[netaron]

Hugh,
My sincere thoughts are with you for your recent loss. I have experienced this difficult time myself, it took me a while to come to terms with it, but a few years later, my son was born, so I guess nature has an interesting way of conducting itself.

Now with my elementary questions:

Does a sealed enclosure present a higher impedance drive for an amp, compared to a vented type or any other for that matter?

The speakers I am building will have a sealed enclosure, but I am augmenting the lower octaves with a pair of subs, unfortunately the speaker impedance will be a nominal 4Ohms, which is not the greatest for my AKSA 55N+ or my tube amp.

My second option would be to drive the speakers through the plate amp by utilizing the plate amps low pass filter, but am afraid it may color the sound.
Any thoughts appreciated

Thank you,

Haron
 

[andyr]

Hi Haron,

I can't answer all your Qs but I'll give you answers to those that I can! 

As I understand it, a sealed enclosure is "faster" than a vented.  For a system which plays music and not dinosaur footfalls, this is good!     IMO, good subs are also "servo-assisted" - eg. the ones from www.rythmikaudio.com

Nominal 4 ohm speakers should be fine with your 55N+ ... however, forget it, if you are thinking of can driving your subs with a 55N+ ... that's completely a waste of a gorgeous 55N+!!     What you want for subs is purely "muscle" - not finesse - so plate amps (modified to suit the sub driver) are fine!

You say you are "augmenting the lower octaves (of the speakers you are building)" with a pair of subs.  I totally agree with you that a pair of subs is better for a high quality music system than just one ... because they cause less room-interaction problems.  However, you have 2 ways of adding the subs to your main speakers.  I suggest the subs need to have at least an 18dB LP filter with a -3dB "crossover frequency" of 70Hz or less.

You then have a choice of:
1.  Running your main speakers full-range.  This has the advantage of not "colouring the sound" of the main speakers (by forcing their signal to pass through an active bass HP filter) but has the disadvantage that you are necessarily taking the bass driver of the main speakers down into its LF distortion region.
2.  Using a HP filter to roll off the main speakers at some frequency to suit the LP frequency which you have selected for the subs.

The Rythmik sub which I have has a 24dB sub LP slope (which is adjustable) with a fixed 12dB HP slope, which you specify when ordering.

Regards,

Andy

[AKSA]

Thank you Haron,

Much appreciated, and yes, it will take a while to adjust.

Does a sealed enclosure present a higher impedance drive for an amp, compared to a vented type or any other for that matter?

In a vented box, the resonance of the cone and diaphragm produces a very sharp impedance rise.  This is normally flattened with a Zobel, a C and R in series across the voice coil, which neutralises the reactive peak of the speaker voice coil and mechanical assembly acting together.

In a sealed box, this peak is much reduced because there is no resonant conjunction of the cone and the vented aperture.  Damping is far tighter because the mass of air within the box is locked in and acts as a spring, preventing resonant overshoot.  This reduces the impedance peak, as the excursion of the voice coil is far less and thus back EMF is less.  Thus the Zobel can be much milder.

Andy describes it well.  In subjective terms, the vented enclosure has a flatter response, but the resonant peak makes the bass somewhat muddy, less defined, more 'one note'.  The sealed box, however, needs more and more power with reducing frequency, and the slope, which is around 12dB/octave, starts around 2 x fs.  But it's more musical - swings and roundabouts.

Some here know more about the speaker mechanisms, and might like to correct/add to my comments.

As an aside, we might well ask what the problem is with changes in speaker impedance.  Seldom do I see explanations of why this important.......

Rises and falls of impedance on a driver reflect the inductance of the voice coils, the phase changes introduced by the passive crossover, and the back emf induced in the voice coils by reflected mechanical energy finding its way back into the system after the initial event.  This is highly complex, in a real and an imaginary sense, if you get my drift (picture the 'imaginary' resistance of an Argand diagram used to depict reactive voltage and current flow).  However, for an amp using global negative feedback, changes in impedance are very serious.

Impedance is AC resistance, and it is a vector, described by both voltage and current, which act at an angle to each other, called phase.  Picture a resistor.  As voltage across it varies, so too does current, and they are in synchrony.  A cap is different, however.  When we apply voltage to a discharged cap, current is immediately very high as it charges, but voltage across it builds slowly as it charges.  Thus we say that current leads voltage;  current is maximal when voltage is minimal, and vice versa, since when it's fully charged, current through it is zero.  An inductor has precisely the opposite effect;  a large current change through it causes a large voltage to appear across it;  this is the principle of spark ignition systems in automobiles;  the current through the coil is interrupted, a large voltage appears by Lenz' law across the primary, and the transformer action of the secondary is then used to apply a very large voltage across the sparking plug.

When the phase angle of a feedback amplifier is changed, as it is at the speaker terminals by the reactive nature of the transducer, the feedback mechanism is subtly modified.  Feedback amps, of which both the AKSA and the Lifeforce are examples, require overall negative feedback to operate correctly, and this means that the feedback must be exactly 180 degrees out of phase with the input.  If this requirement is altered by a reactive speaker, then what was negative feedback can become positive, and this causes massive, irreversible and destructive positive feedback, which must be avoided at almost any cost.

Thus while some phase shift is acceptable, large changes in speaker impedance is asking for trouble, and an attempt to flatten the impedance curve will reduce this phase shift, making the speaker generally easier to drive.  In subjective terms, a well designed speaker will return much better spatial characteristics, that is, the imaging will be much superior.

There are some speakers in the market which show alarming impedance changes across the audio range.  One notable speaker I know of ranges from 4 ohms to 42 ohms, an absurd range, and yet costs thousands of dollars.  This sort of poor design is quite common.

I hope this goes part way to explaining the question, Haron, but one could fill a book on this stuff....

Cheers,

Hugh

[David Ellis]

One notable speaker I know of ranges from 4 ohms to 42 ohms, an absurd range, and yet costs thousands of dollars.  This sort of poor design is quite common.

Ouch!  That is quite extreme.  It makes my speakers look good.   



Also, I would like your opinion on something.  One very reputable (previously DIY) manufacturer of loudspeakers conveyed the importance of flattening the impedance, but only flattened the midrange impedance.  He did nothing do flatten the impedance in the bass region.  This made very little sense to me because the propensity of amplifier energy is conveyed in the bass region.  I kinda' thought that flattening the impedance curve would be most advantageous in the bass region.  Of course, the component values are much larger and more expensive.  What do you think?

I have not flattened the impedance of the 1801 because changing the port/tuning OR changing air pressure/elevation will shift the impedance bumps in the bass region.  So, the only part of the impedance hump that is fixed is the midrange hump.  I didn't think this mattered much - especially with the dampening factor of most amps.

Dave

[AKSA]

Hi David,

You have the knack of taking me right to the limits....  I will attempt to do justice to your question, but might disappoint!   

Also, I would like your opinion on something.  One very reputable (previously DIY) manufacturer of loudspeakers conveyed the importance of flattening the impedance, but only flattened the midrange impedance.  He did nothing do flatten the impedance in the bass region.  This made very little sense to me because the propensity of amplifier energy is conveyed in the bass region.  I kinda' thought that flattening the impedance curve would be most advantageous in the bass region.  Of course, the component values are much larger and more expensive.  What do you think?

Let's take a typical 8" driver, the Peerless 830868, which has a voice coil inductance of 1.5mH.

The inductive reactance of this coil at resonance, which is 32Hz (factory spec, NOT measured value!), is 0.3R.  This is dwarfed by the DCR at this point, which is 5.7R, some 19 times greater.  So the impedance peak generated at resonance (assuming the system resonance is around this figure also) is certainly not attributable to the inductive reactance of the voice coil;  this does not start to become significant until around 200Hz (1.9R).  Rather, the peak is attributed to the back emf induced into the voice coil by its movement inside a strong magnetic field, with a small contribution from reflected box energy directed into the back of the cone.  Anything we can do to make the box less reflective will reduce this peak marginally, but actually most of the impedance rise is created by the movement of the voice coil within the driver magnet. (This is tough to visualize;  we move the voice coil with an applied voltage from the amp, but this very movement 'generates' a back emf which works against the amp and is reflected back into it, with a phase change largely dictated by the mechanical and magnetic construction of the driver motor.  This also happens in DC electric motors.)  This back emf will change the phase of the applied voltage/current, and should be ameliorated with a Zobel so that the feedback mechanism of the amp is not phase shifted towards the possibility of destructive positive feedback.  I believe that even small phase deviations from the nominal 180 degrees of negative feedback reduce the differential accuracy of the feedback mechanism and thus cost detail;  it is in the detail that the imaging information (which is largely made up of environmental audio cues at very low level) is found.

As we go higher, towards the midrange, the impedance of the woofer voice coil begins to increase, and by 2500Hz, a typical crossover frequency for a two way, it stands at 23.5R for a 1.5mH voice coil, more than four times the DCR of 5.7R.  This increased impedance means that the phase angle is changing, too, and this will adversely affect the amp.  Furthermore, with an increased impedance, less current will flow through the voice coil, and thus the electro-acoustic sensitivity begins to drop.  However, the tweeter is starting to chime in at this frequency, and since its voice coil impedance is around ten times less, the primary factor influencing its current at this frequency will be its DCR, which is typically 2R9 (rising to a max of 19R and thus defining an inductance around 0.15mH for a Vifa XT25TG30).  Since the tweeter is fed from a high pass filter and the woofer from a low pass, then the reactance components driving them are respectively capacitive and inductive, and these conveniently have complementary characteristics, pulling phase angle back somewhere around the crossover point.  But at the extremes, the lower frequency phase angle is seriously affected by back emf on the coil, and at high frequencies by the predominantly capacitive reactance of the high pass filter on the tweeter in conjunction with the inductance of the tweeter voice coil.

So, I would suggest that the amp will be happier with a flatter impedance curve, not just in the midrange, but also at the bottom end, because of the issues of decreasing sensitivity with increasing impedance, and because of control of the phase shift so necessary to amplifier stability.  As a general rule, my speaker designer suggests a range no more than minimum 4 ohms, and maximum 15 ohms.

Hope this is helpful, and if there are flaws in the argument, let's discuss.  Nothing more dangerous than one man's opinion......

Cheers,

Hugh

[netaron]

Hi Hugh,

So if the impedance "peak" is reduced in a sealed enclosure, what happens to the lower impedance curve?

Does this present a better drive for the amp itself if a sealed box is used, from you explantion I gather, in a sealed enclosure, there is more resistance to the cone going backwards, so it would show as a higher nominal impedance for the amp?

I am mainly intrested in the relationship of sealed vs. vented and how the nominal low impedance is effected as far as the amp is concerned, in all honesty, the answer to my question may lie within your technically sound response, bu I am having a hard time discerning that fact since my knowledge is limited (sorry for being redundent).

Thank you,

Haron

[AKSA]

Hi Haron,

Your questions are highly informed and very tough;  don't apologise!

So if the impedance "peak" is reduced in a sealed enclosure, what happens to the lower impedance curve?

Does this present a better drive for the amp itself if a sealed box is used, from you explantion I gather, in a sealed enclosure, there is more resistance to the cone going backwards, so it would show as a higher nominal impedance for the amp?

I am mainly intrested in the relationship of sealed vs. vented and how the nominal low impedance is effected as far as the amp is concerned, in all honesty, the answer to my question may lie within your technically sound response, but I am having a hard time discerning that fact since my knowledge is limited (sorry for being redundant).

The low frequency impedance peak is reduced in a sealed enclosure because there is more resistance to the cone going backwards, just as you identify.  To explain further, this increased resistance means less movement at the cone, and in turn this means that there is less back emf induced by the magnet in the voice coil, and so more current flows into the system from the amplifier.  This shows as a LOWER impedance peak (not a higher peak!), since same voltage at the input means more current flow.  Thus it helps flatten the peak, which evens the amplitude response somewhat.

You are having a hard time understanding my 'technically sound response', can you imagine how difficult it is for me to articulate something I don't fully understand!!!   

David, help, I need your input here!!   

Cheers,

Hugh

[David Ellis]

David, help, I need your input here!!

I don't really see where help is needed.  Your assertions are all very solid - especially the assertion concerning back-emf.  In this arena, I believe your explanation was extremely well articulated.  I certainly don't see any flaws.

I must also offer candidly that you again overachieved in your response.  Your comments were much more thorough than I had expected.  I was simply curious if you had subjectively tested the impact of flattening the impedance curve or if you were somehow able to measure the impact on the amplifier during live music.  In this latter issue, I am totally clueless how live amplifier testing might be accomplished with a real musical signal.

I'll provide a some further backround.  @ 5 years ago I asked Dennis Murphy to test the impact of flattening the impedance of the 1801 in the midrange.  He accomplished this, and listened to the results.  Subjectively Dennis found no impact.  The amplifier used was a Marsh A200.  I am not certain of the dampening factor with the 1801, but assume the dampening factor is fairly good.  With other amplifiers the impact of a common loudspeaker impedance swing (i.e. 1801) might be more significant. 

I am very aware that extremely few commercial loudspeakers have a very flat impedance.  Could this happen because a flat impedance curve only matters to those folks running SET amplifiers having a high output impedance.  The answer appears to be "yes". 

I'll probably remain curious about the relationship between amplifiers and loudspeakers for a considerable time. 

My hunch is that if a speaker guy actually built a world class speaker that also had 90+db @ 8 ohm sensitivity AND flat impedance... THEN paired it with an amplifier designed for such a speaker... something special might happen.  Unfortunately, the state of common speakers is very different .  Currently, amplifier guys are forced to tolerate the wide variations in loudspeaker nominal impedance and loudspeaker impedance swings.  I am fairly certain this forces unfavorable design compromises for those amplifier guys  .

Anyhoo, I have probably rambled entirely too long on my soap-box.

Dave