Have a look at this, Ohm's law at work. https://www.stereophile.com/content/primaluna-prologue-premium-power-amplifier-measurements
Last edited:
It is not damping (dampening - making moist), it is about making the impedance resistive. An impedance corrector. The actual network will most probably be too high in impedance to have any appreciable effect on the damping.
But I think I have an answer. Since the speaker still sees the same driving voltage as without the impedance corrector, it will still 'play' the same with or without the corrector. The amplifier will see a resistive impedance which also means that it sees a lower impedance at the speaker resonance points. So the amplifier job gets just a bit harder, but probably not catastrophically so.
So it appears that impedance correction does indeed make is easier to design passive xovers with negligeable disadvantages, except a few extra dollars cost.
Jan
Jan, download Boxsim, it's plenty of fun to see what happens as you change values.
When it comes to the tweeter it IS about damping, because the stored energy interferes with roll off. Otherwise the rise in impedance isn't a problem.
There are some designs out there where they use a 2nd order crossover, put sharply roll off by the use of numerous LCR's after the desired point of mixing. I might have to try it sometime.
When it comes to the tweeter it IS about damping, because the stored energy interferes with roll off. Otherwise the rise in impedance isn't a problem.
There are some designs out there where they use a 2nd order crossover, put sharply roll off by the use of numerous LCR's after the desired point of mixing. I might have to try it sometime.
Ok, I think Jan and most here are talking about impedance peaks at the resonance frequency of a driver.
But some posts also make a reference to the natural impedance rise with frequency, which made me think about the correcting networks at the output of a power amp. How important are they? What practical difference do they make?
Reading this now: https://www.neurochrome.com/taming-the-lm3886-chip-amplifier/stability/#Zobel
But some posts also make a reference to the natural impedance rise with frequency, which made me think about the correcting networks at the output of a power amp. How important are they? What practical difference do they make?
Reading this now: https://www.neurochrome.com/taming-the-lm3886-chip-amplifier/stability/#Zobel
Just took a quick look at boxsim. It seems to have pretty good frequency domain features, but where many speaker designs seem to be lacking is in time domain response. That is to say, aspects of performance related to waterfall plots, and impulse response. What do people like to use for that?
Jan: I am slightly confused her. I first ready your question about flattening out the driver effects to the crossover network sees a constant impedance, but the responses seem to suggest that its making the amplifier see a constant impedance? Which was it?
If the former my first reaction would be that this is an EE response to an electromechanical problem for 2 reasons
1. You run the risk of thinking electrical slopes not acoustic slopes
2. In normal crossover design you use the impedance of the driver to help achieve 1.
But then I realised that crossover design is rarely that simple and there are all sorts of gotchas. But for a masterclass in lucking out look here. Zaph|Audio
If you meant the latter then apologies for wasting everyones time
If the former my first reaction would be that this is an EE response to an electromechanical problem for 2 reasons
1. You run the risk of thinking electrical slopes not acoustic slopes
2. In normal crossover design you use the impedance of the driver to help achieve 1.
But then I realised that crossover design is rarely that simple and there are all sorts of gotchas. But for a masterclass in lucking out look here. Zaph|Audio
If you meant the latter then apologies for wasting everyones time
Interesting read at zaph audio. He designed two crossovers for that speaker, one of which reduces the impedance peak at 3KHz from 40 ohms to 20 ohms.
Quoting from that page: "Once again, I offer a tube friendly version of the crossover without the impedance swing in the midrange."
I suppose the tube amp with its significant Zout, low damping factor, has trouble with the impedance peaks. What trouble? Doesn´t control the resonances, doesn´t damp, and the speaker will have a much different frequency response driven by a tube amp rather than the SS class AB that was probably used during measurements.
(Related to this, I recently heard an improvement removing two 0,22 ohms resistors I had at the output of a bridged chipamp. I suppose my speaker has some impedance peaks. Would like to measure all of this.)
Quoting from that page: "Once again, I offer a tube friendly version of the crossover without the impedance swing in the midrange."
I suppose the tube amp with its significant Zout, low damping factor, has trouble with the impedance peaks. What trouble? Doesn´t control the resonances, doesn´t damp, and the speaker will have a much different frequency response driven by a tube amp rather than the SS class AB that was probably used during measurements.
(Related to this, I recently heard an improvement removing two 0,22 ohms resistors I had at the output of a bridged chipamp. I suppose my speaker has some impedance peaks. Would like to measure all of this.)
John Krutke didn't have much time for tube amps, but did on occasion do the tube friendly version to reduce the bigger frequency deviations that high Zo would give you. If you read carefully there were 3 versions of the crossover.
Brad recommended Esa's book, can't get much better thumbs up than that!
Brad recommended Esa's book, can't get much better thumbs up than that!
Markw4, to get things like waterfall plots or impulse response there's only one tool; that is used by everyone. We call it a microphone.
Boxsim has the ability for phase information input, and can even do polar plotting. You have to put in your information for whatever box you want etc. There's a few things it can't understand dealing with ports, but that's the biggest limit I've found so far. Well, and it doesn't care or know about room gain, so you have to plan for 2-3db of room gain in the corresponding frequency spectrum, based a bit on things about the speaker like BL, box type and what not.
Boxsim has the ability for phase information input, and can even do polar plotting. You have to put in your information for whatever box you want etc. There's a few things it can't understand dealing with ports, but that's the biggest limit I've found so far. Well, and it doesn't care or know about room gain, so you have to plan for 2-3db of room gain in the corresponding frequency spectrum, based a bit on things about the speaker like BL, box type and what not.
Last edited:
Jan: I am slightly confused her. I first ready your question about flattening out the driver effects to the crossover network sees a constant impedance, but the responses seem to suggest that its making the amplifier see a constant impedance? Which was it?
If the former my first reaction would be that this is an EE response to an electromechanical problem for 2 reasons
1. You run the risk of thinking electrical slopes not acoustic slopes
2. In normal crossover design you use the impedance of the driver to help achieve 1.
But then I realised that crossover design is rarely that simple and there are all sorts of gotchas. But for a masterclass in lucking out look here. Zaph|Audio
If you meant the latter then apologies for wasting everyones time
You are right, these are two issues; I mixed them up as they both came up in my discussions off-line.
The 1st statement was that impedance correction would make xover design easier. It does; much easier to design a xover if you know the load to be resistive and frequency-independent.
The 2nd statement was that a flat frequency-independent speaker load would make thing 'easier' for the amp. I don't see that; as far as I see it it only causes the load to draw more current at the frequencies where the peaks used to be which are now equalized down. You do get rid of most of the phase shift between Vout and Iout but apart from possible SOA issues it would not impact reproduction quality I believe.
For the 1st case, I thought that the disadvantage is that the output current from the xover divides itself between the correction circuit and the speaker (both are in //) and that division is of course frequency dependent. So, I thought, you just exchange one problem for another: now you have an easy xover design but you need to address that frequency-dependent load current division.
I am mainly interested in the last paragraph - am I right or not?
Jan.
Ah, that's a relief
Yes, but a speaker impedance varies all over the place with frequency, and your xover design assumes a constant impedance over frequency. So your super duper xover behaves different then what you think it does.
That is the reason for the impedance correction in the first place.
Jan
Sure, for measurement. Same tool as needed for measuring frequency response.
But, I was asking about design and simulation tool(s). For example, it appears that boxsim is a simulator that can be used to model and predict frequency response. What about a tool for modeling and predicting time domain (transient) response, rather than only frequency response?
The reason I asked is because designers naturally may tend to focus attention on aspects of performance for which design tools are available. Daniel Kahneman said, WYSIATI (What You See Is All There Is). In the case of speaker design, if there are good frequency response design tools and no time domain design tools, one might expect to see a lot of designs with pretty good frequency response performance, and poor, or fairly random time domain (transient) performance.
Last edited:
Bill I think we are not communicating. Of course the speaker is NOT constant impedance, that is the reason for the idea to use a // complex network as impedance compensator so that the combination of this network and the speaker - as seen by the xover - is resistive and freq independent.
Jan
Yes, but if you look at the frquency response both as a magnitude and (often neglected) phase response, you are looking at the time domain in anther guise. The two are linked.
Jan
- Status
- Not open for further replies.