Series vs. parallel sound quality

[slicemaster101]

Hi all, I haven’t been here in a wile but I decided to get back into this community that I was very active in a wile back.

Anyways, I just wanted to ask a fairly simple question about wiring speakers. I was wondering if there is any sound quality difference between wiring speakers in series or parallel and vice a versa?

Signed,
Slice

 

[bogoes]

There should be absolutely no difference in sound quality as the drivers are still receiving the exact same signal. The only thing I can think of that would affect the sound quality is that when you connect 2 drivers in parallel, you cut the impedance in half, which usually results in a small amount more distortion coming from the amp...but if you can tell the difference between 0.1% THD and 0.2% THD, then you have better ears than me.

 

[rdf]

Speakers in parallel see the full damping factor of the amplifier, in series they see a high series impedance changing with frequency (damping factor = 1.) High frequency drivers are less affected by drive impedance than woofers, and cabinets can be designed to take it into account in the case of woofers, but unless your amp can't take the load I'd go parallel.

 

[Svante]

Speakers in parallel see the full damping factor of the amplifier, in series they see a high series impedance changing with frequency (damping factor = 1.) High frequency drivers are less affected by drive impedance than woofers, and cabinets can be designed to take it into account in the case of woofers, but unless your amp can't take the load I'd go parallel.

If the drivers are identical, this reasoning is wrong. As an example, two driver with a Qts of 0.4 can be seen as one driver with the a Qts of 0.4 (but doubled Vas etc.). Since the drivers have the same impedance, the voltage across each speaker will be exactly half the driving voltage for all frequencies. It does not matter how a voltage is delevered to a given driver, and the voltage alone determines the sound radiated from the speaker.

If the drivers are slightly different, the result will still be about the same. This is because the sum of the sound pressures is proportinal to the sum of the voltages across each of the drivers and this sum is the total driving voltage.

If the drivers are very different, the driver with the largest impedance at a given frequency will "take over" at that frequency. But the sum of the two will still be proprotinal to the driving voltage.

 

[rdf]

Your reasoning ignores driver back EMF - the voltage generated by the speaker as it's coil moves through the magnetic field - and the amplifier's job damping it. System Q (Qec') is almost completely determined by electrical Q, defined by Qec'=Qec(Re + Rs)/Re where Qec is driver electrical Q, Re is driver resistance and Rs source. Yet your reasoning makes sense, damn if I can resolve the paradox at the Thiel-Small level.

 

[Svante]

Nope, my reasoning does not ignore the back EMF. If one of the speakers had been a series resistor, I would have ignored it. Since the two driver's electrical impedances and back EMFs are identical, the back EMF of one driver is just right to cancel the back EMF from the other and vice versa. This keeps the voltage over each of the drivers to exactly half the total driving voltage. This would not have been the case with a series resistor or if you were to put the hand on one of the drivers to keep the cone at standstill.
But I think that this way of thinking is more complicated than the simple voltage divider with two identical impedances.

So, the bottom line is; connecting identical drivers in series is OK, it does not affect Qts.

Edit: you can also see it from the equation for Qes:

Qes=2*pi*fs*Mms/Res, Res=(Bl)^2/Re

If you add a driver in series and see the two drivers as one, B and fs would stay the same, Mms would double, l would double, and Re would double. Net effect: Qes stays the same.

 

[moamps]

.....Since the two driver's electrical impedances and back EMFs are identical, the back EMF of one driver is just right to cancel the back EMF from the other and vice versa..... [/B]

Back EMFs are in phase. (inductors in series)

What JBL says about this....strange...

 

[rdf]

No, Svante appears to be correct, though I was going to argue Qes is not the same as Qes'. Here's a 2000 Usenet post from a man who knows speakers, a professional consultant in the field for decades.

In article <kQiX5.1753$qs2.401109@dfiatx1-snr1.gtei.net>,
Bill Whitlock <bill.whitlock@verizon.net> wrote:
>Perhaps I confused you with my answer. To restate, I believe it is OK to
>wire two 4-ohm speakers in series to present an 8-ohm load to the
>amplifier, provided that the speakers are identical.
>
>Before reading the article by Richard Clark, I believed that doing so
>would have a detrimental effect on damping factor (which can affect the
>speaker's frequency response, especially at low frequencies). Others in
>the audio field still believe this is a controversial issue.

And it is only a belief, unfounded by any physics.

The notion is thatputting one speaker in series with another can
adversely effect that damping. This conclusion is based on a
highly flawed model of what the damping in the speaker REALLY
is and what it is due to. To dispel this falsehood, it's
necessary to do a more detailed analysis of what's really going
on using the actual parameters involved.

The damping of the speakers is due to very specific mechanisms,
primarily mechanical (due to frictional losses in the
loudspeaker suspension, and some minor absorbtion losses) and
electrical (due priomarily to the DC resistance of the voice
coil). The amount of damping is measured using the "Q" factors
due to each contributing source.

The measure of mechanical damping or Q, Qms is:

Mms
Qms = 2 pi Fs -----
Rms

where Qms is the Q factor due to mechanical losses, Fs is the
resonant frequency, Mms is the moving mass of the system, and
Rms is the mechanical (frictional) losses. Notice that there is
no implicit term relating how the Qms is affected electrically,
i.e., how it is connected to the amplifier. Now, with two
drivers, we have double the mass, Mms, because we have two cones
instead of one. But we have also have twice as much suspension
and thus twice as much suspension loss, Rms. Thus the ratio of
Mms to Rms remaines the same, and thus the mechanical Qms
remains the same.

The measure of electrical damping, or electrical Q, Qes is:

Mms
Qes = 2 pi Fs ----- Re
2 2
B l

where Qes is the Q factor due to electrical losses, B is the
flux density in the magnetic gap, l is the length of the voice
coil wire immersed in the magnetic field, and Re is the DC
resistance of the voice coil. Hooking two such system in series
does three things:

1. It doubles the moving mass Mms,

2. It doubles the DC resistance Re,

3. It doubles the length of wire in the magnetic field l.

So while the intuitive leap might be to suggest that because the
resistance Re is doubled, and therefore the Qes is similarily
doubled, doing so completely ignores the other two effects.

Let's look at the ENTIRE analysis. Let's double all the
necessary quantities and plug them back into the equation and
see what happens. For Mms, we'll substitute 2*Mms and so forth:

2 Mms
Qes = 2 pi Fs --------- 2 Re
2 2
B (2 l)

and:

And we can simplify by gathering the factors together. In the
numerator, we have 3 instances of factors of 2, for a total of
8, and there is a single factor of 4 in the denominator:

8 Mms
Qes = - pi Fs ----- Re
4 2 2
B l

Simplifying one step further, ye factoring out the common factor
of 4 now in both the numerator and denominator, we end up with:

Mms
Qes = 2 pi Fs ----- Re
2 2
B l

which shows that the electrical damping for two systems in
series is identical to the electrical damping for a single
system alone. Q.E.D.

This contradicts the seemingly intuitive notion that two drivers
in series must be severely underdamped as a result of the
additional series impedance. It's not the first time that
intuition has failed to coincide with physical reality in audio.

--
| Dick Pierce |
| Professional Audio Development |
| 1-781/826-4953 Voice and FAX |
| DPierce@world.std.com |

I went back to the Thiele Small derivations for the first time in twenty years and they bear this out, the components defining this behaviour appear in series in a driver's equivalent circuit and sum in a linear manner as described. Completely counter-intuitive to me and with some interesting consequences, but physics is physics.
Series those puppies up.

 

[Paradise_Ice]

There should be absolutely no difference in sound quality as the drivers are still receiving the exact same signal. The only thing I can think of that would affect the sound quality is that when you connect 2 drivers in parallel, you cut the impedance in half, which usually results in a small amount more distortion coming from the amp...but if you can tell the difference between 0.1% THD and 0.2% THD, then you have better ears than me.

I think there is a difference on the amplifier for a start.

Running lower loads on amplifier is not good for the damping factor for a start, and the distorsion can be much higher than 0.2%, more like 5% at high volume levels, this depends on the amplifiers design but am talking average.

I had an old Creek CAS 4040 and i had 2 sets of speakers on the poor little amplifier in parallel , it sounded loader with 2 sets of speakers but when i went back to 1 pair it sounded better and more focused.

Who wants louder over quality of sound?

 

[Timn8ter]

May we take this one step farther? I have two identical drivers. I get nearly the same sound quality wired in parallel along with a series resistor that I do when I wire them in series without the resistor. I'm assuming increased Qe. Correct?

 

[Svante]

Back EMFs are in phase. (inductors in series)

What JBL says about this....strange...

Well, then JBL is wrong. My explanation with the cancelling EMFs may suffer from bad wording, so I'll give it another try:

As I said I very much prefer the explanation that the impedances are identical, and this ensures that the voltage across each driver is exacly half the driving voltage. Still, let's try to look at each impedance as an electrical impedance Ze and an EMF. When two such identical circuits are connected in series:

V+ ------Ze1-------EMF1-------*------Ze2-------EMF2-------GND

...and the voltage V+ is applied, there will be a voltage drop over each of the Ze:s and the two EMF:s will appear. Now, if there had been no EMF1, the voltage at the midpoint (*) would have deviated from V+/2 (due to the presence of EMF2). My point here is (which I unclearliy called cancellation in my previous post) is that EMF1 assures that the midpoint assumes the voltage V+/2. The fact that we have a mechanism that forces the voltage to V+/2, makes the circuit equivalent to having two separate speakers driven by two separate voltage sources of V+/2.

If I really were to defend my statement that the EMFs cancel, I could redraw the schematic above like this:

V+ ------Ze1-------(+)EMF1(-)-------
GND ------Ze2-------(-)EMF2(+)------*

Then we see that if EMF1 pulls the midpiont voltage upwards, EMF2 would pull it downwards by the same amount. But then, that would just be a poor excuse for my bad explanation. Sorry about that.

 

[Svante]

May we take this one step farther? I have two identical drivers. I get nearly the same sound quality wired in parallel along with a series resistor that I do when I wire them in series without the resistor. I'm assuming increased Qe. Correct?

In the first case (R in series with the two paralleled drivers), the effective Qes (and Qts) of the drivers will increase. In the second case (two drivers in series) it will not. So the sound will be different (and efficiency will be lower in the first case).

 

[moamps]

Well, then JBL is wrong....

Svante,

I tend to agree with you that the JBL must have made a mistake. A simplified explanation of the problem may go more along these lines:

Electrical Speaker Dampening is determined only by current Iemf, which is induced by voltage Vemf. The value of Iemf is determined by the total impedance in the circuit. If we assume that the output impedance of an amp is zero and that the connection cables are ideal, then the only impedance in the circuit is that of the speaker coil Ze. Iemf is then:

Iemf=Vemf/Ze

If two identical speakers are in series, then the total voltage VEMF is 2xVemf because these voltages are in phase. The total impedance is then ZE=2xZe and IEMF is calculated as:

IEMF=VEMF/ZE=2*Vemf/(2*Ze)=Vemf/Ze=Iemf

It follows that each speaker, given the same Vemf, generates the same Iemf. However, as this applies only to ideal cases, and we know speakers cannot be 100% identical, the voltages Vemf generated by the speakers cannot have the same value. Thus, it can be said that the damping factor is a bit lower for serial speaker connections but the serial connection cannot cause the factor to drop below 1, IMO.

Regards
Milan

 

[Svante]

It follows that each speaker, given the same Vemf, generates the same Iemf. However, as this applies only to ideal cases, and we know speakers cannot be 100% identical, the voltages Vemf generated by the speakers cannot have the same value. Thus, it can be said that the damping factor is a bit lower for serial speaker connections but the serial connection cannot cause the factor to drop below 1, IMO.

Actually, for a given frequency, and if the drivers are not identical, one driver will see a damping factor slightly less than infinity, and the other will see a negative damping factor of the same magnitude. And still, the sum of the two outputs, will still be proportional to the driving voltage. In other words, a slight difference between the drivers will still produce the same acoustic output, but one of the speakers will contribute slightly more to this sum, than the other.

So, if we can live with that one speaker delivers 51% of the sound pressure and the other delivers 49%, serial connection of two slightly different speakers is still OK.

 

[Daveis]

May we take this one step farther? I have two identical drivers. I get nearly the same sound quality wired in parallel along with a series resistor that I do when I wire them in series without the resistor. I'm assuming increased Qe. Correct?

A similar question I have to this, but I am using active amplifiers and crossovers..

Assume my amplifier can handle 2 ohm loads or less (big Threshold Class A).

Which will sound better or measure better
a) speakers wired in parallel
b) speakers wired in series

I had a physics teacher in high school measure the output of a single driver and two drivers in series, and two in parallel. We played a sine wave through them. I don't remember if he placed an oscilloscope or microphone to measure them. But the waveform I saw was not a sine wave. It appears like a sine with another wave of lower amplitude riding on it

My PHD physics teacher kind of mocked series speaker connection.
Now this was a smart guy. And I remember what I saw.

I thought he was measuring the two series speakers with a microphone. And I admit at the time (I was 17 years old) that I thought the series connection sounded a bit fuzzy and less clear than the single driver.

Has anyone actually measured the THD of two drivers in series versus two drivers in parallel? Played at the same volume.

I guess I just need to try it for myself before I'm a believer that series connection isn't a sonic compromise.

Yeah I know guitar cab speakers and line array speakers are often series connected. But guitar cabs aren't a model for fidelity and honestly I don't like the sound of line arrays. They always sounded a bit off to me... kind of harsh sounding even at low volume levels.

 

[TimA]

Quote...
"Hooking two such system in series
does three things:
1. It doubles the moving mass Mms,
2. It doubles the DC resistance Re,
3. It doubles the length of wire in the magnetic field l."


Surely, when two drive units are connected in series the Qes must be calculated for each speaker individually. Unless I am missing something the Mms and length of wire in the magnetic field for each unit remains unchanged while the Re is doubled, resulting in altered Qes.

Tim.

 

[CLS]

I myself ran into this problem recently.

I built a pair of 3-woofer OB bass, while I got no amp to handle 2.x ohm load properly, so I wired them in series to match one of my SE tube amp.

Nominal impedance should be 3x8=24 ohm, I just hooked them to the 16ohm tap. This virtually raises the load impedance of the output stage of that SE tube amp and decreases the output power. (Well, I don't mind some loss in power... )

Problem is, this is bass module and very much of its operating range falls around the resonant impedance peak region. So the summed impedance of these 3 woofers is skyrocketing when the frequency approaches the resonant point, maybe hundreds of ohm!

So, when the summed impedance rises into 3-digit range, the operating point of amp falls off the optimum furthur (may I say it's just wayyyyy too far). With so high the load impedance, so less current through the output stage, I wonder how all these combinations can work properly together.

Except for the back EMF's of drivers, now the amp also joins in the crowded party.

To make things even worse, these 3 woofers are not the same. 2 are Eminences, 1 is EV. So probably their different back EMF are fighting each others. (I think of a picture -- 3 different coil springs with different masses attached to them in a string, bouncing up and down, pushing each other in an unexpected messy manner.... )

So, I did one thing last night, I paralleled 33ohm resistor to each woofer. I guess this does 2 things:

1. Total 99ohm is paralleled to the nomial 24ohm woofer-string, bringing down the summed impedance to about 19ohm, which is closer to the 16ohm tap and much more "stable". No more skyrocketing to 3digit number.

2. Locally paralleled resistors help damping each woofer's back EMF. (or, do they?)


I have no sophisticated equipment to measure all these. By ears, the bass seems more controlled and less boomy, but it could be mental, I'm not so sure.

Any thoughts?

 

[Graham Maynard]

I would think, and as has been observed here by trying, that the floating connection between two drivers in series will follow half of the applied drive potential only when both drivers, plus their individual loadings, are identical.

Asymmetrical driver positions, or different damping material effects, within a common cabinet, or different external driver positions wrt a nearby reflecting plane or within a room, leads to the development of individual driver impedance peaks, which series interact and affect the centre potential, thus each idividual driver can mutually affect the drive and damping for the other.

Cheers ............. Graham.

 

[Pan]

Surely, when two drive units are connected in series the Qes must be calculated for each speaker individually. Unless I am missing something the Mms and length of wire in the magnetic field for each unit remains unchanged while the Re is doubled, resulting in altered Qes.

Tim.

As Svante wrote:

Edit: you can also see it from the equation for Qes:

Qes=2*pi*fs*Mms/Res, Res=(Bl)^2/Re

If you add a driver in series and see the two drivers as one, B and fs would stay the same, Mms would double, l would double, and Re would double. Net effect: Qes stays the same.

You must realise that an active device like a speaker driver in series with another is not the same thing as putting a resistor there instead.


/Peter

 

[sreten]

Hi,

There are practical considerations (not basic theoretical) that will
mean there is a difference, not least amplifier loading / damping.

But lets consider effectively identical cases except one is 2 16
ohm drivers in parallel the other two 4 ohm drivers in series.

Would it matter significantly which one you choose ?

The answer is yes it can, for at least one mechanism I'm aware of,
reported from high level driver testing, related to cone jumping.
Instability of cone position is a known possible problem with drivers
and it was reported by the tester that the series connection showed
instabilities not apparent in the parallel case, especially in regard
to apparently (IIRC) opposite (!) direction cone jumping.

What is certainly true is back EMF's do not interact electrically
(except due to acoustic coupling) for a voltage amplifier in the
parallel case, and they must do in the series case.

/sreten.