Slightly depending how low you cross. SB A's impedance peaks is only twice the nominal impedance. If you cross three octaves above Fs, how much peak matters?
This tweeter has impedance peak 6 ohm @ 1000Hz, 4 ohm nominal. Quite flat overall.
Last edited:
not sure there is a simple answer to that
my approach is to try and deal as many issues possible
but within reason, and what makes sense
may not all be needed
though, when using 'simple means' you never know what will happen
but I am beginning to see a 'pattern'
or maybe its just a 'pattern' in my choices
and mostly I still cant resist to try a crazy idea
lower ressonance peak might indicate a weaker or less effective design
might still sound good tho
good looking tweeter, absolutely
It's all a bit of trial and error if you ask me.
Thought I'd share a fun day on the DIY Monitor upgrades front. Spent a bit of cash on a decent £55 multimeter which can measure INDUCTANCE, CAPACITANCE an' all that TECHNICAL stuff!
Also some speaker cable and plugs and all the usual nuts and bolts and wood bracing and stuff. Got the 680nF/8R2 7W Zobel network plus 100 ohm 10W HF damping resistors which I'm a bit vague about, but still waiting for better 3.3uF audio capacitors from Maplin. Too tired tonight to do anything, but cooking with GAS, no?
Thought I'd share a fun day on the DIY Monitor upgrades front. Spent a bit of cash on a decent £55 multimeter which can measure INDUCTANCE, CAPACITANCE an' all that TECHNICAL stuff!
An externally hosted image should be here but it was not working when we last tested it.
Also some speaker cable and plugs and all the usual nuts and bolts and wood bracing and stuff. Got the 680nF/8R2 7W Zobel network plus 100 ohm 10W HF damping resistors which I'm a bit vague about, but still waiting for better 3.3uF audio capacitors from Maplin. Too tired tonight to do anything, but cooking with GAS, no?
So we agree, that the goal of the filter is to adjust the response while successfully navigating the impedance curves. This can require a somewhat complicated singular circuit to achieve.
The whole idea of Zobel correction in this context is to simply break up this complex circuit into two more simple parts: dealing with the impedance and rolling off the response. Despite the parts count, the end result is the same.
You're correct of course about the LCR interacting with the tweeters impedance. It is also commonly thought that if you correct the impedance with an LCR so it 'looks' resistive, it isn't truly like a resistor because the LCR and the tweeters impedance are exchanging energy (ringing) behind the scenes.
This is also true, but so is the fact that a tweeter will respond to the voltage found at its terminals. Thus, if we were driving an actual resistor instead of a tweeter, the signal voltage would remain steady due to a lack of ringing, so seeing that the voltage remains steady should therefore be the goal.
I'd like to show how impedance correction can be a positive step toward this end...
This circuit is the equivalent of a tweeter (just the resonance). The voltage plot shown below it is indicative of the impedance peak.
The transient response in the next image is the result of a sine wave stimulus at resonance (above), and just off resonance (below). They should be straight sinewaves of constant level, but the non-constant levels indicate energy storage. Their variance up and down indicates ringing.
The transient response in the next image is the result of a sine wave stimulus at resonance (above), and just off resonance (below). They should be straight sinewaves of constant level, but the non-constant levels indicate energy storage. Their variance up and down indicates ringing.
Attachments
This circuit has an RLC added. The impedance is virtually flat as a result.
The RLC is accepting energy in a conjugate manner (equal and opposite). Looking at the transient plots (same frequencies as before), the voltage which is driving the tweeter remains steady, as it should.
Furthermore. I've been driving this circuit with a 1k resistor to indicate the impedance. This resistor also isolates the circuit showing that it is behaving itself without assistance from amplifier damping.
The RLC is accepting energy in a conjugate manner (equal and opposite). Looking at the transient plots (same frequencies as before), the voltage which is driving the tweeter remains steady, as it should.
Furthermore. I've been driving this circuit with a 1k resistor to indicate the impedance. This resistor also isolates the circuit showing that it is behaving itself without assistance from amplifier damping.
Attachments
I have simulated crossovers this way and can reproduce same on request. I have built each and my listening results do not suggest any concernable difference to me. I also performed nodal analysis on this with pen and paper (before I owned a computer).
The simulated transient response is of particular interest to me, and I have no general reason to doubt the sim.
You can make your crossover design easier with selection of drive units too. This is a comparison of the £33 Morel Classic CAT 298 and the vented £46 CAT 308, available in 94 and 104mm mount, which uses a bigger magnet but the same replaceable voice-coil/dome:
An externally hosted image should be here but it was not working when we last tested it.
This is almost a perfect example of how a bigger enclosure lowers the Qts, but given you can do something about the rising inductance with a parallel Zobel RC network, the lower Qts is going to help too. The vent and enclosure also lowers the Fs a tidge and increases efficiency, though the bigger magnet hides that somewhat. You'd guess the vented 308 sounds better too. Thought I'd share...
Those impedance blips at about 1800 hz look worrisome. However, the Morel site's spec sheet for the 308 doesn't show any anomolies in the response. I'd like to see some independent tests before getting too excited.
Last edited:
I note the word 'concernable'
but I hear you
the 'differences' I hear wouldnt bother any of my friends
they would be happy with the worst of my achievements
it definately is questionable how much time should be spent on something that isnt needed
might be there are much worse 'issues' to take care of
I only said 'concernable' because I didn't want to pretend I'd done any lab grade listening tests.
The Zobel and tweeter are exchanging energy that just happens to come out to the right amount. Sounds dodgy, but we are happy to put a driver in a box and that's more or less the same kind of back and forth. If you find another circuit that creates the same effect, it's because it is introducing the same amount of reactance.
The Zobel and tweeter are exchanging energy that just happens to come out to the right amount. Sounds dodgy, but we are happy to put a driver in a box and that's more or less the same kind of back and forth. If you find another circuit that creates the same effect, it's because it is introducing the same amount of reactance.
I'm with you there.
I'm loving my added Zobel network, discussed elsewhere:
From AllenB, discussing equalising the Fs resonance with an RLC:
That's FANTASTIC, AllenB! Love it. Question though...because I'm wondering if you've made an all-pass filter there, what does it do to square waves around the Fs resonance? Any delay or ringing? For people who don't know about all-pass, it is that their sinusoidal frequency response is flat, but phase and group delay is still affected...
@speakerdoctor: That little blip at 1800Hz you notice on the Morel tweeter is, IMO, probably an artifact of the voicecoil gap needing more venting or damping as in the yellow trace for an Audax TWO34 unmodified tweeter. There are limits to how much you can do, but Troels Gravesen discusses this stuff here:
TW034
An externally hosted image should be here but it was not working when we last tested it.
From AllenB, discussing equalising the Fs resonance with an RLC:
That's FANTASTIC, AllenB! Love it. Question though...because I'm wondering if you've made an all-pass filter there, what does it do to square waves around the Fs resonance? Any delay or ringing? For people who don't know about all-pass, it is that their sinusoidal frequency response is flat, but phase and group delay is still affected...
@speakerdoctor: That little blip at 1800Hz you notice on the Morel tweeter is, IMO, probably an artifact of the voicecoil gap needing more venting or damping as in the yellow trace for an Audax TWO34 unmodified tweeter. There are limits to how much you can do, but Troels Gravesen discusses this stuff here:
TW034
Attachments
When the impedance is offset correctly, square waves will be plenty square.
The point of this was to demonstrate that the RLC is harmless. This doesn't mean that it is a requirement. Try to see the the high pass filter and the RLC like two extension cords in series used to power a television. As long as one end fits the wall, and one accepts the TV, then you can watch it. The speaker can appear in its raw form with its naturally reactive impedance, and any crossover can be applied as long as it 'fits' the impedance. Then at the other end, the amp will 'see' the reactive high pass filter regardless of whether an RLC was used.
This means that the RLC doesn't benefit anything but the design process. Sometimes it is the only practical way to deal with a problem. But there is no benefit to adding Zobel correction to a properly working crossover. Even adding Zobel, then 'correcting' the crossover back to were it initially was, has no benefit.
Where Zobel correction is applied to a properly working crossover and it gives a positive result, this is referred to as 'equalisation'.
The point of this was to demonstrate that the RLC is harmless. This doesn't mean that it is a requirement. Try to see the the high pass filter and the RLC like two extension cords in series used to power a television. As long as one end fits the wall, and one accepts the TV, then you can watch it. The speaker can appear in its raw form with its naturally reactive impedance, and any crossover can be applied as long as it 'fits' the impedance. Then at the other end, the amp will 'see' the reactive high pass filter regardless of whether an RLC was used.
This means that the RLC doesn't benefit anything but the design process. Sometimes it is the only practical way to deal with a problem. But there is no benefit to adding Zobel correction to a properly working crossover. Even adding Zobel, then 'correcting' the crossover back to were it initially was, has no benefit.
Where Zobel correction is applied to a properly working crossover and it gives a positive result, this is referred to as 'equalisation'.
Attachments
Last edited:
Just to remind ourselves, we are looking at correcting the Fs resonance here. Sometimes causes issues when low order tweeter crossovers still drive the tweeter appreciably near Fs, and Qts is increased by the high impedance of the crossover at Fs.
That's pretty convincing on the square wave driven by 1K source, but I wonder what happens with 6 ohm and zero ohm source impedance to the RLC correction? Because there is an element of energy storage regardless. But maybe it does all come out in the wash!
Uploaded with ImageShack.us
I'm just thinking that these things always have an element of Sod's Law. It looks too good to be true! FWIW, I think the Zobel works by giving my ordinary voltage amplifier a nice load as much as anything, but there IS an element of self-damping introduced into the speaker too, and maybe a small LC resonance too. Well known that amps frequency response is load dependent at the top-end in practise. Hence the 8R/100nF Zobel usually added for stability at the output.
An externally hosted image should be here but it was not working when we last tested it.
That's pretty convincing on the square wave driven by 1K source, but I wonder what happens with 6 ohm and zero ohm source impedance to the RLC correction? Because there is an element of energy storage regardless. But maybe it does all come out in the wash!
An externally hosted image should be here but it was not working when we last tested it.
Uploaded with ImageShack.us
I'm just thinking that these things always have an element of Sod's Law. It looks too good to be true! FWIW, I think the Zobel works by giving my ordinary voltage amplifier a nice load as much as anything, but there IS an element of self-damping introduced into the speaker too, and maybe a small LC resonance too. Well known that amps frequency response is load dependent at the top-end in practise. Hence the 8R/100nF Zobel usually added for stability at the output.
Feedback amps also need to steer clear of positive feedback.
Using Zobel correction on a driver doesn't give the amp a resistive load when it is hiding behind a filter. You can correct the entire system before the crossover, if you want.
You're right about there being resonance, but without any correction there is an obvious resonance. So where's the problem?
Using Zobel correction on a driver doesn't give the amp a resistive load when it is hiding behind a filter. You can correct the entire system before the crossover, if you want.
You're right about there being resonance, but without any correction there is an obvious resonance. So where's the problem?
a small zobel on tweeter can have a positive effect
but the opposite could also happen, and result in negative effect
I have experienced both, even with the same speaker
and only because I changed other other parts of the xo
apart from impedance, its important to note that a zobel also affects phase
impossible to add anything without affecting phase
sometimes its an advantage, sometimes not
but no matter how hard you try, there will always be remains of small phase issues