Which Capacitors for Audio?

[abbey road d enfer]

Are we talking about these kinds of things?

No. These are WiFi/BT speakers using conventional speaker arrangements and heavy EQ.

 

[abbey road d enfer]

I stated no visible movement with a specific 15" Midrange (for front horn loading) driver when driving the speaker with a signal containing bass, which would cause very large displacement with classic 15" Bass Driver like EVM15B or JBL 2226.

I clearly stated LF or bass. Midrange has many other problem but excursion and Bl variation is not one.

So you are just being deliberately obstuse.

Perhaps because you are deliberately opinionated.

Clearly you are mistaken, IF for the added mechanical damping electric damping is reduced.

Read again. I clearluy stated "optimization conducted by passive means". Current drive is an active process (unless you want to put a 1kohm resistor in series).

No, but current drive means Re is no longer material. Current drive cancels Are variations contribution in the driver's acoustic output.

At the cost of increased power draw.

A conductive voice coil former with the correct resistance provides electromechanical self damping of the driver, so driving it with an AC current source doesn't cause LF peaks and poor transient response, due to the absence of electrical damping from the driving amplifier.

Such created damping is purely electrical; as such it does not compensate the effects of Bl loss, which are quite well taken into account by providing a low source impedance.
It's a xfmr which primary is the voice-coil and the secondary a shorted turn. It's a common trick used to compensate the increase of impedance due to the inductive element (Le)

Basic laws of physics? Lentz among others?

Yes, and Lenz says that there is energy dissipated. If damping is superior, energy demand increases.

Once the voice coil leaves the air gap output drops off so rapidly, we do not add appreciable LF output anyway. But cooling of the voice coil is comprised, potentially creating a thermal avalance condition that delaminates the voice coil glue.

When the voice-coil temperature increases, thus its DCR, current drive will result in faster thermal avalanche than voltage drive. I'm very familiar with the destructive limits of loudspeakers.

Interesting, what was the HD under those conditions? And compression?

I would say it's almost irrelevant. The audience wants high spl and they want it for many hours in a row.

At Xdam the BL is likely 10-20dB down on standard Xmax BL.

Then come PHL loudspeakers.

A servo loop could be used to make this usable in practice, again, Sub use only. The protection circuitry against thermal or mechanical avalanche conditions would be interesting. I'd not volunteer to do that without a serious DSP maybe not even with.

I designed an analogue protection system that worked well enough to ensure the speakers would work consistently from the beginning to the end of a festival night, and from day one to day x. Later, this system has been ported in digital platforms by Linea reserach (now Martin Audio) and Powersoft.

At the same time, why not specify a longer voice coil, drop midband efficiency (ultimate LF Output is down to Xmax and cone area anyway) and you have a Sub that has better inherent linearity with exactly the same 32Hz (or whatever frequency you pick) output?

Because a long VC results in large waste of power. We want both efficiency and spl.

Or, why not use an industrial rotary motor or a a set of linear motors a do a modern version of Danleys "ServoDrive". You can easily get a 1sqm carbon fiber honeycomb panel and get way past +/-1" linear excursion.

I'm familiar with Servo Drive, only good for shaking trousers. Also familiar with M-Force. Quite better.

 

[moamps]

Removed

 

[living sounds]

I have been using various Burr-Brown / TI "Advanced Segment DAC" parts, under software control and with some use of non-documented features and slightly non-standard analogue stages.

Other than that the CS43131/198 is really good as DAC. Especially if used dual mono and possibly using multiple DAC's per channel.

ADC I have been using TI parts, the PCM4222 is nice. The TLV320ADCX140 range is good for lower end gear.

I have been unimpressed generally with ESS.

Something I'd like to try is to build an ADC/DAC using the industrial 20-Bit / 1 MspS ADC/DAC Chips from Linear, these tend to be 50 Bux plus a piece, but we could get a 768kHz sample rate 20 Bit ADC/DAC without digital filter.

Add an FPGA and trade sample rate and bit depth with simple filters, so averaging two samples give us 384k/21 Bit, while averaging four gives 192k/22Bit.

With likely 300 USD+ in chip's alone this would be not a cheap device.

Something like this I might call:

Atlantique Macrosonics

Model T.R.E.S

And try to make it look appx like this:

View attachment 111394
Hehehehe...

Thor

Interesting. I've played quite a lot with the PCM4222 evaluation board, even tried different input stages. With the digital low cut filter turned off and DC coupled it was the best performing AD converter in null tests. In listening tests I always found it to sound a little blunted / liveless dynamically and somewhat hazy on top.

 

[living sounds]

That sounds like a case of designing the filters to look good on paper (specs) while cutting some corners.

It's my impression that that is what AA filters in converters and plugins today mostly are designed for. I use discrete DACs that allow for using your own filters (within the limits of its processing power, of course), and then one I use (not my design) because it sounds by far the least colored, cuts gently and a lot lower than what is almost always used today. For AD I use a Lavry Gold, which has more latency than the competition, and my guess is that this is down to a reconstruction filter with a bigger window. With the usual filters it should have very little latency, since from the way THD behaves at different amplitudes it seems not to be using delta sigma modulation.

 

[JohnRoberts]

I'm familiar with Servo Drive, only good for shaking trousers. Also familiar with M-Force. Quite better.

(Intersonics) ServoDrive was remarkable for what it was, especially when it was (1980s). I'm pretty sure those cones moved.

Coincidentally I drove down to Arlington, VA, from CT in the early 70s to visit the patent office and do a physical search through the shoes (drawers) full of physical patents. I don't recall exactly what I was looking for. Probably several things. I recall finding one patent from Phillips(?) with a motional feedback device coupled to the bass driver.

Speaking of the ServoDrive Tom Danley has made numerous notable achievements in full range loudspeaker design (mostly in box/horn design and novel driver alignments). Other industry giants have made advancements in DSP.

IMO loudspeakers are another mature audio technology. There are still modest incremental advancements being made. I recall back in the 70s playing with negative impedance (?) active drive to generate dynamic bass boost, making tiny boxes with small drivers sound big, until their driver voice coils melt.

JR

 

[thor.zmt]

Perhaps because you are deliberately opinionated.

Perhaps, except I'm as straight a shooter as I can manage with sniper grade ammo in a Drag, or 3 subsequent rounds on a playing card at 500m, within 10 seconds.

Read again. I clearluy stated "optimization conducted by passive means". Current drive is an active process (unless you want to put a 1kohm resistor in series).

Voltage drive is an active process.

Let's take a BJT Pair, no feedback, class A. Connected as grounded collector, we have voltage drive. Switch the ground point to grounded emitter and we have current drive, no other change.

At the cost of increased power draw.

Incorrect.

Such created damping is purely electrical; as such it does not compensate the effects of Bl loss, which are quite well taken into account by providing a low source impedance.

Wrong again, as the coil former remains fully within the gap and magnetic field for sensible excursions (bump the backplate and lengthen the former if not) there is no BL loss to start with.

And BL loss with excursion is simple math.

It's a xfmr which primary is the voice-coil and the secondary a shorted turn. It's a common trick used to compensate the increase of impedance due to the inductive element (Le)

Nope. That shorted turn is commonly (in fact afaik exclusively) part of the static magnet system, never of the moving assembly, with a handful (at most) outliers.

Yes, and Lenz says that there is energy dissipated. If damping is superior, energy demand increases.

If damping is equal but from different sources, energy demand is equal.

Hence my repeated qualification of " all else being equal".

When the voice-coil temperature increases, thus its DCR, current drive will result in faster thermal avalanche than voltage drive.

Correct. However it will cancel power compression up to the point of self destruction. Pick your poison. Design for the choice.

I would say it's almost irrelevant. The audience wants high spl and they want it for many hours in a row.

In concert/venue settings, yes. That said, I think this overstated. I took several incredibly loud clubs where screaming into the barman's ear to order was useless, rearranged the systems and dialed down SPL's a lot, while maintaining emotional involvement.

My big moment was when a huge Hong Kong Canto Pop Star walked into the venue in Beijing that I had worked at, looked around dazed and confused (he was a regular with his own tables and loads of champus), took out his ear plugs and said "I don't need these anymore".

He asked who had finally sorted out the sound and we got introduced. I still get the occasional mail every few years asking for "sound advise" and more (pun intended) in Asia and jobs.

Actually, it's not absolute SPL, it's experience and adrenaline/oxyticine intoxication they want.

Then come PHL loudspeakers.

555

I designed an analogue protection system that worked well enough to ensure the speakers would work consistently from the beginning to the end of a festival night, and from day one to day x.

So did I. Not that I'd ever doing again.

Because a long VC results in large waste of power. We want both efficiency and spl.

NOT AT THE LF CORNER! Above that, perhaps, but it also is cleaner, a lot.

Again, pick your poison.

I'm familiar with Servo Drive, only good for shaking trousers. Also familiar with M-Force. Quite better.

In the early 90's I did live sound with an outfit that used ServoDrive Bass Tech 7 and Community tops.

It was the hardest and loudest and cleanest (not just bass) I ever got from a system that fitted into a single truck. It was ridiculous. It got chest crushing at the mixer booth. Probably violated strategic arms limitation treaties In force back then.

I also worked with Turbosound, Funktion One and Meyer. Not even close.

I am sure you disagree.

Thor

 

[JohnRoberts]

you guys are wearing me out....

JR

 

[thor.zmt]

Interesting. I've played quite a lot with the PCM4222 evaluation board, even tried different input stages. With the digital low cut filter turned off and DC coupled it was the best performing AD converter in null tests. In listening tests I always found it to sound a little blunted / liveless dynamically and somewhat hazy on top.

Try a pure transformer frontend, (stepdown, use low distortion units) and limit the signal to around 80kHz passive and be sure to filter both common and differtial mode

Major biggie, use an active driver for Vref, low noise, low impedance (> 100 milliohm for Audio range). Op-Amp's can work, NJM5534 is pretty good. This is more important than the actual signal path, believe it or not.


I like half diamond Sziklai on top of a cascode CCS instead, 2N4401/03 for the diamond, BD140 to get the Sziklai. Noise is much lower than any commercial op-amp I'd care to use.

And watch powersupplies, add 10nF 0603 C0G bypass on all supplies, use multiple 1uF 0603 X7R to replace the larger size, lower value decoupling caps.

Put oversized Panasonic Os-Con as main rail decoupling caps. Use low noise power supplies.

Again, I like the diamond/Sziklai structure, TL431 as reference, then RC filter at 0.1Hz or less (with a bridging diode to speed up PSU ramp up).

If you can, make a new PCB, 6 layers. Use 3 layers as digital ground, VCC for logic and routing wedged between the layers.

Under absolutely no conditions add inductors anywhere on the VCC, make it a solid image plane like GND, anyone who suggest different is an exit. Look at the current loops involved with push-pull CMOS logic.

Remaining layers handle analogue, using classic analoge layout for lowest interference. Digital ground plane is also general analogue ground plane for anything RF, signal ground is a separate plane or just traces (star ground ) linked at Vref ground pin of the ADC IC.

Clock quality also matters a lot, I prefer to run clocks via Coax and SMB connectors with an isolated and rise time optimised transformer coupled clock.

This eliminates a lot of trouble sources. Using simple lan magnetics (1000 Base T) and unbuffered inverters from a low noise, fast rise time series is usually good.

Of course, this all applied as much to DAC as to ADC.

here clock distribution in a high quality DAC:




The "golden box" is a compliant suspended and contains the clock crystals and driver circuitry with power supplies.

It is not ovenised, but still quite well stabilised. Makes most if not all "rubidium" clocks look bad. Can sync with external 10Mhz clock without adding noise.

Using a burried stripline is probably as good, but then we need 8 - 10 layers. Just use Coax instead.

Thor

 

[MicUlli]

WTF have all the posts in common with the question at the very beginning of the thread?
The question was:

Can someone help provide some idiot guidance on which are the better quality film type caps for audio use?

I personally don't believe that "better" is the correct wording. Quality film type caps should have:

1. tight capacitance tolerance (5% or smaller),
2. at least 2x voltage rating in relationship to the DC operating voltage,
3. small ESR (in the requency band of interest),
4. small DF,
5. at least 3x AC current rating in relationship to the RMS opreating current,
6. small tempco (if you want to create military products)

Check the cap you have chosen within your circuit (prefer meter reading vs. golden ear disbelief)

BR MicUlli

 

[JohnRoberts]

It is the nature of the internet that after the immediate question is answered people will keep bloviating all that they know about the subject, and other subjects.

mea culpa.

JR

 

[Ricardus]

WTF have all the posts in common with the question at the very beginning of the thread?
The question was:

Can someone help provide some idiot guidance on which are the better quality film type caps for audio use?

I personally don't believe that "better" is the correct wording. Quality film type caps should have:

1. tight capacitance tolerance (5% or smaller),
2. at least 2x voltage rating in relationship to the DC operating voltage,
3. small ESR (in the requency band of interest),
4. small DF,
5. at least 3x AC current rating in relationship to the RMS opreating current,
6. small tempco (if you want to create military products)

Check the cap you have chosen within your circuit (prefer meter reading vs. golden ear disbelief)

BR MicUlli

Watch Mr. Carlson's video on youtube about capacitor replacements, and then go to Mouser and don't spend more than $5 on any cap. You'll be fine.

 

[abbey road d enfer]

1. tight capacitance tolerance (5% or smaller),

It depends what you expect. If the capacitor is used to determine the LF frequency response of a whole unit by defining a dominant pole, its correct. If it is used in an intermediary stage (like the coupling caps between PI and output tubes), it's less important as the -3dB LF sould be significantly lower than the desired overall response.

2. at least 2x voltage rating in relationship to the DC operating voltage,

Again it depends very much of the circuit. Is leakage an important factor? Is the voltage coefficient important (it relates to distortion).

3. small ESR (in the requency band of interest),

ESR is in most cases indifferent, particularly in film caps. Contrary to electrolytic, the ESR is stable and not modulated by signal, so does not introduce distortion. Low ESR is a requirement for electrolytic caps used in smps.

4. small DF,

Dissipation results in increasing damping of tuned circuits. It may or may not create issues in filters, but is inconsequential for caps used in coupling position.

5. at least 3x AC current rating in relationship to the RMS opreating current,

Yes.

6. small tempco (if you want to create military products)

Or simply a stable oscillator.

There are other factors, like the type of dielectric (which governs distortion), construction type (stacked or rolled), compactness...

I respect your choices, but as you say, there's no "better". Each one has their own preferences, and each circuit requires analysis.

 

[ct.waveform]

Tell me how a cone that doesn't move can produce loud bass, except if loaded with a particularly efficient horn.

What the Hell is the problem with understanding this?! Thor is, certainly on a single read, right in everything he says. I've had a current driven MFB loudspeaker on my list of projects I plan to do for years now. And if someone would pay me I'd start work on it right now. But I think you've got to shape the input signal for the impedance curve. And the other thing I would do differently is have the microphone inside the cabinet because I would want to keep the transfer function of the bass low pass faithful and uncontaminated by the mid output and in room noise. Of course you have the radiation resistance problem but that's trivial. There may also be some unwanted feedback issues but that's practical engineering. (I may well be disabused of both those positions but I'll find out in due course. )

The shorted, or semi-shorted, VC is a truly lovely idea but I don't think you can bring down Qm all the way for quite a few reasons, but you probably can take it part-way. You are going to lose a helluva lot of practical power handling if you take it all the way, and need more amplifier power. You lose the impedance hump, which probably has a mean of 30 ohms (and might have a peak of 100) which has now been brought down to 6 ohms, or even 4, so your amp is now in a totally different ballpark, in a region close to the peak spectral content of music. (Yes, I realise those figures will be lower for PA speakers.) I reckon there's every chance you'll fry the former before you fry the coil. LOL! 600 deg C. You've got them both producing heat that they don't usually do, and we can work out almost exactly how much - nothwithstanding that it does get over very real Qm and excursion problems. I would doubt you can even solder the braid, except perhaps on a prototype. The other problem I foresee, though this would need a bit of thought as to exactly what happens, is what it's going to do the magnet and metalwork. Presumably there'll be some cancellation, but what if it snaps in and out of saturation, varying the inductance instantly (a source of distortion in any case, which you have now made worse, when you thought you were improving it) and varying the Qe. You could say it's current driven so Qe doesn't matter, and we've got feedback, but this could be out of band and impractical to correct after you've tweaked the system to ignore cabinet resonances or whatever else arises. Two sine waves with a few degrees between them is sod all difference in pressure but that stops being a few degrees further up in frequency. (Before some belligerent tosser tells me that's not how negative feedback works, yes, I know.)

I don't doubt for a second that it works, is probably excellent, and might even be practical domestically with over-specced drivers and now that we have free power with Class D (I wouldn't want to be the transformer in an A/B amp driving this), but I suspect that the power handling penalty would be too much for concert use - not that that's my area AT ALL. In either case you're putting your driver through hell and I doubt I'd expect a long life. And when the enamel finally gives up the ghost with the amp delivering 15A, that'll be the amp gone too, and probably represents a fire hazard to boot. (Actually, what about Curie temperatures, if it lasts that long?) This is fun! But I understand that it's such a nice idea that it's a siren call. I also like it because it backs up my long held position against the bollocks over low-loss rubber surrounds being better. The last thing you want is a high Qm; not only does it vary by itself but the higher it is, the more changes in Qe show up in Qt. These drivers badly need damping and having them generate back EMF to be gobbled up by the amplifier might even be thought one of the worst ways to do it. So I'm doubtful of its commercial future, but there's nothing wrong with it in principle (and with good home insurance). Or I could be wrong and it'll appeal to the kind of people who buy JBL Everests (a pretty good speaker, incidentally) and want an even bigger amp!

But anyway, none of this should be difficult to understand.

 

[moamps]

WTF have all the posts in common with the question at the very beginning of the thread?

You're right, that's why I deleted my offtopic post.
But your post and some posts before actually show how this place is getting more and more hostile and the S/N ratio is getting worse and worse.

 

[thor.zmt]

I've had a current driven MFB loudspeaker on my list of projects I plan to do for years now.

I did two prototypes, one a Subwoofer with 6 X 10" Drivers, sealed and one a vented box where the the MFB was similar to the Meyer solution.

And the other thing I would do differently is have the microphone inside the cabinet because I would want to keep the transfer function of the bass low pass faithful and uncontaminated by the mid output and in room noise.

Well, think this through some more. The Mic based systems I used helped a LOT with room effects.

Of course you have the radiation resistance problem but that's trivial. There may also be some unwanted feedback issues but that's practical engineering. (I may well be disabused of both those positions but I'll find out in due course. )

The biggest issue is to get a sufficiently linear sensor. This is surprisingly non-trivial.

The shorted, or semi-shorted, VC is a truly lovely idea but I don't think you can bring down Qm all the way for quite a few reasons, but you probably can take it part-way. You are going to lose a helluva lot of practical power handling if you take it all the way, and need more amplifier power. You lose the impedance hump, which probably has a mean of 30 ohms (and might have a peak of 100) which has now been brought down to 6 ohms, or even 4, so your amp is now in a totally different ballpark, in a region close to the peak spectral content of music.

Really? What is the impedance hump? It is the electrical representation of a mechanical resonance. If we do not short-circuit the voice coil and feed such a speaker from a current source we get a huge LF response hump and a very resonant system.

How is that mitigated using a low output impedance Amplifier? By injecting a current that opposes this resonance. These blind currents are out of phase with the voltage but are very real. Except in this case the dissipated power is "destructive", that is, it "destroys" some of the

Let us for a moment ignore the shorted turn damping and apply some strictly mechanical damping solution, be it the grease-pots of the original 1930's eckmiller coaxial, the Goodmans ARU or the Dynaudio "Variovent".

What is the result of the correct tuning of this element? The impedance "hump" flattens to the desired system Q. Does that make the speakers blow up?
Damping must be supplied somewhere. Doing so requires energy to be absorbed.

How this is done varies, but ultimately the same energy is required to be input into the system to get a given acoustic output and for an equally damped system it will be equal. Preservation of energy et al.

I don't doubt for a second that it works, is probably excellent, and might even be practical domestically with over-specced drivers and now that we have free power with Class D

As remarked, we had drivers with such voice coil formers existed in the 80's and worked very well in practical terms. And no, they did not overheat or blow up amplifiers. So this is not a problem.

Don't forget, the voice coil is wound around a solid mild steel (or similar) core and surrounded by a large ring of solid mild steel etc. That are already two shorted turns and are one of the reasons for the poor electrical efficiency of moving coil (or moving magnet) speakers, something that could be corrected for example using more efficient motor systems (electrostatic which we now get as MEMS, piezoelectric etc.).

I also like it because it backs up my long held position against the bollocks over low-loss rubber surrounds being better. The last thing you want is a high Qm; not only does it vary by itself but the higher it is, the more changes in Qe show up in Qt.

Well, yes and no. You want, in my view a Qm that is equal to Qt with linear damping of the spring/mass resonant system.

Typically spiders and surrounds the produce low Qm are not very linear. This does have some potential practical benefits, in some specific situations.

These drivers badly need damping and having them generate back EMF to be gobbled up by the amplifier might even be thought one of the worst ways to do it.

Yup. Then again, in a lot of ways having an internal combustion engine drive a car is a way worse solution than electrical drive, for a wide range of reasons (reliability, longevity, efficiency).

Yet they dominated the whole sector for over 100 Years, for reasons outside of the scope of the direct technical solution.

So I'm doubtful of its commercial future, but there's nothing wrong with it in principle

If the principle is implemented correctly, it will allow relatively commodity grade systems to be materially improved. That applies to the kind of lower end audio systems all over the range, be it portable sound system, monitors or home audio.

No doubt high end systems could also be improved, but less so, or similar performance could be attained for lower budgets.

Thor

 

[terry setter]

Thor, you mentioned the Dyneaudio Variovent. I used those "on spec" once and never was able to do sufficient testing to arrive at a final opinion about their usefulness. Do you know why they didn't remain popular? My experience was that I preferred the sound of the Variovent-designed cabinet when I sealed it entirely.

 

[ct.waveform]

I did two prototypes, one a Subwoofer with 6 X 10" Drivers, sealed and one a vented box where the the MFB was similar to the Meyer solution.

Before I start my reply I wanted to ask you whether Ansar are still in business? Their website appears to date from 2019 and is not that functional. In about 1992, which I see must have been very soon after they started, I needed some parts quickly (we didn't have standard next-day delivery back then) so drove up to Cricklewood Electronics where I spotted these capacitors. I can't remember what it was I was buying, but since I had been getting samples of all sorts exotic capacitors to audition from people like Leclanché (sadly now defunct, for capcitors at least) decided to grab a handful of them. They came out very well. It's astonishing how well one can remember impressions, even from 30 years ago, but my only complaint was they were just missing a tiny bit of sparkle. Otherwise they were right at the top. Ironically, with my more mature priorities today, which don't strive for tinsely detail as much easy detail with some body to it, they would have come out on top. In fact, some of them ended up in the tweeter section of a lovely system I used at home for non-critical, non-stressful, listening for about 20 years. Only that sort of longevity tells you that something was absolutely right, and you come away with a real love for systems like that.

But back to Meyer. I'm not as impressed with their idea of what's original as they are. And their 138dB figure brings a new dimension of meaninglessness to SNR figures. But I am encouraged that you, and they, have managed to do it. What has put me off doing it already is the prospect of getting bogged down in some aspect that requires limitless thinking. This has happened to me so many times before and it's often better to leave it to time to provide you with some insights or better understanding. Which it often does. In this case there are so many things I want to put into it that it was bound to get me caught up in the weeds. As it turns out, there are a few areas that I have sorted out and have come to properly understand - or at least have an outline or path towards the solution - by bashing away at them over time, or because they arose in some other work and needed solving there. One of which is immediately below.

Well, think this through some more. The Mic based systems I used helped a LOT with room effects.

What happens in a room at low frequencies is a minefield and there are numerous aspects to it which I can't (or won't) cover here. I have read reams and reams on this topic and I'm very unimpressed with what's out there in general, with some of it being just plain wrong. In fact, one of the fundamentals - universally agreed on, AFAICS, and which I had previously taken as obvious - has a glaring mathematical error in it. The problems weren't solved by the Eureka project (the one done in the '80s by KEF and B&O into room effects ), though it possibly could be if one recalibrated the data. That data could validate what I believe the situation to be, but it would be a mountain of work and in some ways is only doing in real life what I would do in software. (Analytical software, not in-action DSP). From what I have seen in things like people's use of Directivity Index and from loudspeaker designs that I know about, I don't think anyone has got this quite right, though I know that Andrew Jones has been looking at this for a very long time, from first realising that the bass he had been expecting when measuring behind a speaker wasn't there! A few people have got aspects of it, and I know that for certain, but I'm not sure anyone has the full picture. (Well, nor have I because I haven't yet written the software and I don't know what might come up as I do so.) Maybe Andrew does, but he usually gives away everything he knows (if you have ears to listen) and I don't think I've heard him speak excitedly on the topic. But I may have missed it if I hadn't got to understanding it fully when it was mentioned.

The subwoofer use case is a good deal simpler than that of a bass unit in a loudspeaker system. It's not just the wavelengths but also the integration with something that has different directivity. I don't mind saying here that the usual solutions to the baffle step are categorically wrong, but anyway I'm going WAY further than that. I probably would put a microphone on the outside with a sub because of the room pressurisation and tricky things like absorbtion coefficients of walls which are nearly impossible to estimate. Those get solved automatically.

But in the system case, suffice to say that I think I know what I want to put into the room - and it then has to get averaged out in different room settings. There may need to be some adjustment for things like proximity to a rear wall where there'll be obvious dip, but in general I'm hoping not. Although I have an elegant solution for a short coil in a long gap, I don't want my linearity or the limited dynamic range of a domestic loudspeaker used up by someone opening a door, and at the moment I think that output of the other speaker (though some of it will come through anyway) is probably something I don't want in my feedback loop. But I could change my mind as I haven't done this yet and there are plenty of unknowns.

Anyway, those are my reasons for wanting the mic inside rather than outside the speaker. One of the things tipping the scales is that I prefer the idea that my speaker will try to keep the cone stationary with a gust of air rather than flapping around. This could be beneficial when it comes to doing outdoor measurements - which is the only way you can do definitive measurements at these low frequencies. (And for the hostiles who perked up at an opportunity to snipe, nearfield does NOT cover this.) Meyer seem to me to be selling a flaw as a virtue here. I don't see how you can know your system is delivering the target response you wanted with a mic on the outside, especially with another speaker in the room. We ASSUME it's giving us that because it's feedback and we have a reference on the other side of the op amp, but that isn't actually the case. (For those who aren't keeping up, the thought experiment is what happens if you put a subwoofer in the room?)

The biggest issue is to get a sufficiently linear sensor. This is surprisingly non-trivial.

Yup. And I'm making it worse by putting it in a high SPL space with a rising response as you go down in frequency. Fortunately it flattens off below resonance but this is still the aspect that worries me the most. I know it can work because there are some Low Frequency measurement systems that do it this way (and it's a lovely solution to ever diminishing signal level) but assuming they run at 2.83V, that's not the same as party level in a large domestic monitor.

I probably need something with a nice high resonant frequency but I've thought of a few potential solutions like putting it in a tube, behind fine mesh to act as a resistor, a MEMS mic, accelerometers... But you probably know what the scale of the problem is, having put something near to the cone's surface already. You've probably at least prompted me to stick a microphone inside a speaker and work out what the spl is going to be.

Really? What is the impedance hump? It is the electrical representation of a mechanical resonance. If we do not short-circuit the voice coil and feed such a speaker from a current source we get a huge LF response hump and a very resonant system.

Yeah, yeah. I'm actually a little closer to your position than I was, not least because I remembered that I had done a fully equalised bass section (at least on one of the bass units) and at the time I viewed this as generally good for the amp as it brought down the total impedance hump. And, in spite of it being less than half of two drivers in parallel, it never caused a problem to any amp.

I'll come to rest of this a bit later as I'm running a bit out of steam. But also because I want to reconsider what those opposing currents do across the output transistors. Is the impedance hump giving the amp a bit of a let-off somewhat near the peak of the spectral content, which tended to be my characterisation of it before we started looking at this and before you neatly pointed out what was going on (which is a much better encapsulation than my mental image of this kinda all happening in the voice coil but not really involving the amp, except inasmuch as it now has to look at a reactive load) or is the amp just as well off delivering the current it would otherwise have had to? Or is there an optimum? It would be nice to be kind to the amp.

There are a few things in this area (which I have left below) that I never disputed, but I still don't think dissipating it near the coil is a great idea and I would rather use the surround as that helps with the mechanical impedance mismatch between the cone and the surround. I don't think spiders usually dissipate much, else we wouldn't have QMs of 5 or 7, but they too can have an impedance mismatch messing up what would otherwise be a perfectly pistonic region in any driver, so the more the better.

As for ICE dominance, the reason is portable energy density, which hasn't been overcome and will hopefully bring about their failure, along with the rest of the climate horseshit.

How is that mitigated using a low output impedance Amplifier? By injecting a current that opposes this resonance. These blind currents are out of phase with the voltage but are very real. Except in this case the dissipated power is "destructive", that is, it "destroys" some of the

Let us for a moment ignore the shorted turn damping and apply some strictly mechanical damping solution, be it the grease-pots of the original 1930's eckmiller coaxial, the Goodmans ARU or the Dynaudio "Variovent".

What is the result of the correct tuning of this element? The impedance "hump" flattens to the desired system Q. Does that make the speakers blow up?
Damping must be supplied somewhere. Doing so requires energy to be absorbed.

How this is done varies, but ultimately the same energy is required to be input into the system to get a given acoustic output and for an equally damped system it will be equal. Preservation of energy et al.



As remarked, we had drivers with such voice coil formers existed in the 80's and worked very well in practical terms. And no, they did not overheat or blow up amplifiers. So this is not a problem.

Don't forget, the voice coil is wound around a solid mild steel (or similar) core and surrounded by a large ring of solid mild steel etc. That are already two shorted turns and are one of the reasons for the poor electrical efficiency of moving coil (or moving magnet) speakers, something that could be corrected for example using more efficient motor systems (electrostatic which we now get as MEMS, piezoelectric etc.).



Well, yes and no. You want, in my view a Qm that is equal to Qt with linear damping of the spring/mass resonant system.

Typically spiders and surrounds the produce low Qm are not very linear. This does have some potential practical benefits, in some specific situations.



Yup. Then again, in a lot of ways having an internal combustion engine drive a car is a way worse solution than electrical drive, for a wide range of reasons (reliability, longevity, efficiency).

Yet they dominated the whole sector for over 100 Years, for reasons outside of the scope of the direct technical solution.



If the principle is implemented correctly, it will allow relatively commodity grade systems to be materially improved. That applies to the kind of lower end audio systems all over the range, be it portable sound system, monitors or home audio.

No doubt high end systems could also be improved, but less so, or similar performance could be attained for lower budgets.

Thor

 

[thor.zmt]

Thor, you mentioned the Dyneaudio Variovent. I used those "on spec" once and never was able to do sufficient testing to arrive at a final opinion about their usefulness. Do you know why they didn't remain popular?

The Variovent aka Aperiodic damping as a system improves transient (and frequency response), but in the process reduces the feeling of "pressure".

I find that many people do not judge HiFi (or Studio Monitoring) by how realistic it sounds, how close to listening to real music they take you.

To quote the late and great J. Gordon Holt from 1962:

"High fidelity may be a science, but it isn't an exact science. There are enough things about it that aren't understood to leave room for a goodly amount of educated opinion.

.....

This raises the question of whether high-fidelity can, or should be, better than the real thing.

.....

Sound recording may eventually become a creative art in its own right, producing musical sounds that bear no relation to any natural sounds. Indeed, some branches of it—pops and so-called electronic music—are already well on their way in that direction. This is not high fidelity, though, and there's no sense pretending that it is."

Why Hi-Fi Experts Disagree

My experience was that I preferred the sound of the Variovent-designed cabinet when I sealed it entirely.

I suspect it was more true to the source.

Thor

 

[thor.zmt]

Before I start my reply I wanted to ask you whether Ansar are still in business? Their website appears to date from 2019 and is not that functional. In about 1992, which I see must have been very soon after they started, I needed some parts quickly (we didn't have standard next-day delivery back then) so drove up to Cricklewood Electronics where I spotted these capacitors.

Ansar are pretty old. Much older than '92, they used to make mil spec capacitors. They used to be very accessible and flexible, but by the '10's had become hard to deal with. I have no idea what happened after that.

I used to go to Cricklewood Electronics a lot, often followed by a Kashmiri Curry at the Pink Rupee nearby.

What happens in a room at low frequencies is a minefield and there are numerous aspects to it which I can't (or won't) cover here.

Well, it's MOSTLY very simple, at really low and low frequencies.

At really low frequencies we have a leaky sealed box. Pressure rises according to the enclosed volume but is limited by air leakage.

Above that we have the "modal region", where room modes are dominant. Here it is also very simple. At each of the walls there is always a pressure maximum. Between walls there is least one or several pressure-minima where in turn air velocity is at maximum.

By controlling the pressure near walls microphone MFB speakers are maximally coupled to room modes and by controlling pressure in the maximae position they are effectively cancelled.

The experience of hearing really low bass at high SPL with low distortion while room modes are largely suppressed is extremely unusual. It sounds a lot more like a big classical auditorium, more realistic, but it is NOT impressive at first listen.

In fact the bass, while not anemic does not stick out all. It is all there, but not obviously so, but naturally so.

From what I have seen in things like people's use of Directivity Index and from loudspeaker designs that I know about, I don't think anyone has got this quite right

Try Joachim Kiesler's design, the MEG RL-901. Not sure if KMR in London still stocks them, if they do and you happen to have time on your hands while visiting London, drop in.

musikelectronic geithain gmbh - RL 901K

but I'm not sure anyone has the full picture.

The problem is that Audio mixes not just electronics and electromechanics but adds electroacoustics, common acoustics, musicology, philosophy and physiology with a dash of interior design.

There are an incredible number of dimensions.
And then there is preference and I have found that many enthusiasts prefer a "sound" that is not, strictly accurate, but rather impressive (and almost compelling engagement with the music).

So do we pursue accuracy? Engagement and emotional engagement? Something else?

The subwoofer use case is a good deal simpler than that of a bass unit in a loudspeaker system.

Subwoofers are especially complex, if we go past "airpump" types, where it is simple to state that that two 12" Subwoofers are a better choice than one 15" subwoofer and that's about the maximum I will tolerate in my "den" area, though four pieces would be better (use "SWARM" principle).

The Swarm Subwoofer System

It's not just the wavelengths but also the integration with something that has different directivity.

It doesn't NEED to have a different directivity. Cardioid Sub's can be give directivity that matches the lower mids and the whole speaker can be made with flat on Axis response in room (or Harman curve, I ended up with a similar target) and smooth falloff with moving off angle in the bass to midrange and additional falloff at higher frequencies (that is constant directivity for bass and midrange and narrowing directivity towards higher frequencies.

An example of such a speaker may use a monopole stacked with a dipole for (Sub)Bass and use a relatively wideband Mid/High speakers (Manger Schallwander) with a rear "acoustic sump" that makes for cardoid lower mid directivity.

The Mid-Hi speaker operates "wideband" with only a simple (passive) high-pass and is critically damped at resonance by the rear acoustic device, to be driven by the customers Amplifier.

The (sub)woofer crosses over using a transient compensated 3rd order crossover that matches the Mid-Hi and is equalised to to operate the system as cardioid across the rooms modal range and switches to an equalised sealed sub (the dipole part is cut off) for the region below the modal region.

Such a speaker sounds very unusual, I would argue very realistic and with minimal interaction with the listening room, it tends towards "open window" into the performance.

This is not what most listeners are either used to from speakers or expect from speakers, so initial reactions are commonly negative. If the listener can overcome the initial negative reaction and get used to the Speaker system, it becomes very obvious just how many colourations and distortions from speakers we have gotten used to and even expect.

I find it is not conductive to sales to make realistic or natural sounding speakers, MOST people dislike them.

For commercial use the application of multiple sound profiles targeting different demographics who may select them according to taste, including "Normal people", "Bass head", "Audiophile", "Neutral to a fault" tend Pullman Saint Paul, be all things to all people.

I don't mind saying here that the usual solutions to the baffle step are categorically wrong, but anyway I'm going WAY further than that.

What baffle step?

It is a "Test Equipment Ghost" caused by either using an anechoic chamber or pseudo anechoic measurements that assume the energy that bends around the enclosure is "lost" which is only true in open space, not in more or less enclosed areas.

I probably would put a microphone on the outside with a sub because of the room pressurisation and tricky things like absorbtion coefficients of walls which are nearly impossible to estimate. Those get solved automatically.

Bingo. Move forward six spaces and collect 500 Brownie points!

Although I have an elegant solution for a short coil in a long gap, I don't want my linearity or the limited dynamic range of a domestic loudspeaker used up by someone opening a door, and at the moment I think that output of the other speaker (though some of it will come through anyway) is probably something I don't want in my feedback loop. But I could change my mind as I haven't done this yet and there are plenty of unknowns.

Think it through in full detail.

Anyway, those are my reasons for wanting the mic inside rather than outside the speaker.

Better to embed an acceleration sensor in the driver. It is the way it's commonly done.

One of the things tipping the scales is that I prefer the idea that my speaker will try to keep the cone stationary with a gust of air rather than flapping around.

This is an issue. In a PI regulator as such a MFB subwoofer is, you simply limit the loop gain at very low frequencies, but there is always a tendency for cones to flutter when pressure changes (door opening) happen.

It's inherent, as technically this pressure change is "distortion" and will be attempted to be corrected.

This could be beneficial when it comes to doing outdoor measurements

Microphone feedback on the box outside is explicitly for use in enclosed rooms and with good coupling too room modes and room gain (e.g. put speakers into corners where the Missus/E.R. Indoors wants them).

With the correct Mid/Hi system the result is excellent, as even the main speakers problems ("baffle step" directivity, non-flat LF response) are corrected by the "subwoofer".

Meyer seem to me to be selling a flaw as a virtue here.

Really?

I don't see how you can know your system is delivering the target response you wanted with a mic on the outside, especially with another speaker in the room.

It will do so AT THE MICROPHONE POSITION (which is true of anything that measures response). One must then create a room placement that ensures that the acoustic result in room closely resembles this, which, up to a few 100Hz is fairly trivial.

But yes, it needs to be considered.

I probably need something with a nice high resonant frequency but I've thought of a few potential solutions like putting it in a tube, behind fine mesh to act as a resistor, a MEMS mic, accelerometers... But you probably know what the scale of the problem is, having put something near to the cone's surface already.

It is huge.

Normal Mic's with J-Fet overload in a most nasty way. The result depends on how good your protection system is when the mic looses it. So you first need a mic that can deal with 145dB+ SPL's in the bass with low distortion and has a predictable LF response).

While closing the loop may seem a challenge, this is the 21st century and modern mathematical simulation tools can easily handle creating a stable loop.

If using a sealed enclosure, having it really SEALED and not leaking air was critical to loop stability...

Mems accelerometers can handle this better, but the output is not directly proportional to sound pressure the way a microphone is. Extra processing is needed.

Most Accelerometers these days are digital output, so we end up with a fully digital system AD on analogue speaker level and line level input, DSP chip to which we feed the Accelerometer Delta sigma output as feedback signal and all else is software.

Definitely the future. Also works for mic feedback especially with (say) a ring of high SPL Mems Mic's (say 8 PC's) around the driver circumference, compensated for acoustic delay and set up as nearfield array that rejects far field sounds.

If we had more bright young things in STEM and audio (as opposed to onlyfans, tiktok, YouTube and in economics majors we'd already have such systems.

Or if large western firms would still Invest in actual design, instead of buying the cheapest OEM/ODM solution from china they can find and then charge like light brigade...

It's not just Audio BTW, it's every except defense.

Offshoring combined with chinese business ethics and sense (or rather the crass lack thereof) has created a perfect storm of cheap toxic garbage. End of rant.

is the amp just as well off delivering the current it would otherwise have had to? Or is there an optimum? It would be nice to be kind to the amp.

It all depends how the Amp is designed.

For a Bass system it is IMPORTANT to remember, that T/S parameters are SMALL SIGNAL lumped parameters and not valid at high levels. So if we compensate the resonance impedance through parallel conjugates (which at first blush MAY seem a good idea, especially for "current drive"), we need to do so at high power levels.

I know one specific speaker with LF Impedance compensation that used to make a rather powerful Amplifier I had designed clip at fairly low levels. After removing the impedance compensation it was all well (and sounded better even at low levels). The first thing we found on analysis was that perhaps through parameter drift during production or perhaps run in of drivers or whatever the impedance compensation was mistuned to start with, it got worse as levels rise so that dynamic impedance at high levels dropped very low.

This kind of problem of course doesn't happen with mechanical means.

There are a few things in this area (which I have left below) that I never disputed, but I still don't think dissipating it near the coil is a great idea and I would rather use the surround as that helps with the mechanical impedance mismatch between the cone and the surround. I don't think spiders usually dissipate much, else we wouldn't have QMs of 5 or 7, but they too can have an impedance mismatch messing up what would otherwise be a perfectly pistonic region in any driver, so the more the better.

Spiders can be non-linear. That is, their spring action gets stronger beyond a certain excursion. On first blush it would seem a terrible idea to introduce such a nonlinear item in a speaker, right?

Design correctly and the spider will grow more resistive to movement of the cone at around the excursion where the voice coil starts to leave the air gap. The force that tries move the voice coil reduces and force opposing movement increases.

We have in effect arrested the cone for low bass, it stops moving (well, it never could do much low bass in the first place, only at low SPL), but higher frequencies will still work fine.

Of course, the bass will be distorted, but because this is mechanical "soft clipping" and usually in a speaker driver causes high levels of H2, the mid bass output and up remains clean.

This is a little similar to using a Alu Voice coil former that acts as short circuit, but is SPL dependent and allows very low Bass to reproduced with modest SPL.

The transient high LF distortion with LF peaks naturally reduces the peak LF SPL and adds harmonics, which has interesting results with music...

Missing fundamental - Wikipedia

I found that using this intentionally in driver design allows a small (2-Way, no sub) speaker, that subjectively plays both loud and low and remains clear. In demo's most often people kept looking for subwoofers that were not there.

Naturally, this approach has limits and you will not make a 2 X 2" Bluetooth Speaker go as loud amd low subjectively as a pair of 15" 2-Way PA Speakers, but within the limits it is useful, ESPECIALLY as it is essentially cost neutral and can use of the shelf parts (intentionally non-linear spiders and rubber surrounds exist) in reasonable production quantities.

Appy Polly Logie for another super long Meta of Meta post, but as I remarked, audio, contrarywise to common perception, is arguably incredibly complex.

Mind you, give me big speakers and a big class A amp and a decent DAC to stream to and for purely listening for enjoyment I'll be happy enough.

Thor