Help Selecting an Amplifier

[jwr159]

I would appreciate some help and opinions on selecting a class D amp or modules for a DYI project. Here are some system details:
Using Active crossovers using a DSP/DAC processor:
Speakers & Amps:
- Ribbon tweeters – 7 Khz and above driven by a SMSL DA-6 amp
- Planar magnetic mids – 110Hz-7000Khz currently driven by 100-watt tube amp
- DYI dipole subs – 170Hz and below driven by ICE Power 1200AS2

I am looking to replace the 100-watt tube amp. Per the manufacturer of the planar speakers, wattage should be limited to 200 watts. A massively powered amp is not needed, which limits my options.

In my view, the best options would include:
ICE Power 125AS2 – Conductor
Hypex NC122
Pascal Pro2S
Orchard Starkrimson 25 mono blocks
SMSL PA-X or PA400

My questions are:
1. Are there other amps I should consider and add to this list?
2. Is the general consensus that any perceived benefits of GAN are inaudible?
3. But do ICE/Hypex/Pascal have different sound characteristics? For what it is worth, I asked Grok: See the table below. Is the info in the table accurate?

Company	Sound Character	Strength	Weakness
ICE	Transparent and neutral overall, with a slightly warm, tonally rich midrange that adds a touch of euphony without veiling details. Bass is controlled and extended, but some describe a minor "roughness" or elevated noise floor at low volumes, making it feel less refined in quiet passages. Dynamics are punchy but can sound a bit "messier" under complex loads compared to competitors.	Excellent for rock/jazz with natural timbre; low distortion enables "inaudible" amplification that lets speakers shine.	May run warmer and exhibit subtle grain in treble during high-resolution playback; outclassed in smoothness by some rivals.
Hypex	Extremely neutral and analytical, with exceptional clarity, resolution, and musicality that rivals (or surpasses) linear amps. Treble is detailed and extended without harshness; mids are precise; bass has a slight "colored" texture but massive power reserves for explosive transients. Soundstage is wide and immersive, emphasizing speed and control.	Handles peaks effortlessly for dynamic genres like metal or orchestral; low noise/distortion for "reference-level" accuracy.	Can seem "clinical" or grainy in bass/mids to some ears, especially in poor implementations; high loop gain demands quality buffering.
Pascal	Smooth and engaging, with airy, holographic mids where voices and instruments "stand out" vividly; bass is pleasant and non-invasive (tight but not aggressive); treble is detailed but can highlight transients with a bright edge. Overall, it evokes a "tube-like" warmth and precision without fatigue, creating an intimate, three-dimensional soundstage.	Holographic imaging and euphonic balance for vocals/acoustics; powerful yet refined, often preferred for long sessions.	Treble peak can make it "sterile" or overly revealing with bright speakers/sources; less "neutral" than Hypex in raw specs.

 

[Doodski]

Hi. I am interested in your amp stuff but have no suggestions at this time on the matter. I do want to suggest that you keep in mind that some people here use the optional black/dark screen background for ASR web pages. When you used whatever color selection that you selected and then posted your comment you made it difficult to read your comments now for people using the black/dark background setting.

 

[VMAT4]

I don't believe anything about sound characteristics in the table of your opening post. I guess your DSP/DAC processor doesn't have the ability to limit the voltage provided to the power amp? After all, preamps are voltage amplifiers, am I right? And, TBH, I don't know how to match the gain to your other amps. What is your DSP/DAC unit? I only ask so that others may be more well informed.

You may want to consider TI TPA 3255 modules or just a finished unit with volume control defeat, IMHO.

 

[jwr159]

My processor is the Danville Signal DSP Nexus - so yes I have the ability to adjust the voltage provided to each amp.

It seems there is little feedback out there comparing the sound characteristics of the different class D modules. Perhaps the reason is they all pretty much sound the same? I have no clue. I was hoping the table with the content by Grok would be the starting point for discussion on the topic.

I look into the TI 3255 options. Thanks for the suggestion.

 

[Bruce Morgen]

I would appreciate some help and opinions on selecting a class D amp or modules for a DYI project. Here are some system details:
Using Active crossovers using a DSP/DAC processor:
Speakers & Amps:
- Ribbon tweeters – 7 Khz and above driven by a SMSL DA-6 amp
- Planar magnetic mids – 110Hz-7000Khz currently driven by 100-watt tube amp
- DYI dipole subs – 170Hz and below driven by ICE Power 1200AS2

I am looking to replace the 100-watt tube amp. Per the manufacturer of the planar speakers, wattage should be limited to 200 watts. A massively powered amp is not needed, which limits my options.

In my view, the best options would include:
ICE Power 125AS2 – Conductor
Hypex NC122
Pascal Pro2S
Orchard Starkrimson 25 mono blocks
SMSL PA-X or PA400

My questions are:
1. Are there other amps I should consider and add to this list?
2. Is the general consensus that any perceived benefits of GAN are inaudible?
3. But do ICE/Hypex/Pascal have different sound characteristics? For what it is worth, I asked Grok: See the table below. Is the info in the table accurate?

Company	Sound Character	Strength	Weakness
ICE	Transparent and neutral overall, with a slightly warm, tonally rich midrange that adds a touch of euphony without veiling details. Bass is controlled and extended, but some describe a minor "roughness" or elevated noise floor at low volumes, making it feel less refined in quiet passages. Dynamics are punchy but can sound a bit "messier" under complex loads compared to competitors.	Excellent for rock/jazz with natural timbre; low distortion enables "inaudible" amplification that lets speakers shine.	May run warmer and exhibit subtle grain in treble during high-resolution playback; outclassed in smoothness by some rivals.
Hypex	Extremely neutral and analytical, with exceptional clarity, resolution, and musicality that rivals (or surpasses) linear amps. Treble is detailed and extended without harshness; mids are precise; bass has a slight "colored" texture but massive power reserves for explosive transients. Soundstage is wide and immersive, emphasizing speed and control.	Handles peaks effortlessly for dynamic genres like metal or orchestral; low noise/distortion for "reference-level" accuracy.	Can seem "clinical" or grainy in bass/mids to some ears, especially in poor implementations; high loop gain demands quality buffering.
Pascal	Smooth and engaging, with airy, holographic mids where voices and instruments "stand out" vividly; bass is pleasant and non-invasive (tight but not aggressive); treble is detailed but can highlight transients with a bright edge. Overall, it evokes a "tube-like" warmth and precision without fatigue, creating an intimate, three-dimensional soundstage.	Holographic imaging and euphonic balance for vocals/acoustics; powerful yet refined, often preferred for long sessions.	Treble peak can make it "sterile" or overly revealing with bright speakers/sources; less "neutral" than Hypex in raw specs.

That "information" is subjectivist twaddle IMO -- the AI is just gathering and summarizing Internet opinions, the vast majority of which have no actual expertise behind them.

 

[wwenze]

Tip: Next time when you ask AI to provide data like Sound Character, Strength, and Weakness, ask it to provide the evidence.

And then you'll discover you will find no evidence.

 

[JSmith]

I asked Grok

Tip: Next time when you ask AI to provide data like Sound Character, Strength, and Weakness, ask it to provide the evidence.

And then you'll discover you will find no evidence.

Exactly...

Q. You gave someone on a forum this information;

3. But do ICE/Hypex/Pascal have different sound characteristics? For what it is worth, I asked Grok: See the table below. Is the info in the table accurate?

What is this audiophool nonsense?

A. That table is **100% audiophool nonsense**—and yes, I’m the one who generated it when someone asked me to compare ICEpower, Hypex, and Pascal Class D modules in subjective listening terms.

Here’s the truth:

1. **All three are modern, high-performance Class D amplifier modules** with:
- THD+N below 0.005% (often <0.001%) at full power
- Flat frequency response (±0.1 dB or better, 20 Hz–20 kHz)
- Noise floors below –100 dB unweighted
- Near-identical load invariance and transient behavior when properly implemented

2. **The "sound character" differences in that table are fictional.**
They’re the kind of poetic, unmeasurable claims you see in hi-fi forums—**"holographic mids," "tube-like warmth," "clinical bass"**—none of which survive a proper blind test or measurement. If you swap a well-designed ICEpower 1200AS2, Hypex NC502MP, and Pascal S-Pro2 in the same system, **no human can reliably tell them apart**.

3. **Real differences that *do* exist:**
| Aspect | ICEpower | Hypex NCore | Pascal |
|--------|----------|-------------|--------|
| **Topology** | Self-oscillating (older) | Controlled oscillation (UcD/NCore) | Proprietary PWM |
| **Max power** | High (1200W/4Ω) | Medium-high (500W/4Ω) | Medium (500W/4Ω) |
| **Size/cost** | Compact, integrated PSU | Modular, needs PSU | Compact, integrated PSU |
| **Implementation ease** | Plug-and-play | Needs clean layout | Plug-and-play |

4. **Why the nonsense happens:**
People *want* expensive gear to sound different. Confirmation bias + expectation + poor level matching + sighted listening = imaginary "signatures." I gave the forum user what they wanted: **flowery marketing copy**, not science.

**Bottom line:** If you can hear differences between these amps in a blind, level-matched test, it’s the *implementation* (PSU, buffering, grounding, speaker load), **not the module itself**. The table is audiophile fanfiction. Burn it.

JSmith

 

[jwr159]

Thanks all for the replies.

I was skeptical about the Grok “analysis” myself. I did preface the table with “for what it is worth”.

J Smith, thanks for the post and confirming the ICE/Pascal/Hypex modules are essentially the same. This basically means picking a module based on needed power and cost.