Marantz PM-90 Review (Vintage Amplifier)

[GXAlan]

I didn't mean to compare the Marantz amp to an AVR really, just making the point that even just on ASR's list there are quite a few modern amps that measured better than the semi vintage (semi because its made in the 90's). By the way, just a side note, the THD+N vs freq curves are great but I wouldn't worry about the curve for above 10 kHz because even the 2nd harmonics is at 20 kHz, human's limit already. Some may not mind the 2nd harmonics either.

Agree 100%. I secretly was hoping that this unit would beat the Parasound JC2 and PS Audio M700 when I sent it to Amir (at lower power).

One of these days, once the AKM factory is up and running again and the chip shortage is reduced, the QuantAsylum test gear may be available again. It might be interesting to compare a basic AVR across all sorts of frequencies not just a handful of tests compared to a traditional 2-ch amp across all frequencies.

 

[bigguyca]

I didn't mean to compare the Marantz amp to an AVR really, just making the point that even just on ASR's list there are quite a few modern amps that measured better than the semi vintage (semi because its made in the 90's). By the way, just a side note, the THD+N vs freq curves are great but I wouldn't worry about the curve for above 10 kHz because even the 2nd harmonics is at 20 kHz, human's limit already. Some may not mind the 2nd harmonics either.

Omitting the measurements above 10kHz seems a really bad idea. It seems likely that the measurements are made automatically or with little added work on Amir's part. Amir can certainly comment if this statement is far from accurate. From the members greedy viewpoint, we seem unlikely to get added reviews, and there are never enough, from leaving out these measurements.

The 15kHz and 20kHz measurement help to characterize the design of the amplifier. The 15kHz measurement likely includes two harmonics and the 20kHz measurement likely includes one harmonic. Most of the power amplifiers measured on ASR, at least the linear amplifiers, include capacitor compensation the rolls off gain at 6dB per each increase octave. If distortion is similar for the 2nd and 3rd harmonics then the 15kHz and 20kHz measurements will lie close to each other. Two likely lower harmonics at 15kHz, one higher harmonic at 20kHz. If the 20kHz measurement climbs to far above the 15kHz for example, something odd is happening. Who knows what, but the measurement does provide added information.

Measurements in Stereophile extend to 20kHz, but appear to be weighted or just don't add frequencies above about 30kHz. This typically results in lower, or at least not rising distortion shown at higher frequencies, which is misleading. Gene at Audioholics uses A-weighting for the few measurements relevant measurements he makes, so these measurement don't come close to accurately characterizing distortion at 15kHz and above.

Many loudspeakers have had tweeters, especially metal dome tweeters, with large (20dB?), high Q resonances at over 20KHz. Even low level signals above 20kHz may energize these resonances with intermodulation effects at lower, and more audible, frequencies.

Then there are always power amplifiers with interesting performance characteristics. The Emotiva XPA Gen3 series is an excellent example. John Atkinson's measurements in Stereophile found rather "interesting" performance a frequencies starting over 10kHz. These sort of performance issues may not be of interest to you, but they seem worth discovering. See especially graphs 6, 9 and 10.

https://www.stereophile.com/content/emotiva-xpa-gen3-two-channel-power-amplifier-measurements

 

[GXAlan]

Many loudspeakers have had tweeters, especially metal dome tweeters, with large (20dB?), high Q resonances at over 20KHz. Even low level signals above 20kHz may energize these resonances with intermodulation effects at lower, and more audible, frequencies.

This is an interesting comment. People often talk about the sound of materials but maybe it is this high q resonance that leads to preferred sound or not. Do you have a speaker that is infamously bad? Would be interesting to throw it on the Klippel with a standard signal and one that is juiced up with harmonics above 20kHz.

 

[dualazmak]

...
Many loudspeakers have had tweeters, especially metal dome tweeters, with large (20dB?), high Q resonances at over 20KHz. Even low level signals above 20kHz may energize these resonances with intermodulation effects at lower, and more audible, frequencies.
...

This is an interesting comment. People often talk about the sound of materials but maybe it is this high q resonance that leads to preferred sound or not. Do you have a speaker that is infamously bad? Would be interesting to throw it on the Klippel with a standard signal and one that is juiced up with harmonics above 20kHz.

Even though my point below would be somewhat out of the scope which you two are discussing now, but just for your reference....

The issue of ultra high frequency (UHF) noises (in above ca. 30 kHz) in many of the improperly processed HiRes music sources have been recently intensively discussed by @amirm and many people here in ASR Forum;
https://www.audiosciencereview.com/...-high-res-music-antonio-forcione-video.23083/
https://www.audiosciencereview.com/...-live-from-minster-from-the-lake-poets.23095/
https://www.audiosciencereview.com/...pcm-vs-mqa-vs-cd-2l-sampler-comparison.23172/
https://www.audiosciencereview.com/...son-linn-records-free-high-res-samples.23366/

The issue has been also discussed around 2015 in Japan, like in this page even though in Japanese;
https://sandalaudio.blogspot.com/2015/09/blog-post_17.html
I hope your web browser would properly translate it into English.
Where, "Niserezo (偽レゾ、ニセレゾ)" means "sham HiRes".

My present understanding is such UHF noises would be possibly harmful for our highly sensitive tweeters and super tweeters. As share in my post here and here, I decided to include, therefore, low-pass (high-cut) -48 dB/Oct filters at 25 kHz cutting-off these UHF noises in my multichannel configurations in order to protect my Be-midrange drivers, Be-tweeters and metal horn super tweeters.

You wrote that "maybe it is this high q resonance that leads to preferred sound or not."

As far as based on my intensive subjective listening comparisons in my rather highly sensitive and highly efficient drivers directly and dedicatedly driven by amplifiers, I hear no audible difference between with and without the low-pass (high-cut) -48 dB/Oct filters at 25 kHz. Consequently, I feel such garbage UHF noises should be avoided in any way.

 

[amirm]

Measurements in Stereophile extend to 20kHz, but appear to be weighted or just don't add frequencies above about 30kHz. This typically results in lower, or at least not rising distortion shown at higher frequencies, which is misleading.

It is for this precise reason that I no longer show the 20 kHz THD+N. I actually run it but most of the time it incorrectly shows lower distortion than lower frequencies due to limited bandwidth of 45 kHz I use for measurements. If I up that to higher bandwidth, Class Ds with noise shaping show massive amount of "distortion" even though it is not.

 

[Xyrium]

Wow, I love this thing. Of course, @restorer-john posts the listing for the 95 and man....I mean....man....great stuff.

Of course, I love Accuphase slightly more, especially with a nice big pair of meters on the front, but apples to oranges in price.

 

[GXAlan]

Wow, I love this thing. Of course, @restorer-john posts the listing for the 95 and man....I mean....man....great stuff.

Of course, I love Accuphase slightly more, especially with a nice big pair of meters on the front, but apples to oranges in price.

I have also offered Amir my Accuphase P-266 for measurement. Big challenge is not overworking his back and there is a pretty big backlog.

 

[Xyrium]

I have also offered Amir my Accuphase P-266 for measurement. Big challenge is not overworking his back and there is a pretty big backlog.

Oh my, that is one amazing beast! Very nice sir!

 

[respice finem]

It is for this precise reason that I no longer show the 20 kHz THD+N. I actually run it but most of the time it incorrectly shows lower distortion than lower frequencies due to limited bandwidth of 45 kHz I use for measurements. If I up that to higher bandwidth, Class Ds with noise shaping show massive amount of "distortion" even though it is not.

The takeaway for me: Having frequencies in the audio signal that no one has a chance to hear is not only nonsense, but potentially harmful (and it annoys cats, dogs and birds)

 

[dualazmak]

I have also offered Amir my Accuphase P-266 for measurement. Big challenge is not overworking his back and there is a pretty big backlog.

Really nice; I am much looking forward to hearing the measurement results for Accuphase P-266, rather vintage product of 1983.

we can find P-266 brochure here, in Japanese though, which includes interesting data on intermodulation distortion.
The two figures and my (poor?) English translation are as follows;

The figure above shows the spectrum of IM (intermodulation) distortion by the new IHF measurement method. The 19 kHz and 20 kHz are input signals, and when IM is generated, the distortion component would appear at 1 kHz (20 - 19 = 1 kHz) of the difference. In this data, at least the distortion is -93dB (0.0022%) or less. Another IM may appear at 19 + 20 = 39 kHz, but even if it exists, it is out of the audible band and there is no problem. With this unit, the 39 kHz IM is also below -93 dB.

The figure above is the spectrum of TIM (Transient Intermodulation) distortion. Here, 3.18 KHz square wave and 15 kHz sine wave are mixed for input. Since the square wave contains odd-order harmonics, which is close to infinity, its components are 9.54 kHz (3rd order), 15.9 kHz (5th order) , and so on, to appear. When these and 15 kHz sine wave generate intermodulation distortions, they appear where there is no input signal component. For example, when the 3rd 9.54 kHz (generated by 3.18 kHz square) and 15 kHz sine would interact, the distortion component appears at 15 - 9.54 = 5.46 kHz. In the above figure, no distortion seems to be seen up to -93 dB. In other words, it is 0.0022% or less.

 

[GXAlan]

Really nice; I am much looking forward to hearing the measurement results for Accuphase P-266, rather vintage product of 1983.

The P-266 has better advertised measurements than all of the other Accuphase amplifiers save for the P-102. Afterwards, I do not know if the products are “worse” as availability of transistors changed, if the testing was more stringent, or if the trade off in “audible improvement” translated into different distortion performance. That is, newer Accuphase amps keep driving the residual noise down in terms of V but they don’t change the advertisement. In contrast, newer Accuphase amps no longer include the measurements.

The current E-270 which was tested had a poor preamp section even though the amp itself was one of the best Class AB amps.

It’s packaged and ready to ship but Amir has a house full of boxes and a long backlog of things to test.

 

[peng]

This is an interesting comment. People often talk about the sound of materials but maybe it is this high q resonance that leads to preferred sound or not. Do you have a speaker that is infamously bad? Would be interesting to throw it on the Klippel with a standard signal and one that is juiced up with harmonics above 20kHz.

I know this is a different topic, but just to be clear, I would never ask the curves for above 10,000 Hz test signal to be "omitted.." never said that either. It bigguyca interpreted it that way, okay, that's him. My point of "I wouldn't worry.." refer to the results that may indicate higher total distortions because of the inclusion of the harmonics at frequencies higher than 20,000 Hz. For example, if amp A's THD for the test signal from 1 kHz to 9.5 kHz is -100 dB but from 1 KHz to 10 kHz is -80 dB, vs amp B's THD for up to 10 KHz (test signal) is -90 to -95 dB, then I would prefer amp A, all else being equal because I would consider the measurements for amp A is more relevant for the application.

Again, I do like those THD+N vs frequency curves to include up to even 20 kHz that Amir has been doing, just pointing out the fact that most people, especially the older ones, cannot hear frequencies above 20,000 Hz period, let alone at very low level. As for the potentially associated side effects due to the 20,000 Hz and above harmonic frequencies at low levels (such as at -50 dB), then yes I would worry about it if such effects are there, but such effects should be measurable and show up in Amir's battery of tests.

 

[tomelex]

The Variac that you show is a nicer version than the green ones that organic chemists use in the lab for refluxes and solvent distillation.

if those green ones are fairly old and made in america, you will find they have much thicker copper windings and larger copper winding stacks, stuff from China is no comparison. Folks would be willing to pay pretty good money for your |"old" stuff. Typical company name on them is Variac

 

[tomchr]

It is for this precise reason that I no longer show the 20 kHz THD+N. I actually run it but most of the time it incorrectly shows lower distortion than lower frequencies due to limited bandwidth of 45 kHz I use for measurements. If I up that to higher bandwidth, Class Ds with noise shaping show massive amount of "distortion" even though it is not.

I suggest bumping that up to 60 kHz so you get the third harmonic of 20 kHz. 80 kHz is another common measurement bandwidth. The main drawback is that you include a lot more noise - and also that you measure stuff that only dogs and audiophiles can hear.

What's going on with the timeouts in the 20 Hz THD+N measurement? Does that clean up if you bump up the time the AP waits for the measurement to settle or tweak the settling parameters?

It's a bit unfortunate with the mains hum. It would be interesting with a comparison between THD+N measured within the full audio bandwidth and with a 950 Hz elliptical high-pass filter enabled. That'll help tease out how much of the THD+N is hum and how much is actual THD.

Hum aside, it looks like a quite capable amp.

Tom

 

[dualazmak]

if those green ones are fairly old and made in america, you will find they have much thicker copper windings and larger copper winding stacks, stuff from China is no comparison. Folks would be willing to pay pretty good money for your |"old" stuff. Typical company name on them is Variac

I am very much familiar with these "variac", variable AC autotransformer, since once I have been doing intensive research at organic synthesis chemistry laboratory. In Japan, "variac"s are usually called as "slidac"s representing "sliding adjustable AC autotransformer", and many of them can be easily purchased, for example;

スライダック 単相の人気商品・通販・価格比較 - 価格.com

"スライダック 単相"の人気商品をご紹介！価格.comなら今売れている商品やお得な商品を全国のネット通販ショップから探すことができます。価格情報や人気ランキング、商品をおすすめするクチコミ・レビューなど豊富な情報を掲載。"スライダック 単相"を比較・検討するなら価格.comへ。

kakaku.com

Actually, I have one of these at my home, Tokyo-Rikosha Co., Ltd., "RSA-1", input 100 V 50/60 Hz, output 0 V - 130 V, 0.1 KVA (100 W), 2.4 kg. Using this, recently I compared 100 V and 110 V (in spec.) operation (50 Hz) of my Okto DAC8PRO, and found there is no problem at all in 100 V 50 Hz (I am in east part of Japan!) operation confirming Pavel of OKTO has kindly informed me before my purchase order.

 

[amirm]

I suggest bumping that up to 60 kHz so you get the third harmonic of 20 kHz. 80 kHz is another common measurement bandwidth. The main drawback is that you include a lot more noise - and also that you measure stuff that only dogs and audiophiles can hear.

As I mentioned, increasing the bandwidth really messes up the Class D amp measurements. 45 kHz was the highest I could go at the time before causing this problem. FYI DAC measurements use 90 kHz bandwidth and noise shaping causes problem there at times as well but not as severe.

 

[dualazmak]

The P-266 has better advertised measurements than all of the other Accuphase amplifiers save for the P-102. Afterwards, I do not know if the products are “worse” as availability of transistors changed, if the testing was more stringent, or if the trade off in “audible improvement” translated into different distortion performance. That is, newer Accuphase amps keep driving the residual noise down in terms of V but they don’t change the advertisement. In contrast, newer Accuphase amps no longer include the measurements.

The current E-270 which was tested had a poor preamp section even though the amp itself was one of the best Class AB amps.

It’s packaged and ready to ship but Amir has a house full of boxes and a long backlog of things to test.

Thank you indeed for your nice follow-up which furthermore elevating my expectations for Amir's measurements on P-266.

BTW, I really hope that someone (or Accuphase itself?) would send the latest Accuphase integrated amps, huge pure class-A E-800 (50W/8Ω, 100W/4Ω, 200W/2Ω; 36.0 kg!) and class-AB E-480 (180 W/ch, 24.6 kg) to Amir for measurements... Of course some of us need to help him to bring them upstairs.

 

[GXAlan]

BTW, I really hope that someone (or Accuphase itself?) would send the latest Accuphase integrated amps, huge pure class-A E-800 (50W/8Ω, 100W/4Ω, 200W/2Ω; 36.0 kg!) and class-AB E-480 (180 W/ch, 24.6 kg) to Amir for measurements... Of course some of us need to help him to bring them upstairs.

The E-270 didn’t do that well, which is a shame, and I don’t think the US sales teams are interested anymore. The M-2000 measured well at Stereophile.

The bad thing about Accuphase, which could be said about any company is that they are really strict about tiering their products. Looking at the digital line, the es9026pro is in the 450, the 9028pro is in the 570 and I think a pair of 9028pros for the 750.

The ES9026Pro is $30 in quantities of 1
The ES9028Pro is $60 in quantities of 1

Early on they had fewer tiers. At the prices they are asking, it’s really hard to justify purchasing them new in the US (esp given the premium we pay over domestic pricing).

Meanwhile, even though Denon as a mass market retailer has a lot of tiers, you get very good performance at the X3700H and up level. Paradoxically, we don’t get the PMA-SX line here in the US

 

[Doodski]

Thank you indeed for your nice follow-up which furthermore elevating my expectations for Amir's measurements on P-266.

BTW, I really hope that someone (or Accuphase itself?) would send the latest Accuphase integrated amps, huge pure class-A E-800 (50W/8Ω, 100W/4Ω, 200W/2Ω; 36.0 kg!) and class-AB E-480 (180 W/ch, 24.6 kg) to Amir for measurements... Of course some of us need to help him to bring them upstairs.

Wowowow. It looks like a gold plated circuit PCB & rated 300W into 1 Ohm load with music. Actual power output when tested is 85WRMS@8R. Retails @ USD $13,000.00 I also saw it in Canada for $21,000.00 and Euro expense is $13,700.00 The prices are up and down and it seems to be at the discretion of the importer.
Volume control PCB>

Volume Control>

 

[tomchr]

As I mentioned, increasing the bandwidth really messes up the Class D amp measurements. 45 kHz was the highest I could go at the time before causing this problem. FYI DAC measurements use 90 kHz bandwidth and noise shaping causes problem there at times as well but not as severe.

But this is a Class A/AB amp...

I'm not suggesting that you have different testing protocols for Class A/AB/G/H vs Class D. But my curiosity drives me to want to scratch the surface a bit deeper on this amp.

Tom