Hornresp User Manual
Hi Everyone,
A Hornresp user in Sweden has produced a very comprehensive 170 page "Hornresp Manual" that you may find of interest. The PDF file can be downloaded from:
Facebook
The author, Harald Karlsson, has obviously gone to a lot of trouble to produce the document, and he is to be congratulated on his efforts - very impressive indeed!
Kind regards,
David
Hi Everyone,
A Hornresp user in Sweden has produced a very comprehensive 170 page "Hornresp Manual" that you may find of interest. The PDF file can be downloaded from:
The author, Harald Karlsson, has obviously gone to a lot of trouble to produce the document, and he is to be congratulated on his efforts - very impressive indeed!
Kind regards,
David
I should have perhaps mentioned that the Maximum SPL tool can be used to specify a constant input power rather than a constant input voltage. This could possibly make "1W @ 1m SPL" a little more meaningful
.Attachments
My understanding is that Hornresp displays the predicted sound power, not the predicted on-axis sensitivity which is a sound pressure value. That means it shows all of the energy radiated, regardless of angle.
The directivity of the cabinet is not simulated, apart from in a few cases of single-segment horns of certain types. So conversion to sound pressure is difficult, because you need to know the directivity index for the cabinet - which is also a function of frequency, rather than a single number.
Have a look at this:
Sound power level SWL and sound pressure level SPL distance compare acoustic power sound source noise Conversion of sound pressure to sound intensity conversion sound level energy level strength directivity factor coefficient sound intensity SIL - se
Just to clarify, the Hornresp Directivity tools can be used with any finite single-segment non-negative flare horn that does not have a throat adaptor. The tools can also be used with direct radiating single drivers and ports.
Do you read the article and references?
According to Beranek and Mellow, the loudspeakers efficiency is given with the formula down below.
Which has a Re term in there.
fyi, by definition efficiency is always given in percentage.
Next, the loudspeaker sensitivity per watt is given as following;
dB(1 watt) = 112.02 + 10·log(η0)
Which still has that Re variable in there.
The sensitivity per volt is given as following;
dB(2.83 V) = dB(1 watt) + 10·log(8/Re)
Which has even a double Re variable in there.
So I don't know what you're talking about, but all the formulas are right in front of you.
I would highly recommend reading;
- Acoustics: Sound Fields and Transducers, By: Beranek and Mellow
- Loudspeaker Handbook, By: John Eargle
- Loudspeaker and Headphone Handbook, By: John Borwick
Two of these can be even found "somewhere" without to much problems.
Attachments
There is a better explanation at this link: Loudspeaker Sensitivity & Impedance Explained | Audioholics
Quote:"Loudspeaker Sensitivity: a measure of Sound Pressure Level (SPL) at a specified distance for a specified input signal. This is usually specified for a loudspeaker in a non-reverberant environment, in dB SPL and referenced to 1 meter on the reference axis with an input of 2.83 volts RMS, typically at one or more specified frequencies (often 300, 400, 500, 600Hz or the average of these). Sensitivity should always be accompanied by an impedance specification."
The reason for including the impedance specification is given further down the page: - "Whenever sensitivity is quoted, the nominal impedance must also be stated. This will prevent the manufacturer from cheating in cases where the lower impedance speaker is able to draw more current making the speaker appear more efficient. Of course the efficiency has NOT increased, but the sensitivity has, which is why it is important to include nominal impedance into the spec. Plus, all speakers are designed to be driven from voltage source amplifiers and have a flat frequency response when so driven. Therefore, sensitivity does give a direct indication of how much louder one speaker will sound on a direct comparison (disregarding the capability of the amplifier to drive a speaker that cheats on impedance)."
Finally: "Speaker engineers use the term “sensitivity” because it is technically more precise definition of how loud a speaker will play when driven by an input signal of at a specified voltage level. “Efficiency” is usually a more colloquial term used by consumers, to informally describe how much power a speaker needs to achieve a desired loudness level: “Oh, your speakers a real power hogs. They’re not very efficient.” From a technical standpoint, “efficiency” is defined as the sound power output divided by the electrical power input—but that’s not how consumers use the word. So rather than try to turn laypeople into engineers, we’ll just concentrate on sensitivity.
The reason for quoting sensitivity rather than efficiency is that the true efficiency of a speaker varies wildly with frequency because of its impedance variation. The efficiency quote would be fairly meaningless, so loudspeaker engineers measure and refer to “sensitivity” instead."
No, there isn't really one "true" sensitivity.
The sensitivity just has a lower part, were the speaker acts as a piston; see reference above.
As well as a part (roughly 2-3x Fs) were the loudspeaker doesn't really act in the pure piston region anymore and other variables starts to play. (things like demodulation rings etc)
But in general if the word efficiency is being used, one refers to the piston part of the loudspeakers.
Aka, calculated or derived from the T/S parameters.
Fyi; as can be seen in the formulas shown before, for loudspeakers sensitivity as efficiency are directly connected to each other.
The discussion if those T/S parameters are the most reliable way of calculating the true efficiency is debatable because of all kinds of non-linear effects. But that is by definition already the whole issue with T/S parameters in general.
So not only totally different discussion, but one should also be aware of that in general.
In any case, the nominal impedance never has anything to do with all of this.
It never had and it never will be.
Once again, I would highly recommend reading those books firs, it is all perfectly explained.
Those books can be considered a little more scientifically correct instead some random article from audioholics.
I would also really advice on focusing on proper literature as well.
Using the sensitivity in terms of "better" is just extremely silly.
The sensitivity just has a lower part, were the speaker acts as a piston; see reference above.
As well as a part (roughly 2-3x Fs) were the loudspeaker doesn't really act in the pure piston region anymore and other variables starts to play. (things like demodulation rings etc)
But in general if the word efficiency is being used, one refers to the piston part of the loudspeakers.
Aka, calculated or derived from the T/S parameters.
Fyi; as can be seen in the formulas shown before, for loudspeakers sensitivity as efficiency are directly connected to each other.
The discussion if those T/S parameters are the most reliable way of calculating the true efficiency is debatable because of all kinds of non-linear effects. But that is by definition already the whole issue with T/S parameters in general.
So not only totally different discussion, but one should also be aware of that in general.
In any case, the nominal impedance never has anything to do with all of this.
It never had and it never will be.
Once again, I would highly recommend reading those books firs, it is all perfectly explained.
Those books can be considered a little more scientifically correct instead some random article from audioholics.
I would also really advice on focusing on proper literature as well.
Using the sensitivity in terms of "better" is just extremely silly.
Last edited:
Andrew Jones contributed to the content of that "random" Audioholics page I shared. As he has a bit more experience than I do in the field of loudspeaker design, I think I'll take him on his word for how the term "loudspeaker sensitivity" should be defined and used, thanks
I also agree with him here: "Because a speaker is a voltage-driven device, we would be much better off to move away from amplifier power to amplifier voltage, specified in dB relative to 2.83 volts. At the same time, the minimum impedance of the load that the amplifier can maintain at this maximum voltage should be quoted. Using this, we could directly calculate how loud the speaker could go. An example will illustrate this: suppose we take a conventional 100W amp into an 8 ohm load. 100W is 28.3Vrms, which is +20dB relative to 2.83V. If the speaker is 90dB sensitivity, then max SPL is 90+20= 110dB. So, we quote the amp as a +20dB amp. That’s it...we now know how loud it will play with any sensitivity speaker. Easy!"
I also agree with him here: "Because a speaker is a voltage-driven device, we would be much better off to move away from amplifier power to amplifier voltage, specified in dB relative to 2.83 volts. At the same time, the minimum impedance of the load that the amplifier can maintain at this maximum voltage should be quoted. Using this, we could directly calculate how loud the speaker could go. An example will illustrate this: suppose we take a conventional 100W amp into an 8 ohm load. 100W is 28.3Vrms, which is +20dB relative to 2.83V. If the speaker is 90dB sensitivity, then max SPL is 90+20= 110dB. So, we quote the amp as a +20dB amp. That’s it...we now know how loud it will play with any sensitivity speaker. Easy!"
Those calculations are specific to a driver's "reference efficiency", which is not quite what I'm talking about here.
Ok, maybe we can clear this up with a simple thought experiment - take any decent driver, and put it in a maximally-flat vented alignment, let's say a sim in Hornresp. The result would be a flat passband with the driver's impedance varying through the passband, right? Now, would you refer to its calculated SPL response in the passband as in terms of dB/W/M, or in terms of dB/V/M? Which would be more correct?
One can refer both. Totally depends what one is interested in.
The fact that the impedance changes doesn't matter, since it's all related back to the Re
So in other words, as for input it has to be clear if voltage is being used or power.
As far as I know Hornresp has the option for both.
Which is a good thing, because this way one can compare whatever suits.
Does this tell as exactly how much power a driver exactly needs?
No, off course not, especially not around Fs.
For subwoofers for example it is eve EXTREMELY important to also take in active filters into the whole equation. (something 98% of the people seem to forget).
But for fullrange systems the power for input gives quite a decent comparison between multiple systems. It doesn't give the power at any given point.
But the parameter was never intended to be used that way.
In fact, any kind of single number unit will have the same issues.
That is just the nature of the beast.
Things of maximum loudness and such don't have anything to do with this.
The only reason to use the parameters we are talking about, is to understand how much voltage we need (which tells us what amplifier we have to use), or compare two systems with the same reference.
The fact that the reference isn't exactly right isn't important, since it's relative.
With dB/V there is not the same reference anymore, as can be seen with the formulas before.
The fact that the impedance changes doesn't matter, since it's all related back to the Re
So in other words, as for input it has to be clear if voltage is being used or power.
As far as I know Hornresp has the option for both.
Which is a good thing, because this way one can compare whatever suits.
Does this tell as exactly how much power a driver exactly needs?
No, off course not, especially not around Fs.
For subwoofers for example it is eve EXTREMELY important to also take in active filters into the whole equation. (something 98% of the people seem to forget).
But for fullrange systems the power for input gives quite a decent comparison between multiple systems. It doesn't give the power at any given point.
But the parameter was never intended to be used that way.
In fact, any kind of single number unit will have the same issues.
That is just the nature of the beast.
Things of maximum loudness and such don't have anything to do with this.
The only reason to use the parameters we are talking about, is to understand how much voltage we need (which tells us what amplifier we have to use), or compare two systems with the same reference.
The fact that the reference isn't exactly right isn't important, since it's relative.
With dB/V there is not the same reference anymore, as can be seen with the formulas before.
There's only one source input that would produce a flat passband response in a maximally-flat vented alignment that I provided as my thought experiment, and that's voltage. The calculated sensitivity in the passband will not vary, and its value will not depend on what the loudspeaker's impedance is at any point in the passband.
IMO as the source is a voltage source, it therefore stands to reason that the passband sensitivity should be also defined in terms of dB/V, not dB/W, because while the input voltage remains the same, the wattage consumed by the speaker at different points in the passband does not.
If however you looked at the model using constant power rather than constant voltage, the resulting SPL response will definitely not be flat. due to the varying impedance through the passband, the calculated wattage will also vary across the passband. There would be basically no way to define the passband response accurately without also giving the frequency at which the definition is being made.
Hornresp's models default to modeling with a constant voltage source, "Eg". I'm not aware of an option that allows one to model with a constant power source. Now, Hornresp does allow you to calculate the required voltage to produce a given power level at a given impedance, and use that voltage in its model, but that's it . The model will continue to use that voltage in its calculations no matter how much the impedance varies in the model through the passband, so the wattage will also vary. Perhaps David can advise further here.
Now, for real speakers, the SPL levels at a particular voltage depend on a lot more than just what a simple box model suggests , and you're going to have variations in the passband. However, the source that's driving them is still a voltage source, so now we have to be more specific about how the sensitivity is actually being measured (pink noise, bandwidth-limited noise, etc.).
Anyway, I think we've taken up too much of David's thread on this discussion, so maybe at this point we can just agree to disagree and move on
IMO as the source is a voltage source, it therefore stands to reason that the passband sensitivity should be also defined in terms of dB/V, not dB/W, because while the input voltage remains the same, the wattage consumed by the speaker at different points in the passband does not.
If however you looked at the model using constant power rather than constant voltage, the resulting SPL response will definitely not be flat. due to the varying impedance through the passband, the calculated wattage will also vary across the passband. There would be basically no way to define the passband response accurately without also giving the frequency at which the definition is being made.
Hornresp's models default to modeling with a constant voltage source, "Eg". I'm not aware of an option that allows one to model with a constant power source. Now, Hornresp does allow you to calculate the required voltage to produce a given power level at a given impedance, and use that voltage in its model, but that's it . The model will continue to use that voltage in its calculations no matter how much the impedance varies in the model through the passband, so the wattage will also vary. Perhaps David can advise further here.
Now, for real speakers, the SPL levels at a particular voltage depend on a lot more than just what a simple box model suggests , and you're going to have variations in the passband
. However, the source that's driving them is still a voltage source, so now we have to be more specific about how the sensitivity is actually being measured (pink noise, bandwidth-limited noise, etc.). Anyway, I think we've taken up too much of David's thread on this discussion, so maybe at this point we can just agree to disagree and move on
You're very clearly mixing certain things up.
(or I am not capable of explaining it well )
Nobody is talking about a constant power source.
It is still a voltage source, but "referenced" to the same power at Re.
The sensitivity number also doesn't have anything to do with the impedance or freq resp, or any graphs for that matter.
The main reason why I responded here, was because I was reading sensitivity in terms of nominal impedance (resistance). Which is in all cases wrong, or at least very unusual and confusing.
(or I am not capable of explaining it well
)Nobody is talking about a constant power source.
It is still a voltage source, but "referenced" to the same power at Re.
The sensitivity number also doesn't have anything to do with the impedance or freq resp, or any graphs for that matter.
The main reason why I responded here, was because I was reading sensitivity in terms of nominal impedance (resistance). Which is in all cases wrong, or at least very unusual and confusing.
Last edited:
Re is a DC measurement yes?
Impedance is an AC derivation. Yes?
You are sending an AC voltage source that gets changed into current in the coil. Yes I am skipping details.
You can look at this from a few angles and still be perfectly valid as both yourself and Brian are doing. Neither of you are incorrect. Just approaching from a different side of viewpoint.
It's getting pedantic. Move on please. You and Brian are to worth listening to on other useful bits and pieces. The horse is dead. Stop beating it.
Impedance is an AC derivation. Yes?
You are sending an AC voltage source that gets changed into current in the coil. Yes I am skipping details.
You can look at this from a few angles and still be perfectly valid as both yourself and Brian are doing. Neither of you are incorrect. Just approaching from a different side of viewpoint.
It's getting pedantic. Move on please. You and Brian are to worth listening to on other useful bits and pieces. The horse is dead. Stop beating it.
Well, not really.
The reason why I am keep talking about it, is because it seems to be a very confusing subject for many people.
When somebody is talking about constant power sources, that tells me that somebody is missing a very important point.
And that horse is far from being dead yet.
But yes, I do understand that this is going offtopic
The whole point is, there is not just one parameter to describe everything.
like everything in engineering; "it depends"
All these parameters can be useful and all have their shortcomings.
If anything, I would like to see the VA graph (power vs frequency) in the loudspeaker wizard in Hornresp
I work in audio design daily.
And I agree with the two of you. VA is only useful if you really understand what you are doing. Most will not have a clue what to do with VA. And to make it useful they would have to do a further derivation to make some sense of it. And is again the other side of the coin if you know what Eg is doing.
Audio is a journey with many different paths to the same destination.
Sometimes one way will show you something the other failed to do so.
And I agree with the two of you. VA is only useful if you really understand what you are doing. Most will not have a clue what to do with VA. And to make it useful they would have to do a further derivation to make some sense of it. And is again the other side of the coin if you know what Eg is doing.
Audio is a journey with many different paths to the same destination.
Sometimes one way will show you something the other failed to do so.
- VA is useful to know how much power the amplifier needs to deliver as well as how much power the loudspeaker is using at a given frequency *
- Eg is the information you need to know how much voltage the amplifier has to deliver.
For example, in some cases the amplifier only has to deliver 100W (at the highest point in the VA graph), while still has to deliver 26Vrms.
In this case a 100W @ 4ohm amplifier won't be enough.
But non of those two have anything to day with the efficiency or sensitivity parameters we described before.
Not different side of the coin, but a totally different coin, that's what I am trying to make clear the whole time.
* Not taking non-linearities into account and max rated power for speakers is a whole different can of worms.
- Eg is the information you need to know how much voltage the amplifier has to deliver.
For example, in some cases the amplifier only has to deliver 100W (at the highest point in the VA graph), while still has to deliver 26Vrms.
In this case a 100W @ 4ohm amplifier won't be enough.
But non of those two have anything to day with the efficiency or sensitivity parameters we described before.
Not different side of the coin, but a totally different coin, that's what I am trying to make clear the whole time.
* Not taking non-linearities into account and max rated power for speakers is a whole different can of worms.
Apologies for the confusion
.You said:
"For loudspeakers, efficiency and sensitivity are directly related to each other."
I simply posted a link to a web page that I felt supported what you were saying. I was not replying to your message as such, rather just trying to provide further explanatory information that others may have found useful, given that there seemed to be some confusion out there regarding the difference between loudspeaker efficiency and sensitivity.