Expanding the Class D Conversation: How Would You Characterize Their Differences?

Finally, I would mention Westminster.  WestminsterLab is Class A because its output devices conduct continuously through the full waveform (no crossover switching like AB or D), but it achieves efficiency by using a highly regulated, tightly controlled power supply and optimized bias that minimizes wasted current outside the linear operating window. In other words, the signal path remains pure Class A, while the power architecture is engineered to reduce heat and excess dissipation without altering the conduction behavior of the output stage.  Basically a Class A amp that behaves a lot like a Class D.  These things are awesome and I had a pair in for a few months and have heard them on some incredible systems.  

I asked someone associated with Westminster if their design was similar to the KRELL DUO XD lineup of Class A amps and I was told that it was. There is a thread on WBF forum on Westminster where this was answered.

I returned the Meitner MA3i DAC to my friend today and needed a DAC for the office, so I dusted off the RME DAC I had lying around and paired it with the Class D GanFet 6.5 amp to drive the Magnepan Mini. I got this DAC for $500 used.

In comparison to the imersiv D-1 + CODA #16 amp with the Mini or the Meitner MA3i + CODA #16 with the Mini, the $1500 worth of Class D and RME does not embarrass. This in comparison to close to $30k of the best DAC and Class A amp tech around compared to this 5lb Class D GanFet amp. I am really impressed with this amp.

I am going to give the imersiv D-1 DAC + Class D GanFet one more round together but this time I am going to try some of the 2nd Order Harmonic Distortion settings on the imersiv D-1. I have 50 settings that I could apply with this DAC to juice it up.

Disclaimer attempt: Please don’t rake me over the coals too much for trying this, LOL  It was an experiment. 

The results are puzzling if any of it is close to being remotely true.  I queried an AI engine, asking what the real sonic differences a human might hear when comparing a high quality Class AB solid state amplifier design to a high quality Class-D amplifier design running Purifi modules. The last bullet is odd.  Being a midrange focused listener myself, it would be fun to hear from others who might have differing feedback and opinions. I’ve read a lot about Class-D Midrange, in the top shelf amps, hmmm. 

-------------

The biggest audible differences tend to be:

🔹 Midrange texture vs neutrality

Class AB may feel slightly richer.
Purifi will feel more exact.

🔹 Bass character

Class AB: weighty, rounder
Purifi: tight, ultra-controlled

🔹 Emotional perception

Some listeners interpret small harmonic differences as “engagement.”
Purifi aims to remove that entirely.

@earthbound 

So what would Class AB be classified as OP?

If we follow the point I’m trying to make – i.e. that within an engineering design type the sound can vary – then "Class AB" might not fall into a single classification. 

Rather than talk about Class AB as *one* kind of sonic cluster, I looked around for various descriptions of some Class AB amps and then sorted them into two.

Class AB Amps:

Seamless: Organic, Warm, Enveloping, Layered, Intimate Midrange, Blooming Bass

• McIntosh MC452 / MC462 / MC601 
• Parasound Halo JC-1 / JC-1+
• Jeff Rowland Model 6 / Model 10
• Luxman M-900u
• Quad 303 (reissue)
• Classé CA-2300

Articulate — Clean, Detailed, Punchy, Precise, Controlled, Dynamic, Forward Midrange, Taut Bass

• Bryston 4B³ / 7B³ / 14B³
• Benchmark AHB2
• Parasound Halo A 21+
• Krell KAV-2250
• Hegel H590

Again, if Ralph is right – that there is no one "sound" for Class D amps – then we are best off not calling it "the Class D sound" and, as above, assigning different tonal categories to different class D amps.

@hilde45

This should be printed out and nailed to the front door of your nearest Hi-fi store.

I am very much anticipating your comparisons of the NAD M23 vs. the Pass XA25

Thanks.

@koestner Here is a summary of the trials for NAD M23 vs. Pass XA-25.

I’m moving on to AGD vs. M23, starting today.

System & Room Conditions
Listening room treated with diffusion panels, absorbers, and bass traps. Signal chain: Intel NUC streamer (Roon) → Holo Audio Spring DAC → solid state preamp → amplifier. Power conditioning via Audience Adept Response aR2p High Resolution Power Conditioner. Speakers: sealed, large format with JBL 15-inch drivers and Beyma AMT tweeters, mounted on Townshend platforms.

NAD M23 vs. Pass XA-25

Session 1 — February 18, 2026
Conditions: Solid state preamp set to achieve 81.7 dB, C-weighting.
Overall verdict: Slight edge to the Pass.
•    M23 demonstrated superior clarity and microdetail retrieval — in complex passages, background elements were more distinctly resoluble
•    Pass offered better tonality and timbre, with rounded leading edges on higher notes that nonetheless remained truthful and musically involving
•    Soundstage marginally wider on the M23; the Pass was entirely adequate spatially
•    Both handled instrument and voice articulation well
•    Pushing the Pass to approximately 51 on the preamp revealed no harshness — the sound remained magnetic and engaging at elevated volume

Session 2 — February 19, 2026
Conditions: Preamp set to create louder session +4 dB increase from Session 1; 85.5 dB C-weighting.
Overall verdict: Edge to the Pass — less on its own merits than on tonal failings of the M23 at higher volume.
•    M23 clarity remained highly engaging — more hidden notes and rhythmic details surfaced
•    Bass was taut, detailed, and full (string bass especially)
•    Female vocals repeatedly approached a cringe threshold — a hardness or forwardness that prompted a "turn it down" response
•    M23 was rhythmically fun with a wide soundstage
•    M23 was favored early in the session; the cumulative effect of the edge-of-cringe moments shifted the verdict by the end

Session 3 — February 20, 2026
Conditions: Mid-day listening with new program material; volume dialed back to 82.3 dB C-weighting. Solid state preamp at 40 (NAD), 43 (Pass).
Overall verdict: Slight edge to the M23.
•    Stjepan Hauser with the London Symphony Orchestra performing the Nocturne in C-Sharp Minor was markedly more engaging through the M23 — more dramatic orchestral presence, deeper bass weight, and a resonant, detailed cello sound (the body of the instrument was audible)
•    Wide soundstage consistent across multiple tracks
•    Percussion accurately placed and natural
•    M23 surfaced details that the Pass tended to smear or obscure
•    Pass excelled in its characteristic areas: vocals intimate, organic, and fully human-sounding; piano natural, percussive, and accurate
•    None of the harshness or cringe moments from Session 2 appeared — possibly attributable to the lower listening level

Emerging Pattern, NAD M23 vs. Pass XA-25:
Across three sessions a reasonably consistent picture is forming:
•    M23 leads on: resolution, dynamic articulation, bass definition, and soundstage width
•    Pass leads on: tonal naturalness, timbral roundness, and organic vocal and piano reproduction
•    The M23’s central liability appears to be a volume-dependent edge — a tendency toward hardness or brightness at higher levels that the Pass does not share
•    The Session 3 result (slight M23 edge at lower volume, no harshness) suggests the M23’s tonal character may be level-sensitive, which would be a meaningful variable to isolate in future sessions