PS Audio M700 Monoblock Amplifier Review

[GaryMnz]

Nope. Search here for a thread about intersample overs by @ayane.

https://benchmarkmedia.com/blogs/application_notes/intersample-overs-in-cd-recordings

0.8dB extra? I think we can neglect that as a red herring in the argument.

 

[GaryMnz]

Just mark in red in my posting where I was wrong, please.

Pretty much all of it, the crest factor stuff is irrelevant as the methodology accounts for it.

 

[pjug]

Ah, I see. This is what Gary isn't addressing.

Since the noise floor is spread across the frequency spectrum, the value at one frequency dominating in the region where our ear is perhaps less sensitive or the music content is absent in that region does not require the lows at another frequency to be above it to hear it. And it does not have to be simple additive depending on our ear sensitivity to specific frequencies.

So if I my noise floor is dominated by an earthquake rumble (which is known to happen in my parts) in low frequencies, I can happily ignore it without having to raise my music to enjoy it ... unless my music also has an earthquake rumble in it as part of the music. But if my floor noise is from a chainsaw murder going on nearby....

I can see both sides (not to say they are equivalent).

I do think floor noise could be in the region that interferes with music content but may not just be a linear addition on top of the noise floor depending on frequency, so not sure what the right resolution from a measurement perspective to this conflicting goals is. Will leave it to the experts to hash it out.

Here is an idea for how you can get an idea of what dynamic range you need with an easy to do exercise in Audacity. Anyone is welcome to point out flaws in this. It won't surprise me if this is all wrong.
1. Open some track of some music
2. If it is 16 bit convert to 24 bit or 32 bit
3. Find a lively part and start a select a section starting in the middle of this
4. Use amplify tool to attenuate by something (suggest starting with -80dB or -85dB). My version of Audacity requires doing this in two steps, so I did -50dB and -35dB
5. Adjust volume of the louder beginning of the track to as loud as you can stand
6. See if you can hear the attenuated part

With headphones I can hear attenuation of about -85dB to -90dB depending on how loud I've set the volume. This was jazz so maybe I could do better with something else. Likewise it would be worse with more background noise listening with speakers. But my conclusion is that Amir's 96dB does not seem to be extreme overkill.

 

[GaryMnz]

Here is an idea for how you can get an idea of what dynamic range you need with an easy to do exercise in Audacity. Anyone is welcome to point out flaws in this. It won't surprise me if this is all wrong.
1. Open some track of some music
2. If it is 16 bit convert to 24 bit or 32 bit
3. Find a lively part and start a select a section starting in the middle of this
4. Use amplify tool to attenuate by something (suggest starting with -80dB or -85dB). My version of Audacity requires doing this in two steps, so I did -50dB and -35dB
5. Adjust volume of the louder beginning of the track to as loud as you can stand
6. See if you can hear the attenuated part

With headphones I can hear attenuation of about -85dB to -90dB depending on how loud I've set the volume. This was jazz so maybe I could do better with something else. Likewise it would be worse with more background noise listening with speakers. But my conclusion is that Amir's 96dB does not seem to be extreme overkill.

Nothing wrong with that but point us to a recording with this extreme dynamic range. I would give up and go away for a beer if I was made to listen to a piece of music that at one point hurt my ears and at the other required a silent room and me straining to hear. Listening to music is supposed to be a relaxing past time.

 

[LTig]

Pretty much all of it, the crest factor stuff is irrelevant as the methodology accounts for it.

I'd like to disagree. Let me do a few statements and tell me please what you think of them:

• The methodology assumes a fixed crest factor of 12 dB (factor 4). Correct? If you don't agree please show the correct value and its calculation. I like to learn from my mistakes.[EDIT: the factor is 16, we're talking power, not voltage].
• Different types of music do have different crest factors. Correct?
• The methodology therefore does not account for all types of music. Correct? If not please explain. As I wrote I like to learn from my mistakes.

 

[LTig]

https://benchmarkmedia.com/blogs/application_notes/intersample-overs-in-cd-recordings

0.8dB extra? I think we can neglect that as a red herring in the argument.

Make it ten times. AFAIR @ayane measured up to 8 dB. You have to search here, not at benchmarks site.

EDIT: this thread

 

[GaryMnz]

I'd like to disagree. Let me do a few statements and tell me please what you think of them:

• The methodology assumes a fixed crest factor of 12 dB (factor 4). Correct? If you don't agree please show the correct value and its calculation. I like to learn from my mistakes.
• Different types of music do have different crest factors. Correct?
• The methodology therefore does not account for all types of music. Correct? If not please explain. As I wrote I like to learn from my mistakes.

You're still missing the point. The test is done with respect to peak digital. Hence the upper limit is defined. Crest factors are accounted for in the piece of music you chose to play to set your volume control. You can use different pieces of music and will get a different number, but all are relevant in your system. Best perhaps to chose the piece with the highest dynamic range and go from there.

 

[Blumlein 88]

https://www.audiosciencereview.com/...-from-recording-to-listening-room.8205/page-2

Maybe just move this discussion to an old thread I started. Is 120 db dynamic range from recording to listening room possible?

 

[GaryMnz]

Make it ten times. AFAIR @ayane measured up to 8 dB. You have to search here, not at benchmarks site.

Even that is an anomaly, an aberration. And its not in the recording, it is a result of misbehavior in the DAC or digital filter. No point in basing our research on faulty technology.

 

[Blumlein 88]

Here is an idea for how you can get an idea of what dynamic range you need with an easy to do exercise in Audacity. Anyone is welcome to point out flaws in this. It won't surprise me if this is all wrong.
1. Open some track of some music
2. If it is 16 bit convert to 24 bit or 32 bit
3. Find a lively part and start a select a section starting in the middle of this
4. Use amplify tool to attenuate by something (suggest starting with -80dB or -85dB). My version of Audacity requires doing this in two steps, so I did -50dB and -35dB
5. Adjust volume of the louder beginning of the track to as loud as you can stand
6. See if you can hear the attenuated part

With headphones I can hear attenuation of about -85dB to -90dB depending on how loud I've set the volume. This was jazz so maybe I could do better with something else. Likewise it would be worse with more background noise listening with speakers. But my conclusion is that Amir's 96dB does not seem to be extreme overkill.

I can't find it at the moment, but I had a thread where I mixed pink noise in with music at various levels asking people to say when it disappeared over speakers at their normal listening levels. The most common answer was -70 db. With listening levels of 75-80 db SPL average and the profile of pink noise that seems fairly reasonable. Headphones are going to go deeper on this.

 

[GaryMnz]

https://www.audiosciencereview.com/...-from-recording-to-listening-room.8205/page-2

Maybe just move this discussion to an old thread I started. Is 120 db dynamic range from recording to listening room possible?

It probably warrants a new thread as there is much more to be mined here. But I think I'm about done...

 

[LTig]

You're still missing the point. The test is done with respect to peak digital. Hence the upper limit is defined. Crest factors are accounted for in the piece of music you chose to play to set your volume control. You can use different pieces of music and will get a different number, but all are relevant in your system. Best perhaps to chose the piece with the highest dynamic range and go from there.

Yes, it seems our communication is very bad. I miss a point and you cannot answer questions. Time for bed.

 

[GaryMnz]

Yes, it seems our communication is very bad. I miss a point and you cannot answer questions. Time for bed.

I answered your question, very specifically.

 

[Hugo9000]

I can't find it at the moment, but I had a thread where I mixed pink noise in with music at various levels asking people to say when it disappeared over speakers at their normal listening levels. The most common answer was -70 db. With listening levels of 75-80 db SPL average and the profile of pink noise that seems fairly reasonable. Headphones are going to go deeper on this.

There is this thread, where you used white noise:

https://www.audiosciencereview.com/...hat-level-is-noise-heard-in-your-system.1013/

Is that the one you're thinking of?

Unless it was something you posted on that other forum where some of us used to hang out. (Or other other forum, for some of you haha!)

 

[Blumlein 88]

I'd like to disagree. Let me do a few statements and tell me please what you think of them:

• The methodology assumes a fixed crest factor of 12 dB (factor 4). Correct? If you don't agree please show the correct value and its calculation. I like to learn from my mistakes.

No he is using -12 db to prevent damaging equipment with the test tones. He could use -20 db or -30db or whatever. In this case the level of the test tone isn't related to any crest factor. The crest factor doesn't enter into choosing this level.

Now you get to pick which loudest piece of music you use for a reference, and of course crest factors vary. And this isn't meant to be a measure accurate to the last percent. Just one to get you a good general estimate based upon your music, your gear and how loud you like to listen.

• Different types of music do have different crest factors. Correct?

Yes they do, but it doesn't effect the test done this way.

• The methodology therefore does not account for all types of music. Correct? If not please explain. As I wrote I like to learn from my mistakes.

The methodology isn't trying to account for all music, just the music you use yourself. The answer you get for such a test might be a little different than the one I would get if I came over to your house, picked my music, and set the loudest level I would wish to listen. The result would be valid for me with your gear. It isn't an answer all over the world. It wouldn't even be necessarily valid for you with your gear until you do the test for yourself.

 

[Blumlein 88]

There is this thread, where you used white noise:

https://www.audiosciencereview.com/...hat-level-is-noise-heard-in-your-system.1013/

Is that the one you're thinking of?

Unless it was something you posted on that other forum where some of us used to hang out. (Or other other forum, for some of you haha!)

That is it, and thank you for finding it.

I mis-remembered it as pink noise.

It was white noise filtered below and above 3000 hz and 5000 hz.

 

[DonH56]

Contributing to this train wreck/pissing contest seems pointless, but a couple of things I consider in system design:

1. Max SPL desired is a function of speakers, distance to them, and amplifier power. And how much we as listeners can stand (I tend to listen at lower levels than current mix standard of ~80 dB). That usually sets the upper bound on amplifier power. Average to peak levels in music was measured ages ago at around 17 dB, a power factor of 50 (I'd have to dig to find the AES paper if I even have it any more), and more recently extended dynamic range recordings have hit 20 dB (factor of 100 in power) or so. Movies even higher. How often you find recordings that good I could not say. A couple of CDs I looked at a some years back would have peaks on that order (some Sheffield recordings, don't recall exactly which ones, looking at the analog output of my CD player). I have said before that IME people tend to overestimate their average power and underestimate the peaks.

2. The noise floor that bounds the other (low SPL) end of dynamic range depends upon the room as well as the electronics. I don't remember my room's noise floor off-hand, think it was around 20-30 dB SPL but I have not looked at it in a while. My room includes floating walls and ceiling and a minisplit to isolate it from the ouse HVAC. Floor is concrete (basement). It is about as good as most recording studios. Mic self-noise is often in the 20-30 dB range referenced to standard SPL.

We can pull signals out below the noise floor (out brain can process the signal from the noise). Too loud and fatigue if not actual pain occurs. With my speakers, amp, room, and listening position I can achieve 107+ dB SPL at the listening position, giving me a dynamic range around 87 dB plus whatever I can hear below the room noise, plenty enough for me. However, I've friends who can achieve 120+ dB peaks in similarly quiet rooms so 97+ dB dynamic range. I do not hear any hiss at the listening position so the room is setting the noise floor and not the electronics in my case.

As Amir mentioned, the OSHA standards aren't really relevant, as they are for specific frequency bands and long-term exposure to high average levels. OTOH, they are also misused as a "safe" upper bound, since they are meant to serve as guidance for being able to hear conversations and such and not the nuances of music. There is also the issue of deep bass, where we are much less sensitive, and it takes much more power (20-30 dB or more, 100x to 1000x in power) to sound as loud as sounds in the midrange.

What all this has to do with the M700, I don't know, and this thread is essentially useless as a review past the first page or so.

 

[richard12511]

Gun shots? I'm talking about music, at normal domestic listening levels.

Music can have gun shots(or cannon shots) . In fact, the loudest I've ever seen from my main system was ~120db at the end of the 1812 Overture when the cannons are going off. Not sure how much power that took, but I'm guessing it was quite a bit. Measurement was done at ~4m at the mlp.

 

[DonH56]

It's been a while since I measured on a 'scope but percussives can easily hit 20-30 dB above average for things like drum strikes and such. That is using an oscilloscope to measure the instantaneous peaks, not what Vu meter would show.

 

[USER]

I am not an engineer or mathematician, so I have to ask: was this all about calculating something in a way that Amir doesn't care to calculate for already considered reasons? Also, is the argument over a (dynamic range) threshold that engineering-wise has been surpassed by pretty much any state-of-the-art amplifier? To try to think about how this relates to the thread, are we simply trying to not hurt the feelings of an amplifier that uses someone else's modules when considering the field at play today?

I know I'm dumb, but...