Why would you say this? The output is greater in line with the increase in drive Voltage. The resonance damping changes accordingly.
But then we all know that except for low efficency bass systems, the current is very low. And then the distortion in voltage controlled 4 to 8 ohm driver is very low because they are operated far from x max
I only repeat this once more, please read the article and posts in this thread again and do circuit analysis:
Acoustic output depends on current through voice coil, voltage accross voice coil could be anything and have no effect on the output depending what makes the circuit imepedance, what sets the current. If you keep the current constant due to impedance outside of the driver, the voltage could vary all day long and have no effect on acoustic output. This is done in current drive, it is all what it's about, circuit impedance is made so high that current can be held constant no matter what the driver impedance is and side product is that voltage measured at various parts in the circuit varies according to impedance it's measured on.
No.
Current is dependent on Voltage and impedance. The impedance is modified by backEMF. Therefore backEMF counts. Voltage on the driver counts.
Yes, exactly, and you could use very low distortion driver, or multiple drivers as it has even more advantages on home hifi. You could use voltage amplifier but increase circuit impedance above the bass resonance. You can utilize "current drive" or "voltage drive" as you wish, to your advantage. There is no need to use current drive amplifier, as you can get similar performance without just with system design and manipulating circuit impedance as you wish.
Last edited:
Yeah that is if you do not understand that the electrical circuit expands outside the driver, and impedance outside the driver also affects the current on the circuit. When there is no shunt networks but just one loop of circuit then current in the circuit is same on all parts of it simultaneously. Now, if you put 1000Ohm resistor between driver terminals, or high output impedance amplifier, the circuit current is not affected by driver impedance as it is very little in proportion of total circuit impedance and have very little effect on circuit current, and very little effect on acoustic output.
Conversely, if you put voltage amplifier between driver terminals with very low output impedance it is basically just a short circuit. Then the driver impedance dominates the whole circuit impedance, and circuit current, and in that case your sentence is true. Then back EMF voltage has very low impedance to work on and can vary circuit current and have maximal effect on acoustic output.
Last edited:
I enjoyed them all. To me it seems like the benefit og current feedback is highest for high distortion drivers or high currents/voltages like this: https://www.diyaudio.com/community/...-woofer-available-sort-of.231187/post-5724539 even without acceleration feedback the distortion is redused by an impressive amount on that faulty subbass driver
I kinda lost interest in the discussion.
It's shooting in all kinds of random directions atm, while it just only started with me taking a little side track to see what benefits it would bring compared to the passive solution to fix some distortion issues.
It's shooting in all kinds of random directions atm, while it just only started with me taking a little side track to see what benefits it would bring compared to the passive solution to fix some distortion issues.
Sending someone into the woods; "here is a topic/book with 600 pages, good luck with it, it's somewhere in there", seems to be the standard these days 🙁
You might as well not share the source or reference at all at that point.
Especially when it refers to a specific detail.
Sorry, there are several issues involved as you know.
To make/explain the sub they needed to discuss all of them. So not easy to pick one of them to explain the whole system.
And that is the difficulty of discussing why a passive component may help distortion in an active system.
But my personel conclusion is that
To make/explain the sub they needed to discuss all of them. So not easy to pick one of them to explain the whole system.
And that is the difficulty of discussing why a passive component may help distortion in an active system.
- Active component increase impedanse that the speaker sees.
- That is equivalent to using a current sourse at the frequencies of interaction. Z passive component >> Z speaker
- Will current source decrease distortion?
- Give references of therory why current sourse decrease distortion (Large field of theory on speakers and amps)
- Give reference to experiments where current sourse is used and distortion is decreased.
But my personel conclusion is that
- Passive damping of large HF reconances outside passband is easy, inexpencive and gives lower distortion inside passband
- Introdusing a current amp to achive lover distortion in LF is very difficult, but doable
Yeah, the impedance governs the whole electric domain and it feels complex or random if you can't put it all together as one concept, impedance 😀 which literally isn't anything else than relationship of voltage and current.
I've tried to explain the same thing from various perspectives with various wording, Esa's article explains it, any current drive paper explains it, many other people explain it on the forums. It's all about impedance, how voltage and current relate. To understand why distortion is reduced with current drive: With realization that only current makes acoustic sound, and that there is voltage source within the driver whose output voltage depends on cone velocity, and the voltage makes circuit current if circuit impedance is low outside the driver.
It gets hard to understand if one tries to incorporate power amplifier internal workings on the thought process, but gets very simple if the actual power amplifier is abstracted away with just equivalent resistor. Just ignore power amplifiers altogether to understand how the driver emits it's own distortion into acoustic domain through low circuit impedance, or prevent outputting it's own distortion when circuit impedance is high. Basic circuit analysis can be easily applied how much current voltage source within the driver makes through the voice coil. In other words, how big portion driver impedance is of total circuit impedance. This is all illustrated with maths and images in Esa's article.
I cannot say this any more simpler: it is literally how much current back EMF voltage can make into the circuit. This can be analyzed by analyzing what impedance is series with the driver. Z=U/I, or I = U/Z, make Z high and current gets less.
Everyone needs to do their own thought process and circuit analysis to get grasp on it. Draw the circuit to understand whats between driver terminals affects current through voice coil. Please find as much material as you need, as I think everything about passive filters is very easy to understand when it clicks. I'm afraid I have nothing more to give on this, tried in various ways failing miserably 🙂 Too much flooding making it more confusing than it is.
I've tried to explain the same thing from various perspectives with various wording, Esa's article explains it, any current drive paper explains it, many other people explain it on the forums. It's all about impedance, how voltage and current relate. To understand why distortion is reduced with current drive: With realization that only current makes acoustic sound, and that there is voltage source within the driver whose output voltage depends on cone velocity, and the voltage makes circuit current if circuit impedance is low outside the driver.
It gets hard to understand if one tries to incorporate power amplifier internal workings on the thought process, but gets very simple if the actual power amplifier is abstracted away with just equivalent resistor. Just ignore power amplifiers altogether to understand how the driver emits it's own distortion into acoustic domain through low circuit impedance, or prevent outputting it's own distortion when circuit impedance is high. Basic circuit analysis can be easily applied how much current voltage source within the driver makes through the voice coil. In other words, how big portion driver impedance is of total circuit impedance. This is all illustrated with maths and images in Esa's article.
I cannot say this any more simpler: it is literally how much current back EMF voltage can make into the circuit. This can be analyzed by analyzing what impedance is series with the driver. Z=U/I, or I = U/Z, make Z high and current gets less.
Everyone needs to do their own thought process and circuit analysis to get grasp on it. Draw the circuit to understand whats between driver terminals affects current through voice coil. Please find as much material as you need, as I think everything about passive filters is very easy to understand when it clicks. I'm afraid I have nothing more to give on this, tried in various ways failing miserably 🙂 Too much flooding making it more confusing than it is.
Last edited:
Hi, I'm not sure what your post was about but just in case I've commented the bullet points, hopefully helping you out. If it feels confusing ignore it.
Yeah true
Speaker sees impedance that is between it's terminals. Usually active systems have no passive components between amplifier and driver, so it's just amplifier output impedance + any cabling. This is very low with voltage amplifier
true with changing to: Z in series with driver >> Z driver
Yeah, if it makes series impedance with driver high so that back EMF makes very little current in the circuit. In other words when Z in series with driver >> Z driver
For example Esa's article.
For example Purifi article about parallel notch in series with driver found. Or my measurements earlier in this thread, and there is likely many others I cannot remember now. With actual current amplifier without passive network, see Esa's posts in Finnish forum here https://foorumi.hifiharrastajat.org/index.php?threads/särövertailua.93446/ and I believe there is also article with the data here https://acoustics.ippt.pan.pl/index.php/aa/article/view/1780
Yeah true
Are you bringing this over from the example of breakup response level affecting harmonic response levels?.. because that is a different situation.
While the breakup itself is different thing from all this impedance stuff as it is not in electronic domain but in acoustic or mechanical domain, the harmonics in breakup discussion were from same distortion mechanism as here: Le(x) affecting current through voice coil.
Increasing circuit impedance reduces effect of Le(x) on total circuit impedance and thus in current. Increased circuit impedance attenuates the harmonic (current and acoustic output) and now that the breakup still boosts it, it's less than with low circuit impedance. It is nice demonstration how Le(x) makes distortion appear in acoustic domain and how to prevent it.
I could use another wording to describe the same thing, but you really have to figure out it yourself. It is quite clear I have not been able to give such explanation on the subject that it would be clear how it plays out. You are smart person so you should be able to figure it out without my input, if you wish.
Increasing circuit impedance reduces effect of Le(x) on total circuit impedance and thus in current. Increased circuit impedance attenuates the harmonic (current and acoustic output) and now that the breakup still boosts it, it's less than with low circuit impedance. It is nice demonstration how Le(x) makes distortion appear in acoustic domain and how to prevent it.
I could use another wording to describe the same thing, but you really have to figure out it yourself. It is quite clear I have not been able to give such explanation on the subject that it would be clear how it plays out. You are smart person so you should be able to figure it out without my input, if you wish.
Last edited:
Lets do this, if you want me to help you to understand the subject please post questions that I can answer with simple "yes", "no" or "it depends" so that you can do your own reasoning based on your own thought process. Otherwise, don't get offended if I do not answer, I've tried my best to explain which obviously doesn't work, I cannot think for you.
With the breakup example, the produced distortion is not changed. Instead, the harmonic is independently added to or subtracted from, which is a linear process.
I think the back-emf counts in either type of drive. However, I don't think it manifests itself by modifying the impedance, but instead presents itself in opposition to the voltage applied to the voice coil (picture). And with current drive, this voltage applied by the amplifier is simply adjusted such that the coil current tracks the audio signal, in spite of the back-emf or any impedances like Re, Le, Rcable, Lcable etc.
I don't want to get too involved in the discussion but I did do a fair amount of work on the topic back in the 80s, 90s. The question was what was the difference when driving a loudspeaker with a high impedance (near current) source vs. a voltage source and did it make a difference in distortion.
First off, driving with a high impedance looks largely like using a voltage source and then adding an equalizer matching the shape of the impedance curve. So at the motional impedance peak (resonance) the SPL peaks. As the inductance rises the response rises. Rather than SPL falling due to any impedance rise the current is made constant (so voltage increases in proportion to impedance and frequency response follows the impedance curve).
This makes it difficult to compare distortion because you would always be comparing between two cases with very different frequency responses (hence different woofer excursion, etc.).
I got around this by using an ancient B&K oscillator that had a feedback (compressor) loop and let me drive the units slowly against frequency while keeping the SPL flat. The SPL fundamental was filtered out and used as the feedback component. As soon as this was done the bass distortion curves, which had been very different for voltage drive versus current drive, became the same.
This makes sense because all the parameters that cause bass distortion are solely related to excursion. This includes coil length and Bl drop due to that, and magnetic circuit B profiles that vary due to core pole and gap design. So if I am using the compressor loop to maintain a flat 90 dB at 1 m across a range of frequencies then the excursion and therefore distortion will be constant, in spite of the amplifier input needing to vary dramatically between the constant voltage case and the constant current case (inputs will vary, sadly the amplifier outputs will be the same).
Now there was one distortion parameter that did care about driving impedance. That was the iron related midrange distortion. Hysteresis causes a distortion in the current that for midrange frequencies leads to 2nd harmonic (if I recall correctly) distortion. This is typically reduced in high end drivers with copper pole caps, or aluminum shorting rings at the base of the core pole. (See JBL SFG or the McIntosh equivalent, LDHP). If the driver is driven with a voltage source then the hysteresis causes distortion in the current and since current drives the speaker (F=BLi) then some distortion results. If you drive it with a current source then the amplifier keeps the current constant and this one distortion is suppressed. Another way to look at it is that the distortion factor in the iron is changing the instantaneous impedance of the driver through out every cycle. Feed it with constant voltage and the SPL must track the impedance variation. Feed it with constant current and you are putting a larger linear resistor in series with the driver's non-linear resistor.
Now this doesn't put me in the camp of "Lets use constant current amplifiers", primarily because I wouldn't relish the work of equalizing a current driven speaker. I also think that the distortion mentioned can be easily fixed in drivers so that would be the better route. There is also the minor issue of thermal compression, but I have never seen that as a real issue for everyday speaker use.
On that subject and as a historical note, while I was at KEF there was a joint venture with B&O (Ureka project) to simulate soundfields with an array of speakers in an anechoic chamber. Laurie Fincham was worried about thermal drift of loudspeaker sensitivity and had B&O design their amplifirs as high impedance sources so that changes in VC temp wouldn't change speaker sensitivity. It was probably overkill but Laurie liked that sort of thing.
Regards,
David
First off, driving with a high impedance looks largely like using a voltage source and then adding an equalizer matching the shape of the impedance curve. So at the motional impedance peak (resonance) the SPL peaks. As the inductance rises the response rises. Rather than SPL falling due to any impedance rise the current is made constant (so voltage increases in proportion to impedance and frequency response follows the impedance curve).
This makes it difficult to compare distortion because you would always be comparing between two cases with very different frequency responses (hence different woofer excursion, etc.).
I got around this by using an ancient B&K oscillator that had a feedback (compressor) loop and let me drive the units slowly against frequency while keeping the SPL flat. The SPL fundamental was filtered out and used as the feedback component. As soon as this was done the bass distortion curves, which had been very different for voltage drive versus current drive, became the same.
This makes sense because all the parameters that cause bass distortion are solely related to excursion. This includes coil length and Bl drop due to that, and magnetic circuit B profiles that vary due to core pole and gap design. So if I am using the compressor loop to maintain a flat 90 dB at 1 m across a range of frequencies then the excursion and therefore distortion will be constant, in spite of the amplifier input needing to vary dramatically between the constant voltage case and the constant current case (inputs will vary, sadly the amplifier outputs will be the same).
Now there was one distortion parameter that did care about driving impedance. That was the iron related midrange distortion. Hysteresis causes a distortion in the current that for midrange frequencies leads to 2nd harmonic (if I recall correctly) distortion. This is typically reduced in high end drivers with copper pole caps, or aluminum shorting rings at the base of the core pole. (See JBL SFG or the McIntosh equivalent, LDHP). If the driver is driven with a voltage source then the hysteresis causes distortion in the current and since current drives the speaker (F=BLi) then some distortion results. If you drive it with a current source then the amplifier keeps the current constant and this one distortion is suppressed. Another way to look at it is that the distortion factor in the iron is changing the instantaneous impedance of the driver through out every cycle. Feed it with constant voltage and the SPL must track the impedance variation. Feed it with constant current and you are putting a larger linear resistor in series with the driver's non-linear resistor.
Now this doesn't put me in the camp of "Lets use constant current amplifiers", primarily because I wouldn't relish the work of equalizing a current driven speaker. I also think that the distortion mentioned can be easily fixed in drivers so that would be the better route. There is also the minor issue of thermal compression, but I have never seen that as a real issue for everyday speaker use.
On that subject and as a historical note, while I was at KEF there was a joint venture with B&O (Ureka project) to simulate soundfields with an array of speakers in an anechoic chamber. Laurie Fincham was worried about thermal drift of loudspeaker sensitivity and had B&O design their amplifirs as high impedance sources so that changes in VC temp wouldn't change speaker sensitivity. It was probably overkill but Laurie liked that sort of thing.
Regards,
David
@[B]speaker dave[/B]
Thank you very much for this very informative post👏I wonder if those results you saw back in time correspond with measurements made during development of the The Grimm Audio LS1s-DMF subwoofer (I believe, but don't know) and presented in this post in diyaudio. It is the second measurement: "Distortion taken from the sensor (100db spl)" and the green and black curves. Or did you experience a much less decrease in distortion with current amp?
I also wonder about the case where a speaker element is driven by a constant voltage amp, but we have passive components in series with the sub or mid speaker element, limitng the upper bandwith. Do you think the in band distortion will be decreased because of the "current drive" when Zspeaker << Zpassive out of band even without a breakup peak out of band? Or will the in band distortion in that case be dominated by other effects.