Evidence-based Speaker Designs

[QMuse]

How far apart are the exits from the pipes?

I'd be more interested in asking him how he avoided cancellation between the direct radiation of the two woofers!

This is how it looks when you remove bottom foot and turn them over. Tubes are tapperd and of different diameter as they are tuned to different frequencies. Each midwoofer is loading it's own tube. Frankly, I cannot even begin to imagine what a nightmare it was to design this speaker.

 

[andreasmaaan]

This is how it looks when you remove bottom foot and turn them over:

View attachment 64788

Well, the distance is not significant compared to the wavelength of the frequencies coming out of them. Unless the pipes are radically different length, they should sum constructively.

 

[ctrl]

+5dB at 400-500Hz shouldn't be a stress for a good speaker in terms of THD/IMD.

With the Harlech 5'' bass-midrange drivers used, 5dB can be a hell of a lot, especially since due to the non-linearity of the distortions it is simply impossible to predict how +5dB sound pressure in a certain frequency range will affect the overall system.
But if it has a positive effect on sound, why not?

As you can see no major dips exist in 400-800Hz area. My correction was based on 2 responses in the middle as I like it to be as spatially robust as possible.

I just noticed that the speaker is only 0.32m deep. This means that the driver offset is probably 0.2m instead of the previously assumed 0.28m.

If the simulation is corrected with this, the result might reflect your room measurements a bit better.

I'd be more interested in asking him how he avoided cancellation between the direct radiation of the two woofers!

Since this is probably really a 2-way floorstanding speaker, the answer is simple: Not at all
https://www.hifiengine.com/manual_library/castle/harlech.shtml

It should be remembered, however, that in reality it is unlikely to be as dramatic as the simulation, as the rear-located chassis will produce significantly less sound pressure on axis.

 

[QMuse]

Well, the distance is not significant compared to the wavelength of the frequencies coming out of them. Unless the pipes are radically different length, they should sum constructively.

No, length is not equal but it is relatively similar in length.

 

[QMuse]

With the Harlech 5'' bass-midrange drivers used, 5dB can be a hell of a lot, especially since due to the non-linearity of the distortions it is simply impossible to predict how +5dB sound pressure in a certain frequency range will affect the overall system.
But if it has a positive effect on sound, why not?

Check this, it was tested with RTA single sine tone of 40Hz played at 100dB at LP (so 4m from the speakers), and yet, THD is only 1.65%. My Castles are not struggling with distortion at high SPLs.

Since this is probably really a 2-way floorstanding speaker, the answer is simple: Not at all
https://www.hifiengine.com/manual_library/castle/harlech.shtml

This is older version, mine is with carbon fiber midwoofers and with optimised crossover and corrected dip in the 500-700 region.

 

[tuga]

Check this, it was tested with RTA single sine tone of 40Hz played at 100dB at LP (so 4m from the speakers), and yet, THD is only 1.65%. My Castles are not struggling with distorion at high SPLs.

View attachment 64790



This is older version, mine is with carbon fiber midwoofers and with optimised crossover and corrected dip in the 500-700 region.

Impressive.
Can you produce a step response plot?

 

[QMuse]

I just noticed that the speaker is only 0.32m deep. This means that the driver offset is probably 0.2m instead of the previously assumed 0.28m.

If the simulation is corrected with this, the result might reflect your room measurements a bit better.
View attachment 64787

By looking at my EQ filters I can only tell that both speakers seem to share a 4dB dip at 590Hz.

 

[ctrl]

By looking at my EQ filters I can only tell that both speakers seem to share a 4dB dip at 590Hz.

Due to the simplicity of the simulation, it can also be a little off. It is important that the phenomena that occur are reproduced approximately - which is the case.

Check this, it was tested with RTA single sine tone of 40Hz played at 100dB at LP (so 4m from the speakers), and yet, THD is only 1.65%. My Castles are not struggling with distorion at high SPLs.

It is not possible to make reliable distortion measurements at a distance of 4m including all room reflections.
You should measure at a distance of 0.3m to get as few reflections as possible.

To investigate the effect of your +5dB boost, you could do a multitone measurement and/or a two-tone measurement (40Hz + 600Hz).

 

[andreasmaaan]

It should be remembered, however, that in reality it is unlikely to be as dramatic as the simulation, as the rear-located chassis will produce significantly less sound pressure on axis.

Yes, I was asking rhetorically

In any case, I'm not sure I'd simulate it the way you have. The rear woofer is not only offset in the Z plane, but also elevated in the Y plane.

Simulating the arrangement in space rather than time, and adding an 1800Hz LR4 crossover, I get this:

Given the rear woofer is not going to be radiating omnidirectionally at 800-900Hz, as you say, the actual situation is not going to be as bad.

EDIT: wrong crossover setting! Fixed.

 

[QMuse]

It is not possible to make reliable distortion measurements at a distance of 4m including all room reflections.
You should measure at a distance of 0.3m to get as few reflections as possible.

Ok, I'll give it a try, but I don't really see how reflections can reduce distortion.
And as FR is pretty linear at LP harmonics would be measured quite correctly.

At which level should I measure at such close distance?

To investigate the effect of your +5dB boost, you could do a multitone measurement and/or a two-tone measurement (40Hz + 600Hz).

I don't know how to do two-tone distortion measurement with REW.

Can you provide some help?

 

[QMuse]

Simulating the arrangement in space rather than time, and adding an 1800Hz LR4 crossover, I get this:

This makes more sense to me as well. Btw, i didn't notice any phase conflicts between 2 midwoofers. Or maybe that small dip at 600Hz is because of that?

 

[ctrl]

Yes, I was asking rhetorically

Didn't get it

Given the rear woofer is not going to be radiating omnidirectionally at 800-900Hz, as you say, the actual situation is not going to be as bad.

It could even work as a baffle step correction when used correctly (i.e. the chassis spacing is optimized accordingly).

This makes more sense to me. Btw, i didn't notice any phase issues between 2 midwoofers.

Both approaches cause a phase shift and deliver the same result (with identical parameters).

Ok, I'll give it a try, but I don't really see how reflections can reduce distorion.

Due to the different transit times of the reflections compared to direct sound, cancellation or addition can occur in the fundamental as well as in the harmonics.

The reflections can be absorbed to varying degrees depending on the frequency.

Your room itself could provide additional distortion due to resonance.

I don't know how to do two-tone distorion measurement with REW.
Can you provide some help?

Sorry, but I only use Arta for measurements.

 

[Ilkless]

My speakers were designed by Steve Stanley Hewlett and IMHO they represent one of the best designs made in Britain those days. I would be very interested in asking him how he avoided that phase inverted bass coming from twin pipes TL at the bottom of the speaker also don't cause cancellation but unfortunately he passed away from cancer 2 years ago.

Interesting, because IAG (the Quad and Wharfedale guys now) took over Castle, and I assumed production was moved elsewhere because of this.

 

[QMuse]

It could even work as a baffle step correction when used correctly (i.e. the chassis spacing is optimized accordingly).

Isn't a TL design like my Castles actually an "infinite" baffle design?

Due to the different transit times of the reflections compared to direct sound, cancellation or addition can occur in the fundamental as well as in the harmonics.

The reflections can be absorbed to varying degrees depending on the frequency.

Your room itself could provide additional distortion due to resonance.

Isn't pretty much all of that cooked into my EQ for LP?

 

[andreasmaaan]

Isn't a TL design like my Castles actually an "infinite" baffle design?

No, the TL doesn't make it immune from the effects of having a baffle. Or perhaps I'm not understanding what you mean when you say it's an "infinite baffle design"?

I don't know how to do two-tone distortion measurement with REW.

Can you provide some help?

Open the Generator, select "dual tone" from the dropdown menu, check the "custom" option, then select the frequencies and the ratio.

 

[QMuse]

Interesting, because IAG (the Quad and Wharfedale guys now) took over Castle, and I assumed production was moved elsewhere because of this.

It was moved in China, together with all the tools and technology to manufacture drivers, and Hewlett also move there. He even learned language and was able to commincate without translator. IAG initially produced Howard S3 model which is similar to my Harlechs but it had larger midwoofers and upper midwoofer worked only up to 350Hz. Later on they designed Avon series which is also a TL design but with single quarterwave pipe and true ribbon tweeter.

Mine were made in UK, at the time when Great Britain was still great and @Thomas savage was just a beardless teenager.

 

[QMuse]

No, the TL doesn't make it immune from the effects of having a baffle. Or perhaps I'm not understanding what you mean when you say it's an "infinite baffle design"?

From the wiki:

The transmission line (TL) is the theoretical ideal, and one of the most complex constructions, with which to load a moving coil drive unit.[citation needed] The most common and practical implementation is to fit a drive unit to the end of a long duct that is usually open at the far end. In practice, the duct is folded inside a conventional shaped cabinet, so that the open end of the duct appears as a vent on the speaker cabinet. There are many ways in which the duct can be folded and the line is often tapered in crossection to avoid parallel internal surfaces that encourage standing waves. Some speaker designs also use a spiral or elliptic spiral shaped duct, usually with one speaker element in the front or two speaker elements arranged one on each side of the cabinet. Depending upon the drive unit, and quantity and various physical properties of absorbent material, the amount of taper will be adjusted during the design process to tune the duct to remove irregularities in its response. The internal partitioning provides substantial bracing for the entire structure, reducing cabinet flexing and colouration. The inside faces of the duct or line, are treated with an absorbent material to provide the correct termination with frequency to load the drive unit as a TL. The enclosure behaves like an infinite baffle, potentially absorbing most or all of the speaker unit's rear energies.[8] A theoretically perfect TL would absorb all frequencies entering the line from the rear of the drive unit but remains theoretical, as it would have to be infinitely long.

Open the Generator, select "dual tone" from the dropdown menu, check the "custom" option, then select the frequencies and the ratio.

View attachment 64804

Nice, thank you! I have to save that as a file as I cannot play it from PC. What do I do with it? Play it in RTA window?

 

[andreasmaaan]

Nice, thank you! I have to save that as a file as I cannot play it from PC. What do I do with it? Play it in RTA window?

Hit the little button that looks like a floppy disc and says "WAV"

From the wiki:

Ok, there are two things that occur when you place a driver on an (idealised) infinite baffle.

Firstly, the driver radiates into half-space, not full-space.

Secondly, the acoustic impedance at both the front and rear of the IB are equal, and there are no reflections from the rear of the driver.

It's the former effect that results in there being no baffle step.

It's the latter effect that that wiki page is talking about (although it overstates the extent to which it is true in the case of a TL design).

 

[andreasmaaan]

It could even work as a baffle step correction when used correctly (i.e. the chassis spacing is optimized accordingly).

Very good point

 

[QMuse]

Hit the little button that looks like a floppy disc and says "WAV"

I got that part and I have wav file on my NAS.
How do I do IM measurement with it?

Ok, there are two things that occur when you place a driver on an (idealised) infinite baffle.

Firstly, the driver radiates into half-space, not full-space.

Secondly, the acoustic impedance at both the front and rear of the IB are equal, and there are no reflections from the rear of the driver.

It's the former effect that results in there being no baffle step.

It's the latter effect that that wiki page is talking about (although it overstates the extent to which it is true in the case of a TL design).

But aren't my drivers also radiating into half-space as pretty much all of the rear-space radiation is funneled into TL tubes? I mean, you cannot simply absorb rear side radiation - you have to channel it somewhere.