential I2S over HDMI or RJ45, ARC under CEC control, MQA and IR remote code. Actual conversion is by ESS 9038 Pro but I/V conversion again embeds on Serce. The rear panel hosts USB C, coaxial S/PDF, AES/EBU, I2S and ARC on HDMI plus Toslink digital inputs, one set of RCA analog inputs auto-converted to balanced; and class A/B RCA/XLR analog outputs. The touch screen opens a multi-layer menu with extra functionality like three digital filters from Sabre, two from HQ Player (more to come) and volume type or bypass. Published bandwidth is 10Hz-200kHz ±0.1dB, output impedance 22/44Ω RCA/XLR, THD+N -115dB, A-weighted dynamic range 127dB. MQA is auto-sensing. Wandla measures 21.7×20.6x5cm WxDxH and weighs 1.8kg.
Ferrum would remind chip spotters who conflate ESS (or any other brand) with a particular sound that beyond power supply and analog output stage, digital pre-processing by filter and upsampling algorithms as well as post-processing in the current-to-voltage stage are vital sonic contributors. The same goes for volume control execution. Wandla exploits Sabre’s on-chip version for digital, for analog dual-mono Muses PGA which buffer by opamps “whose careful selection proved crucial for best sonics.” Initial I2S pin configuration is in the PS Audio mode which proved perfect for my Singxer SU-2/SU-6 USB bridges. As demands arise, Wandla’s software will add other pin configs. Though MQA declared insolvency by April 2023 and streaming service Tidal announced that they’d be shortly offering standard .flac files in their hi-rez tier with no intentions of acquiring MQA’s assets, Wandla doesn’t care. MQA is aboard yet their lossy algorithm is bypassed for all non-MQA content. Wandla is codec agnostic. Own a large MQA library? That won’t vanish should MQA the company fold. Wandla can play such files as that codec intended. If we don’t own a single MQA file, no issue either. With PCM support up to 32/768 and DSD up to 256, no native file is left behind.
Ferrum’s default filter choice is HQ Apodizing, its default Hypsos setting 24V which is safely adjustable from 22-30V. A lovely display convenience is temporary zoom on the volume figures whenever we change SPL. They get big enough to remain legible from 3-4m in the seat. Menu options are I2S input DSD pin and format, up to ±12dB analog gain trim, up to -12dB digital input trim for 9.5/4.75Vrms to 2.4/1.2Vrms XLR/RCA output voltage, HT bypass, TV control/trigger and various adjustments to display/logo brightness and behavior. When the outputs are set to fixed, the display shows ‘bypass’ to remind us that we now best have volume control elsewhere. The display also confirms the incoming sample rate and bit depth. I checked that Wandla handles 48kHz source material pre-upsampled to 768kHz in Audirvana Studio without issue. Setting digital input trim to -8dB creates standard 2V/4Vrms RCA/XLR outputs. By default, Wandla is set to the ‘pro’ standard of 4.75/9.5Vrms for, cough, “better S/NR”. At 9.5V, most consumer systems expecting the homie standard of 2V will need very stiff attenuation to not get too loud too quick. Many amplifiers in fact hit full output with less than 2V. “Our upsampling multipliers are always integer. All ESS filters upsample x 8 by default. Our HQ filters upsample to 352.8/384kHz. Volume scale is in decibels so 0 = minus 100dB, 80 = minus 20dB. Bypass and digital volume circumvent the Muses chip and its buffers. Analog signal from the anti-aliasing filter at the DAC output past the I/V converter proceeds straight to the output line driver.”
Back on chip spotting, my first sessions with default filter, volume bypass, 4.6V RCA and 24V Hypsos were decidedly – um, toothless. That’s if we reference early (if diehard) ESS perceptions of a detail-über-alles, slightly pixilated, wiry, hyper-enunciated and cold sound. Au contraire, mes amies. Wandla presented warm, soft, dense, dark and ultra-friendly. No bloodthirsty sabretooth tiger but purring lap cat. That kitty can still be stoked by changing Hypsos supply voltage in what Ferrum calls sweet-spot tuning. At 22V the sound leaned out but perked up, at 30V it maxed out the dark weighty silk aesthetic. Compared to punching through digital filter options, setting our preferred Hypsos voltage is a far bigger sonic decider. My best/favorite results came with -10dB digital input trim, 22V, HQ’s Apodizing filter and I2S signal pre-upsampled to 176.4/192kHz in Audirvana Studio. It gave me the spatially deepest most 3D-sculpted imaging and best walkabout visibility. The sound was still properly dense but not as soft, thick and subliminally fuzzy as with far higher power voltage. My settings of course are irrelevant. What’s decisive is your being able to select your own. Trust your ears. Have fun.
Regardless of fussing over adjustments, what I heard as the main attractions were elegant textures and tone. In fact, Wandla was surprisingly comparable to my Cen.Grand DSDAC 1.0 Deluxe. That full-size deck resamples all to DSD 128, 256, 512 or 1024. My iFi DSD Pro Signature can likewise resample to DSD 1024 on the fly; or treat PCM as PCM. Further on the subject, a reader gifted me with a hard drive full of native .dff files. Armed with these formatting tools, my take on the subject is that at 512, DSD’s treble becomes competitive with PCM whereas below, it remains hooded. I also hear DSD as generally sweeter, spatially billowy or ‘reverb’ enhanced thus texturally soft and elastic. By contrast, I hear PCM has packing higher resolution and superior soundstage layering whilst being more damped, edge specific, dry and lit up. Though a €3’800 tab for Wandla/Hypsos x a £6’000 ask for the DSDAC 1.0 Deluxe might suggest otherwise, I really heard these contenders as quite on par. The key offset was Cen.Grand’s stage beginning farther back. That put more virtual distance between it and my chair. The far end of the stage simply didn’t retreat equally. Ferrum’s soundstage extended just as deep so apparently through the front wall but started closer at the speakers. The net effect was a more distanced shallow perspective for DSD über alles, one more forward yet deep for Wandla’s PCM. Just so, the latter exhibited that softer more generous tone with less emphasis on transients and more on decay which I conflate with DSD. Meanwhile economics class would suggest greater labour savings for China than Poland. Ferrum beating Cen.Grand on value was unexpected chicane. True, we won’t buy the same case size/bling but get more featurized with Wandla. What’s more, Ferrum’s young team applied more smartphone think to their user interface. It’s more modern and multi-layered than Cen.Grand’s.
Next, I was curious to contrast Wandla’s analog volume control to that in my Sonnet Pasithea. Unlike Ferrum’s circuit which generates high fixed gain then trims it resistively or in algorithm, Sonnet’s variable reference voltage on their R2R ladders only produces the gain we set. Now signal-to-noise ratio doesn’t shrink as we throttle back SPL. Meanwhile, Wanda’s published S/NR figure only applies to max gain. As we select a standard 2V out, the circuit’s noise remains constant against a smaller signal. That’s resolution loss at higher attenuation despite involving no bit stripping from digital volume. It’s what I suspect was behind the admittedly costlier Dutch offering audibly higher resolution driving our 250W Kinki EX-B7 monos which only need 1.4Vrms for full output. That made Wandla’s balanced gain 61⁄2 times (!) too high prior to pre-trim. Over the years I’ve conducted numerous comparisons between active preamps including very high-gain sorts like Supratek to multi-tapped autoformers for passive-magnetic attenuation. The latter wandels voltage to current rather than burns off voltage like a car’s brake pads heat up to need eventual replacement. From those prior experiments, I suspect that a key enabler of Wandla’s textures and overall warmth is excess gain. High-gain preamps exhibit very similar attributes; and invariably trail a premium passive-magnetic pre on resolution and speed.
That’s back at higher Pasithea resolution. It presents very similar to how a top widebander speaker like a Cube, Rethm or Voxativ behaves versus a premium multi-way Raidho. The latter will express a certain sweetness and tonal fill but feel less electrifying and invigorating. The widebander does it more direct, dynamically twitchy and communicative. The premium multi-way can be prettier but it will be less lively. Wandla packs the textural elegance of the latter but not yet the speed and resolution of the former. Getting both needs a bigger spend. On which, consider Hypsos a Wandla necessity even if you must delay until the wallet recovers. Far more important than sample rates, formats or digital filters, the upgrade power supply lifts this converter to a higher level. The excitement of Ferrum’s marketing department over their “dynamic filtering”—a forthcoming library of user-voted favorite digital filters uploaded by firmware upgrade—misdirects from the fact that the existing PSU makes a far more meaningful difference. Don’t buy Wandla to become part of the dynamic filtering club. That’s just a talking point for PR. Buy Wandla because you want a DSD-type sound. The same holds for I2S vs. AES/EBU vs. coax. Whatever little differences those make, the dominant thing is Wandla’s gestalt of textural/tonal elegance coupled to darkness whose degree is adjustable with the Hypsos power supply.
Cogito, ergo sum. This Roman proverb usually translates as “I think, therefore I am”. Today we might propose a new audiophile proverb:
I think, therefore I Wandla!
Further information: Ferrum
