Dr. Klaus Heinz of HEDD Audio (ex ADAM Audio) - measuring speakers, in particular speaker dynamics

[Krunok]

My claim on high sensitivity-high dynamic with low distortion is what you often counter among very experienced DIY speaker builders. So it’s hearsay, but not of the naive audiophool kind.

Why is it that some speaker designers prefer high sensitivity drivers?

Such claim is very easy to be proved or disproved by measurements. The point is you made that claim as it is something that is proved, not hearsay. If you heard that from "very experienced DIY speaker builders" which also didn't provide any proof that doesn't make it any less hearsay.

 

[svart-hvitt]

Such claim is very easy to be proved or disproved by measurements. The point is you made that claim as it is something that is proved, not hearsay. If you heard that from "very experienced DIY speaker builders" which also didn't provide any proof that doesn't make it any less hearsay.

Maybe I missed a question mark in a sentence. I am sorry for that.

This is a thread on dynamics. The originator of the thread, Mr. Klaus Heinz, says he cannot measure what he believes to be a (sum of?) factor(s), i.e. «dynamics».

Elsewhere, @John_Siau brought the «first watt» up to the fore, writing:

«Speakers are also relatively clean at low power levels...»
Source: https://benchmarkmedia.com/blogs/application_notes/power-amplifiers-the-importance-of-the-first-watt

To me, it sounds like low power is better than high power.

Science is much about taking in information from different sources. Then a sorting process may begin to try and find true patterns, signals in the noise. I think I have heard the statement «sensitivity matters (for dynamics)» often enough to bring it to this discussion on Heinz’ remark on dynamics. Is it just a myth to source high sensitivity drivers if «dynamic» speakers are a goal. Is power a substitute for sensitivity to reach «dynamics»?

 

[Krunok]

This is a thread on dynamics. The originator of the thread, Mr. Klaus Heinz, says he cannot measure what he believes to be a (sum of?) factor(s), i.e. «dynamics».

I don't really know what he meant when he said "dynamics". I'm not sure if anybody besides him knows..

 

[svart-hvitt]

I don't really know what he meant when he said "dynamics". I'm not sure if anybody besides him knows..

This thread is almost 300 posts long and nobody knows or cares about dynamics?

«Dynamics» is an often used term. Sometimes, it’s defined as dynamic range, in which you can focus on «macro dynamics» and «micro dynamics» in recorded material. However, you can also narrow the discussion of dynamics to speakers only, where @Kvalsvoll suggested another definition, «transient reproduction», «sense of power and realism and immediacy of attack on drums and all other sorts of percussive instruments». Maybe one could add «tactile sensation» as part of the picture us well, what you feel in lower frequencies and as SPL is high.

I think this discussion of «dynamics» is interesting. From time to time I see speaker designs that seemingly do everything right and can back it up through measurements. However, can such low volume boxes be «dynamic»? Is volume the way to, a prerequisite for «dynamics»?

Even if Mr. Heinz doesn’t provide an answer to his own question, I think «dynamics» deserves an own discussion. Science is not only discussing what is known, verifiable facts.

 

[Soniclife]

For HF, the situation is different from low freq in that the initial transient peak at the listening position is not what determines how loud or powerful a transient sound appears like. A horn can give a sound-field with much less loss at distance, and at the same time much less sound into surfaces close to the speaker, this gives a sound with louder late reflections, because the high freqs are louder when they reach the back of the room. This is what gives the impression of louder transient peaks. Imagine you clap you hands in a dead space - it will not sound loud at all. Then do the same in a bathroom - loud, powerful sound.

Very interesting. Does this hold also for tweeters in waveguides?

 

[svart-hvitt]

Very interesting. Does this hold also for tweeters in waveguides?

Horn and waveguide; is there a difference?

 

[Kvalsvoll]

This would be an argument to use few smaller drivers instead of a single large one, right?

It is an argument to choose drivers with smaller inductance and motor system that can handle larger currents. Now if you also can design for reduced cone excursion, you improve things further, and extreme xmax is no longer necessary. Such drivers are readily available.

Still, you need adequate excursion, and power handling, and this easily becomes difficult to achive in a smaller package. One larger driver usually outperforms many smaller drivers with same total cone area.

 

[Kvalsvoll]

Very interesting. Does this hold also for tweeters in waveguides?

Waveguide is just another word for horn, often used for horns that are quite shallow in depth, wide pattern. A short, wide horn will have a pattern closer to a direct radiating driver. It is the profile and dimensions of the horn that determiens its radiation properties, does not matter what you call it.

@svart-hvitt , as I see it, just different names, and in the end it is the execution (profile, size..) that matters, not what you decide to call it.

 

[Kvalsvoll]

This thread is almost 300 posts long and nobody knows or cares about dynamics?

«Dynamics» is an often used term. Sometimes, it’s defined as dynamic range, in which you can focus on «macro dynamics» and «micro dynamics» in recorded material. However, you can also narrow the discussion of dynamics to speakers only, where @Kvalsvoll suggested another definition, «transient reproduction», «sense of power and realism and immediacy of attack on drums and all other sorts of percussive instruments». Maybe one could add «tactile sensation» as part of the picture us well, what you feel in lower frequencies and as SPL is high.

I think this discussion of «dynamics» is interesting. From time to time I see speaker designs that seemingly do everything right and can back it up through measurements. However, can such low volume boxes be «dynamic»? Is volume the way to, a prerequisite for «dynamics»?

Even if Mr. Heinz doesn’t provide an answer to his own question, I think «dynamics» deserves an own discussion. Science is not only discussing what is known, verifiable facts.

If there is a difference between 2 speakers, it can be measured. Including dynamics.

One clue is to realize that full-scale sound reproduction requires quite huge spl capacity, and when you compare distortion levels from typical small speakers, it becomes evident that they can not maintain output and sufficiently low distortion levels required for peaks.

Perhaps this discussion should be moved to a thread about loudspeaker dynamics.

 

[svart-hvitt]

If there is a difference between 2 speakers, it can be measured. Including dynamics.

One clue is to realize that full-scale sound reproduction requires quite huge spl capacity, and when you compare distortion levels from typical small speakers, it becomes evident that they can not maintain output and sufficiently low distortion levels required for peaks.

Perhaps this discussion should be moved to a thread about loudspeaker dynamics.

Measuring dynamics? Then you need to define it.

Maybe «dynamics» is simply headroom, i.e. output capacity without distortion?

However, to attain a certain radiation, one needs a given size, which may mean big size.

 

[dc655321]

Measuring dynamics? Then you need to define it.

This.
Thank you for raising the most important point in 15 pages.

What is the (or even "a") definition of dynamics in this context.

When I hear that word I think of a quantity analogous to slew rate in electronics (units of Voltage/time).
What are the dimensions/units for a loudspeaker? dB SPL/time at 1% THD?

Not saying this is anywhere near the mark, but until there is a physical definition (units!) of dynamics, we're talking in circles...

 

[Kvalsvoll]

This.
Thank you for raising the most important point in 15 pages.

What is the (or even "a") definition of dynamics in this context.

When I hear that word I think of a quantity analogous to slew rate in electronics (units of Voltage/time).
What are the dimensions/units for a loudspeaker? dB SPL/time at 1% THD?

Not saying this is anywhere near the mark, but until there is a physical definition (units!) of dynamics, we're talking in circles...

It is possible for a speaker to have a slew rate limitation, and if it has, it can be measured. One relatively simple approach would be to compare frequency response at low spl to high spl - if there is slew rate limitation, that will show up as a reduction in bandwidth at higher spl levels.

 

[Kvalsvoll]

Measuring dynamics? Then you need to define it.

Maybe «dynamics» is simply headroom, i.e. output capacity without distortion?

However, to attain a certain radiation, one needs a given size, which may mean big size.

Yes, what is to be measured must be defined. For a meaningful technical discussion.

But in hifi, it seems nothing needs to be defined, and if there is no physical or technical explanation, that is also fine. Much easier to use in marketing as well, difficult to debunk claims that has no defined meaning in the first place. Hifi has evolved similar to Flat Earth Society. If something is not understood, or the science-based explanation does not suit the agenda, then simply make up something.

But output capacity can be measured, and is sometimes also specified. Even for hifi speakers. Radiation patterns as well. But it is still not easy to see how those specifications relate to what we hear, in relation to dynamics.

 

[pozz]

Horn and waveguide; is there a difference?

Waveguides generally refer to designs which control directivity, while horns are for loading.

 

[andreasmaaan]

Very interesting. Does this hold also for tweeters in waveguides?

It holds for any directional source, as SPL loss per distance is inversely related to dispersion. The inverse square rule holds for an omnidirectional point source only.

And yes, the difference between a “waveguide” and a “horn” is semantic only, as both restrict the radiation of the driver (“waveguide”) and provide acoustic loading (“horn”).

 

[Krunok]

But output capacity can be measured, and is sometimes also specified. Even for hifi speakers. Radiation patterns as well. But it is still not easy to see how those specifications relate to what we hear, in relation to dynamics.

I think that whatever you measure you need to compare it with blind tests so you can correlate measurements with perception. I see no other way to be sure which measured parameters matter and which don't..

 

[Wombat]

This.
Thank you for raising the most important point in 15 pages.

What is the (or even "a") definition of dynamics in this context.

When I hear that word I think of a quantity analogous to slew rate in electronics (units of Voltage/time).
What are the dimensions/units for a loudspeaker? dB SPL/time at 1% THD?

Not saying this is anywhere near the mark, but until there is a physical definition (units!) of dynamics, we're talking in circles...

dv/dt, di/dt, dPower/dt, dPressure/dt, dx/dt - accurate and on demand. Of course in practice, with a transducer, it will be resisted by factors introducing inertia, elastic linearity/limit, phase, etc.

 

[KSTR]

I believe he’s actually suggesting it be done with DSP, but that doing it the way the lineariser does (ie by placing a time-inverse all-pass filter upstream) is functionally equivalent to doing it with DSP-based linear-phase crossover filters in the first place.

The view I’ve been putting forward here is that this is not the case, because the former approach, unlike the latter - exhibits pre- and post-ringing on the design axis.

Actually, there is zero difference whether you apply the phase unwrapping to each driver channel individually or globally in front of the whole XO (XO meaning the acoustic output, not the electical filter alone). Global compensation is more easy to realize, using much less computing resources, especially if it is a clean, analytically derived convolution kernel.

Any linear phase XO, if properly designed (no matter how it's done), does not have pre-ringing on the design axis but will always have some pre-ringing off-axis. This is also true for minimum phase crossovers but there the ringing is post-ringing only, buried in the tail of the impulse response. If you look at the step response of a crossover target function which is the sum of the indivdual responses you'll see that there is always partial cancellation in the sum because any of the individual responses contains a high pass component which *must* go below zero during some time sections, the time integral of any highpass must be zero. On the design axis, those are made to cancel out but off-axis the cancellation will falls apart, both from the time-of-flight offsets and the non-equal directivity pattern of the drivers. The only way to deal with is is by driver construction, using a coaxial design with equal directivity of the sections around the XO point.
EDIT: And of course, proper choice of the target responses, that is, using targets with a soft knee at the transition and no steeper slopes than absolutely required. No way to have low off-axis ringing with eg a high-order (> 3rd order) Butterworth target.
EDIT2: When you phase-unwrap a high-pass (like the woofer highpass) there will be a pre-signal, it's in the math, no way around it. But it isn't ringing, it's more like a exponential rise before the core of the step response, which is interestingly percieved as noise (no tonal signature). That's why compensating the bass roll-off of a speaker to linear phase must be taken with a grain of salt. While you can totally unwrap the phase of, say, a 6th order highpass (4th order of a ported design plus 2nd order of the subsonic filter ususally employed), it doesn't sound good (and introduces a huge kernel with high latency). A partial unwrap to reduce the phase response to 3rd order is a good compromise.

 

[andreasmaaan]

Actually, there is zero difference wether you apply the phase unwrapping to each driver channel individually or globally in front of the whole XO (XO meaning the acoustic output, not the electical filter alone). Global compensation is more easy to realize, using much less computing resources, especially if it is a clean, analytically derived convolution kernel.

Any linear phase XO, if properly designed (no matter how it's done), does not have pre-ringing on the design axis but will always have some pre-ringing off-axis.

Thanks for your reply @KSTR. The second paragraph of the quoted text is clear. What is less clear is the first part.

Perhaps we could just firstly clarify the terminology? By “phase unwrapping”, you mean (for present purposes at least) an all-pass filter designed to linearise the phase of the system - is this right?

Such a filter necessarily introduces pre-and post-ringing. Are you saying that such a filter won’t ring on the design axis?

My view is that the ringing created by such a filter would be equal regardless of location. I would be happy to be corrected though...

 

[KSTR]

Yes, phase-unwrapping means taking the allpass response of the XO (also known as the excess phase), inverting it in time and use it as convolution kernel to undo the original allpass behavior. The system now behaves like a single way with the same magnitude response. No ringing at all, at least on the design axis, as noted. By design, literally.

Assume a simple crossover at 1kHz with ideal drivers (infinite flat response from DC to light).
Let LP(s) be the original lowpass and HP(s) the highpass. LP(s)+HP(s) form the system sum which is an allpass AP(s).
The time inverse allpass be denoted as IAP(s). It is immediately clear that IAP(s)*[LP(s)+HPs(s)] is the same as IAP(s)*LP(s)+IAP(s)*HP(s).