FWIW,
the drivers I'll be using are the Parts Express #269-469 full-range 4in. 150-16KHz, 90dB/W/m, 15W max pwr, 8ohm. I got two cases of 32 at about 70 cents for each driver, including shipping!
My latest thought is to put 32 of them into a rectangular open baffle array of 4x8 on each channel. The middle 16 (rows 3,4,5,6) will be full range while the top/bottom 16 (rows 1,2,7,8) will be rolled off at about 300Hz. These panels should have flat response from about 150-10KHz, 102dB/W/m efficiency, and 200W+ power handling. I imagine it'll end up looking (even sounding?) like a large planar!
(For lower bass I'll add a dipole woofer section of 4 15" drivers per channel, biamped. For HF, I'm planning to cross this array over around 4KHz with a high-efficiency horn tweeter to further cut down on any beaming/comb filtering nonsense...So I guess it'll be triamped.)
The rectangular array should create less comb filtering than a line array and will also spread the Q of the dipole baffle dimension-related fpeak (similar effect to baffle step).
I'd like to wire the array 2(series) by 8(parallel) for both the rolled-off and full range halves. This will yield a nominal impedence of 1-2 ohms (frequency dependent).
So, I guess you could think of the amp in question as a low wattage, high current mid/bass amp. As such, ultimate frequency extension isn't crucial.
Bill
the drivers I'll be using are the Parts Express #269-469 full-range 4in. 150-16KHz, 90dB/W/m, 15W max pwr, 8ohm. I got two cases of 32 at about 70 cents for each driver, including shipping!
My latest thought is to put 32 of them into a rectangular open baffle array of 4x8 on each channel. The middle 16 (rows 3,4,5,6) will be full range while the top/bottom 16 (rows 1,2,7,8) will be rolled off at about 300Hz. These panels should have flat response from about 150-10KHz, 102dB/W/m efficiency, and 200W+ power handling. I imagine it'll end up looking (even sounding?) like a large planar!
(For lower bass I'll add a dipole woofer section of 4 15" drivers per channel, biamped. For HF, I'm planning to cross this array over around 4KHz with a high-efficiency horn tweeter to further cut down on any beaming/comb filtering nonsense...So I guess it'll be triamped.)
The rectangular array should create less comb filtering than a line array and will also spread the Q of the dipole baffle dimension-related fpeak (similar effect to baffle step).
I'd like to wire the array 2(series) by 8(parallel) for both the rolled-off and full range halves. This will yield a nominal impedence of 1-2 ohms (frequency dependent).
So, I guess you could think of the amp in question as a low wattage, high current mid/bass amp. As such, ultimate frequency extension isn't crucial.
Bill
cool idea
Interesting idea!
With 104 db/1w speakers, you would need only 1 watt output per channel from your amp. If you want more headroom, maybe assume a 2 watt power amp. This would give you more available power than you probably will ever use.
Assuming 1W = 104 db. then 2 W should give 110 db. More than enough.
In order to get 2 watts across 1/4 ohm, you would need peak current of: I*I*(.25) = 2watts =: sqr.(8)= 2.83 amps per channel peak.
The Zen amp would seem like a great candidate if it wasn't for its 1 ohm output impedence. With 2.83 amps available, 1.6 watts of the 2 watts output would be lost across the amplifier output impedence, leaving about .5 watt for the speaker.
Thats not factoring in any milliohm impedence added by cables etc.
You will need either an amp with a very low output impedence,(very high damping factor), or tremendous current reserves.
mg16
Interesting idea!
With 104 db/1w speakers, you would need only 1 watt output per channel from your amp. If you want more headroom, maybe assume a 2 watt power amp. This would give you more available power than you probably will ever use.
Assuming 1W = 104 db. then 2 W should give 110 db. More than enough.
In order to get 2 watts across 1/4 ohm, you would need peak current of: I*I*(.25) = 2watts =: sqr.(8)= 2.83 amps per channel peak.
The Zen amp would seem like a great candidate if it wasn't for its 1 ohm output impedence. With 2.83 amps available, 1.6 watts of the 2 watts output would be lost across the amplifier output impedence, leaving about .5 watt for the speaker.
Thats not factoring in any milliohm impedence added by cables etc.
You will need either an amp with a very low output impedence,(very high damping factor), or tremendous current reserves.
mg16
104spl 1w/1m
I see were you came up with 104 db. You started with a 90 db driver and added 3 db for every doubling of the drivers. You must also double the wattage to maintain your 3 db increase. You must double the drivers and the wattage to increase 3 db. So, you wouldn't wind up with 32 drivers with a 104 db spl at 1w/1m, as each driver would require 1w to maintain its original 90 db spl rating. It would be 104db spl at 32w/1m. Bear in mind, at 2m you will lose 6 db, at 4m the loss is 12 db. 104 db is not that loud. Normal conversation is usually in the 80 db range. The formula, (Watts = log x 10) will give you the db increase for whatever wattage you decide to use. This will be added to your 104 spl at 32w/1m. At the risk of offending I am assuming a lot, but bear in mind I'm only trying to help and would like to see you succeed. Good luck.
I see were you came up with 104 db. You started with a 90 db driver and added 3 db for every doubling of the drivers. You must also double the wattage to maintain your 3 db increase. You must double the drivers and the wattage to increase 3 db. So, you wouldn't wind up with 32 drivers with a 104 db spl at 1w/1m, as each driver would require 1w to maintain its original 90 db spl rating. It would be 104db spl at 32w/1m. Bear in mind, at 2m you will lose 6 db, at 4m the loss is 12 db. 104 db is not that loud. Normal conversation is usually in the 80 db range. The formula, (Watts = log x 10) will give you the db increase for whatever wattage you decide to use. This will be added to your 104 spl at 32w/1m. At the risk of offending I am assuming a lot, but bear in mind I'm only trying to help and would like to see you succeed. Good luck.
Depending on the situation 104 db can be both loud and not loud. If I were in a quiet environment 104 db would be loud. Do we really think about sound levels of 90 db when were walking down a busy city street? 3 db represents a volume increase of double, but the human ear can't perceive it. The human ear requires a 6 db increase to perceive any increase at all. It doesn't sound like its twice as loud does it, but in fact 6 db is more than twice as loud, yet it takes as much as 10 db for the ear to perceive it as twice as loud. The decibel is a logrithmic scale as is your ears hearing ability. The difference between 100watts and 1000watts would be 10 db. The 104 db spl also represents the volume level 3' away from the speaker. If you move to 6' it drops to 98 db. If you move to 12' it drops to 92 db. If you have a big listening room and your 24' away, guess what? It's not really that loud after all is it? As to your speaker question, The difference is in the resistance of your different speaker setups. If the resistance is lower for your 4x12's then the amp will attempt to put out more power. Theoretically an amp at 100watts into 8 ohms will put out 200watts into 4 ohms. It's generally less due to a number of things. Hope this clears up my earlier statements.
Near enough...a basement with 4" concrete floors and three out of four walls being cinder block.
Actually, aside from that, I've taught my ears a thing or two. I've said it elsewhere, but can't remember the thread. People who lose their sight find their hearing increasing in acuity to compensate. The trick is to do this without having to lose your eyesight. Try this:
At night, with the lights out, navigate your home by sound alone. No cheating. The first few times you try it, you'll hit the jamb of the door you were trying to go through. (Keep your hands out in front of your face so as not to get a bloody nose.) It's embarrassing; you thought you could hear, but discover that you can't. But keep at it, and you'll find yourself able to <i>hear</i> walls and doorways and such. (Open doorways are quieter than walls--the sound that hits them doesn't reflect and come back to your ears. Not to mention there are often tiny sounds in the other room that are different than the room you're in, another person's breathing, clocks ticking, etc.)
For extra credit, try to hear the furniture. Don't laugh--most furniture these days is loaded with foam and covered with fabric. It's a black hole for sound. If you hear a profound silence in front of you, go slowly, there's a couch there.
It also helps if you're a musician accustomed to playing by ear (jazz or rock here, classical, for once, has to take a back seat).
<i>Concentrate!</i> It will soon become second nature.
Don't blame me if you start hearing flaws in your stereo you never recognized before. It's expensive to learn to hear.
Grey
[Edited by GRollins on 12-01-2001 at 04:03 PM]
Actually, aside from that, I've taught my ears a thing or two. I've said it elsewhere, but can't remember the thread. People who lose their sight find their hearing increasing in acuity to compensate. The trick is to do this without having to lose your eyesight. Try this:
At night, with the lights out, navigate your home by sound alone. No cheating. The first few times you try it, you'll hit the jamb of the door you were trying to go through. (Keep your hands out in front of your face so as not to get a bloody nose.) It's embarrassing; you thought you could hear, but discover that you can't. But keep at it, and you'll find yourself able to <i>hear</i> walls and doorways and such. (Open doorways are quieter than walls--the sound that hits them doesn't reflect and come back to your ears. Not to mention there are often tiny sounds in the other room that are different than the room you're in, another person's breathing, clocks ticking, etc.)
For extra credit, try to hear the furniture. Don't laugh--most furniture these days is loaded with foam and covered with fabric. It's a black hole for sound. If you hear a profound silence in front of you, go slowly, there's a couch there.
It also helps if you're a musician accustomed to playing by ear (jazz or rock here, classical, for once, has to take a back seat).
<i>Concentrate!</i> It will soon become second nature.
Don't blame me if you start hearing flaws in your stereo you never recognized before. It's expensive to learn to hear.
Grey
[Edited by GRollins on 12-01-2001 at 04:03 PM]
Guru
There was a review of an ear training program in this months issue of audio xpress. Claimed to be able to teach frequency recognition as well as a few other things. I like to close my eyes and dim the lights when I do any listening. I like to shut down the senses that I don't need to hear with. I believe it makes a difference. My wife doesn't understand it. She can't hear a 6db difference either.
There was a review of an ear training program in this months issue of audio xpress. Claimed to be able to teach frequency recognition as well as a few other things. I like to close my eyes and dim the lights when I do any listening. I like to shut down the senses that I don't need to hear with. I believe it makes a difference. My wife doesn't understand it. She can't hear a 6db difference either.
Practical and impractical things
I my very humble opinion the whole idea of running 30 drivers in parallel is extremely impractical and not worth considering.
If a person wanted to drive 30 small drivers to build a sound wall it would be better to drive each speaker with its own small amplifier. Some of these small amplifiers would likely need to be driven from phase and frequency compensating networks to attempt compensate for driver location problems.
It also seems to me that there would be no, or a very peaky low end since the resonant frequency of each individual small driver is just likely to be to way to high. Large cone excursions would thus still be required at the very low frequencies. A lot more drive more power would be needed on the low end.
The best I can relate this to is to parallel many tweeters and attempt to make them reproduce the lower midrange. You have more radiating surface area but they still won't do it due to physical limitations. The high and low-end SPL relationships will remain the same whether you use only one or put 30 drivers in parallel.
Here is my advice on this subject.
I would forget this idea altogether and move on.
John Fassotte
Alaskan Audio
[Edited by alaskanaudio on 12-02-2001 at 12:47 AM]
I my very humble opinion the whole idea of running 30 drivers in parallel is extremely impractical and not worth considering.
If a person wanted to drive 30 small drivers to build a sound wall it would be better to drive each speaker with its own small amplifier. Some of these small amplifiers would likely need to be driven from phase and frequency compensating networks to attempt compensate for driver location problems.
It also seems to me that there would be no, or a very peaky low end since the resonant frequency of each individual small driver is just likely to be to way to high. Large cone excursions would thus still be required at the very low frequencies. A lot more drive more power would be needed on the low end.
The best I can relate this to is to parallel many tweeters and attempt to make them reproduce the lower midrange. You have more radiating surface area but they still won't do it due to physical limitations. The high and low-end SPL relationships will remain the same whether you use only one or put 30 drivers in parallel.
Here is my advice on this subject.
I would forget this idea altogether and move on.
John Fassotte
Alaskan Audio
[Edited by alaskanaudio on 12-02-2001 at 12:47 AM]
The notion that you cannot effectively series and
parallel the devices is not correct. To the extent
that the devices are identical and the acoustic
environment of each device in series is also identical,
the frequency reponse and damping are the same as if
each was driven in parallel.
For the most part, the drivers will be identical
enough, and in wiring them you will want to physically
cluster those drivers which are in series.
I've done it plenty of times and it works just fine.
Having said this, I still think an amplifier which will
drive .1 ohm is cool.
parallel the devices is not correct. To the extent
that the devices are identical and the acoustic
environment of each device in series is also identical,
the frequency reponse and damping are the same as if
each was driven in parallel.
For the most part, the drivers will be identical
enough, and in wiring them you will want to physically
cluster those drivers which are in series.
I've done it plenty of times and it works just fine.
Having said this, I still think an amplifier which will
drive .1 ohm is cool.
No Way! This amp could actually be useful? I was just shooting for do-able!
Ok Nelson, Grey, and others who have a clue, how would I proceed?
Let's start small and imagine a 5 watt (or less) amp that could drive .5 ohm all day. What would the circuit look like? How would you address issues already raised on this thread including transistor type, output impedence, current demand, etc.?
Even better, is there already a (simple, cheap) circuit out there that could be adapted with certain tweaks and component substitutions?
Bill
(I'd like to take this opportunity to also thank all of you guys for chipping in, including those who have tried to warn me off. From inkling of inception, this idea never pretended to be practical in any way. If I wanted practical, I would have built a kit
Ok Nelson, Grey, and others who have a clue, how would I proceed?
Let's start small and imagine a 5 watt (or less) amp that could drive .5 ohm all day. What would the circuit look like? How would you address issues already raised on this thread including transistor type, output impedence, current demand, etc.?
Even better, is there already a (simple, cheap) circuit out there that could be adapted with certain tweaks and component substitutions?
Bill
(I'd like to take this opportunity to also thank all of you guys for chipping in, including those who have tried to warn me off. From inkling of inception, this idea never pretended to be practical in any way. If I wanted practical, I would have built a kit
Confound it, this is how it starts.
People think I'm contrary, that if you tell me not to do something, I'll go out and do it just to assert my independence. Not so. Actually, whether someone says to (or not to) do something has nothing to do with it. It's when I get this idea that it might be relevant and I start to think about <i>how</i> to do it that I get into trouble. That one word 'how' has caused me more grief than I could possibly relate. Once 'how' enters the equation, my curiosity gets to moving and there is no stronger motivator in my world than a good, strong shot of curiosity.
Okay, here goes...
As I see it, we've got two options: High (i.e. normal) rail voltages, and low rails (say, on the order of 5V). The advantage to low rails is that you're going to keep your dissipation down. The bias current on this thing is going to be pretty high, hence a lot of heat. The disadvantage to low rails is that I suspect that you'll have a hard time keeping the circuit even remotely linear, as you'll be sitting right on top of cutoff for the gain device 24-7, no matter what you do, you can't get away from it. This tells us that we'll need to operate in class A.
Okay, how about this: Single-ended--bias the circuit half way up the curve so that you're reasonably linear; as far away from the cutoff as you can get. Start with a Zen. Lower the rail from 34V (or whatever it is) to, say, 5 volts, but bias the critter for--pick a value out of thin air--10 amps. Still only 50 watts heat dissipation. Before you complain about the bias current, go look at the SOA for any of the MOSFETs that are likely candidates. Most have a peak current capability around 8 to 10 volts. You can put loads of current through one at that kind of rail.
Up until last night, I'd have said that this was a hands-down, no question about it, don't even ask such a stupid question candidate for a solid state circuit. But it might be possible to use a normal (i.e. ~2-5k plate load) transformer with something like a 6AS7 or 6C33. The 6AS7 has a plate resistance of something like 280 ohms, which would be a drastic mismatch if hooked to a normal output transformer. But...a transformer is nothing but a ratio device. Lower the load presented by the driver and you'll lower the effective load presented to the plates. Given a few moments and a calculator, you could probably find a reasonably good match between one of those tubes and a .5 ohm load or whatever it is that we're shooting for, here.
If someone has a Zen lying about, you might put this to the test. I'd have to build one from scratch, and though I've probably got enough parts in stock, I don't have the time; need to be working on some of these other projects that I've got hanging. Anyone care to put it to the test?
Grey
People think I'm contrary, that if you tell me not to do something, I'll go out and do it just to assert my independence. Not so. Actually, whether someone says to (or not to) do something has nothing to do with it. It's when I get this idea that it might be relevant and I start to think about <i>how</i> to do it that I get into trouble. That one word 'how' has caused me more grief than I could possibly relate. Once 'how' enters the equation, my curiosity gets to moving and there is no stronger motivator in my world than a good, strong shot of curiosity.
Okay, here goes...
As I see it, we've got two options: High (i.e. normal) rail voltages, and low rails (say, on the order of 5V). The advantage to low rails is that you're going to keep your dissipation down. The bias current on this thing is going to be pretty high, hence a lot of heat. The disadvantage to low rails is that I suspect that you'll have a hard time keeping the circuit even remotely linear, as you'll be sitting right on top of cutoff for the gain device 24-7, no matter what you do, you can't get away from it. This tells us that we'll need to operate in class A.
Okay, how about this: Single-ended--bias the circuit half way up the curve so that you're reasonably linear; as far away from the cutoff as you can get. Start with a Zen. Lower the rail from 34V (or whatever it is) to, say, 5 volts, but bias the critter for--pick a value out of thin air--10 amps. Still only 50 watts heat dissipation. Before you complain about the bias current, go look at the SOA for any of the MOSFETs that are likely candidates. Most have a peak current capability around 8 to 10 volts. You can put loads of current through one at that kind of rail.
Up until last night, I'd have said that this was a hands-down, no question about it, don't even ask such a stupid question candidate for a solid state circuit. But it might be possible to use a normal (i.e. ~2-5k plate load) transformer with something like a 6AS7 or 6C33. The 6AS7 has a plate resistance of something like 280 ohms, which would be a drastic mismatch if hooked to a normal output transformer. But...a transformer is nothing but a ratio device. Lower the load presented by the driver and you'll lower the effective load presented to the plates. Given a few moments and a calculator, you could probably find a reasonably good match between one of those tubes and a .5 ohm load or whatever it is that we're shooting for, here.
If someone has a Zen lying about, you might put this to the test. I'd have to build one from scratch, and though I've probably got enough parts in stock, I don't have the time; need to be working on some of these other projects that I've got hanging. Anyone care to put it to the test?
Grey
This thread just won't die will it?
I have had good luck with Mosfets driving the
Raven tweeter directly. In one case we simply
put 4 ohms in series with it and hooked it up to
an X150 and it worked just fine as a current source.
For this case, you should consider a high current
follower. The Mosfets like IRF150 and its cousins
will throw lots of amperage around and will be quite
linear at 2 volts Drain-Source. If you are running
.25 ohm, think about using +-5 volt rails which you
can obtain from two 5 volt regulated high current
computer switching supplies.
These will give you the ability to swing +- 3Volts
peak, which is 18 watts rms into .25 ohm and is drivable
by your average CD player. 3 volts into .25 ohms
means +-12 amp peaks, so you can bias this follower
with a 12 amp current source, which will give a total
dissipation of 120 watts for the circuit and 60 watts
each for the devices, which is a bit high but do-able.
If you don't want to use a current source, you can
bias it with a resistor. For the 5 volts you will
need .16 ohm, so the dissipation of the whole system
goes up to 300 watts. On second thought, maybe a
constant current source is a better idea....
The other alternative is push-pull followers using
the Mosfets such as the IRFP150's and 9150's as
followers with +- 5 volts and a simple front end
bias circuit. With proper adjustment for bias, you
can safely run these without Source resistors, and
this is probably the best approach of all, as you can
run the bias lower than 12 amps, say only 6 for Class A
and 3 for highly biased AB and still get great performance.
[Edited by Nelson Pass on 12-05-2001 at 02:00 PM]
I have had good luck with Mosfets driving the
Raven tweeter directly. In one case we simply
put 4 ohms in series with it and hooked it up to
an X150 and it worked just fine as a current source.
For this case, you should consider a high current
follower. The Mosfets like IRF150 and its cousins
will throw lots of amperage around and will be quite
linear at 2 volts Drain-Source. If you are running
.25 ohm, think about using +-5 volt rails which you
can obtain from two 5 volt regulated high current
computer switching supplies.
These will give you the ability to swing +- 3Volts
peak, which is 18 watts rms into .25 ohm and is drivable
by your average CD player. 3 volts into .25 ohms
means +-12 amp peaks, so you can bias this follower
with a 12 amp current source, which will give a total
dissipation of 120 watts for the circuit and 60 watts
each for the devices, which is a bit high but do-able.
If you don't want to use a current source, you can
bias it with a resistor. For the 5 volts you will
need .16 ohm, so the dissipation of the whole system
goes up to 300 watts. On second thought, maybe a
constant current source is a better idea....
The other alternative is push-pull followers using
the Mosfets such as the IRFP150's and 9150's as
followers with +- 5 volts and a simple front end
bias circuit. With proper adjustment for bias, you
can safely run these without Source resistors, and
this is probably the best approach of all, as you can
run the bias lower than 12 amps, say only 6 for Class A
and 3 for highly biased AB and still get great performance.
[Edited by Nelson Pass on 12-05-2001 at 02:00 PM]
To lower output impedence, would it be worthwhile to parallel a number of outputs?
For a front end, I'm thinking of the ART DI/O. It's a DA/AD converter that's relatively cheap and gets good reviews. Its output is 7 volts, rather hot for line level. Output impedence is specd at 220 ohms.
Would this have the juice to drive a bank of parallel outputs?
Bill
For a front end, I'm thinking of the ART DI/O. It's a DA/AD converter that's relatively cheap and gets good reviews. Its output is 7 volts, rather hot for line level. Output impedence is specd at 220 ohms.
Would this have the juice to drive a bank of parallel outputs?
Bill
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.