Large versus small speakers.

[Glowing Bits!]

Tannoy Mercury M3 speakers sound good with the front port open but sounds knackered with it blocked, it's the only ported speakers I've found that seem to work properly, unlike the rear ported things.

[Pilot Mariner]

Thanks Mark H for that explanation. Having recently sold my kit built IPL S4tl transmission lines, I won't be ruling out rear ported speakers when deciding on the replacements.

Mark

[mole42uk]

Originally Posted by stevehertz

A quickie. Small speakers, if correctly designed and built like the BBC LS35A, suffer from less panel resonance and hence can be very uncoloured and accurate from the lower mid range upwards.

I would add Peter Keeley's 'Keesonic Kub' to that. A superb ported design.

[mhennessy]

Originally Posted by Glowing Bits!

Tannoy Mercury M3 speakers sound good with the front port open but sounds knackered with it blocked, it's the only ported speakers I've found that seem to work properly, unlike the rear ported things.

Of course, you have to wonder what is having the biggest influence here: the tuning of the port or the position of it.

A case in point: the Graham Audio LS6. It has a large rear port. I was invited to comment on a pre-production version, and my only criticism concerned the slightly "slow" bass. Absolutely nothing to do with the position of the port, but everything to do with the system tuning. In response, Derek Hughes lengthened the port to tune it down in frequency, and that helped a lot, but it still wasn't perfect. But given that the box size was fixed by this stage, that was all they could sensibly do. It has to be said that the bass extension is very impressive for the size of the woofer (6") and cabinet - but as always, everything is a compromise. I did suggest supplying a foam bung for the port, which works well with this speaker (doesn't always - just depends on the particular tuning alignment).

Some time later Phil Ward of Sound on Sound magazine reviewed them, and was as complimentary as I was, but also picked up on the bass tuning, and suggested a port bung as well:

https://www.soundonsound.com/reviews...-chartwell-ls6

Not long after, they released the floor-standing version - the LS6f. The much larger volume of air was a better match for the drive unit, and the bass ended up being more extended, and "quicker". This time, the port is front-facing, but it's actually a rectangular slot in the plinth, meaning its output is loaded by the floor, increasing efficiency. In practice, this means you need less output from the port compared to when it was up there with the drive units, and the less you ask the port to do, the better the results in my experience!

[stevehertz]

I'm intrigued by the use of the word 'slow' when describing sound. I'm not trying to suggest audiophoolery here, just genuinely interested in the concept of 'slow' sound. Many years ago in a 'proper' hifi shop, the 'all knowing' (as they generally are) assistant made reference to the Garrard 401 as being a 'slow' deck. I was thinking, just adjust the fine speed control knob and watch the strobe! But I didn't say anything. Seriously, in that instance I think he was blowing hot air and regurgitating a term he'd just seen in Hi-Fi News.

[Radio Wrangler]

Probably too slow in the rate of contribution to his commission...

David

[stevehertz]

That's true, having been thoroughly put off by his bombastic approach I left the shop with arms by my side and brass in pocket. I did walk slow mind.

[mhennessy]

As mentioned, it's transient response.

At the bass end, you have a resonant system. All resonant systems imply energy storage, meaning you have to put some energy in to get it started, and you have to use some energy to stop it again. Just think about pushing someone on a swing...

Sealed boxes are easiest - they have just the one resonant system, formed by the mass of the moving parts, working against the springiness of the suspension and the air in the box. For a given drive unit, there will an specific volume of air to enclose, which will result in the chosen damping ("Q"). You can aim for a Q of 0.7, which is maximally flat, perhaps 0.5, which results in an earlier rolloff, but a slightly lower extension, and slightly better transient response, or a higher value like 1 or 1.5, which gives a bump in the bass before rolling off - but will roll away a bit quicker, and at a higher frequency. The American speaker design textbook that I had on constant loan from the school library always recommend doing that, but ATC tend to be at the other end - IIRC, my SM20SLs are 0.5, although I can't remember if I've verified that myself...

Ported boxes are more complicated, because in addition to the system just described, you have a second resonant system formed by the mass of air enclosed by the port, working against the springiness of the air in the box. In an ideal world you get more extension, but a faster roll-off below. But because there are 2 resonant systems that you have to get going and stop, it's usual to see a poorer transient response.

Worse still are bandpass systems. There are many variations on the theme, but think of them as a sealed or ported box, but with another box bolted on the front with an exit port. I've never tried building one of these, but have heard plenty. In my experience, they're good for rumble, but no good for kick drum...

I've found the images I referred to earlier. They are comparing two speakers:

1. A simple 10" JBL car sub that I threw in an MDF box many years back. The box is a 12" cube.
2. A bandpass sub that I have sometimes borrowed from a friend for PA use. It's a pair of 10" woofers, both firing into a common cavity. Details here.

The first image is the frequency response, measured using nearfield techniques, so for comparison purposes primarily. It's easy to tell which is the sealed box, with its smooth response and gentle rolloff. The bandpass box is living up to its name, with the narrow band of operation and rapid LF rolloff. Note the peak just above 50Hz. The output above 400Hz demonstrates a point I made earlier about ports allowing midrange out. Of course, this would be used with a crossover, so not an issue in practice.

Turning to the transient response, the next images shows the input stimulus in yellow, and the measured output in cyan. A simple toneburst.

First up is the sealed JBL at 50Hz, which is where it is at resonance. Despite that, it stops and starts pretty well, though inevitably you can see the build-up over the first one or two cycles.

The next image is the same unit at 70Hz, so above resonance, and it's very similar indeed.

Turning to the bandpass at 50Hz, it's quite different. You can see it takes many cycles to get going, and a long time to stop. The final image shows the same at 60Hz, where it's quite comical.

Although these effects are quite short-lived in theory, it is quite surprising how audible they can be in practice. Obviously some programme material will show it up more readily than others, so how bothered one might be by all this is highly variable.

I'm no experts on turntables, so it's best I don't attempt to guess at what the salesman was getting at, beyond the obvious. I suppose you could argue that a turntable contains resonant systems too - and that's a very convenient "grain of truth" to attempt to hang an argument on, but I'd want to see it backed up by measurements like I've presented here. I've no idea if anyone has made a test LP with tonebursts on, but if that exists, then it's easier to do the measurements than it was on the speakers - no microphone needed, just a 'scope.

[stevehertz]

Thanks so much Mark for taking the time provide that detailed response to my question. I'm sure a lot of us have learnt a lot. Thanks.

[samjmann]

Thanks so much to everyone that have all contributed so much to answering my question. Just as Steve has said on the above post, I've learnt so much.

It was something that I'd noticed for years when visiting customers that a Hacker/Roberts or maybe a Grundig radio would often be playing in the kitchen and the owner would be in the adjoining room listening to it. The hifi next to the person would be off! I suppose what's happening is that the radio is relatively loud and as mentioned would start to sound good. The hifi would have to be at low level in the lounge and therefore not at it's best. Maybe this is what was happening. SJM.

[Radio Wrangler]

There is a mathematical process called the Fourier Transform. And by using it, any signal which can be described in the time domain can be converted to a fully satisfactory description in the frequency domain, though the DC component often has to be described as an exception.

So any effect we notice that can be explained in terms of waveform can be converted to an explanation in terms of frequency - and vice-versa.

It's quite common for some sort of resonance technique to be used to make a bump to compensate some of the low frequency roll off of a speaker.

Resonances have a characteristic "Q-factor" commonly used in RF electronics. Q can be expressed in many ways. It is the ratio of the centre frequency to the 3dB bandwidth. It is also the ratio between real and imaginary impedances in a resonant circuit or a modelled resonant circuit. It is also the ratio of the energy in a resonator to the amount of energy lost per radian of phase worth of time at the resonant frequency.

This last one is the one we want here. Q sets the rate of the exponential decay of the amplitude of an oscillation in a resonance. Any resonance.

If I make a resonant bump to prop up the bass roll off of a speaker, I want to make a low-Q resonance. This makes it broader adn spreads the happiness across a larger range of frequency, and it also shortens the time the resonance takes to build up or die down.

Less happily, it also makes the bump less large in amplitude. The geometry of a speaker cabinet/port/driver can limit what can be done in taming the Q of the bump. We wind up with a narrow but tall bump so the bass extension is less... um... extensive. Worse, the bass output in that part of the spectrum takes time to build up and time to decay. Time that wasn't in the original instrument's response.

This can be noticeable and gives what has been called 'slow bass'. Actually it can happen at any frequency where there is a resonance in play.

So, think of a 40Hz sinewave. Arbitrary choice of frequency, nothing special. On an oscilloscope it gives you a simple sinewave. On a spectrum analyser you get a single response showing at 40Hz and nothing anywhere else. Simples.

So now I gate this signal to be on for one second only. To make life easy the on and off gating happens exactly on a zero crossing and so we get just exactly 40 cycles of our bass tone.

Our scope shows a zero volts horizontal line with the block of 40 cycles of sinusoid embedded in the middle of it.

The spectrum analyser shows something much more complicated. It still shows us the 40Hz term, but it has grown sidebands around it. If you think our off/on/off switching is 100% amplitude modulation, and the modulating wave is a single rectangular pulse. To reproduce our burst of 40Hz we need a lot of bandwidth... those off/on and on/off edges have harmonics stretching up to infinity, but getting smaller.

So if we have a speaker with a bass resonance with non-zero Q, then it will limit the frequency range of these modulation sidebands.

What happens if the modulation sidebands are restricted? Then the rate of growth and rate of decay of bass outpout for frequencies near the resonance are slowed.

It gets called 'slow bass' but it's actually narrow modulation bandwidth which gives a clearer mental image of what's going on. It's a time to respond thing.

The same thing happens in radio tuned circuits and filters. An IF filter might be centred on 455kHz, but if it's only 6kHz wide, then the audio the set can receive is going to roll off above 3kHz.... But the filter can let 455kHz through calls out someone from the back of the room. With modulated signals, then bandwidth becomes important.

Hope this helps clear things a little.

Oh, and those coupled resonator speakers are not only quite high Q, there are also two poles of resonance tightening the filtering further. Their response to modulation is especially sluggish.

Transmission lines can be made very low Q, but you have to use artfully distributed stuffing to kill off their throughput before frequency gets high enough for destructive addition to become significant.

DAvid

[Cruisin Marine]

Well written Wrangler- Bravo!

[GrimJosef]

Originally Posted by Radio Wrangler

There is a mathematical process called the Fourier Transform ...

Thanks for drawing attention to this. The transformation between time and frequency can give quick, deep insights into what's really going on.

So now I gate this signal to be on for one second only ... The spectrum analyser shows something much more complicated. It still shows us the 40Hz term, but it has grown sidebands around it ...

Sadly with transient signals Monsieur Fourier's transform runs out of steam - it's most use with signals which don't have any discontinuities either in the signal itself or in its temporal derivatives. Fortunately there's another transform - Monsieur Laplace's - which, if we need to, we can use on the discontinuous signals. If I remember rightly (it was a long time ago that I last used this in anger) it deals particularly well with transients.

Cheers,

GJ

[Radio Wrangler]

They ALL deal with ALL sorts of signals. It's just that for some signals one form rather than another can make things clearer.

An eternal sinewave makes a single line spectrum on a spectrum analyser. On an oscilloscope it makes a sinewave which goes on forever - an infinite data set.

A single fast pulse can be represented on a finite scope screen, but it has an infinite spectrum.

Connect an antenna to a scope and you can't untangle all the radio stations, SMPS etc. A spectrum analyser makes it easy.

For the Laplace domain, with real and imaginary components of frequency, then a vector signal analyser comes into its own.

What the transforms (Fourier, inverse-fourier, Lapalace) allow you to do is to hop from one domain to another, quite easily, so you can do your thinking in whichever domain makes that thinking easiest.

Logarithms are a transform.

Let's say you can handle addition and subtraction, but aren't keen on multiplication and detest division.

Transform your numbers into the logarithmic world, do addition or subtraction, then transform the result back. Lo and behold, you've done that multiplication!

So you must find exponentiation and rooting quite impossible.

Transform to the log world and you need multiplication or division. But you don't like those! so transform to log again. Like on ladders, don't look down. You're now in the loglog world which would be scary if you allowed yourself to think about it. But here you need to only add or subtract. That done, do the antilog transform to get back into the plain old log world, then antilog again to get back into the normal linear world where we started. The only difference is that the nasty exponentiation/rooting has been done.

Transforms are maths equivalent to wormholes and hyperspace. Mathematicians are amazingly lazy and love short cuts.

David

[Herald1360]

Brilliant! Loved that last line.