I am returning to the world of DIY speakers after almost 30 years away! Not much seems to have changed - closed box, ported box, crossovers, etc. - but one new development is open baffle designs which simply are not mentioned in my DIY speakers books from that time. Something else has changed - cheap drivers have got even cheaper!
While I have managed with commercial small tuned boxes + subwoofer for many years, my interest this time is to provide music at a Summer garden party a long way from any electric supply. I looked into super-efficient design amp & speakers to run off batteries (& perhaps solar power) and it appears that the Chinese Tripath amps selling on eBay for around £15 are just the ticket, getting almost 90% efficiency into 8 ohms (nearer 80% into 4 ohms). Of course, these 'class D' amps did not exist either in my youth. I settled on a Lepai Ta2020-based unit because 1) it has (bypass-able) bass & treble controls which may be of benefit 2) the 2011 revised version has addressed much of the criticism of the previous model and 3) everyone says they sound great.
I chose an open baffle design also because everyone raves about the openness, transparency and that 'in the room with you' realism, because I have not yet heard any and because they are very easy to make. The figure of 8 polar pattern also appealed for their directivity in an outside scenario. I settled on a line array because they break the inverse square law, i.e. they exhibit cylindrical rather than spherical directivity so the SPL drops off in linearly with distance rather than the square of the distance - handy when people may be scattered over the large area of a garden rather than sitting in the ideal listening position.
For drivers, I found these Chinese units listed on many UK shopping sites. I have trawled and trawled and could not find any reviews or tests of them or designs using them. They are listed as either Electrovision or Soundlab 204mm (8 inch) full-range speakers and seemed like good candidates. Here is their full spec.
Colour: Black
Diameter: 205mm
Dimensions: 86 x 205 x 205 mm
Frequency Response: 40Hz-18KHz
Impedance: 8Ω
Magnet Weight: 0.34kg
Peak Music Power: 20W
Power RMS: 15
Resonant Frequency: 40Hz
Sensitivity: 96dB
Weight: 0.997 kg
(from SoundLab 204mm (8") Round Speaker 15W ONLY 6.36)
Now 15W RMS may not seem that much but just look at that 96dB sensitivity. I bought ten units for £60.76 (under $10 each) with free delivery from DFB Sound & Light Warehouse. My project is to put four units, configured two in series x two parallel to maintain the ideal 8 ohm load for the Tripath amp in a vertical line array. This should yield a whopping 102dB SPL @ 1W, 1m! This means I can keep the output of the amp well down to avoid distortion, have loads of headroom and run the system all day off 10 AA rechargeable batteries. The line array also helps to mitigate manufacturing tolerances by averaging out the different peaks and troughs between the drivers.
Of course the great weakness of open baffle designs is the weak bass. An open baffle has to be at least the size of the half a wavelength of the lowest frequency, e.g. a minimum of 3.5m at 50Hz, or 7m at 25 Hz, to benefit from the 'boundary effect' 3dB lift. I looked into H-frames and U-tubes but in the end developed a simple design to use the boundary effect in reverse, i.e. to take it away from all but the lowest frequencies and thereby provide additional low bass support.
Above around 1KHz, corresponding to the half-wavelength equal to the width of the drivers, the dispersion should be highly directional and benefit from the 3dB boundary effect lift. This is fine as high frequencies lose their power faster over distance than low frequencies. By putting the drivers on the smallest possible baffle, i.e. the same width as the drivers, we deny the boundary effect until the difference in direct and reflected sound (from the ground) becomes small relative to the wavelength. The difference is maximal right next to speaker and lessens with distance. With a bit of simple trigonometry and a spreadsheet, the distance from the ground to the centre of the speakers should be the order of 1-2 m for a listening distance between 2m and 8m to reinforce only the bottom end of the response range. The final height will be determined by experimentation as I cannot be confident that these drivers will go smoothly down to the claimed 40Hz.
I sourced some reclaimed scaffolding timber which is 9" wide and 35mm thick for the baffles. Cleaned, stained and waxed it looks stunning. The drivers are due tomorrow, the amp is on its way from China and the 10 AA battery box arrived today. It is two weeks to the party - wish me luck!
Any thoughts? Have I missed anything?
Tim.
While I have managed with commercial small tuned boxes + subwoofer for many years, my interest this time is to provide music at a Summer garden party a long way from any electric supply. I looked into super-efficient design amp & speakers to run off batteries (& perhaps solar power) and it appears that the Chinese Tripath amps selling on eBay for around £15 are just the ticket, getting almost 90% efficiency into 8 ohms (nearer 80% into 4 ohms). Of course, these 'class D' amps did not exist either in my youth. I settled on a Lepai Ta2020-based unit because 1) it has (bypass-able) bass & treble controls which may be of benefit 2) the 2011 revised version has addressed much of the criticism of the previous model and 3) everyone says they sound great.
I chose an open baffle design also because everyone raves about the openness, transparency and that 'in the room with you' realism, because I have not yet heard any and because they are very easy to make. The figure of 8 polar pattern also appealed for their directivity in an outside scenario. I settled on a line array because they break the inverse square law, i.e. they exhibit cylindrical rather than spherical directivity so the SPL drops off in linearly with distance rather than the square of the distance - handy when people may be scattered over the large area of a garden rather than sitting in the ideal listening position.
For drivers, I found these Chinese units listed on many UK shopping sites. I have trawled and trawled and could not find any reviews or tests of them or designs using them. They are listed as either Electrovision or Soundlab 204mm (8 inch) full-range speakers and seemed like good candidates. Here is their full spec.
Colour: Black
Diameter: 205mm
Dimensions: 86 x 205 x 205 mm
Frequency Response: 40Hz-18KHz
Impedance: 8Ω
Magnet Weight: 0.34kg
Peak Music Power: 20W
Power RMS: 15
Resonant Frequency: 40Hz
Sensitivity: 96dB
Weight: 0.997 kg
(from SoundLab 204mm (8") Round Speaker 15W ONLY 6.36)
Now 15W RMS may not seem that much but just look at that 96dB sensitivity. I bought ten units for £60.76 (under $10 each) with free delivery from DFB Sound & Light Warehouse. My project is to put four units, configured two in series x two parallel to maintain the ideal 8 ohm load for the Tripath amp in a vertical line array. This should yield a whopping 102dB SPL @ 1W, 1m! This means I can keep the output of the amp well down to avoid distortion, have loads of headroom and run the system all day off 10 AA rechargeable batteries. The line array also helps to mitigate manufacturing tolerances by averaging out the different peaks and troughs between the drivers.
Of course the great weakness of open baffle designs is the weak bass. An open baffle has to be at least the size of the half a wavelength of the lowest frequency, e.g. a minimum of 3.5m at 50Hz, or 7m at 25 Hz, to benefit from the 'boundary effect' 3dB lift. I looked into H-frames and U-tubes but in the end developed a simple design to use the boundary effect in reverse, i.e. to take it away from all but the lowest frequencies and thereby provide additional low bass support.
Above around 1KHz, corresponding to the half-wavelength equal to the width of the drivers, the dispersion should be highly directional and benefit from the 3dB boundary effect lift. This is fine as high frequencies lose their power faster over distance than low frequencies. By putting the drivers on the smallest possible baffle, i.e. the same width as the drivers, we deny the boundary effect until the difference in direct and reflected sound (from the ground) becomes small relative to the wavelength. The difference is maximal right next to speaker and lessens with distance. With a bit of simple trigonometry and a spreadsheet, the distance from the ground to the centre of the speakers should be the order of 1-2 m for a listening distance between 2m and 8m to reinforce only the bottom end of the response range. The final height will be determined by experimentation as I cannot be confident that these drivers will go smoothly down to the claimed 40Hz.
I sourced some reclaimed scaffolding timber which is 9" wide and 35mm thick for the baffles. Cleaned, stained and waxed it looks stunning. The drivers are due tomorrow, the amp is on its way from China and the 10 AA battery box arrived today. It is two weeks to the party - wish me luck!
Any thoughts? Have I missed anything?
Tim.
Attachments
Last edited:
Eight inch diameter drivers--even full range ones--will have very limited high frequency capability because of comb line effects. It is associated with center to center spacing. Please read my white paper for insight to line array design. I also wonder if you really will have enough bass without a large baffle.
ArraySpeakers Forum - New Home for 'Near Field Line Array White Paper' - Jim Griffin, April 19, 2008 at 00:47:55
http://www.greatplainsaudiofest.com/Photos/CSSLineArray01.jpg
Many thanks to Jim for the reply and link to your paper. There is plenty in there for me to chew over and I am grateful that you have included the maths and graphs too.
Every loud speaker design is of course a compromise. My favourite sounding speakers have always been wide-range driver boxes (usually + sub). I don't know why that is, perhaps something to do with my brain, but all the speakers I have liked the most, upon investigation, have turned out to have wide-range drivers. I know there is a 'full-range' community but also plenty of reasoning why a single driver can never adequately reproduce the whole 10 octaves of the hi-fi spectrum. I think my brain must not like the effect of crossovers.
So I accept that the higher frequencies will be hurt by comb-filtering and not just the higher ones. However, to me, this is a price worth paying for the lack of crossovers and the dispersal patterns for outdoors where SPL could be lost very quickly. I think my brain is better able to cope with comb-filtering, something that we constantly rely upon to locate us in a physical space, than changing phase shifts which I think are the aural equivalent of sea sickness 🙂
Every loud speaker design is of course a compromise. My favourite sounding speakers have always been wide-range driver boxes (usually + sub). I don't know why that is, perhaps something to do with my brain, but all the speakers I have liked the most, upon investigation, have turned out to have wide-range drivers. I know there is a 'full-range' community but also plenty of reasoning why a single driver can never adequately reproduce the whole 10 octaves of the hi-fi spectrum. I think my brain must not like the effect of crossovers.
So I accept that the higher frequencies will be hurt by comb-filtering and not just the higher ones. However, to me, this is a price worth paying for the lack of crossovers and the dispersal patterns for outdoors where SPL could be lost very quickly. I think my brain is better able to cope with comb-filtering, something that we constantly rely upon to locate us in a physical space, than changing phase shifts which I think are the aural equivalent of sea sickness 🙂
Hi,
An error people make is open baffle is simply a roll-off in the bass, it isn't,
and OB bass outdoors will be utterly and totally pathetic. Its hard enough
outdoors as it is, and OB cannot use rear walls or corners for affect.
Outdoor design is far different to indoor design, OB's a complete no-no.
The reason for this is its baffle loss, not roll-off, roll-off changes
the excursion of the driver, loss does not, but effectively makes
the drivers size, or excursion limits far smaller with the more loss.
9.5" wide baffles ?????!!!!!! don't what planets theory your coming
from but not a chance in hell even domestically for decent bass.
You won't even get upper bass + lower mids, nevermind bass.
There is very, very big hole in your understanding of the situation.
A very poor point to be planning on building the thing from.
rgds, sreten.
You need full TS specs to know what is going on, Fs = 40Hz,
sensitivity = 96dB is all I can find. Given that and some
intuition put the line array in a box vertically offset
from the centre line, around 50 - 75 L internal.
Add vent tuning 40 - 45 Hz.
Last edited:
zero baffle ground design experiment
I didn't mean to offend anyone with my speaker design. I am just a re-enthused speaker builder from 30 years ago.
As I tried to explain, my plan is not to support the midrange or bass until the last couple of octaves. Hence zero baffle and use of the ground reflection only to boost the bottom. This means withdrawing the drivers to a height above the ground until the boundary effect only supports the lows, rather than supporting the mid range and high bass as in conventional open baffle designs. I am giving up 3dB in SPL from 1Khz downwards to around 50Hz precisely so that the bottom end gains 3dB. Everything is relative. Perceived bass response can be achieved by taking away something above.
O.K., this is now the 'zero baffle ground design' experiment. I am a little taken aback by the replies so far. I was only trying to share my ideas in the hope of refining them for mutual benefit. There is more opinion than i bargained for. As a result of the feedback, I have ordered parts for a measurement mic and will now be doing tests which I will share here.
Tim.
I didn't mean to offend anyone with my speaker design. I am just a re-enthused speaker builder from 30 years ago.
As I tried to explain, my plan is not to support the midrange or bass until the last couple of octaves. Hence zero baffle and use of the ground reflection only to boost the bottom. This means withdrawing the drivers to a height above the ground until the boundary effect only supports the lows, rather than supporting the mid range and high bass as in conventional open baffle designs. I am giving up 3dB in SPL from 1Khz downwards to around 50Hz precisely so that the bottom end gains 3dB. Everything is relative. Perceived bass response can be achieved by taking away something above.
O.K., this is now the 'zero baffle ground design' experiment. I am a little taken aback by the replies so far. I was only trying to share my ideas in the hope of refining them for mutual benefit. There is more opinion than i bargained for. As a result of the feedback, I have ordered parts for a measurement mic and will now be doing tests which I will share here.
Tim.
I believe there is large silent majority excited about your project... Keep going...
Positive encouragement
Hello Tim,
Don't let the tone of the replies negate the information in them, or your desire to achieve a positive outcome. Since you stated that you did do DIY in the past, I would assume you've dealt with the following scenario: I put all this work into this and I didn't get the results I wanted. The "enthusiasm" that you received so far looks (to me) like fellow DIY'ers trying to steer you away from that disappointment.
As someone who dived into DIY speakers with a FAST OB design, I still smile every time I listen to a concert recording and think about what I put together and the results it has given me. I love the open sound.
I have been a sound engineer for numerous outdoor events, but I've never used OB for sound reenforcement in any type of venue. I encourage you to proceed with your measuring and sharing your results.
Good luck and remember to have fun!
later,
G
Hello Tim,
Don't let the tone of the replies negate the information in them, or your desire to achieve a positive outcome. Since you stated that you did do DIY in the past, I would assume you've dealt with the following scenario: I put all this work into this and I didn't get the results I wanted. The "enthusiasm" that you received so far looks (to me) like fellow DIY'ers trying to steer you away from that disappointment.
As someone who dived into DIY speakers with a FAST OB design, I still smile every time I listen to a concert recording and think about what I put together and the results it has given me. I love the open sound.
I have been a sound engineer for numerous outdoor events, but I've never used OB for sound reenforcement in any type of venue. I encourage you to proceed with your measuring and sharing your results.
Good luck and remember to have fun!
later,
G
Many thanks for the support of gv & gvimhoof! I am so grateful and privileged just to be able to take part in such a discussion.
My Lepai amp arrived today. I have married it, with snips and a soldering iron, to the 10 x AA battery box and they look very happy together. I also commissioned the baffle maker. Here is the design. My very rough sketch is not to scale!
The drivers should arrive Tuesday.
Bestist,
Tim.
My Lepai amp arrived today. I have married it, with snips and a soldering iron, to the 10 x AA battery box and they look very happy together. I also commissioned the baffle maker. Here is the design. My very rough sketch is not to scale!
The drivers should arrive Tuesday.
Bestist,
Tim.
Attachments
Very interested in this project because I actually have a pair of these drivers - mine have a chrome coloured whizzer. Always thought it would be worth playing with them in an open baffle...
Hi, with 4 of these drivers, you'll probably have sufficient cone area to reproduce some reasonable bass.
If you find them to be lacking, a small (4ohm or less - you'll need to experiment) series resistor can, by altering the T/S parameters, give more bass on open baffles.
Be prepared to experiment with room placements and things like that.
Best of luck,
Chris
If you find them to be lacking, a small (4ohm or less - you'll need to experiment) series resistor can, by altering the T/S parameters, give more bass on open baffles.
Be prepared to experiment with room placements and things like that.
Best of luck,
Chris
Come on Sreten, tear him down without offering any ideas of making it better?
Tim,
The width of the baffle (or more correctly the distance from back of driver to front of driver) determines where the dipole roll-off starts. With a 9.5" baffle, the EQ needed to flatten the response will likely reduce efficiency to more like 80 dB than 100. Not to mention the dramatic amount of excursion.
Also to follow up on Jim's comment, with 4 8" drivers in a row, the issues in the HF is going to start really low. Well into the midrange.
With 10 drivers i can think of better ways to get the most out of them (it would end up being a 2-way and use 5 drivers/side).
You can go ahead with your scheme, be a good learning price, but i'm all too afraid that it will disapoint.
dave
If there's too little bass, you can always add wings to the sides of the baffle later. Attaching them with piano hinges would allow them to be folded for slightly easier transport and also let you play with the angle to get the best sound.
It would have been good practice to do some EDGE simulations before cutting wood 🙄
Anyway - here is what you can expect:
Both sims account for the floor/ground reflection by assuming a mirrored baffle with four added drivers. The diagrams only show the baffle loss in the bass area. You have to subtract the bass loss of the drivers from that too.
If you pool the four drivers in the upper part of the baffle, you will have a dip around 500 Hz. If you distribute the drivers along the length of the baffle, this dip will vanish, but there will be more dips above 2 kHz. Choose your poison.
Anyway - here is what you can expect:
Both sims account for the floor/ground reflection by assuming a mirrored baffle with four added drivers. The diagrams only show the baffle loss in the bass area. You have to subtract the bass loss of the drivers from that too.
If you pool the four drivers in the upper part of the baffle, you will have a dip around 500 Hz. If you distribute the drivers along the length of the baffle, this dip will vanish, but there will be more dips above 2 kHz. Choose your poison.
Attachments
As a separate thought, in order to sort out the potential HF problems...
A reasonably sturdy compression driver, crossed in at, say, 4kHz might alleviate some of the comb filtering problems.
The only way to find out is listen, though. You might find that you're happy with the "smooth" (ie, rolled off treble) sound that all 4 would have to offer.
You could try crossing the lower 2 over, so they're not reproducing HF that would interfere with the output from the upper drivers.
Chris
A reasonably sturdy compression driver, crossed in at, say, 4kHz might alleviate some of the comb filtering problems.
The only way to find out is listen, though. You might find that you're happy with the "smooth" (ie, rolled off treble) sound that all 4 would have to offer.
You could try crossing the lower 2 over, so they're not reproducing HF that would interfere with the output from the upper drivers.
Chris
That is what I am hoping for. The total area should be approaching the equivalent of a 16 inch driver.
Thanks for that. I don't understand why though. Is this specific to open baffles?
Here is the little Lepai amp powered up from the batteries. There is a reassuring click of a relay around half a second after switching on. I will mount the amp & battery pack on the rear of one of the baffles to make the speakers 'active'. The source will be an old iPod mini connected with a 1/8" lead.
The last time I made speakers was before CDs let alone iPods, class-D amps and line arrays.
I have just bought 100m 79-strand 2.5 sq mm speaker wire for £13 (from Wilkinson Plus). I am sure this was almost £1 a metre 30 years ago. With a nominal rating into 8 ohms of 3.2KW (!) it should not interfere with the transients.
Many thanks!
Attachments
Chris, Dave, Godfrey and Rudolf,
Many thanks for your input. Actually those frequency response curves don't look too bad to me! I didn't actually know that such modelling software existed so that is fascinating and thanks again. These are low-cost drivers and are very much an unknown beast. The limited parameters that the manufacturer supplies cannot really be relied upon so it will very much be trial and error. I hadn't considered moving the drivers apart but the modelling is very interesting. I have given the project my best initial shot. In the end there will almost certainly be some modifications to compensate for the actual weaknesses of the system - perceived, measured or both.
I was thinking that if the top end is lacking, I may add some piezo tweeters (this will probably provoke even more outrage!) The reason being that I could still avoid using a crossover. However, my hearing falls off quickly over around 14Khz so these drivers may just be OK for me as they are. The comb filtering effects should lessen with distance but that also means they will start to behave like a point source and the SPLs will drop off - perhaps this is good so as not to annoy the neighbours we haven't invited. Old as I am, almost everyone at the party will be even older. This is turning out to be quite a specialised project - a music system for old people at a garden party without a power outlet 🙂
I just wish the drivers would arrive. DFB don't have the fastest delivery service. They may be on a slow boat from China.
Tim.
Many thanks for your input. Actually those frequency response curves don't look too bad to me! I didn't actually know that such modelling software existed so that is fascinating and thanks again. These are low-cost drivers and are very much an unknown beast. The limited parameters that the manufacturer supplies cannot really be relied upon so it will very much be trial and error. I hadn't considered moving the drivers apart but the modelling is very interesting. I have given the project my best initial shot. In the end there will almost certainly be some modifications to compensate for the actual weaknesses of the system - perceived, measured or both.
I was thinking that if the top end is lacking, I may add some piezo tweeters (this will probably provoke even more outrage!) The reason being that I could still avoid using a crossover. However, my hearing falls off quickly over around 14Khz so these drivers may just be OK for me as they are. The comb filtering effects should lessen with distance but that also means they will start to behave like a point source and the SPLs will drop off - perhaps this is good so as not to annoy the neighbours we haven't invited. Old as I am, almost everyone at the party will be even older. This is turning out to be quite a specialised project - a music system for old people at a garden party without a power outlet 🙂
I just wish the drivers would arrive. DFB don't have the fastest delivery service. They may be on a slow boat from China.
Tim.
Last edited:
Both sims account for the floor/ground reflection by assuming a mirrored baffle with four added drivers. The diagrams only show the baffle loss in the bass area. You have to subtract the bass loss of the drivers from that too.
I am still studying this modelling. It is fascinating. Sadly Edge is Windows only. Does anyone know of anything similar for the Mac or is web-based?
My understanding that this is showing the effects of 1) constructive & destructive interference due to there being 4 drivers + ground reflections, 2) baffle boundary and diffraction effects at various frequencies and 3) the dipole interference, i.e. rear waves interfering with front waves. Is this correct?
I expect that the response curve alters drastically - e.g. to the opposite peaks & troughs shown here - depending upon precise listening position. As Rudolf pointed out - this does not show the effects of the actual driver response - a sobering thought. The major complications are that a listener has two ears in significantly different places, listeners will be in a wide range of positions and there will be two arrays in different places. In most recorded music, there will be single elements of the mix panned somewhere between right & left - hence more interference. This acute localisation may be one of the perceived qualities of an open baffle array. The psychology of sound quality is much more complex than its engineering!
We know that linear arrays suffer terribly from interference in order to achieve their overall cylindrical dispersion pattern. Yet they work. I am not bothered in the least by a 6 dB trough or peak in one modelled position. The brain is remarkably adaptable to this. I think by removing the boundary effect lift from around 1 Khz down to around 60Hz (through a minimal baffle and ground reflection), and avoiding crossover-induced phase shifts, the design principles are sound even though the calculated (and actual) frequency response in any one position may look horrific.
The proof is in the pudding. I wish you could come to the party!
BTW I tried out the Lepai amp with small box speakers today. It sounds absolutely great! There was no perceivable noise and oodles of power. I have had some expensive hi-fi in my time but this is the best sounding amp I have ever listened to. Soooooooo looking forward using it with the new speakers.
I will have two 8" drivers left over. These are candidates for the bookshelf version!
Tim.
Last edited:
Done right a piezo can be a quite good. And you won't be eliminating an XO, just using the one built-in caused by the capacitive nature of a piezo impedance. The reason piezos get trashed is from those instances whem people do not add additional XO components,
dave
Hi,
No. It simply won't work at all and the application of "theory" is simply wrong.
Contrary to what you say, I posted how to use the drivers for the application.
Somebody "complaining" about their ideas being criticised isn't even listening.
I'll repeat, it simply won't work at all as it stands. The drivers are very likely
too low Q and lack excursion capability. By all means build it if you want to
pretend it will be allright, it wont, it will be terrible for the intended purpose.
rgds, sreten.
With a 10" wide baffle effective cone area will plummet in the bass. a 15"
driver in a 24" baffle is about equivalent to a single 8" in a box, here you
will end up in the bass with the equivalent of 4 x 2" drivers, it will not work.
Last edited:
- Status
- Not open for further replies.