The same twist applies to the vintage DAC chip audiophiles treat as sacred.
A tube shortage should have been a serious problem for a designer whose flagship CD player relied on one. For Ancient Audio’s Jarek Waszczyszyn, it became an accidental test of what audiophiles think they are paying for.
His listeners compared the old 6H30 tube stage with a new MOSFET version and did not give the expected answer.
And while that result did not settle every tube debate, it did expose a weak spot in how buyers judge expensive digital players.
The Accidental Swap
Isolating what shapes a CD player’s sound is difficult, but Waszczyszyn stumbled into a natural test. When the supply crisis forced him to replace the 6H30 tube, he built a MOSFET version that listeners could switch against the tube circuit during playback.
However, the swap was driven by necessity, not curiosity, so nobody expected it to become a referendum on one of audiophiles’ favorite assumptions.
Experienced listeners then compared both versions without knowing which topology was active.
Two details matter here:
- The listeners could not clearly identify the tube or MOSFET stage, which challenges the idea that a tube output stage should sound obviously different on sighted expectation alone.
- When they did hear small differences, they preferred the transistor version.
A 6moons review of the Lektor Joy pointed in the same direction.
It found that “fast switching showed no big differences” between tube, transistor, and IC output stages, though longer listening with emotional rather than technical attention revealed subtler differences.
The test proves nothing on its own. The sample was self-selected, the conditions were casual, and Waszczyszyn shaped both circuits being compared.
But the result still matters because it ran against his own commercial history with tubes: when forced to abandon the part his flagship was built around, he did not end up with a worse player by default.
Five Decades of the Same Awkward Result
Waszczyszyn’s listening sessions wouldn’t survive peer review, but its result fits a pattern that has embarrassed audio assumptions for decades.
When electronics are level-matched and hidden from view, the “obvious” sound of a topology often becomes much harder to prove.
In 1978, QUAD founder Peter Walker offered £5,000 (~US$6,400) to anyone who could reliably distinguish his Quad II valve amplifier from the solid-state 303 and 405.
A panel of six, including audio journalists and a BBC representative, tried and failed.
Still, that result does not prove tubes and transistors always sound the same, and it does not directly test Waszczyszyn’s CD player. But it does show why his MOSFET outcome is less shocking than the parts mythology suggests.
And once the identity of the circuit is hidden, listeners may struggle to hear the dramatic difference they expected from the tube itself.
This means the output stage is only one part of a larger system. If the power supply, clocking, filtering, grounding, and analog layout are doing more of the audible work, then paying for a famous tube or a romantic topology tells only a small part of the story.
Worshipping the Wrong Silicon
The tube question is only half the mythology Waszczyszyn wants to dismantle, and the other half lives in the DAC chip.
Philips TDA1541, a multibit R2R converter discontinued in the mid-1990s, has become an audiophile relic.
Some designers strip away the separate digital oversampling filter and run the chip in NOS mode, converting digital samples directly to analog without the interpolation that modern delta-sigma chips use.
But Waszczyszyn warns that the NOS approach demands an analog filter with “extremally high” precision and aperiodic response.
“Its sound can be very well if made by master, or awful if made by beginner,” he told Dagogo.
The counterargument comes from SW1X Audio Design, a manufacturer that sells NOS R2R DACs, which argues that modern delta-sigma chips create “completely new samples” through interpolation that aren’t part of the original recording.
Even SW1X concedes tradeoffs, acknowledging that its preferred architecture can suffer from a “perceived lack of resolution.”
For the Lektor Joy, however, Waszczyszyn took the opposite path. He runs dual ESS ES9038Pro chips in mono mode with separated power supplies and grounds, choosing built-in digital filtering for its reproducibility rather than the hand-tuned analog filter that NOS implementations require.
This choice challenged the idea that any chip carries sonic magic on its own, before the filtering, power supply, grounding, and analog stage are even considered.
Inside the Parts Nobody Argues About
Inside the Lektor Joy, the most complex engineering sits where no audiophile forum would think to look.
Power and timing, not tubes and chips
The Lektor Joy’s power supply consumes the most power of the entire CD player and uses 20 times more components than most solid-state amplifiers. Waszczyszyn treats it as the foundation for the rest of the machine, because voltage noise can travel through every stage of the analog circuit.
Standard regulators like the uA 7815 and 7915 were not quiet enough for the design. Their complex pulse response produced a hundred times more noise than his circuit required.
Clocking gets the same attention. Consumer CD players typically let the transport’s internal clock govern the system, which forces the DAC to follow the timing generated by the disc mechanism.
A single Tent Labs clock governs the transport, display, keyboard, and DAC from one source in the Lektor Joy.
That setup reduces the inter-clock jitter that can appear when multiple independent oscillators are trying to coordinate the same digital signal.
Same parts, different player
The Gryphon Ethos makes the execution thesis concrete, since both the Ethos and the Lektor Joy use the same CD-Pro8 transport from Stream Unlimited and the same ESS ES9038Pro DAC chip family.
Waszczyszyn himself calls the Gryphon a “direct competitor…however two times more expensive.”
But the overlap ends at the parts list. Gryphon runs eight ES9038Pro chips in a dual-differential, dual-mono topology where Waszczyszyn uses two in simple mono mode.
Gryphon employs dual independent temperature-compensated crystal oscillators while Waszczyszyn’s single Tent Labs clock serves the entire system, and the Ethos retails for $39,000 against the Lektor Joy’s €15,000 (~US$16,200).
Shared transport and shared DAC family do not produce identical sound, because the power supply architecture, analog output stage, clocking philosophy, physical layout, and filtering approach all differ between these machines.
A spec sheet that lists “CD-Pro8” and “ES9038Pro” captures the ingredients while saying nothing about what the engineer did with them.
Six Years Searching for the Right Meter
During six years at Poland’s Institute of Electronics, Waszczyszyn searched for a measurement that could predict how a CD player sounds. He tested many complex measures and came away convinced that the usual numbers miss too much.
speech”Measures of CD players are completely unrelated to sound. I spent 6 years during my job at Institute of Electronics to find any serious relation, by many complex measures,” he told Dagogo./speech
His research is unpublished in English, so it should be treated as his position rather than established fact. The numbers he cites still explain why he distrusts the usual spec-sheet race.
Traditional errors like frequency response within ±0.2 dB and THD at 0.002% fall far below the threshold of human hearing, which suggests every competent CD player should sound the same.
In his own tests, he says wavelet transformation analysis showed 5 to 6 dB differences between players that standard metrics did not capture.
That belief shaped the Lektor Joy. Its noise floor sits below -100 dBA, beneath the resolution of the measurement equipment itself.
The result is a CD player built around the parts buyers rarely ask about: the power supply, clocking, filtering, grounding, analog stage, and physical layout.
Basically, a famous tube or vintage DAC chip can make a product easier to talk about, but the work worth paying for is the engineering around those parts.