What a mess this all turned into. Forget about sound acoustic engineering fundamentals, basic physics didn‘t even get a fair shake.
I’m not surprised they sound bad……the rear wave slapping back on the cone from such a shallow enclosure must make for a nice plate reverb effect…..and goes nicely with that horrendous cone breakup of nearly every Mark Audio driver.
How you folks let a journeyman member build a 10cm deep enclosure for a dynamic driver is beyond me…….another example of why forums often are more harmful than helpful.
I’m not surprised they sound bad……the rear wave slapping back on the cone from such a shallow enclosure must make for a nice plate reverb effect…..and goes nicely with that horrendous cone breakup of nearly every Mark Audio driver.
How you folks let a journeyman member build a 10cm deep enclosure for a dynamic driver is beyond me…….another example of why forums often are more harmful than helpful.
That's unlikely to be a problem -- the backwave simply expands sideways.
OTOH, plugging a full-range driver directly into a class-AB receiver with no passive XO parts to reduce the damping factor will probably sound harsh. Many people jump straight to 2 or 3-ways without first figuring out the nature of the problem (I know I did).
The inductance at the mid-to-high frequency end of the speaker's range tends to get modulated by low frequencies. The physical displacement of the coil and its changing position relative to the magnet poles (which are imperfectly saturated, so the impedance and current changes dynamically: LF signals end up rapidly modulating the gain of HF signals) is just one mechanism that adds IMD. There are also others.
The key, which seems pretty reliable at least for the top half of a speaker's range, is to add some other impedance in series between the speaker and the power amplifier. A resistor (typical values may be 10 or 22 ohm, etc.) is probably the simplest option. Another one is to roll-off the high frequencies a little (or a lot) with an air-cored coil.
Even if the frequency response is not flat after doing those things, it doesn't matter: it can be tuned with active EQ. One might ask, why not just tune the response with active EQ in the first place? Because the shape of the frequency response is not what's causing the harshness. Distortion can cause a harsh sounding tone without altering the EQ much, if at all.
OTOH, plugging a full-range driver directly into a class-AB receiver with no passive XO parts to reduce the damping factor will probably sound harsh. Many people jump straight to 2 or 3-ways without first figuring out the nature of the problem (I know I did).
The inductance at the mid-to-high frequency end of the speaker's range tends to get modulated by low frequencies. The physical displacement of the coil and its changing position relative to the magnet poles (which are imperfectly saturated, so the impedance and current changes dynamically: LF signals end up rapidly modulating the gain of HF signals) is just one mechanism that adds IMD. There are also others.
The key, which seems pretty reliable at least for the top half of a speaker's range, is to add some other impedance in series between the speaker and the power amplifier. A resistor (typical values may be 10 or 22 ohm, etc.) is probably the simplest option. Another one is to roll-off the high frequencies a little (or a lot) with an air-cored coil.
Even if the frequency response is not flat after doing those things, it doesn't matter: it can be tuned with active EQ. One might ask, why not just tune the response with active EQ in the first place? Because the shape of the frequency response is not what's causing the harshness. Distortion can cause a harsh sounding tone without altering the EQ much, if at all.
Thanks for the comment. Could you clarify what exactly you have in mind. Which design principles is being broken here? Would help to improve the design or maybe revisit my expectations/design goals.
Actually they sound fine 🙂. The lack of bass is by design (they will be crossed over to a sub below ~100Hz), and the high highs is tamable (playing with AVR EQ brought those down).
I was actually afraid of that as well, hence why I build those only for experimentation out of spare wood I had (figured worst case I was losing an afternoon and less than 100e).
IIUC, the rear wave slaps back regardless of the depth of the enclosure right? I thought bracing/absorbing inside the cabinet would help mitigate that.
Wrt to the cone break-up, I have not been offended by it. Maybe if I hear it next to another speaker which is more linear I would pick it up, but as is, it is not distracting.
Could you clarify that point? Why adding adding that resistor would reduce IMD? (sorry my background in electronics is near zero)
Edit: my understanding was that IMD was mostly caused by the mechanical behavior of the driver, whereas adding a resistor would mostly affect the incoming signal, not the mechanical behavior of the driver.
Last edited:
Note the right hand side of the image. +/-7mm may not be accurate, but the point is the impedance bounces up and down, based on low frequency vibrations, which could make a mess at the ~300Hz to 20kHz range in the example.
If you add 22 ohm, at 1kHz the variation would be between 30 to 36 ohm, or 17%, which is a much lower percentage than 8 to 14 ohm, or 43%.
If you add 22 ohm, at 1kHz the variation would be between 30 to 36 ohm, or 17%, which is a much lower percentage than 8 to 14 ohm, or 43%.
I guess every midrange and tweeter with a rear cup is awful? And smaller cone drivers having about 90% of the cone's rear output area obstructed by their own frame and magnet geometry is irrelevant?
Beyond that, if you push the internal resonance/standing wave mode up high enough in frequency you have a much better chance of actually absorbing it nearly completely with a practical amount of acoustical foam or other absorber in the enclosure.
Until you've built something like this it's hard to predict the actual sound. One can pontificate from a distance about engineering purity, but the reality of what someone wants to build takes precedence in the real world. If a person wants to build a shallow full-range enclosure, I'm going to try to help them while also setting expectations and encouraging experimentation before committing to a big build/large cash outlay.
I've built wideband enclosures with much, MUCH tighter side/rear walls than this and never heard/measured anything untoward. I was worried about it at first from a purist standpoint, but it was a non-issue in practice.
Last edited:
On th esubject of shallow enclosures, it can make for an issue. `
We have done 4 shallow enclosures for 4ish” drivers. In PAWO, the FE127e decidedly benefitted from increasing the depth by about 25mm with a supreBaffle. The Ellipsa crosses the driver at 200 Hz, but it works really well.
Mileva, our response of a properly designed box in the shape of PAWO, did not suffer.
The jury is out yet on the Hawking Memorial Obelisk.
dave
We have done 4 shallow enclosures for 4ish” drivers. In PAWO, the FE127e decidedly benefitted from increasing the depth by about 25mm with a supreBaffle. The Ellipsa crosses the driver at 200 Hz, but it works really well.
Mileva, our response of a properly designed box in the shape of PAWO, did not suffer.
The jury is out yet on the Hawking Memorial Obelisk.
dave
This is an interesting point. For a fixed volume, I would think that having one dimension small will force to make another dimension relatively large. This would lead to move one cabinet mode up, but another down. In the case of my particular build, the largest internal dimension is 46cm, so a mode at 762 Hz. I would still think that enough internal absorber would make that a non issue (using this calculator).
However, can that be an issue for very large cabinet? I was thinking of a ~ 1m x .5m x .25m box for the sub, for which the cabinet would have a mode at 343 Hz, a frequency not well absorbed by common absorbers (or would require to almost completely fill the inside of the cabinet).
Edit: actually I guess bracing should partially solve that issue right ?
@planet10 could you detail what kind of issues were caused by your shallow speakers experiments? (wondering if I can make things better).
Sorry not a native english speaker. What is cupiness? FR goes down in midrange?
And what do you think causes that when the cabinet is shallow?
And what do you think causes that when the cabinet is shallow?
Cup your hands (mske s megaphone) overyour mouth snd speak.
It is early reflection coming off the back of the box.
dave
It is early reflection coming off the back of the box.
dave
As Dave points out, the acoustic wave output from the rear of the cone reflects back through the cone itself which is partially acoustically transparent.…..Mattstat attacked the premise…..but it’s a SERIOUS problem with Longer wavelengths of midbass and bass. Yes, it works just fine with closed back and chambered drivers of upper midrange and treble frequencies. These wavelengths are short in relation to the distance and aren’t a problem…….but with a metal cone fullrange driver?…..well……I’m just plain sorry you came to specialty DIY forum and received such poor guidance.
The ONLY a reasons for a speaker box is to either develop low frequency extension through resonance or suspension OR to absorb out of phase rear wave output…..best box is no box at all if positioning supports it……your case does not so the only real alternative is an in wall application.
Ok I see thanks. I guess this could amplify (high-freq) or dampen (low freq) output sound based on if the backwave is in-phase or out-of-phase right?
Since you mentioned low-freq, I assume the out-of-phase case is the biggest issue right?
I actually did notice something like that after the driver burn-in period. I left the speakers play for ~50 hours, without a cabinet. I tried to put them on a table top (so sitting on the magnet) and the sound was horrible, like an old radio post. When I lifted the speaker off the table (about 20cm), it was sounding just fine. This made me anxious about the shallow speaker at first, but I still gave it a try, thinking that the backwave might just bounce around inside the cabinet instead of coming right back to the front. And it seems it turned out fine (I don't hear the same old radio post effect from the speaker).
Here is what I thought was happening (let me know if I am wrong, I am just trying to understand 🙂). If we assume the back wave goes in all direction inside the cabinet, I think there would be two things happening:
- when the first reflection is directly behind the cone and comes back to it. As Matt pointed out, most cone area (if not all) is backed by the spider/magnet anyway, so I assume that the driver-backpanel distance did not matter for that first reflection case (let me know if that is wrong).
- when the first reflection hits the backpanel, bounces X number of times and eventually gets back to the cone. As you mentioned this should be an issue with low frequencies. However I would assume this is independent of depth and more linked to volume since low frequencies have wavelength much bigger than cabinet dimension (e.g. 100Hz is 3.43 meters). In other words, if the frequencies are low we are below the Schroeder Frequency of the cabinet (which depend on volume) and any mode below that frequency would be an issue. My understanding was that is mitigated by making dimensions not integer multiple of one another (to avoid avoid amplification of modes in all directions), and using enough dampening material so that the cabinet resonance dies out faster.
Would'nt it have the same issue, since the drywall will be at <10cm from the concrete wall? Also, would the in-wall application rely on a cabinet inside the wall, or would it use the whole space behind the wall as a cabinet?
What does the cross-section look like? With 10cm I guess the magnet would be almost up against the back wall, but with a small gap where the sound waves wrap around the magnet. If the pole piece is not vented, you could try this:
That way the wave expansion is a lot smoother. You would have to figure out the details like whether to use some kind of glue or caulk, or make it a dry connection.
That way the wave expansion is a lot smoother. You would have to figure out the details like whether to use some kind of glue or caulk, or make it a dry connection.
The cross section look like page 3 of Dave's design. The internal depth is 5.5cm and the magnet is not touching the backpanel but is in contact with the brace.
Regarding first reflection, I am still not clear why it is more important than Nth order reflections for lower frequencies.
In particular:
Am I missing something ?
Regarding first reflection, I am still not clear why it is more important than Nth order reflections for lower frequencies.
In particular:
- those lower frequencies will be standing waves inside the cabinet anyway, so at least the few first Nth reflections (and all Nth reflections for modes) should be out of phase.
- at those frequencies (e.g. 100Hz) I would expect the few fist order reflection to be barely damped given how little room there is for absorption (the cabinet is almost filled with denim insulation).
Am I missing something ?
When you have such close proximities, (described as proximity effect in microphones…same basic principles apply since a diaphragm is excited) you don’t really develop ‘standing’ waves which would produce resonant modes of significant amplitude but instead, a near infinite number of partial modes that become an overall time smear……cumulative so to speak. It’s why we measure CSD. If you were to take measurements at say 10cm cabinet depth and 20cm respectively, you’d see a broad increase in CSD across almost the entire passband. Add in a metal cone to the excitement and we’ll…..you get the idea.
With an in wall approach, you could mitigate a few different issues by coming as close to the ideal infinite baffle system….inside the cavity for all intents and purposes you‘ll only be dealing with the earliest first reflection from the back wall…..at 30cm horizontal and vertical with damping material the smear is no longer a wide band audible component and assuming sound construction, little to no destructive resonances with a fixed peak amplitude.
There‘s two ideal scenarios…..absolutely no baffle and an infinite baffle……in which case your ear/brain mechanism can’t differentiate so it reverts to searching for a point source…..back to the ideal transducer. In mix/mastering studios, you‘ll often find the baffle wall setup when the listening position and gear create a significant problem in the window for free space enclosure………too much short time based reflected energy to listen through…creates confusion which leads to early listening fatigue.
I‘ve always considered the advantages of DIY as a method to design and build something custom for an application and an environment…..something that commercial products can rarely do on both fronts. Sure, you can trade off the value of your time and effort to mitigate costs (if you have it to spare….in reality most folks could be more productive doing something else) but if you transform that effort into maximizing performance in YOUR environment?…..now there’s a value based approach with measurable performance gains.……for say a $400 investment for 2 decent quality 2 way in walls and your time to install, well…….you fullrange Mark Audio driver in those boxes would be eclipsed on just about every front. I get it…..you didn’t learn anything in the process or get the feeling of accomplishment by looking at your work on a 3D plane……but that’s a different set of goals…….more artistic development than performance based. Now, after doing this stuff for a while, i always ask the ’could i’ and ‘should I’ questions first. When my kids were young, I’d get so frustrated at sports practice where more than half the time the kids were exercising instead of practicing fundamentals of the game……..as some say in America…..ain’t nobody got time for that! YMMV…….I’m an older curmudgeon with skewed values! Lol
With an in wall approach, you could mitigate a few different issues by coming as close to the ideal infinite baffle system….inside the cavity for all intents and purposes you‘ll only be dealing with the earliest first reflection from the back wall…..at 30cm horizontal and vertical with damping material the smear is no longer a wide band audible component and assuming sound construction, little to no destructive resonances with a fixed peak amplitude.
There‘s two ideal scenarios…..absolutely no baffle and an infinite baffle……in which case your ear/brain mechanism can’t differentiate so it reverts to searching for a point source…..back to the ideal transducer. In mix/mastering studios, you‘ll often find the baffle wall setup when the listening position and gear create a significant problem in the window for free space enclosure………too much short time based reflected energy to listen through…creates confusion which leads to early listening fatigue.
I‘ve always considered the advantages of DIY as a method to design and build something custom for an application and an environment…..something that commercial products can rarely do on both fronts. Sure, you can trade off the value of your time and effort to mitigate costs (if you have it to spare….in reality most folks could be more productive doing something else) but if you transform that effort into maximizing performance in YOUR environment?…..now there’s a value based approach with measurable performance gains.……for say a $400 investment for 2 decent quality 2 way in walls and your time to install, well…….you fullrange Mark Audio driver in those boxes would be eclipsed on just about every front. I get it…..you didn’t learn anything in the process or get the feeling of accomplishment by looking at your work on a 3D plane……but that’s a different set of goals…….more artistic development than performance based. Now, after doing this stuff for a while, i always ask the ’could i’ and ‘should I’ questions first. When my kids were young, I’d get so frustrated at sports practice where more than half the time the kids were exercising instead of practicing fundamentals of the game……..as some say in America…..ain’t nobody got time for that! YMMV…….I’m an older curmudgeon with skewed values! Lol
Last edited:
CSD is basically a collection of short frequency response measurements, so you get both frequency and time on one graph, but with limited resolution in both. If you want better frequency resolution, the FFT windows need to be longer, but then the information on 'when' a certain sound occurs gets blurred because each window spans a longer overall time. (A physical limitation that may go some way to explaining why some people hear things differently.)
In any case, if you look at TLs and back loaded horns etc., the back wave gets funnelled along a tube of some sort, which may gradually expand or contract, and the fewer sharp kinks in it the better (although sometimes a well placed reflecting surface may cancel out something else).
In this case we've got something resembling a 'stub' behind the magnet. A resonant hole that doesn't affect low frequencies, and right in the middle behind the magnet where the brace is, would be the focal point, acting as a solid wall where reflections occur.
Hence my earlier post about putting a circular wedge in there. It would probably have to be cut into pieces if the only access is through the speaker hole.
In any case, if you look at TLs and back loaded horns etc., the back wave gets funnelled along a tube of some sort, which may gradually expand or contract, and the fewer sharp kinks in it the better (although sometimes a well placed reflecting surface may cancel out something else).
In this case we've got something resembling a 'stub' behind the magnet. A resonant hole that doesn't affect low frequencies, and right in the middle behind the magnet where the brace is, would be the focal point, acting as a solid wall where reflections occur.
Hence my earlier post about putting a circular wedge in there. It would probably have to be cut into pieces if the only access is through the speaker hole.
Last edited:
Ok let's take a step back and be analytical 🙂
It seems we are assuming that there is a problem with the backwave. I would like to check that first before talking about solutions (if no problem, who needs a solution? 🙂).
Dave suggested I would hear it (cupping). Given that I heard something along those lines when the driver was baffleless/enclosureless, and I am not hearing it with the enclosure, I guess at least it is not severe.
Now maybe I have a backwave issue to some limited extent (and I am not trained to pick it up). How would I check that (other than subjective hearing)? If I was to buy a measurement mic (e.g. UMIK-1?) and measure the CSD would it show up?
It seems we are assuming that there is a problem with the backwave. I would like to check that first before talking about solutions (if no problem, who needs a solution? 🙂).
Dave suggested I would hear it (cupping). Given that I heard something along those lines when the driver was baffleless/enclosureless, and I am not hearing it with the enclosure, I guess at least it is not severe.
Now maybe I have a backwave issue to some limited extent (and I am not trained to pick it up). How would I check that (other than subjective hearing)? If I was to buy a measurement mic (e.g. UMIK-1?) and measure the CSD would it show up?
Last edited:
A signal generator could help narrow down the frequency band even without a mic. If your system is/can easily be hooked up to a computer (or at the minimum a smartphone if you can find some decent apps), one idea could be to use Audacity to generate chirps. (A music keyboard would also work).
Home in on the most audible bumps and warbles that shouldn't be there, and find out what the frequencies are. Then convert those to wavelengths and half-wavelengths and take the detective work from there. My 2c
Home in on the most audible bumps and warbles that shouldn't be there, and find out what the frequencies are. Then convert those to wavelengths and half-wavelengths and take the detective work from there. My 2c
I'm not assuming anything, the backwave IS creating time based delay.....whether or not you can hear it or if it matters is up to you. The critical analysis has been done countless times over decades and the resources are available to you for confirmation. Sounds like what your asking for is subjective confirmation?......you already received that earlier and bought some Mark Audio fullrange drivers based on that,
Not sure how we wound up hyper focused on this when your predominant issue is the terrible 10khz breakup that makes you want to turn it down that you mentioned earlier. I would suggest a measurement mic and freeware so you can quantify it with your eye and then notch it with a passive component filter.
Please excuse me if i'm coming off as snarky but i'd rather not waste my time trying to assist you if in the end, subjective group think drives your efforts.....not my kinda thing. Fullrange drivers have their place and if you accept the laws of physics, then one could support the notion that their best use is near on axis, nearfield listening at low to moderate volume and that's it..........anything else is purely some misguided purist notion that crossovers are inherently evil........typically the same folks that believe a vibrating stylus tracking a piece of cut vinyl on a spinning platter is audio nirvana.