If the impedance curve of a passive crossover crosses that of the driver something very bad is going to happen, right?
e.g. at 1kHz the original driver is 5 ohm but after applying the crossover the impedance actually drops and becomes 4 ohm.
You would expect a crossover to only increase the impedance, would you?
Is this what people refer to as an "unstable" filter?
Thanks in advance.
(I guess this is a similar question to why a passive filter can sometimes boost a signal above 0dB.)
e.g. at 1kHz the original driver is 5 ohm but after applying the crossover the impedance actually drops and becomes 4 ohm.
You would expect a crossover to only increase the impedance, would you?
Is this what people refer to as an "unstable" filter?
Thanks in advance.
(I guess this is a similar question to why a passive filter can sometimes boost a signal above 0dB.)
Last edited:
Hi Cyber!
This usually happens from shunt components not having a resistor in series. Often on another driver. 🙂
That is, in a 2-way, a shut in the bass can cause a reduced impedance in the tweeter.
Can you post a specific example, and/or use XSim to show the impedance curve?
I have seen specific crossovers from Focal which deliberately reduce the impedance, apparently to make the speakers seem more discerning.
Best,
Erik
This usually happens from shunt components not having a resistor in series. Often on another driver. 🙂
That is, in a 2-way, a shut in the bass can cause a reduced impedance in the tweeter.
Can you post a specific example, and/or use XSim to show the impedance curve?
I have seen specific crossovers from Focal which deliberately reduce the impedance, apparently to make the speakers seem more discerning.
Best,
Erik
Most of the impedance of capacitors and inductors is known as reactance and can be positive (inductors) or negative (capacitors). The reactive part of the impedance at any given frequency is in "imaginary" ohms -- which means it is at an angle other than zero (+ or - 90, the impedance of resistors always has an angle of 0 and is called "real" -- real and imaginary are just names here, both are genuine impedance quantities). Loudspeaker driver impedances are all three kinds, part inductive in some places, part capacitive in some, resistive in a few spots. Put something inductive in series with something capacitive and the reactances will cancel each other some --making the total impedance smaller than it would be without that component in series. It can even make a driver play louder (and draw more current) at that frequency.
The trick is that it can only happen over narrow frequency bands, since the inductive reactance is increasing with frequency while capacitive is dropping with frequency, so they can't be near equal for very long.
The trick is that it can only happen over narrow frequency bands, since the inductive reactance is increasing with frequency while capacitive is dropping with frequency, so they can't be near equal for very long.
Last edited:
I am ramming the low pass section with a high Q chebyshev.
So if this is just reactance canceling out each other it does not sound like a real concern.
The surprising part to me was that sometimes I can make the impedance even lower than the DC resistance of the driver, and correspondingly I can get the filter response above 0dB, as if it is creating strength out of thin air. From what I've read, not sure if I read correctly or not, was that this is ringing which is the electrical equivalent of a terrible waterfall plot. If there is some resistance in series as Erik said, even if it is just the DC resistance of the drivers themselves, then eventually that ringing will dissipate away on its own.
My biggest take away is the impedance curve must not drop below the nominal impedance the speaker system is designed for. (Nowhere in the curve shall be 5 ohm for a 8 ohm nominal speaker, for instance.)
The rest I don't know. I just wish to learn if there are other pitfalls I would need to be careful about.
Thanks everyone! And XSim is the most amazing piece of software on the planet, improving sound quality across the globe.
So if this is just reactance canceling out each other it does not sound like a real concern.
The surprising part to me was that sometimes I can make the impedance even lower than the DC resistance of the driver, and correspondingly I can get the filter response above 0dB, as if it is creating strength out of thin air. From what I've read, not sure if I read correctly or not, was that this is ringing which is the electrical equivalent of a terrible waterfall plot. If there is some resistance in series as Erik said, even if it is just the DC resistance of the drivers themselves, then eventually that ringing will dissipate away on its own.
My biggest take away is the impedance curve must not drop below the nominal impedance the speaker system is designed for. (Nowhere in the curve shall be 5 ohm for a 8 ohm nominal speaker, for instance.)
The rest I don't know. I just wish to learn if there are other pitfalls I would need to be careful about.
Thanks everyone! And XSim is the most amazing piece of software on the planet, improving sound quality across the globe.
Hi cyberstudio,
You shouldn't be able to make the impedance go below the DC resistance with any series components (other than negative resistors, which aren't included in Xsim!). Can you show the circuit that is doing that? (In the plot you show, it appears that the impedance is continuing to drop below 10Hz, so it isn't getting down to the DC resistance yet).
BTW, it isn't unusual at all for the impedance of speakers to fall below the nominal value -- you can find LOTS of "8 ohm" speakers that drop below 5 ohms. Nominal is just nominal, someone named the impedance that value as a general guide for what amps should work for it.
You shouldn't be able to make the impedance go below the DC resistance with any series components (other than negative resistors, which aren't included in Xsim!). Can you show the circuit that is doing that? (In the plot you show, it appears that the impedance is continuing to drop below 10Hz, so it isn't getting down to the DC resistance yet).
BTW, it isn't unusual at all for the impedance of speakers to fall below the nominal value -- you can find LOTS of "8 ohm" speakers that drop below 5 ohms. Nominal is just nominal, someone named the impedance that value as a general guide for what amps should work for it.
Sorry I made a mistake. It did not drop below DC resistance. It only rised above 0dB (e.g. +0.01dB). This can be properly explained by the same reactance cancellation explanation.
But I did get the impedance really really low while playing with a Zobel network. That was parallel though, and Erik's link is great reading material.
But I did get the impedance really really low while playing with a Zobel network. That was parallel though, and Erik's link is great reading material.
Measured impedance of your woofer actually indicate it is a nominally 4-ohm speaker, with 4-ohm minimal impedance (between 100 and 250 Hz). When designing the crossover try to stay above (or equal to) the minimal impedance (4 ohms) AND not above the drivers measured frequency response.
Agreed.
This is a somewhat advanced level to be starting with. A person should first understand the behaviour of the component types. Voltage leading or trailing current, reactance calculations, time constants etc. otherwise the learning curve is quite steep.
I'm not sure that's a problem, done reasonably.
It's no problem for the system impedance to drop below that of the driver. In a lot of cases it is required to allow a filter Q which gives you the desired acoustic alignment.
Just make sure it doesn't go too low (3ohm or less) otherwise your amplifier might not be too impressed 🙂
Just make sure it doesn't go too low (3ohm or less) otherwise your amplifier might not be too impressed 🙂
Instability can be described as finite output for zero input, which cannot happen in passive components. Passive crossovers will always be stable.
Yes. There is a misconception that if the output is greater than the natural response then something is losing control, but this is not true beyond the normal confines of resonance. The response will show what can be expected.
Instability is a regenerative condition that requires gain, hence it is often used with regard to amplifiers. Output increases up to the limit of the device due to positive feedback.
Instability is a regenerative condition that requires gain, hence it is often used with regard to amplifiers. Output increases up to the limit of the device due to positive feedback.
To expand on bwaslo's comments:
It is a common misconception to think that the effective impedance of a circuit is the arithmetic sum of the individual impedances e.g. 3 + 4 = 7.
However, the alternating voltages across the capacitors and inductors in a circuit are not in phase with the supply current. Each individual voltage may be considered to act at an angle to the current. This results in the supply voltage being the vector sum of the individual voltages.
Consequently, in a circuit comprising capacitors and inductors, the effective impedance is the vector sum of the individual impedances e.g. 3 + 4 = 5.
(Think of adding 3 and 4 when they are at 90 degrees to each other. The vector sum is 5, represented by the hypotenuse of a right angled triangle.)
It is a common misconception to think that the effective impedance of a circuit is the arithmetic sum of the individual impedances e.g. 3 + 4 = 7.
However, the alternating voltages across the capacitors and inductors in a circuit are not in phase with the supply current. Each individual voltage may be considered to act at an angle to the current. This results in the supply voltage being the vector sum of the individual voltages.
Consequently, in a circuit comprising capacitors and inductors, the effective impedance is the vector sum of the individual impedances e.g. 3 + 4 = 5.
(Think of adding 3 and 4 when they are at 90 degrees to each other. The vector sum is 5, represented by the hypotenuse of a right angled triangle.)
- Status
- Not open for further replies.