Joe,Let me give you at challenge, look at the circuit below:
Call this inductor-drive. It has both the benefits of current-drive AND voltage-drive. At the most critical frequencies, up to 20dB reduction in distortion at critical frequencies, with the inductor forming a 3KHz low-pass filter.
Are you muddying the water with purpose or unintentionally?
Esa measures acoustic distortion with a microphone yet in the drawing above you show the audio analyzer input as the voltage across the current sensing resistor.
As it is it, is all balled up, no one knows what you are talking about.
Thanks DT
Hello All,
P17WJ-00-08
Calibrated SPL Two Tone tests at 300 and 2500 at 2.83V and 400uV to better see noise floor with analyzer and amplifier fans running.
Thanks DT
P17WJ-00-08
Calibrated SPL Two Tone tests at 300 and 2500 at 2.83V and 400uV to better see noise floor with analyzer and amplifier fans running.
Thanks DT
Been a few weeks since I was here, but just noticed this thread had over a million views!!!
I understand, but... Absolutely not.
In fact, if what I was saying by that diagram was fully understood, it is both good news (solution) and also leads to a conclusion that is counter-intuitive. Please keep thinking about it, and via a continued and respectful dialogue, maybe the beauty and simplicity will become apparent. I do admit that I am playing with words when saying 'inductor-drive' and so I plead guilty. But 'voltage-drive' and 'current-drive' are also playing on words. Did you notice that in the 2016 paper by Esa, he calls it current-control and voltage-control?
But even those are questionable terminology. In fact I am thinking that the whole idea about drive and control by the amplifier is not even desirable. Please keep in mind, and you guys here may not realise that I sometimes agonise what words to use because I get misunderstood so often. Now nobody is saying, including myself, that current does not need to be controlled. Of course it does. But is the amplifier where this control should take place? I know this is a dangerous thing to say because it sounds counter to much of the current (present) thinking.
Sometime back a designer friend of mine (I won't mention his name, as I will then be accused of name-dropping, but I am talking about somebody who is Australian, but has designs sold all over the world) was designing some active speakers. Everything looked like it was coming together nicely, but something was not right. The fix was surprising (but not to me). Normally in active systems the drivers are directly connected to the invidividual (two) amplfiers and hence we have the perception that both the midbass and tweeter are controlled by the amplifier. The fix? Was to add something between the amplifier and the tweeter, a resistor. The sound improved perceptually. In my own mind I believe I know why it worked the way it did, for the same reason that my 'inductor-drive' idea works and why a parallel 8R resistor with many speakers also work.
The reason why all these there scenarios can work (and often does to a surprising degree) is that they have something in common.
What is it? Three solutions? They must be lowering distortion that comes out of the speaker. How?
And yes, the amplifier has a job. But what is it?
In my experience, putting things between the amplifier can be a good thing, yet that means less control? That's not intuitive, but once understood why, it can be unlearned.
Last edited:
Which is quite correct. But that is then just a low-pass filter with no additional benefits. But if we can put together the pieces, then there will be additional benefits.
Remember that Esa and I came to the solution for current-drive to become effective, the source to load impedances ratio should be 5:1 minimum. If you note, the ratio of 0.5mH series inductor, to the ratio of the driver's 0.1mH is 5:1 and I happen to know that this driver will then be -6dB @ 3KHz with that value.
I have come up with the same ratio of 5:1 and it will have all the same advantages under 3KHz that having a current source with 40 Ohm would have (5*8R = 40R) and fully compatible with voltage-drive. Bring in a tweeter above 3KHz where the tweeter will be -6dB @ 3KHz and FR will sum flat. No downsides and yet all the same aadvantages as current-drive. Keep in mind, to do the above you need the right driver, a very good driver. No, you won't make the driver better, but it will produce lower distortion? Sounds like a contradiction? Maybe, but it is not. You will still be listening to the flaws of the driver, but some of those flaws (not all) will not be amplified via the interface of a voltage source.
If you have get your head around why this works this way and why we have lower distortion, then you will fully know what I am talking about. No kidding.
I just gave you a Like and just surprised myself.
You and I have had our differences, and it galls me that you accuse me rampant commercialism. So unfair. If you knew about the sacrifices I have had to make, then you would not say such a thing. I still owe a friendly benefactor $4000 for getting Waveguides made and I don't even have any guarantee that I will even break eve.
TNT: Temperature dependent linearity problems?
I assume we are talking about the 8R parallel resistor and why it should have those known better result. There has to be an answer. I think I have figured it out.
I thought long about the issue being temperature related, that perhaps the DC resistance changes (as we know) and hence voltage sensitivity will change (we know that is not the same as efficiency). I even spouted the low temperatures in the VCs as related to the good dynamics. The Elsinores have that jump factor so absent in many speakers.
In times my view changed. I agree with Earle Geddes that it is to slow to show up at regular frequencies, hence limited to very low LF. I agree that small drivers do not have the ease of dynamics of larger drivers with larger (and lower temperature VCs) - but that it is mechanical/excursion and motor flaws are worse with small drivers. This is where Purifi has come in and shown us the way. And Doppler (FM) is not the problem either. But AM distortion is. The key to large drivers is that with less than great motors, the amplitude/excursions are lower and hence less demands on the motor. These are of course my opinions, but fairly sure I am on solid ground.
I can easily get -/+ 10% changes in inductance why simple changing the position of the VC is the gap, up 10% pushing in 3-4mm and down 10% pulling the VC out. You don't even need a cone attached. That means that the speaker is not linear in relation to the current of the amplifier.
Note the underlining, to me this is a huge clue. We are talking about impedance modulations. I put it to Lars Risbo that this was the same as current modulations of the current of the amplifier! This is a non-linear behaviour. This is according to Puri AM distortion. And here it is linked to the current of the amplifier.
TNT: Slowly delayed varying compression probably...
Compression is very real, no doubt about that. The question is too open-ended. Probably means fishing. Not meant as an insult, I am used to fishing. Now feel I am past that stage. But something delayed? Interesting.
I think we have to look at what is happening electrically. The resistor has no affect on the FR. The current through the resistor is very dependable. The current through the VCis not.
You are no doubt an intelligent person, but what did I just say? Can you figure it out?
Please, let us have a respectful discourse. If not, then don't expect any response from me. I am tired of all the aggro.
Over to you.
Regular schedule resuming... an update on the "BAC" version of the Elsinores.
Keep in mind, there is no guarantee that this will become an "official" version yet. The version is done in cooperation with two persons from a university in Melbourne and they work in the physics department. They have the same box and drivers already (you may have seen somebody posting about this a few months back). We shall see how things work out, the "BAC" uses aluminium drivers and hence "hard" cones. I never thought I would do this, but these drivers, both behave rather differently that other drivers with hard cones. The Tweeter is also aluminium but it has a ceramic coating, which is claimed to make it sound smoother than just aluminium.
All ten drivers have been extensively burned-in on my "machine" and the ceramic Tweeters have been mounted onto the Waveguide. Brief description of how it was mounted, it uses an O-ring that is brought under compression by the four nuts and bolts. The sharp edge of the throat of the Wave hits hard against the three 'spokes' that is part of the phase correction part, so the O-ring is not compressed beyond that point. The four holes in the Tweeter needs to be enlarged and then the bolts line up.
The drivers I am hoping will be mounted into the boxes over the next few days. That will enable me to take a shot of the boxes 'populated' and hence see what the final look will be.
Then the next stage will be LOTS of in-box measurement, also one day all the equipment will be set up so that we can do the farfield measurements outside, on a day with no wind, no rain and around 20C. Farfield measurements, nearfield measurements, on and off angle measurements, electrical measurements, when all this data is collected and verified (and all the equipment calibrated) and then we get into the computer modelling stage. This is where the Crossover takes shape. That will now indicate what the Crossover will look like, then build it and now tests are made with brand new Crossover connected and verified that it works as intended. Then we have the final tweaking and adjustment phase, and of course listening and finding out how "hard" Elsinores will sound. If all this comes together by the end of the month, that might be a minor miracle. We shall see, remember the ULD took near six months... so no guarantee.
Keep in mind, there is no guarantee that this will become an "official" version yet. The version is done in cooperation with two persons from a university in Melbourne and they work in the physics department. They have the same box and drivers already (you may have seen somebody posting about this a few months back). We shall see how things work out, the "BAC" uses aluminium drivers and hence "hard" cones. I never thought I would do this, but these drivers, both behave rather differently that other drivers with hard cones. The Tweeter is also aluminium but it has a ceramic coating, which is claimed to make it sound smoother than just aluminium.
All ten drivers have been extensively burned-in on my "machine" and the ceramic Tweeters have been mounted onto the Waveguide. Brief description of how it was mounted, it uses an O-ring that is brought under compression by the four nuts and bolts. The sharp edge of the throat of the Wave hits hard against the three 'spokes' that is part of the phase correction part, so the O-ring is not compressed beyond that point. The four holes in the Tweeter needs to be enlarged and then the bolts line up.
The drivers I am hoping will be mounted into the boxes over the next few days. That will enable me to take a shot of the boxes 'populated' and hence see what the final look will be.
Then the next stage will be LOTS of in-box measurement, also one day all the equipment will be set up so that we can do the farfield measurements outside, on a day with no wind, no rain and around 20C. Farfield measurements, nearfield measurements, on and off angle measurements, electrical measurements, when all this data is collected and verified (and all the equipment calibrated) and then we get into the computer modelling stage. This is where the Crossover takes shape. That will now indicate what the Crossover will look like, then build it and now tests are made with brand new Crossover connected and verified that it works as intended. Then we have the final tweaking and adjustment phase, and of course listening and finding out how "hard" Elsinores will sound. If all this comes together by the end of the month, that might be a minor miracle. We shall see, remember the ULD took near six months... so no guarantee.
The SB17nbac are very good speakers, but I'm afraid that the low order filter you want to use will not be enough to filter out the resonances affecting the sound of this midwoofer. In the case of this driver, using a fairly low crossover point with an LR4 filter is the optimal filtering, which Troels also chose in his design.
Hello,
I do not believe that you are adverse to confusion but that you like to be thought of as counter intuitive. Not at all so.
Too many words and to few definitions. All this talk of current as if it is unrelated to voltage is much like getting the tire wrapped around the axel of the machine.
Go ahead and use all the passive devices you like both in series and parallel with the complex load, the driver. Also use voltage feedback, current feedback, feedforward, DSP and error correction as you please. All these tools may reduce the perturbations of the driver as seen by amplifier to a much smaller version of its self. Much the same as using a shunt regulated power supply.
My expectation is that some amplifiers are much better than others at resisting the perturbations of the driver, in a way very similar to Power Supply Resistance Ratio. I have seen it called Reverse Power Supply Resistance Ratio.
Thanks DT
Edit: I have done this on my bench.
https://www.omicron-lab.com/fileadm...e_Transfer/App_Note_Reverse_Transfer_V2_0.pdf
Thank you for making this point. I know Troels and have visited him and I have also watched him over a period of time gravitate towards low and 1st order filters. But of course he does so selectively and some drivers he feels are suitable for 1st order and seem to like it when he feels the driver is good for it. But I set the bar a little differently, I want to find drivers that can get away with a single series inductor, so that what I am doing. I am willing to accept flaws that he might not, because I feel that correcting might in fact be over-correcting and there might be a risk of throwing the baby out with the bathwater.
I look at the potential driver slightly differently from Troels, but I also respect what he does. In fact, I am doing the "NAC" version (from now on) is because of something Troels did.
Mainly because of each and the CSD (Waterfall) plots he posted and the fact that the aluminium cone version had a 'soft' sound when using MLS (which sounds like a noise rushing out of the driver). Both Troels and I have a preference for 'soft' cones like paper, fibre and poly cones. But here we have a 'hard' aluminium cone that seems to buck the trend. Now I got interested, also because of something else happening in Melbourne, so I made the decision.
Will I be able to get away with that single series inductor? That will depend, but I am not excluding a narrow LCR up around 9-10KHz, which is well away from the crossover frequency which I envisage will be 3-4KHz with that Alu/ceramic tweeter. All will reveal itself over time.
When I look at the driver, I almost ignore the on-axis response and I look to 15° and 30° off-axis responses. Are there signs that persists both on and off axis. Check the CSD. Are there flaws that I don't think are deal-breakers and that a single series inductor can get away with? It takes a knowing eye, but it is clear that a large number of drivers are clearly totally unsuitable. The ones left over, they can be studied and a decision made. There is quite an art to driver selection.
So Troels or myself or any other, we are going to agree on some things and not on others. And that is OK. You have to do what you feel comfortable doing.
I look for drivers where I can get away with a series inductor. I don't know if anybody is doing that, but I have a number of designs behind me and that have users that love them, because they let the music come through and out of the box. Over-correcting drivers is a big issue IMO. Trying to fix something and then somewhere else something even worse can happen? Absolutely. Human experience tells us that this happens often and the solution is to change and adapt your design. Not always an easy path, but the end results is the only thing that matters.
[underlining added]
You are on to something... keep going. Definitely on the right track, especially about the amplifier reacting to the load. Let me give you a key word, tight coupling between amplifier and driver is being preached incessantly so that we have been almost brainwashed into the idea that the amplifier has some kind of grip on the driver and that this can be done using voltage and a low impedance source (but since 70% of the VC is outside the gap, that alone destroys the idea, ouch)
Rather than 'gripping' we actually want something else... isolation.
Once again, I am sure to be misunderstood, it comes with the territory of being a maverick and maybe I am. But a very famous scientist said that we should all be mavericks because we should seek to make things better than they are.
That word also helps describe why an 8R resistor across the terminals of a speaker can easily hear with many speakers with the Z plot I have shown above. You can easily hear an improvement that can be ascribed to lower distortion. But you won't measure any distortion decrease on the voltage amplifier. Note carefully what I am saying next, if not on the amplifier's voltage, it must be on the amplifier's current that is distorting. Voltage of the amp, fine. Current of the amplifier? Maybe not so fine. So how does a parallel 8R resistor do that.
Time to put your detective hat on.
I can tell you that you are on the right track, but keep in mind, to explain what that 8R does, be aware, you may come to a conclusion that the mainstream will likely be resisting.
And it is not just what that 8R is doing. What in common to a parallel 8R resistor, a series impedance (as in current-drive) and a series inductor (as used in the Elsinores, what do they have in common? Those three things, or methods, have something in common alright. What was the word I used above?
Hello,
I am going to tell an illustrative story.
I have been using tri-amped speakers for decades. My favorite midrange drivers are 10 inch cones (JBL2123H). If you cross them over near 400Hz the vast majority of the large voice coil excursions are directed to the woofer and woofer amplifier. Most of the mechanisms that cause the worst and most audible midrange distortions are excluded. These mid range drivers have a slowly rising Frequency Response so I include a small amount of series inductance to flatten the increasing FR. These midrange drivers are also very efficient, they are very loud at just one watt of input. In the world of amplifiers operating at just a watt of output the THD+Noise is problematic. At 5 watts amplifier output the THD+N is much reduced. For this improvement in THD+N I use a fixed voltage divider. In the traditional way of thinking that voltage divider is called a L-Pad (resistor in series with the driver and a second resistor in parallel ). I would say that this in not your father's crossover.
Not counter intuitive at all, if you think about it, it makes perfect sense. Plus the word current was not used. Yet I have done everything that you have been talking about. Series inductor, series resistor and parallel resistor.
Thanks DT
I am going to tell an illustrative story.
I have been using tri-amped speakers for decades. My favorite midrange drivers are 10 inch cones (JBL2123H). If you cross them over near 400Hz the vast majority of the large voice coil excursions are directed to the woofer and woofer amplifier. Most of the mechanisms that cause the worst and most audible midrange distortions are excluded. These mid range drivers have a slowly rising Frequency Response so I include a small amount of series inductance to flatten the increasing FR. These midrange drivers are also very efficient, they are very loud at just one watt of input. In the world of amplifiers operating at just a watt of output the THD+Noise is problematic. At 5 watts amplifier output the THD+N is much reduced. For this improvement in THD+N I use a fixed voltage divider. In the traditional way of thinking that voltage divider is called a L-Pad (resistor in series with the driver and a second resistor in parallel ). I would say that this in not your father's crossover.
Not counter intuitive at all, if you think about it, it makes perfect sense. Plus the word current was not used. Yet I have done everything that you have been talking about. Series inductor, series resistor and parallel resistor.
Thanks DT
Kinda missing the whole point there.
Would you elaborate on that? How?
I am not saying it isn't true, just how?
And would that have nothing to do with current?
Would you elaborate on that? How?
I am not saying it isn't true, just how?
And would that have nothing to do with current?
Sure it isn't a lack of clarity? I think it would be a more plausible explanation given the rather basic nature of the discussion... it isn't rocket science level...is it? Now , I'm gonna catch you in your immediate reflex to answer "no - and thats what so strange that no one gets it..." - before you do that - think one more round around the whole affair.
//
That's about the only thing I agree with you about. Have used the same phrase many times.
The idea is so saturated that the amplifier somehow controls what the speakers do, that they have this 'grip' - that a voltage source with a low Z can achieve that. Yet the pure science actuially weorks against this fallacy. The amplifier's job is to supply current and it is a question how well it does that. The fact that it is a voltage source makes it less in control of the current. If you want 100% control of the voltage, does that not mean that the amplifier has no control over the current, but the load does?
Do you agree with this or not?
If you don't answer that question, then there is no discussion to be had with you. It is your choice. Please, answer that question, so that at least I will know where you stand. Surely that is only fair.
You see, the thesis that you bring to the table, is fuzzy, so I cant answer it. It's the use of the word "control" that is the culprit. Sorry. So no discussion with me - it's OK.
//
//
Joe,
Yes you are kinda missing the whole point. There have been too many words and too much confusion generated about current and voltage. Yes they are important, however there are other major distortion mechanisms also at play.
You do not like it much when people link other stuff. You can find it.
Pufifi posted that after they had worked out and corrected Force Factor/Amplitude Modulation and other distortion mechanisms, including Hysteresis in the construction of their drivers, they were disappointed that much of the measured distortion remained. It was back to the Lab for them.
The New Patented Purif suspension/surround was the result. Suspension distortion and Sd modulation distortion (cone area) have been improved by 20+dB's.
I am a evidence based sort of Engineer. I like a theory that is supported by the measurements with the emphases on the measurements.
Thanks DT
For the Mini Elsinores on my bench I am likeing the 5" SB15NBAC30-8 woofer and SB26ADC-000-4 tweeter with the VISATON WG 148 R waveguide. The SB26ADC-000-4 with the VISATON WG 148 R waveguide has low distortion and the FR is smoother and flatter.
I have just casually followed this discussion and I sure ain't no electrical engineer so please pardon my naive curiosity; but isn’t a voltage-controlled current source circuit, as the name implies, a device where a small amount of voltage across the input will proportionally control the current flow across the output?
Purifi actually say something quite interesting regarding current drive in the mid band.
"Even if we don’t think it helps against FFM, we use a lot of shorting. This is mostly done to eliminate mid-band distortion caused by hysteresis in the remaining iron parts.
As anyone who has experimented with current drive can attest, its most spectacular effect is a reduction in distortion for pure midrange signals. A 15dB reduction in, say, the 300Hz-2kHz region is quite normal. This shows that for this frequency band the dominant distortion source is something that specifically affects the electrical impedance. This turns out to be the iron in the magnetic circuit.
We’ve recorded and plotted some actual waveforms to illustrate just how peculiar hysteresis distortion is compared to ordinary nonlinearities.
Shorting rings are very effective here. They shield the iron parts from the AC field emanating from the voice coil. Not at low frequencies of course, but there we had bigger fish to fry.
The current HD plots in our data sheets can help the user estimate how much hysteresis distortion is left."
https://purifi-audio.com/2021/10/14/some-speaker-problems-that-needed-solving/
The waveform plots are in this post
https://purifi-audio.com/2020/04/28/dist/
"Even if we don’t think it helps against FFM, we use a lot of shorting. This is mostly done to eliminate mid-band distortion caused by hysteresis in the remaining iron parts.
As anyone who has experimented with current drive can attest, its most spectacular effect is a reduction in distortion for pure midrange signals. A 15dB reduction in, say, the 300Hz-2kHz region is quite normal. This shows that for this frequency band the dominant distortion source is something that specifically affects the electrical impedance. This turns out to be the iron in the magnetic circuit.
We’ve recorded and plotted some actual waveforms to illustrate just how peculiar hysteresis distortion is compared to ordinary nonlinearities.
Shorting rings are very effective here. They shield the iron parts from the AC field emanating from the voice coil. Not at low frequencies of course, but there we had bigger fish to fry.
The current HD plots in our data sheets can help the user estimate how much hysteresis distortion is left."
https://purifi-audio.com/2021/10/14/some-speaker-problems-that-needed-solving/
The waveform plots are in this post
https://purifi-audio.com/2020/04/28/dist/
Ain't that the truth!
Read it.
I always said that if the driver was perfect, then you would not need current-drive. Even Esa knows this.
So always get the best driver possible. Know what you are choosing and what will be the likely result.
And I am not?
Measurements? Be patient and you will get LOTS!
But a measurement must have an aim before the contrive it. Curious, what do you think the test will aim to prove.
Were you not supposed to do some current tests using a current sense resistor on a Vifa P17WJ00-08?
I thought you were going to do this. If you did, then I would likely be able to point out a common mistake that is made about these measurements. The measurement is fine, but they make a rather fundamental interpretation of the result. They mis-identify the source of the current and that it does not come from the load (invoking Kershcoff's Law). They conflate two things that just shouldn't be done.
The world has been bamboozled with the idea that a low Z voltage source somehow control the speaker. But hello, a low Z voltage source cannot control current. Reviewers of amplifiers have also swallowed this... and on and on it goes.
So re the current sense resistor? Maybe if you went down that road, you might get an inkling what I am talking about.
Trust me, I am the one risking my name and reputation here. If you think that I am not evidence based, then you have no idea who I am. I feel that I am on incredibly solid ground.
I know it well. Some use waveguides to lower the crossover point because it makes it easier for them. I resist that. Waveguides should be used to lower distortion and they can do that massively. The Elsinores is all about distortion. The tweeter has been likened by some to electrostatics (one of whom actually builds electrostatics and very surprised). Also, use an LC null filter on the tweeter resonance. Best to use a single cap. Look at the high-pass in the Elsinores. The LC Q can be used to shape the response in the critical octave below (and around) the crossover frequiency.
What if I decided to bring out a design and call it Mini Elsinores?
What low pass are you going to use? Is this going to be passive?
Please, do those current-sense measurements.