Moving to the edges

November 19, 2022
 by Paul McGowan

It might not make much sense when we see an amplifier's frequency range specified as 10Hz to 60kHz. After all, this kind of extreme frequency range is unusable to humans. Our hearing is limited to 20Hz and 20kHz (and that's on a good day).

But here's what might be missing. To be flat at the outer edges of what we can hear the amplifier must extend way beyond the target frequency range. One of the reasons this is true has to do with how specs work. When the industry specs an amplifier's frequency response they are using a measurement known as the -3dB point. For example, if an amplifier is specified by the general consumer electronics industry as having a frequency range of 20Hz to 20kHz, this typically means its response is down 3dB at those outer edges.

Probably not what you or I might be interested in.

Put another way, how excited are you to buy an amplifier who's performance includes a 3dB drop in volume at 20Hz and 20kHz? And to make matters worse, if the amp is down 3dB at 20kHz it is very likely a dB or so down at 10kHz.

This is just one good reason why we should be careful when looking at specs. This is also the reason we at PS Audio spec our products a bit differently than accepted industry standards. Take for example a BHK 300 amplifier. Its frequency response specs out like this: 10Hz – 20kHz +/- 0.1dB, 10Hz – 200kHz +0.1/-3.0dB.

Knowing clearly how your amp will perform within the range of human hearing is a lot more valuable than just simple specs.

Subscribe to Paul's Posts

28 comments on “Moving to the edges”

  1. Interesting to know…I never cared for specs so far. Looking at those of my preamp says 7Hz-400 kHz with no further details. So if those are the -3dB points I know now. It sounds as it sounds anyway 😉

  2. More recently I've seen some speakers specced to -6dB. Some of those Chinese amps are spec their power output at 10% THD rather than the generally accepted 1% THD.

    I suspect a lot of people have no idea what they can hear as they don't have their hearing tested. I have a few times and was not surprised. 20-20,000 went out the window a long time ago, if it was ever. It's not about having good day, you might as well say you can run 100m in 11 seconds on a good day.

    The other thing is that we have very little sense of tonality above about 5 kHz. Most acoustic instruments' fundamental frequencies don't go above 2 kHz and next to none are above 5 kHz. We do the vast majority of our critical listening in the midrange, where our hearing is most sensitive, which is why speakers that are hyper-accurate in the midrange are so pleasing and popular. The extremes of the sonic spectrum may give some people an audiophile rush of blood, I'm not sure they mean much to most people's enjoyment of recorded music and speech.

    Because we are relatively so sensitive in the midrange, especially with voice because that's how humans communicate, it is unsurprising how many engineers think listening to a recording of a familiar voice is the best start to any speaker evaluation. I go with that.

    1. Compared to the frequency graph of a loudspeaker,
      the FG of an amplifier is, by comparison, ruler flat.

      A whitewash in Melbourne on Tuesday & then 'The Ashes' in Blighty, 2023.

    2. I know the human voice example is a well spread theory among Hifi folks and maybe true, but I see the following circumstances which to me seem to make it rather less than better suited for evaluations:

      Only extremely few may be familiar with a live voice (under similar room acoustic dependencies) vs. its recorded version. Rather more folks will be familiar with the sound of a Steinway live vs. canned.

      So given, most, by “a familiar voice”, understand the recorded voice of a popular singer played back on whatever equipment, don’t have any hint how it should sound. They can only compare its playback by different speakers without knowing which is right.

      The only one who could make a meaningful comparison is a studio engineer, comparing the live performance with his recorded version on different speakers. But as his microphones already compromise the voice, he also doesn’t know, which speaker would be closest to the truth, if mics would have been perfect. He only knows which speakers are best, compensating the mic’s flaws.

      To me a lot of this kind of Hifi press theory is spiel. At the end we buy what sounds good to us. We shouldn’t try to base out taste on whatever theory in retrospect. There are very few who could make just a small part of meaningful comparisons for absolute judgements, if any.

      1. It’s a very good first step. Doing it with spoken voice is far preferable. I’ve done it and it is a simple and effective way of telling if a speaker is fundamentally bad, and quite a few are.

        It is not something I’ve read about in the audio press, it’s what I’ve heard from several speaker designers.

        1. Yes sure, I know…voice is good…but as I said only for those who know this voice live and not in a concert hall while comparing it to speakers in a living room, but also the live voice in that living room.

          IMO the live comparison check (not only for speakers) rarely makes much sense, as there’s too much in the chain before, that’s imperfect, so you never know which of those imperfections you’re compensating when choosing a speaker that sounds most real (even in case you know this voice live in your room). Even then you just choose the speaker that compensates best with your current gear and the one engineers equipment who recorded/processed the recording you chose for the test. Take a different recording and everything’s different again.

  3. There’s always the talk of hearing range and nominal numbers. There are always exceptions. I was one who’s hearing when I was in my 20’s went beyond 20k. I learned that thru the pain of hearing an IBM card sorter alignment that was being done where I happened to work at the time. (I don’t remember the exact freq, but it was above 20K but below 25k). My boss was an audio enthusiast and when he verified the frequency with the repair tech he made me leave. (On the condition I show up at his house later that evening for steaks, PBR, and an opinion on what I thought of the high’s on his sound system). Since he was my boss, the steaks, beer, and sound were great. (As I recall the steaks and sound were very acceptable)

    I’m kind of surprised a 12AU7 works that well that far out in frequency. But then again I haven’t dabbled with tubes in a long time.

    If your going to read specs, then you have to know enough about what your reading (what’s presented versus what’s not)

  4. I’ve known for most of my life that frequency response is defined at the extremes of the 3dB down points and yet it seems if a an Audio Enthusiast owns a quality piece of equipment e.g. a power amplifier and it sounds accurate to the human ear, there must be a reason for leaving the old definition in place.

    On the other hand, most of our community has equipment of high caliber so why you shouldn’t a quality audio manufacturer take pride like PS Audio does by listing the exact frequency response in a similar manner? I’m sure these manufacturers take pride in their final product so it doesn’t make sense to me that they wouldn’t list the true, extended frequency response which is a very important specification for most consumers. This makes no sense to me that the audio industry sticks to the original definitions which seem to be outdated.

  5. A little off topic - but still on amps.

    Nice write up in this months TAS regarding the S300’s. Congrats

    The “from the editor” in TAS the month was also eye opening

    1. Not a subscriber to TAS
      these days(daze)—— what was the gist of the S300 write up and “from the editor.”???

      I usually change the word specs to statistics In my mind. Paraphrasing Mark Twain -there’s specs, damned specs and then lies.

  6. Everything that Paul said in today's post is true. However, I want all of my gear to have a frequency range that goes well above 20 kHz. Why? Because all the sounds that we hear practically everyday ( including music ) have frequency content above 20 kHz and that extended frequency content determines what we hear. Now, how can that be? I am in my 70's ( the same as Paul ) and I know my hearing of fundamental tones goes out to about 12 kHz ( when I was in my 60's it went to 14 kHz, when I was in my 50's it went to 16 kHz, etc. ). Hearing fundamental tones is different than hearing complex sound signals.

    Would you like to hear a sound signal that has frequency content well above 20 kHz, Snap your fingers. That snap sound is a good way to create an impulse signal that has lots of frequency content above 20 kHz. All sound signals are made up of a series ( a series is a sum of a many simple items ) of fundamental sine waves. Each sine wave has a different frequency and amplitude. We hear these complex series of sine waves as a single sound ( like the finger snap, or a drum strike, or a cymbal strike, etc. ).

    So if you want finger snaps on your records to sound sharp and crisp and you want instruments in your music to be well delineated as to their position be sure that all of you gear has wide frequency bandwidth so that it does not filter out some of the frequencies that make the sound complete.

    1. So that’s means everything in the recording process from the microphone and on to the speakers have to have the ability to reproduce complex sound signals. So why the fixation with mostly amp specs for bandwidth?

      Here’s the BHK pre frequency specs. -3dB 0.1Hz – 200kHz

      If it’s down 3db at 200KHz and feeding the BHK 300’s that are down 3db you do the math.
      There bottom end is even more telling…. .1 vs 10Hz

      Or better yet answer at what point do higher frequencies no longer matter?

      Is it just a chest thumping spec - or is .5MHz -3db a better sound?

      1. Mike, It is a bit strange the way it works. I heard a vinyl record that had the a singer snapping his fingers to the beat. The record was being digitized using a 192 kS.p.s. sampling rate and they were running a spectrum analyzer in real time. Every time the singer snapped his fingers the spectrum analyzer had a frequency spike out to 96 kHz. The vinyl record came from an analog RTR tape deck. Now it would take a page or two for me to explain to you that it is physically impossible to record a 96 kHz fundamental tone on any magnetic tape. I would have to magnetic domains, particle size, and how quickly you can flip the domain and stuff like that. Yet, there was 96 kHz frequency content on the record.

        People get all wrapped up in this 20 kHz fundamental tone hearing limit. They should focus on what is the highest frequency content you ewant to allow. I would say a bandwidth that is "flat" out to 100 kHz is a good idea.

        1. Where I was going with this Tony….

          Look at the DSDAC MK2 Frequency response Spec’s…. 20-20KHz +/- .25db

          Obviously if music or other sounds have very high order harmonics that result in frequencies well beyond the ‘human range of hearing’ to create timbre and allow us differentiate a tone from a note, a middle c from a piano versus a middle C from a trumpet and so on. Then why isn’t everything spec’d to the same standards?

          Because that’s the way it’s always been done? That’s not an acceptable answer in the PSA recording world anymore. ✌️

          1. Mike, Sorry for not getting back sooner. What you are asking for is a control board. They have these in France, Italy and California for wine, cheese, olive oil and a few other thinks. Amongst the things that a control board does is tell makers of a particular item how it must be labelled. They have the legal power to do this. In California if you label a red wine Cabernet Sauvignon it must contain at least 70% Cabernet Sauvignon wine juice.

            You would like to have an Audio Control board that would have the legal power to tell people like Paul how he must specify his products. I personally would also like that but I doubt that it will ever happen.

            1. No problem Tony. I’m not big on regulation or control boards, but sometimes they are needed.

              Common sense, complete honesty and consistency by any given manufacturer should be the norm.

              Like you, I remember all the buzz about power ratings back in the day. Music, peak, peak to peak, rms and what ever else some entity came up with. So then some minimal standards had to be set. It’s also up to a purchaser to research and educate themselves to make informed decisions.

  7. I’m in the camp that believes that even though we may not “hear” frequencies around and above 20K a tweeter that rings 10-20db in the range from 15K on up can make for a very fatiguing speaker even if you can’t pinpoint why. It’s also why I don’t use metal diaphragm compression drivers in my designs, they sound harsh to me.

    1. I live about 150 yards from a church that belts out bell music three times day over PA speakers. I have no idea what kind of diaphragm they have in thier compression drivers, but they sure are harsh!

    1. Oh, thanks for the link. But I had most interesting correspondences with Ralph Glasgal and Prof. Choueiri much earlier! 🙂 Thus these findings of the reviewer of TAS aren’t new for me.

    1. Thanks for posting that link Russ.
      I’m 76 years and at full volume I can still (barely) hear a 12 kHz sine wave tone in my left ear but my right can’t hear above 11 kHz. This was at full volume. An enlightening experience.

  8. Correct Tony! Music/tones/sounds incorporate odd and even harmonics that extend several octaves above the fundamental pitch. Timbre and tonality recognition of that pitch's harmonic sonic signature gives the listener the aural clues to tell us what we are listening to!

    Even with limited high frequency hearing beyond the fundamental tone, the first 3 to 4 harmonics (covering 2 octaves above fundamental) greatly assist the listener in identifying the specific sound source. The tonality of a a 2K fundamental pitch can be easily recognizable, as these first few harmonics (commonly called overtones in music) extend up to (and beyond) 8K Hz.

    Beyond limitations of your hearing capability, the additional overtones that exist in sounds are recognized in psychoacoustics as perceivable through the physiology of bone conduction (sounds bypassing the eardrum and being perceived with a functioning cochlea)!

    Like Tony, although my upper hearing limits are around 14K Hz (69 years young), I still want a Full Bandwidth Music presentation from my home components and room acoustics!

  9. Once again, measurements don’t tell the whole story. Hearing is believing. Sit back, relax and close your eyes. Let the music wash over you. That is how I learn and critically evaluate things.
    Graphs and charts give a bit of an idea when it comes down to tuning preferences. I can take some stock into that. 🙂

  10. Just because one might not be able to hear a 16Hz or 20Hz pure tone does not mean one cannot hear the effect of high frequency tones mixed with lower frequency tones. The superposition principle of sound waves dictates that high frequency waves interact with lower frequency waves, changing the resultant wave forms. Therefore one hears the effect of high frequencies without hearing the individual frequencies themselves. This is a reality of nature.

Leave a Reply

© 2022 PS Audio, Inc.

linkedin facebook pinterest youtube rss twitter instagram facebook-blank rss-blank linkedin-blank pinterest youtube twitter instagram