Ariel's

[James D]

Just re-read through the above and it isn't all correct after all

I'll correct it tomorrow...

J

[petercom]

When you do take into account the following:

Some valve amplifiers have a single secondary with taps taken off it - this is what I meant by 2.83v across the 8 Ohm tap and 1.41v across the 4 Ohm tap. This is worst case. Luckily all 'decent' valve amps have dedicated secondaries for designated impedances.

The humps that you see in speaker impedance below 300Hz are due to the resonant nature of the driver in the cabinet. Generally we don't want to push power into the driver through the resonant peak area as a little power goes a long way when you excite a high Q resonance! So it is the minimum impedance in this area that we are concerned with.

However amplifier damping factor through these resonant peaks is useful. I am not sure how output transformers cope with the mismatch in impedance and abrupt change in phase angle in this respect.

A rise in impedance through the crossover region is typical and is often accompanied by abrupt changes in phase angle. Luckily for a two way there is not much power being delivered above 1kHz. A three way speaker, however, may make considerable demands on amplifier power delivery through the 200 - 400Hz band if the crossover is, as is likely, in this area.

[James D]

Haven't time to correct my maths today.

I thought I had covered the 'single secondary with taps off it' case in my reply.

more later

James

[Paul Barker]

Hi Peter,

say you take a 3k primary and a tapped secondary first take off is the 4 ohm at a turns ratio of 27.39:1 so 230v test on primary yields 8.4v final take off is at ratio of 19.35 for 8 ohms = 11.88v, so you don't expect to see half the 8 ohm voltage on the 4 ohm tap.

Although the better practice is to wind multiple secondaries and connect them according to impedances you want so that all of the wire in the transformer is always in use, in practice it isn't too much of a compromise to tap offan 8 ohm winding for 4 ohms or a 16 ohm for 8 ohms. The difference in sound is fairly negligable and it only affects hf performance in the range above 20khz.

If you are specifying transformers for a future design you could have prototypes made in both winding methods and audition the results. There are other influences on the sound of transformers that make more significant difference like core material and just the quality in which the layers are built up, whether the whole width is used for every layer (which entails careful prototyping to achieve a turns ratio wire guage and bobbin to make the best of all these demands). It is just as much a compromise to get that wrong and not fill every layer to meet the turns ratio needs as it is to just tap the secondary for impedances. ALways better to fill the layer so if that upsets the turns ration let it, impedances which vary by 20% are not audible, but a badly filled bobbin is (poor HF performance).

[James D]

Rather than go through my previous post I decided to start again but with the correct numbers (or what I believe are the correct numbers). Also I suspect that my meaning hasn’t made it into my words – judging from Peter’s last post. I apologise for taking this back to basics and restating things that are obvious. I think it is necessary to communicate what I think I understand about this.

Let me start by stating what I understand to be the questions – please shout if I haven’t answered the ones you wanted answered (anyone not just Peter).

1)What happens to a Valve amplifier with OPT when it is loaded with different impedances regarding power delivery?
2)Is this different from a conventional solid-state amplifier and if it is different how is it different and why is it different?
3)What does this mean for the stated sensitivity of loudspeakers and how should this be interpreted?
4)How does the impedance curve and current/voltage phase angles affect a valve amplifier driving a complex loudspeaker?
5)What effect does Damping Factor have on the amplifier/loudspeaker interface and power delivery.

I hope that covers it all.

An attempt at answers:

1)I guess I should start out by stating that there are three operational regimes that an audio power amplifier can operate under regarding output impedance and power delivery into a load and these have different reactions to varying the load, these are:

• Voltage source regime, where the power amplifier will attempt to maintain the same voltage across the load under all conditions. This results in a doubling of power into the load each time the load is halved. This is the conventional way of designing an audio power amplifier and so is the assumed behaviour for conventional measurement schemes. Under this regime if 2.83V is generated across an 8 ohm load for 1W of power and the load is reduced to 4 ohms then we still have 2.83V across the load but now 2W of power is delivered.
• Power source regime, where the amplifier will attempt to maintain a constant power delivery into the load under all conditions. This results in the same power being delivered into the load each time the load halves. This is the way a conventional valve amplifier works via output impedance taps on the OPT. It is also the way a Class A triode amplifier works to a first approximation on the same tap. Under this regime if 2.83V is generated across an 8 ohm load for 1W of power and the load is reduced to 4 ohms then we now have 2V across the load and we still have 1W of power
• Current source regime, where the amplifier attempts to maintain the same current into the load each time the load is halved. This results in a halving of the power delivery each time the load is halved. Under this regime if 2.83V is generated across an 8 ohm load for 1W of power and the load is reduced to 4 ohms then we now have 1.41V across the load but only 1/2W of power.

So what happens with a valve amplifier depends what type of valve amplifier it is. Most valve amplifiers are power sources and maintain the same power into the load providing the OPT has taps that allow the load to be connected to the appropriate winding. Some Valve amplifiers do not have output impedance taps, as Peter has said, and so have only a limited ability to cope with impedances that vary from the designed load. If say a Triode Class A amplifier has just a single 6ohm output then it will cope well with an 8 ohm or 4 ohm load or anything in between but it won’t be so happy with impedance variation away from this interval i.e. a 4ohm speaker that was actually 2ohms for a significant part of the bass region is only going to get about half of the power that a 6 ohm speaker would (assuming the 6 ohms was through the bass region or some large part thereof).

So for the example of Ariel’s and their 4 ohm rating run from a valve amplifier, then providing they are connected to a 4 ohm tap on the OPT they will get the full power that the amplifier is capable of producing…

...more to follow below

[James D]

Taps & Windings. It doesn’t make a difference, to a first approximation, between reconfiguring separate secondary winding to achieve the output impedance differences or having multiple taps on a single secondary winding. In both instances the turns ratio changes to reflect the expected load impedance back as the operating point primary impedance for the valve. The sound will change slightly and the efficiency of the transformer will change as the coupling between primary and secondary is effected but the turns ration will be as expected so power delivery is unchanged – to a first approximation which in this case is a less than 3% change.

2)A conventional solid-state amplifier is a voltage source so it does behave differently from a valve amplifier (see answer 1 above). The transistor amplifier will double its power delivery into halved load impedance. So Ariels at 4 ohms would get twice as much power at the same volume setting as an 8 ohm speaker, since the volume control regulates the voltage the amplifier delivers.


3)I think this is the crux of Peter’s query and in many ways I think it is a problem with the assumptions and definitions of the way loudspeaker sensitivity is measured and used.
I guess it starts simple – loudspeaker sensitivity is a function of power delivery into the speaker. Simple so double the power and double the measured sound output from the speakers (not the same as doubling how loud it sounds to a listener). Then the problems start – how do you measure the power being delivered to the speaker? School physics would tell you that you measure the voltage across and the current passing through, the load and multiply them but that only really applies to a resistive load with a dc current applied. For the ac case and a changing with frequency non-resistive load it starts to get complicated. Hence the convention of measuring 2.83Volts RMS across an 8 ohm load speaker and measuring the sound output. Fine as long as all the measured speakers are 8 ohms but if, like Ariels, they are 4 ohms what do you do? Do you scale the output for the equivalent loudness of an 8 ohm speaker when driven with the same power? Or the same voltage? Or the same current? Well here we start with some definitions:

·Loudspeaker sensitivity is given as its voltage sensitivity and is the sound output level for 2.83V across its input and it is assumed that the amplifier is a voltage source and will generate that 2.83V required regardless of the impedance of the loudspeaker. It is further assumed that for a given loudspeaker the measured sound output varies with applied voltage only so that as long as the same voltage is applied to the speaker then as the frequency varies the current and power delivered at that frequency might vary but the sound output level wouldn’t… (This for the pass band of the loudspeaker).

·Loudspeaker Efficiency is given as power efficiency and is specified as the acoustic output in dB for 1W input measure at 1m from the loudspeaker. This is hard to measure but is easy to understand as a characterisation of the loudspeaker and easy to understand when comparing different loudspeakers with different amplifiers. BUT it is damned difficult to measure in a consistent manner

The problem comes when trying to relate measurements under one set of assumptions to the other… For solid state amps, sensitivity is quite a nice way to compare between them but it doesn’t apply to valve amplifiers as they don’t meet the assumptions behind the measurement system so is not valid for them. On the other hand efficiency works rather nicely with valve amplifiers and actually can be measured fairly straight forwardly by measuring the voltage and current signal changes on the anode of the power output valve (or valves for PP or PSE). If you know the OPT efficiency then it is easy to calculate the output power into the loudspeaker. But this doesn’t work for voltage source amplifiers…

So we have a clash of measuring schemes and amplifier types and what suits one doesn’t suit the other… For that reason I always stick to Loudspeaker Efficiency when specifying efficiency or sensitivity but I might call it sensitivity - so I’m sloppy with terminology but not with measuring schemes.

How does this bear on the Ariels? Well Lynn claims 92dB/1m he doesn’t say for 1W that I can see so it might be for 2.83V across the nominal 4 ohm impedance of the Ariels in which case it is for 2W and it really is only 89dB/1W/1m – which is Peter’s point. I guess we can settle for that. Drive them from the 4 ohm tap and uses at least 8W a channel and preferably 16W. And that is what LO recommends so I guess it all makes sense…

Oh on websites I have seen claims of 93dB and 94dB sensitivity – but again not with any specified impedance, voltage or power level…
4)This question is a beauty! And I’ll answer it later as I will 5).

J

[petercom]

Thanks James for the elucidation so far. Next is how is the efficiency of power transfer affected by what appears to be a mismatch with the specified secondary impedance. I have seen graphs which show power delivery falling off as rapidly at, say, three times the specified impedance. What I don't know is how far up or down the valve characteristic load line you move with such a mismatch and how this affects the operating characteristics and distortion of the valve?

/quote
Oh on websites I have seen claims of 93dB and 94dB sensitivity /unquote

There are all sorts of ridiculous claims for speaker sensitivity these days and, as Noel's measurements in HFW show, very few of them are met in practice. Even then, if the sensitivity is based at 1kHz is this any indication of the subjective efficiency of a speaker with a falling bass response?

There is a speaker advertised in HFW with a claimed 99dB/W sensitivity which actually measures 91dB over its power bandwidth for a 4 Ohm load. This is very misleading to valve amplifier owners.

Some common sense needs to be used by the buyer. The laws of speaker physics are fairly straightforward. In a closed box a small diaphragm speaker (under 20cm) will have difficulty achieving 86dB for 1W/8 Ohms. In a reflex box the efficiency could, theoretically, be raised by 3dB if the drive unit has a rising output with frequency but, generally, 87 to 88dB is the norm.

Increase the diaphragm size above 20cm and you begin to squeeze out a few more dB. 89dB at 1W/8 Ohms is perfectly possible for a 22cm or 25cm driver in a closed box whilst still retaining hi-fi bandwidth and quality.

So are there other ways of achieving higher efficiency? One way is to reduce moving mass of the diaphragm but, with a strong motor system, you lose bass output. So this has to be 'strengthened' by some form of resonant enclosure - a quarter wave pipe for example. And this shouldn't be highly damped, so an amplifier with high output impedance (or thin speaker wires) will give an apparently stronger bass output (but with less control over the drive unit). Another way is to fit a massive motor system to a large diaphragm. Again, though, bass output suffers and you need a very, very large bass reflex enclosure to bolster it (this is the basis of high efficiency pro speakers).

The only real way of increasing speaker efficiency is to improve the coupling (energy transfer) of the diaphragm to the air. This is the function of the horn which compresses the air in the region of the diaphragm to better match the impedance of the driver. The dimensions of the horn then expand gradually until the pressure wave exiting the mouth of the horn meets the air in the room at nearly the same impedance. This method can easily yield 100dB at 1W/8 Ohm from a fairly small driver given a big enough horn. Once again, though, you need a large horn (read enclosure) for a good bass response.

So there is no way round it. You either have a large box and not much amplifier power, or a small box and lots of amplifier power.

[speakerman19422]

Being a new member to the group I found this thread interesting. The 1st speakers I built had bextrene cones back in 1979. There is still an intersest in Kef bextrene drivers.. Fried used bextene in many of his earlier designs. The Dalesford units were not as widely known. Fried had custom made high output Dalesford drivers used in his early t-line designs.

[hal55]

How I long for the good old days when Peter was a regular contributor to the forum.

Hal55