Propeller Head Plaza

Technical and scientific discussion of amps, cables and other topics.

Return to Propeller Head Plaza

Drivers is series is bad. Parallel good. Or is it? (Long)

207.216.246.51
Posted on October 15, 2009 at 11:59:47
Presto
Audiophile

Posts: 2946
Location: Canada
Joined: November 10, 2004
 I propose that the claim that series wiring of pseudo-identical drivers is an inferior method to parallel wiring be re-examined. I offer theoretical reasoning why I believe that serial wiring is no less inferior to parallel wiring when drivers are different. I invite others to comment on this reasoning and provide supporting evidence or conjecture. I have no intention to flame believers of the "Serial is bad" camp, but do ask them to come forward with some logical explanations of why they believe this to be the case that goes beyond the commonly accepted (and insufficient) explanation that "the back EMF of the bad driver influences the good one".

Background: When two impedances are in parallel, the voltage across them is the same. When the impedances are the same (vectorally), the currents will also be the same, and in phase with one another. The net gain is 6db (3db for adding the 2nd driver and 3db for halfing the net impedance). When two impedances are in series, the opposite is true. The voltage across each will be a function of it's vectoral impedance and the net current which is the same for BOTH impedances because they are in series. The current will be a function of the applied voltage to the net (summed) impedance of the two serial impedances. The net gain is 0db (3db for adding the 2nd driver and -3db for doubling the net impedance).

An easy way to remember it is:

"Parallel - voltage the same, currents differ"
"Series - current the same, voltages differ"

The claim: When drivers are IDENTICAL in all mechanical and electrical aspects, serial connections are not worse than parallel - just different. But when drivers are different (and they commonly are) parallel is better. This is because in serial connections, the different back-emf's of the different drivers will interact with one another resulting in some form of distortion.

The counter-claim: When drivers are NOT identical, NEITHER connection is superior. The nature of what is causing the difference needs to be indentified to determine whether serial or parallel connection is the better method.

a) Drivers that are different have different T/S parameters. Things that COMMONLY make drivers different are differences in the mechanical suspension systems. These differences are often attributed to spider inconsistencies, different amounts of adhesive used, and other quality control variations. Motor strength (flux in the gap) may also differ. Electrical impedance of the coil, coil geometry and coil inductance are all seemingly the easiest of these factors to control.

b) Drivers that are different will (indeed) present a different back-emf to the applied emf. In parallel circuits, this is why the current through each impedance is different. In serial circuits, this is why the vectoral voltage drop across each impedance is different.

c) The question comes up: Which is the better case when two drivers are non-identical? To have the same VOLTAGE applied to each? (parallel connection) or to have the same CURRENT running through each? (series connection). In the first (parallel) case, there is waveform distortion because the out of spec driver is seeing a different current than the in-spec driver thus causing a distortion of the acoustic waveform. In the second (serial) case NEITHER driver is seeing a current that corresponds exactlty to the total driving force (applied voltage), and despite the fact both drivers are seeing the exact same current (average current) their acoustic outputs are still different because the drivers are different. (driving the same current THROUGH different drivers is as ineffective as applying the same voltage TO different drivers).

d) Back emf simply counteracts the applied voltage. It is seen by the applied voltage as an increase in overall impedance. In series, each reactive load will present a back emf to the portion of the voltage (drop) it is experiencing according to it's relative impedance to the other load.

Series Vt = V1 + V2, It = I1 = I2 (V1 = V2 only when Z1 = Z2)
Parall Vt = V1 = V2, It = I1 + I2,(I1 = I2 only when Z1 = Z2)

(Where all values above are vectoral quantities)

Claims that in series "one driver affects another" is not untrue. But is nothing to do with back emf. The voltages divide across the impedances such that VT = V1 + V2 and are different simply because the impedances are different. Thus, neither V1 NOR V2 are the "correct" voltage even if ONE of the two drivers is in spec. NEITHER driver will produce the net acoustic output, but both drivers will produce the same (average) acoustic output. The net resultant (combined) acoustic output of both drivers will be halfway between the correct output and the incorrect output. The more "different" the drivers are, the more away from spec the NET resultant acoustic output will be. The acoustic output will represent the average of the in-spec driver and out-of-spec driver no matter how different they are, but the MORE different they are the MORE the net acoustic output will be different from that expected.

But let's look at parallel. It's claimed that one driver does not affect another. Both drivers see the same voltage. But is this a cure? No and here is why: The PROBLEM in the parallel connection IS the fact the voltages applied are identical. What will then differ is the currents going through each coil (and ultimately it is current, not voltage, that results in the electro-motive force and resultant mechanical excursion). The in spec driver will see the correct voltage and the correct current and produce the correct acoustic output. The out-of-spec driver will see the correct voltage and the INCORRECT current. The net resultant (combined) acoustic output of both drivers will be halfway between the correct output and the incorrect output. And again, exactly as per the serial connection, the acoustic output will represent the average of the in-spec driver and out-of-spec driver no matter how different they are, but the MORE different they are the MORE the net acoustic output will be different from that expected.

Conclusion: Neither serial and parallel connections can account for driver differences and both methods result in "averaging" of the acoustic output between the in an out-of-spec drivers. Serial connections are not necessarily worse. A serial or parallel connection may be preferred depending on WHICH SPECIFIC T/S parameters are out-of-spec. The importance of ensuring drivers are within specified tolerances is emphasized here, and how series OR parallel connections will BOTH result in currents (and acoustic output) that does not conform to the applied voltage, thus resulting in distortion of the expected waveform. A crude approximation of the different scenarios is as follows:

Parallel: [(Driver 1) + (Driver 2 + error)] / 2 = final output
[D1 + (D2 + E)] /2 = FO
(D1 + D2 + E) / 2 = FO

Series: [(Driver 1 + error/2) + (Driver 2 + error/2)] /2 = final output
[(D1 + E/2) + (D2 + E/2)] /2 = FO
(D1 + D2 + E/2 + E/2) /2 = FO
(D1 + D2 + 2E/2) /2 = FO
(D1 + D2 + E) / 2 = FO

So, whether ONE driver reproduces ALL of the error (parallel connection) or BOTH drivers each produce HALF of the error, or the average error (series connection) the summed acoustic output is identical in both cases! The error of the one driver is STILL averaged into the final output in both cases. The serial case may have an advantage over parallel because both drivers are seeing the same CURRENT. 'Identical current through' may well trump 'identical voltage applied' as it is current that results in the mechanical force in a moving coil (dynamic) loudspeaker - not the applied voltage.

The theory presented here sounds good. Believe it? Or not? And why?

I leave the floor open to contributions or comments by others...

Cheers,
Presto

Let me restate the premise., posted on October 15, 2009 at 21:04:03
Presto
Audiophile

Posts: 2946
Location: Canada
Joined: November 10, 2004
 I think you guys did not get my main idea here.

Yes, I acknowledge that impedance matching to commonly used amps is important, but this was not the main concept here. ie., it would not make sense to put two 8 ohms speakers in series for a 16 ohm load nowadays. Most manufacturers who use dual 8 ohm drivers put them in parallel, and force their customers to obtain a 4-ohm capable amplifier. This is not the crux of the series/parallel debate though.

What my point was is that if a designer could choose between two "Equally good" woofers (say) for a TMWW design, and could choose between two 16 ohm woofers (parallel for 8 ohms) and two 4 ohm woofers (series for 8 ohms), there is a contingent that says the parallel connection is the ONLY way to go for audiophile speakers. This is due to what they claim is an "inequal loading effect caused by back emf of potentially different speakers within the same lot".

That "theory" (that there is an inherent flaw in the series connection) was the concept I was trying to explore. Not impedance selection considerations for commonly available amplifiers.

Cheers,
Presto

RE: Drivers is series is bad. Parallel good. Or is it? (Long), posted on October 15, 2009 at 20:15:59
kurt s
Audiophile

Posts: 68
Joined: October 12, 2009
 >>'Identical current through' may well trump 'identical voltage applied' as it is current that results in the mechanical force in a moving coil (dynamic) loudspeaker - not the applied voltage.<<

Do not look at it like this. This is just not the case as Ohms law works fine on a speaker load. POWER drives the speaker, which can be V^2/Z or I^2*Z. Just because the coil is there does not make it current driven only just as an electrostatic is not voltage driven only.

In a speaker, there is lossy residual coil R plus the R component that is lost in magnetic hysteresis plus the actual air resistance that is power converted to air vibration and that part is the actual R load that is productive work done by the drivers. Voltage energy potential is needed to do work. Or current into the load is needed to do work. Either way is the same way, P=V*I.

The engineering is how you want to apply that array of speakers, and both are good ways. You can design the speaker system to be high impedance and match to high output impedance amps for max power output. Or it can be a voltage amp expected speaker. Or a current amp expected speaker. They just have to interface.

Serial connection just raises impedance and does not change efficiency, only what type of power it seeks. Parallel connected lowers impedance and falsely states it to be more sensitive by implying all amps have to be voltage amps, or low output impedance and drains the reserves from such amps. 90dB/1W/1m is a real sensitivity spec that is honest. 93 dB/2.83V/1m is a bad sensitivity figure until you get a good impedance chart. It implies you will be using SS amps for all speaker sensitivity comparisons. But in the store, most will be SS amps, and side by side, the "loudest" speaker is probably the lower impedance speaker. And people will buy based on that demonstration.

I like higher impedance speakers for any amp. They tend to sound better for SS or tube. Max volume is another issue. So I say series connected woofers can be better. Just get the right amp that will make it shine.

-kurt
The primary reason . . ., posted on October 15, 2009 at 18:13:02
caspian@peak.org
Audiophile

Posts: 386
Location: Oregon
Joined: January 12, 2008
 . . . for running two identical drivers in parallel is the 6dB increase in sensitivity (assuming your amp can sink enough current to drive the halved impedance). In series, there is NO net gain in sensitivity, but the impedance is doubled, which can make an easier load for the amp to drive.

In a line array, you would typically combine series and parallel connections. Say you want to use eight 8-Ohm nominal drivers in an array. Four of the drivers in parallel gets you a 12dB increase in sensitivity over a single driver, but yields a nominal 2 Ohm load, challenging for most amplifiers. But if you put them in series with another set of four paralleled drivers, you're back up to a 4 Ohm load, which most modern SS amps can drive easily. You don't gain any more sensitivity, but you don't lose any, either.

All of this, of course, assumes that the drivers are closely matched in essential specs. Then the variances should, as you note, average out.

RE: Drivers is series is bad. Parallel good. Or is it? (Long), posted on October 15, 2009 at 17:37:24
Tony Lauck
Audiophile

Posts: 3048
Location: Vermont
Joined: November 12, 2007
Contributor
  Since: February 24, 2009
 If the two drivers are identical and are symmetrically placed with respect to the enclosure and room then a simple argument of symmetry supports your conclusion that the two approaches will produce equivalent results, if you consider the speaker on its own. However, the speaker is part of a system that includes at least the amplifier and speaker cable. When these are taken into consideration the two approaches may differ. For example, running speakers in parallel is the best way to go if the amplifier has good current capability but has limited voltage capability.

Tony Lauck

"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar

Page processed in 0.023 seconds.

[ Top Top of Page ] [ Contact Us ] [ Support/Wish List ] [ Copyright © 2009, Audio Asylum, all rights reserved ] [ Go Green ] [ Take5 ]
[ General ] [ Speakers ] [ Tubes ] [ Vinyl ] [ Digital ] [ Hi-Rez ] [ Video Asylum ] [ Cables ] [ Tweaks/DIY ] [ Music ] [ Films ]