Opus Likes this place

[jsherman999]

I use frontier LLM for coding, this is Opus's take on a real question I had, looks correct to me but thought I should bounce it off this forum?

Prompt:

"I'm trying to map the performance+cost of current class-D amplifiers like those based around the Texas Instruments TPA3255, for example the WiiM Amp, to past Hi-fi stereo systems. In particular I'd like to know what stereo systems from the mid 1970s would match it in performance, and find out how much they would cost in adjusted dollars. The WiiM Amp can be found in this ASR SINAD chart. SINAD values relate to historical THD like: -SINAD = THD+N https://www.audiosciencereview.com/...-d-monobloc-amplifier-review-2023-png.257210/"


Answer:

I'll work through this in pieces. Let me start by getting the actual specs.I have enough on the WiiM specs. Let me get inflation-adjustment figures and a couple of 1970s reference points.I have what I need. Here's the analysis.

The framing problem first​

The translation you proposed (-SINAD = THD+N) is right, but it hides a trap that's central to your question. The WiiM Amp measures ~89 dB SINAD at 5W (PFFB-less original) — that's about 0.0036% THD+N. The Pro/Ultra hit ~100 dB (~0.001%).

Here's the catch: no mid-1970s consumer amplifier could match that, because nothing in that era was built to those numbers and most measurement gear couldn't even verify them. A typical good 1975 receiver was spec'd at 0.1–0.5% THD; the genuinely excellent ones reached ~0.05% (≈66 dB SINAD); the absolute statement pieces of the late-'70s "THD wars" got to ~0.02–0.03% (≈70–74 dB SINAD). So you're looking for amplifiers that came up 15–25 dB short of the WiiM and cost vastly more.

That gap is the actual story. Class-D + modern feedback didn't just match the 1970s — it leapfrogged the best of it at a tenth of the price.

So the question becomes "what did 1970s money buy you in approach to this performance"​

Using the CPI factor of ~6.0× (1975→2025; $100 in 1975 ≈ $597 today), here's the mapping. The WiiM Amp is ~$299, ~60W/8Ω, ~89 dB SINAD.

1970s contender	Era spec (THD)	≈SINAD	~Power	1970s price	≈2025 $
Pioneer SX-1250 (1976)	~0.05%	~66 dB	160W/ch	~$900	~$5,000
Marantz 2270 (1971)	~0.1%	~60 dB	70W/ch	~$600	~$4,000+
Sansui AU-9900 (1974)	~0.05%	~66 dB	90W/ch	~$650	~$3,900
Yamaha CA-1000 (1973)	~0.05%	~66 dB	70W/ch	~$600	~$3,600
Technics SU-9070 + SE-9060	~0.05%	~66 dB	60W/ch	~$700	~$4,200

(Note the forum anecdote in my sources is consistent with this: someone spent about $320 plus tax on a Yamaha CR-420 in 1978 or 1979 — a mid-tier receiver — which is ~$1,600 today for performance well below the WiiM.)

The cleanest single answer: a Pioneer SX-1250 or Sansui AU-9900-class unit is the spiritual match — the "I want it quiet and powerful" flagship of its day — and it would run you $4,000–5,000 in today's dollars while still measuring ~20 dB worse than the $299 WiiM on SINAD (though with far more raw 8-ohm power).

The one caveat that matters for fairness​

SINAD is a single-tone bench number. The WiiM's weaknesses — load-dependent frequency response without PFFB, power-supply-limited dynamic burst behavior, and the ~25W loss under multitone stress noted in testing — are exactly the areas where a big 1970s linear amp with a massive transformer would feel more effortless into a difficult speaker. So "performance" splits: the WiiM wins distortion/noise by a mile and by 15× on price; the vintage flagships win on into-the-load brute authority.
Want me to push to a specific target — e.g. find the cheapest 1970s amp that broke ~70 dB SINAD (the late-'70s THD-war pieces like the Technics SU-V or Sony TA-N88), or build this into a proper SINAD-vs-inflation-adjusted-cost scatter plot?

 

[jsherman999]

I should include the follow-up, I have no idea if the Krell-era is correct, I heard they had bad sinad.

Prompt: "summarize the situiation from each decade since, how did the relative performance improve in the 80s, 90s, 2000-2010, couple choices of what would come close or surpass in each decade?"

-------


I'll trace the SINAD/price trajectory decade by decade. The key reference points are solid (SX-1980 at 0.03%/~70 dB SINAD, ~$5,400 adjusted; the broad arc of THD specs over time). Here's the summary.

The throughline​

The WiiM Amp sits at ~89 dB SINAD (~0.0036% THD+N) for ~$299. The story across decades isn't really "when did distortion get good enough to hear" — by any reasonable standard the 1970s were already past audible-transparency for THD. The story is two separate curves: measured performance climbing, and the price of reaching a given performance level collapsing. They cross in the 2010s, which is why a $299 box now outmeasures four-figure gear from every prior decade.

The 1970s baseline (recap)​

Best flagships landed at ~0.03% THD (~70 dB SINAD): the Pioneer SX-1980, rated under 0.03% THD at 270W/ch, retailed at $1,295 in 1978, roughly $5,450 in 2021 dollars (~$6,000+ now). So: ~19 dB short of the WiiM, at ~20× the inflation-adjusted price. Nothing from the decade matches or surpasses it on SINAD.

The 1980s — the spec wars peak, then plateau​

This is when THD numbers went genuinely silly on paper. The early-'80s "0.0003% THD" Japanese integrateds (Technics, Pioneer, Sony, Yamaha's high-feedback designs) chased vanishingly low single-tone distortion — partly a marketing arms race, since, as one engineer in my sources put it, heavy feedback gives great THD numbers and the Japanese were known for "the spec game." On the bench, the best of these claimed ~0.001–0.0005% THD, which on paper is 90–100 dB SINAD — i.e. the first era that, by spec, meets or beats the WiiM.
The caveat: those were optimistic rated figures at favorable conditions, not the all-in SINAD (noise included) that ASR measures. Real-world A-weighted SINAD on a good '80s integrated was more like ~80–85 dB. So the honest verdict:

• 
  
• Comes close on paper / approaches in reality: Technics SU-V series (e.g. SU-V8/V9, "New Class A"), Pioneer A-x0 series, Sony TA-F555ES. Spec-sheet THD ~0.003–0.0007%. Real SINAD ~80–85 dB. Original price ~$700–1,100 → ~$2,000–3,000 today.
• 
  
• Verdict: A top '80s ES-series Sony or Technics is the first thing that lands in the WiiM's neighborhood (within ~5 dB), but still costs ~7–10× as much adjusted, and you're trusting rated specs that modern bench testing often knocks down a few dB.

The 1990s — quietly transparent, big price spread​

The '90s stopped bragging about THD (the wars were over) and the mainstream settled into genuinely transparent territory. A mid-'90s Denon/Onkyo/NAD integrated did ~0.02–0.005% THD honestly. The high end (Krell, Mark Levinson, big Yamaha/Accuphase) pushed real measured SINAD into the ~90–95 dB range — now legitimately matching or beating the WiiM, but at brutal cost.

• 
  
• Matches/surpasses (real measured): Accuphase E-407, Krell KAV-300i, Mark Levinson No.383. These genuinely measure ~90+ dB SINAD. Original prices $4,000–8,000 → ~$8,000–16,000 today.
• 
  
• Comes close, affordably: A Rotel RA-971 / NAD / Denon PMA mid-tier unit, ~0.005% THD (~80 dB SINAD), $400–600 → ~$900–1,300 today.
• 
  
• Verdict: The 1990s is the first decade where gear genuinely surpasses the WiiM on the bench — but the stuff that does is $8k+ adjusted. The affordable stuff is transparent but still ~5–10 dB behind on SINAD.

2000–2010 — the crossover begins​

Two things converge. Cheap, excellent op-amps and delta-sigma DACs make ~95 dB performance routine in mid-priced gear, and the first good Class-D modules (Hypex UcD/nCore, early ICEpower) arrive. By the late 2000s a Hypex-based amp hits ~100+ dB SINAD — clearly past the WiiM — for hundreds, not thousands.

• 
  
• Surpasses, mid-price: A 2008-era Hypex nCore / ICEpower-based amp, or a clean integrated like the NAD C375 / Rotel RB-1080. Measured SINAD ~90–100 dB. Price ~$1,000–2,500 → ~$1,500–3,800 today.
• 
  
• Surpasses, high-end: Benchmark AHB2 (arrived 2014, but its design lineage is here) and top Bryston/Mark Levinson gear push past 100+ dB. Multiple thousands adjusted.
• 
  
• Verdict: This is the decade where matching the WiiM's SINAD stops requiring exotic money — but it still costs 5–12× more than $299. Class-D is the technology that's about to erase the gap.

The 2010s → now — collapse​

The TPA3255 + PFFB (and Hypex/Purifi) make ~90–100 dB SINAD a commodity. The WiiM Amp, Fosi V3, Aiyima, etc. deliver in a $200–400 box what cost $8,000+ adjusted in the '90s and was literally unattainable at any price in the '70s. Performance flattened (we're at the audible-transparency ceiling and well past it); price did all the moving.

The compressed picture​

Decade	Best realistic SINAD	What it cost to match/beat WiiM (adjusted)	Closest representative
1970s	~70 dB	Nothing matches it	Pioneer SX-1980 (~$6k, still 19 dB behind)
1980s	~80–85 dB real (~95 spec)	~$2,500–3,000, approaches but trusts specs	Sony TA-F555ES, Technics SU-V8
1990s	~90–95 dB (high end)	~$8,000–16,000 to truly beat it	Accuphase E-407, Krell KAV-300i
2000–2010	~95–100 dB	~$1,500–3,800 to beat it	Hypex nCore amp, NAD C375
2010s–now	~100+ dB	~$200–400 (parity)	WiiM Amp / Fosi V3 itself

The inflection is clear: each decade the performance needed to match the WiiM got easier to find, but the price of that performance fell by roughly an order of magnitude per ~20 years — from "unattainable at any 1975 price" to "matched by a flagship costing 20× as much" to "beaten by a $300 commodity box."

 

[LTig]

There is a thread regarding the usage of AI at ASR - you should read it first, especially as a new member in his first posting.

Master AI (Artificial Intelligence) Discussion/News Thread

I don't need to tell you that AI is upon us big time. It is important that we all keep up with its developments, issues, etc. Instead of having multiple threads on it, I thought I create this single one so that people who are not interested, can just ignore it. And others know where to find...

www.audiosciencereview.com

 

[kyuu]

My advice is: if you don't know enough to fact check your LLM's output yourself, it's not something you should be using the LLM for.

 

[amirm]

The AI is giving reasonable answers for the questions asked. We basically have a bell shaped curve. Performance peaked around 1980s and then started to decline as costs went up. The ship righted itself around 2010 to 2015 with sharp acceleration in the last few years. A parallel universe still exists of mediocre measured performance with very high retail prices.

 

[Ruediger]

Now get some reality-based used market prices for those units - if available - and put in perspective if any of the gear from the 70s and 80s is still worth a buy today.

Some people don't mind the size, weight, efficiency, heat and prefer the looks.

 

[Mart68]

I have no idea if the Krell-era is correct, I heard they had bad sinad

KSA100 (1980) around 66dB SINAD

KSA50S (1994) is around 70dB SINAD.

 

[solderdude]

With music distortion levels of 0.1% are very difficult to detect.
At full power (100W@8ohm, 200W@4ohm, 400W@2ohm, 800W@1ohm) over a 20kHz BW distortion rises to 0.5%.
At those power levels the speakers will be distorting a lot more than that though.

So yes, distortion is rather high but BW is more than enough (1Hz - 150kHz -3dB) and can drive any load down to 1 ohm.

I would consider that 'audibly transparent' despite the (at this day an unacceptable) 'high' (but inaudible with music) distortion number.

 

[jsherman999]

My advice is: if you don't know enough to fact check your LLM's output yourself, it's not something you should be using the LLM for.

I've built entire systems around fact-checking LLM output, and one thing tha data tells us in spades is that human SMEs are still vital.

I wanted to see what humans in this forum thought of this output, because often the public facts don't match the zeitgeist, and the zeuitgeist can hold useful, hidden info the LLMs miss.

 

[Svend P]

There is a thread regarding the usage of AI at ASR - you should read it first, especially as a new member in his first posting.

Master AI (Artificial Intelligence) Discussion/News Thread

I don't need to tell you that AI is upon us big time. It is important that we all keep up with its developments, issues, etc. Instead of having multiple threads on it, I thought I create this single one so that people who are not interested, can just ignore it. And others know where to find...

www.audiosciencereview.com

I think you mean this thread

Official policy on use of AI

We have had recent issues come about regarding use of AI in member posts. I personally think this technology can be useful to us. And at any rate, hate having to police members left and right on what they post. That said, here are my thoughts: 1. If you are going to use AI, make it...

www.audiosciencereview.com

 

[DVinylfan]

There is a deep irony in using an LLM to produce an answer drawn from its training on the countless hours of committed research and writing of real people, who will never be credited, never compensated for their work, by whichever billionaire whom its sole purpose is to enrich, regardless of the environmental and human costs, and then come on a forum to ask other humans to check its work.

 

[LTig]

There is a deep irony in using an LLM to produce an answer drawn from its training on the countless hours of committed research and writing of real people, who will never be credited, never compensated for their work, by whichever billionaire whom its sole purpose is to enrich, regardless of the environmental and human costs, and then come on a forum to ask other humans to check its work.

I see a much higher chance WRT audio that the LLM has trained on 99% BS found in almost all audio related fora and audiophile brand ads and just mirrors the same BS back. Hence strict rules apply to not swamp ASR with overwhelming number of requests for correction.

 

[jsherman999]

The AI is giving reasonable answers for the questions asked. We basically have a bell shaped curve. Performance peaked around 1980s and then started to decline as costs went up. The ship righted itself around 2010 to 2015 with sharp acceleration in the last few years. A parallel universe still exists of mediocre measured performance with very high retail prices.

Quick napkin breakdown of the components suggests that the pricing structure back in the 1970s wasn't that dishonest, because to get big class A, A/B amp components that will put out big clean sound just simply would cost that much. Speakers have always looked fairly honest (I think,) you always have to pay to get physical quality and always will.

I see a much higher chance WRT audio that the LLM has trained on 99% BS found in almost all audio related fora and audiophile brand ads and just mirrors the same BS back. Hence strict rules apply to not swamp ASR with overwhelming number of requests for correction.

Ironic, since it's clear the response fully ingested and used ASR data

My usage of the LLM: No real surprises, but the real value was the scatter plot, which I think would look better as a 3-axis plot or the stacked rolling bar graph and with more data points, because it will show the 90s suddenly jumping up in SINAD while keeping the 80s/70s prices, other stuff

The one thing I was surprised about was Krell, I had always had this false impression that they were low-perf but huge and powerful

 

[EJ3]

I see a much higher chance WRT audio that the LLM has trained on 99% BS found in almost all audio related fora and audiophile brand ads and just mirrors the same BS back. Hence strict rules apply to not swamp ASR with overwhelming number of requests for correction.

AND the same BS answers based on 'how many think that something is correct" (because it seems that what is, in fact, incorrect: has a consensus) is in the automotive engineering field due to the LLM being trained on 99.9% BS. (Cylinder head port flow modification is an example that I will present [after Eric Weingartner thoroughly explains the test parameters used at 9:45 he flat out says that CHAT-GTP is wrong, the internet (consensus) is wrong, it's just wrong):