speakers in series

With amplifiers when you put two caps in series you have put a small resistor on each cap so each cap gets the same voltage. It sounds like you are saying that there is no way to do this with speakers of different t/s parameters.
Though the reasoning for shunting series DC capacitors with resistors in an amplifier is different to a speaker crossover, you're on the right track with your thinking and it is possible to achieve two different speakers in series with flat amplitude versus frequency.

What you want to do is to add impedance flattening networks in parallel with each driver so that the impedance of each driver becomes a constant impedance versus frequency (i.e. a resistor). These are also commonly referred to as conjugate networks/filters or zobel networks/filters. Typically you will have an RLC network to flatten the bump in impedance due to the low frequency mechanical resonance (Fs) and an RC network to flatten the voicecoil inductance (impedance rising at higher frequencies). See: Speaker With Conjugate Network

I would recommend playing around in Xsim with a single driver to flatten its impedance, then do it again with another driver. Then place both drivers/networks in series and you should see that the frequency response of each driver isn't impacted.

Impedance flattening is also useful for making passive high and lowpass filters behave more ideally, though sometimes a compromise has to be made due to the networks requiring high component values which are physically large and expensive. Usually flattening the mechanical resonance is only practical at high frequencies (>500Hz) due to this, or the speaker designer may choose to only partially flatten the impedance in order to use smaller and cheaper components.
 
Last edited:
Have you tried Xsim? It is a crossover simulator, but you can place different drivers in series and get combined frequency response and impedance. You need to have .frd and .zma files for the drivers, but even if you do not have these for your own drivers you can use other drivers to get an idea of the effects. Dayton has that info for nearly all of their own drivers.


The impedance curves of different drivers in series will interact to change the frequency response of each. The greater the mismatch in the impedance curves the larger the effect will be.


Line arrays are a special case, since the drivers in a line will be of the same model. Ideally they would be identical in all respects, but manufacturing tolerances make this unrealistic. Over most of the frequency range this variance might typically amount to just a fraction of a decibel difference. But if the drivers have a wide variance in Fs this has the potential for a bigger problem. Suppose most of the drivers have an Fs of 50 Hz, but one in the line is at 40 Hz. That driver will produce a greater share of the output at 40Hz, and would reach its excursion limit before the others in the line. Connecting multiple drivers together in series-parallel (such as the 3-by-3 arrangement mentioned above) can help mitigate this by averaging the impedances together in the parallel groups. Also, a resistor in parallel with each group will lower the impedance peak, though it will also lower the overall impedance and waste some power. I might use such a resistor if I had just a few drivers (the woofer/subwoofer section) on an open baffle and I need to ensure equal excursion near Fs.

Does your advice to use a Zobel network on two different drivers in series change if the drivers are full range and as such, do not have a crossover behind them? Will a Zobel still work? Will a/multiple zobels work on a line array to the desired effect??
 
Does your advice to use a Zobel network on two different drivers in series change if the drivers are full range and as such, do not have a crossover behind them? Will a Zobel still work? Will a/multiple zobels work on a line array to the desired effect??



That was from a different post, not my idea.


The main issues with using different full range drivers:
Do they have the same sensitivity?
Do they have the same impedance?
Do they have similar frequency response and phase behavior?
Do they have similar excursion capabilities?


If there is a poor match in any of these characteristics then you will probably need to compensate somehow in a crossover or response shaping network. maybe you simply need to match levels, and that can be accomplished with a few resistors. But there is no requirement that you use both as full range drivers.
 
Though the reasoning for shunting series DC capacitors with resistors in an amplifier is different to a speaker crossover, you're on the right track with your thinking and it is possible to achieve two different speakers in series with flat amplitude versus frequency.

What you want to do is to add impedance flattening networks in parallel with each driver so that the impedance of each driver becomes a constant impedance versus frequency (i.e. a resistor). These are also commonly referred to as conjugate networks/filters or zobel networks/filters. Typically you will have an RLC network to flatten the bump in impedance due to the low frequency mechanical resonance (Fs) and an RC network to flatten the voicecoil inductance (impedance rising at higher frequencies). See: Speaker With Conjugate Network

I would recommend playing around in Xsim with a single driver to flatten its impedance, then do it again with another driver. Then place both drivers/networks in series and you should see that the frequency response of each driver isn't impacted.

Impedance flattening is also useful for making passive high and lowpass filters behave more ideally, though sometimes a compromise has to be made due to the networks requiring high component values which are physically large and expensive. Usually flattening the mechanical resonance is only practical at high frequencies (>500Hz) due to this, or the speaker designer may choose to only partially flatten the impedance in order to use smaller and cheaper components.

Thanks for your great suggestions and explanations. I have been playing with the crossover programs I have already to see how it would look and though I have some work left to do with them, I have some more questions which my programs don't address.
1) can I use a Zobel between a tube amp and two full range drivers in series with each other that do not have a crossover? In other words, for a Zobel to work properly what does it need to precede it? Crossover? transformer? Or does it matter at all?
2) At what frequency does the Zobel start working? Is is variable? Can the frequency be chosen?
3) assuming the above, where would I put a Zobel on two different drivers in series? does each driver get a Zobel and then put in series, or do the drivers go in series first but with only one Zobel for them both?
Thanks again.
 
Zobels are just networks of passive components which have a complimentary impedance to the speaker driver they have been designed for - when the speaker driver's impedance goes up, the impedance of the zobel goes down and vice versa. When connected in parallel with the driver, the overall impedance is constant.

Speaker driver impedance without zobel:
7CixlHW.png


Zobel impedance:
t3rkmXQ.png


Speaker driver impedance with zobel:
TjTjySb.png


They don't have to be used in conjunction with a 'crossover' (low/high/bandpass/notch filters). Typically you place the zobel in parallel with the speaker driver and you now treat the whole lot as a new speaker driver which has a flat impedance (and no other changes). It is up to the speaker designer to determine if it is most appropriate to place the zobel filter(s) before or after the crossover filters or if they are even required.

I would recommend using software like Xsim to design the zobel because if improperly designed you could accidentally load your amplifier with too low of an impedance and damage it.

Sometimes a zobel is placed in parallel with a finished speaker (either mult-way or single driver), just to flatten its impedance and make it behave better with a tube amp. This is because tube amps have significant output impedance (due to the output transformer), and they will produce less output voltage when the impedance is low compared to when it is high. Therefore if your speaker spikes up from 8ohm nominal to 100ohm at a certain frequency, the tube amp will produce a slightly higher voltage and therefore the system will play that frequency louder than other frequencies where the speaker impedance is instead 8ohm.
You don't need this with a solid state amp as they have a very low output impedance and are therefore insensitive to speaker impedance; it will produce almost exactly the same output voltage whether the speaker impedance is 8ohms or 100ohm.

The reason you use zobels with passive crossover filters is the same as for tube amps - the passive filters are sensitive to changes in load impedance so will only produce an ideal response when the speaker driver impedance has been flattened. Often the non-ideal filter response is actually desirable, or the difference is so little that the zobel can just be left out.

Since placing two different speakers in series has implications due to the impedance of the drivers, designing and placing a zobel in parallel with each speaker will resolve that issue.
 
Last edited:
A Zobel network can be used pre, mid, post or sans crossover for a plethora of reasons.

Yep ... although I would strongly suggest opening the Current panels in your simulator and keeping a close eye on the added load placed on your amplifier.

I have seen several crossovers, even commercial ones, that impose such a load as to overheat the amplifier they are connected to.