I have spent 1000's of hours researching loudspeaker design and as a result when i got to design my own 2way I come across many sacrifices which must be made. Yes i know loudspeaker design is a game of tradeoffs. My question is what has the most value in creating good sound. In other words what qualities should i attempt to incorporate into my design. Please do your best in ranking the following in order of importance based on your experience and knowledge and feel free to include any comments explaining yuor decision or the affect of the attribute.
1. First order crossover for phase correct design
2. Time aligned drivers as a result of baffle design.
3. Narrow baffle to increase imaging and reduce diffraction.
4. Open air tweeter design where the tweeter sits on top of a front baffle to avoid reflections.
5. Very rigid box with thick front baffles, sand loaded bases and extensive bracing.
6. Offset tweeter for smooth response.
7. Low group delay (below what time at what frequency)
8. Proper damping/stuffing. Where and how much.
9. Rounded baffle for smoother response.
10. Flush mounted drivers.
11. Close distance between drivers.
12. Dampened cabinet walls.
I know theres lots im missing so feel free to add.
1. First order crossover for phase correct design
2. Time aligned drivers as a result of baffle design.
3. Narrow baffle to increase imaging and reduce diffraction.
4. Open air tweeter design where the tweeter sits on top of a front baffle to avoid reflections.
5. Very rigid box with thick front baffles, sand loaded bases and extensive bracing.
6. Offset tweeter for smooth response.
7. Low group delay (below what time at what frequency)
8. Proper damping/stuffing. Where and how much.
9. Rounded baffle for smoother response.
10. Flush mounted drivers.
11. Close distance between drivers.
12. Dampened cabinet walls.
I know theres lots im missing so feel free to add.
I think good speakers have been made that break almost all the rules, though not at the same time! My #1:
1) Quality drivers with the desired response and no breakup or misbehavior well beyond the crossover points.
I've never been able to turn a bad driver into a good driver with any amount of eq, zobels, or other electrical magic, and there are an awful lot of bad drivers out there. It would seem that the market for buzzers and sirens is larger than I would have predicted.
1) Quality drivers with the desired response and no breakup or misbehavior well beyond the crossover points.
I've never been able to turn a bad driver into a good driver with any amount of eq, zobels, or other electrical magic, and there are an awful lot of bad drivers out there. It would seem that the market for buzzers and sirens is larger than I would have predicted.
I don't think ranking is really possible because engineering is all about trading good for bad to reach a happy medium, and many things interact- and not many folks can agree what makes one speaker better than another, and often even if one speaker IS better than another.
Here are a few of my comments on your criteria.
1. First order crossover for phase correct design
Really not that important. I'm listening to some beautiful sounding drivers with a really steep filter, and I honestly can't hear the phase.
2. Time aligned drivers as a result of baffle design.
This is helpful, but can be overcome somewhat in crossover design.
3. Narrow baffle to increase imaging and reduce diffraction.
Some will argue that "the best baffle is no baffle"- as in, building an infinite baffle to flush mound the speakers on a wall. I think that the most practical thing to do is to simulate the effect of the baffle, and lay out your drivers accordingly- then perhaps use some felt to absorb some of those high frequencies as the spread out over the cabinet face.
4. Open air tweeter design where the tweeter sits on top of a front baffle to avoid reflections.
I think that managed directivity is very important, and I'm not sure that this approach serves this goal. In the case of Linkwitz's Pluto speaker, he did something like this with the tweeter, but the very low crossover frequency allowed the entire collective speaker to stay nearly ominidirectional across the entire crossover.
5. Very rigid box with thick front baffles, sand loaded bases and extensive bracing.
Sand loaded bases seem like a waste of time to me. Overly thick front panels can cause weird cavity resonances in the area between the back of the woofer cone and the main body of the box. Rigid boxes are good, bracing is good, and even better is the rule of thumb that you build stiff for bass and build heavy for mids/highs. The key is that your cabinet can ring like a bell, as long as its fundamental frequency is never excited by the driver or any of its harmonics. Separate cabinets to divide up the frequencies strikes me as a great idea for this, as it also cuts down on the weight of the largest piece you'll have to move.
6. Offset tweeter for smooth response.
Again, this is something that needs to be measured and simulated. A centered tweeter has an effect, but other things might have effects that cause this to be desirable, or at least less objectionable.
7. Low group delay (below what time at what frequency)
There have been some good articles written on this if you google around for them. If memory serves, most mid/high crossovers don't suffer from perceptible group delay. The main area of controversy that I keep hearing about is woofer to sub crossovers.
8. Proper damping/stuffing. Where and how much.
Trial and error- it's different for every speaker. An open baffle speaker doesn't need any. Ported boxes are a touchy thing, and it's best to avoid stuffing lest you damp out the desired port response.
9. Rounded baffle for smoother response.
Eh, depends- you can simulate and/or measure the consequences. You can compare it to chamfered edges. You can consider just covering the face with heavy felt.
10. Flush mounted drivers.
Overall, this looks to be fairly important. This is also something that is measured easily enough, and if flush mounting was really not an option, something that could be accounted for in a crossover.
11. Close distance between drivers.
Yes, looking at the wavelength of sound at the crossover frequency will tell you how close they need to be. Within 1/16th wavelength at crossover is about as good as it gets, greater than 1/2 wavelength is seriously getting in trouble.
12. Dampened cabinet walls.
I've said it before, I'll say it again- "dampen" means "to make moist", as in- with water. Do you mean that the walls have mechanical damping so that resonance in the wall panel is dissipated quickly? This is important if you're forced to leave a panel resonance in or near the bandwidth of the driver in that cabinet. If you mean acoustic absorbing material like foam or fiberglass, yes- this is important too.
That's just my two cents. There are still a lot of myths out there, and a lot of things that get repeated without real understanding. I'm not a guru or anything, but I try to know what I think I know.
Here are a few of my comments on your criteria.
1. First order crossover for phase correct design
Really not that important. I'm listening to some beautiful sounding drivers with a really steep filter, and I honestly can't hear the phase.
2. Time aligned drivers as a result of baffle design.
This is helpful, but can be overcome somewhat in crossover design.
3. Narrow baffle to increase imaging and reduce diffraction.
Some will argue that "the best baffle is no baffle"- as in, building an infinite baffle to flush mound the speakers on a wall. I think that the most practical thing to do is to simulate the effect of the baffle, and lay out your drivers accordingly- then perhaps use some felt to absorb some of those high frequencies as the spread out over the cabinet face.
4. Open air tweeter design where the tweeter sits on top of a front baffle to avoid reflections.
I think that managed directivity is very important, and I'm not sure that this approach serves this goal. In the case of Linkwitz's Pluto speaker, he did something like this with the tweeter, but the very low crossover frequency allowed the entire collective speaker to stay nearly ominidirectional across the entire crossover.
5. Very rigid box with thick front baffles, sand loaded bases and extensive bracing.
Sand loaded bases seem like a waste of time to me. Overly thick front panels can cause weird cavity resonances in the area between the back of the woofer cone and the main body of the box. Rigid boxes are good, bracing is good, and even better is the rule of thumb that you build stiff for bass and build heavy for mids/highs. The key is that your cabinet can ring like a bell, as long as its fundamental frequency is never excited by the driver or any of its harmonics. Separate cabinets to divide up the frequencies strikes me as a great idea for this, as it also cuts down on the weight of the largest piece you'll have to move.
6. Offset tweeter for smooth response.
Again, this is something that needs to be measured and simulated. A centered tweeter has an effect, but other things might have effects that cause this to be desirable, or at least less objectionable.
7. Low group delay (below what time at what frequency)
There have been some good articles written on this if you google around for them. If memory serves, most mid/high crossovers don't suffer from perceptible group delay. The main area of controversy that I keep hearing about is woofer to sub crossovers.
8. Proper damping/stuffing. Where and how much.
Trial and error- it's different for every speaker. An open baffle speaker doesn't need any. Ported boxes are a touchy thing, and it's best to avoid stuffing lest you damp out the desired port response.
9. Rounded baffle for smoother response.
Eh, depends- you can simulate and/or measure the consequences. You can compare it to chamfered edges. You can consider just covering the face with heavy felt.
10. Flush mounted drivers.
Overall, this looks to be fairly important. This is also something that is measured easily enough, and if flush mounting was really not an option, something that could be accounted for in a crossover.
11. Close distance between drivers.
Yes, looking at the wavelength of sound at the crossover frequency will tell you how close they need to be. Within 1/16th wavelength at crossover is about as good as it gets, greater than 1/2 wavelength is seriously getting in trouble.
12. Dampened cabinet walls.
I've said it before, I'll say it again- "dampen" means "to make moist", as in- with water. Do you mean that the walls have mechanical damping so that resonance in the wall panel is dissipated quickly? This is important if you're forced to leave a panel resonance in or near the bandwidth of the driver in that cabinet. If you mean acoustic absorbing material like foam or fiberglass, yes- this is important too.
That's just my two cents. There are still a lot of myths out there, and a lot of things that get repeated without real understanding. I'm not a guru or anything, but I try to know what I think I know.
First on your list should be driver selection! The lowest distorting drivers sound the best because they change the input signal the least.
www.zaphaudio.com
www.zaphaudio.com
Use good xo components. They can only be judged accurately after breakin unfortunately. Wire with good wire, and a star ground is recommended. I prefer drivers with clean cumulative spectral display (waterfall) graphs. Be wary of differing scales when comparing. I like to see at least a 30 dB vertical scale. For me, CSD is the best correlation to sound. I REALLY prefer a mid driver with a smooth high end rolloff. I prefer a trap (if necessary) to a steeper rolloff filter design.
First of all --> Choose the correct transducers
Second --> Match them, make the right frequency cut (DIFFICULT!) and attenuation.
Third --> Have a good cabinet, foam inside, thick wood, good air duct calculated considering phase, impedance, and frequency response.
Fourth --> Well aligned transducers.
In my opinion, 3 way systems are BETTER.
In 2 way systems, the mid frequencies reproduction are critical, due to the cone moviment caused by the lower ones, resulting in mid distortion.
I would go with a 3 way design, and cut the first woofer about 120Hz.
Then you can use a nice mid woofer and the tweeter selection is less critical as the mid woofer can go higher than it would go in a 2 way design. You can choose then a more cheap, light and smooth tweeter.
Second --> Match them, make the right frequency cut (DIFFICULT!) and attenuation.
Third --> Have a good cabinet, foam inside, thick wood, good air duct calculated considering phase, impedance, and frequency response.
Fourth --> Well aligned transducers.
In my opinion, 3 way systems are BETTER.
In 2 way systems, the mid frequencies reproduction are critical, due to the cone moviment caused by the lower ones, resulting in mid distortion.
I would go with a 3 way design, and cut the first woofer about 120Hz.
Then you can use a nice mid woofer and the tweeter selection is less critical as the mid woofer can go higher than it would go in a 2 way design. You can choose then a more cheap, light and smooth tweeter.
Hi Rounder, I don't want to sound too negative, as you have obviously been thinking about this a lot, but all your list items are what I would call secondary considerations that can be addressed much later in the design process.
First things first, define the size of box you, (or your significant other), can live with. Then, find a low distortion mid bass driver with a good FR you like and work out the optimal design, (TL, BR, IB, aperiodic), to get where you want. Then look at tweeters to suit the high end roll off of your mid bass. Once you have suitable candidates, build a test box and measure. Then you can start worrying about crossover type, speaker positioning, etc, listening and measuring as you go.
First things first, define the size of box you, (or your significant other), can live with. Then, find a low distortion mid bass driver with a good FR you like and work out the optimal design, (TL, BR, IB, aperiodic), to get where you want. Then look at tweeters to suit the high end roll off of your mid bass. Once you have suitable candidates, build a test box and measure. Then you can start worrying about crossover type, speaker positioning, etc, listening and measuring as you go.
thanks for everyones input.
Considering how this topic is starting to focus a little more on my personal design ill mention that the parts that i am working with are the peerless hds tweeter 810921 and the hds 6.5 exclusive peerless woofer 830883. Most likely a small tower with roughly .7cuft and an 8 inch wide front baffle. I havent decided anything for sure other than drivers which i have purchased... i would like to use a first order crossover if i can make it sound good... i was thinking roughly 1500hz on the woofer and1800hz on the tweeter... with a notch on the woofer to take care of the peak. I like a bit of a bass kick so ill probably tune it for a bit of a bump at 50hz or so.
Considering how this topic is starting to focus a little more on my personal design ill mention that the parts that i am working with are the peerless hds tweeter 810921 and the hds 6.5 exclusive peerless woofer 830883. Most likely a small tower with roughly .7cuft and an 8 inch wide front baffle. I havent decided anything for sure other than drivers which i have purchased... i would like to use a first order crossover if i can make it sound good... i was thinking roughly 1500hz on the woofer and1800hz on the tweeter... with a notch on the woofer to take care of the peak. I like a bit of a bass kick so ill probably tune it for a bit of a bump at 50hz or so.
rounder ....come over to the dark side
and give up on all that stuff.Here's my take on importance in design...(tongue in cheek towards the other suggestions, but this has worked for me...)
Based on your biases, the OB way is definetly not the way to go, but I had to state it...
really, it is all personal. And there are good/great DIY versions of all types. with crossovers, keep the X-over points at least a full octave away from the resonant frequencies of each driver/tweeter pair that you are considering. And all Xovers create some phase/group delays. Another option may be to do as Linkwitz and others do and use an active crossover and bi-amp your speakers...
and you have not listed your amplification, room size and musical tastes. They are all important.
and give up on all that stuff.Here's my take on importance in design...(tongue in cheek towards the other suggestions, but this has worked for me...)
- room setup, but you CAN make a pretty crumby room sound better , as long as the SOAF is high
- no enclosures...just a baffle.
- no crossovers
- single driver
- Quad ESL , Wharfedale SB3 sized baffles (as in the JE Labs "style)
Based on your biases, the OB way is definetly not the way to go, but I had to state it...
really, it is all personal. And there are good/great DIY versions of all types. with crossovers, keep the X-over points at least a full octave away from the resonant frequencies of each driver/tweeter pair that you are considering. And all Xovers create some phase/group delays. Another option may be to do as Linkwitz and others do and use an active crossover and bi-amp your speakers...
and you have not listed your amplification, room size and musical tastes. They are all important.
pinkmouse said:
I always like to start with a midbass/midrange of sufficiant capability that it can be listened to in a system all by itself.... and then if you need more keep the crossovers out of the 300-5kHz range at a minimum.
A speaker should never be considered without the room & driving amplifier as part of the equation.
dave
SY said:
No, not always by any means and I should have been more clear. It didn't help that I overlooked the 2 way part...
These comments only apply to very capable drivers in a three way system. In a two way, you don't have anything like the same degree of freedom. It does take a significant driver set; not many are capable; it is however less taxing than full range applications. Putting in a second pole an octave out from the crossover point helps quite a bit, like the older Thieles, and is a bit less demanding of component quality for the second pole. For the mid's lower end, using a closed box does of course restrict cone excursion... for the tweeter, a high xmax and low resonant frequency are needed. In the current project, the fs and xo are separated by almost 3 octaves.
In a real room, I may be deficient, but do not notice the lobing. However, in a 3 way the tweeter crossover frequency tends to be higher, which makes lobing issues less audible.
Re: Re: What really matters? Quality loudspeaker design.
Good replies by all, my favourite being the one from Joe Carrow.
In my experience, appreciating the effects of different methods and practises requires a learning curve. In the end, one might discover that certain things affect others, and that others are at the opposite ends of the scale. On my learning journey, I am currently concerned about the following points:
I have not discovered a need to align acoustic centres, but if I were to do it, I would align them a long a vertical axis (physically) as opposed to electronic delay. Alignment by electronic time delay only works on-axis.
These four are interrelated.
3. Narrow baffle to increase imaging and reduce diffraction
I am not convinced that a narrow baffle decreases diffraction, or that it is responsible for better imaging.
6. Offset tweeter for smooth response
Offsetting the tweeter may yield a smooth on-axis response, but it does not solve the problem of diffraction.
9. Rounded baffle for smoother response
Yes, absolutely. This results in less harsh diffraction. But rounding (radius) must be large enough to be effective.
10. Flush mounted drivers
I always do this. I believe that some drivers need it more than others, though.
I believe that narrow baffles got the credit for producing the best imaging only because this forces the drivers to be centrally mounted, and therefore produce a symmetrical power response. It is the symmetrical power power response, I believe, that produces good imaging. With this reasoning, I think that wide baffles can also produce good imaging.
Again, offset mounting of the tweeter can achieve the smoothest on-axis response for that baffle, but would result in asymmetrical power response.
Power response should be smooth off-axis. No sudden excursions. The rule in audio is Nice and Easy does it. The same counts for group delay: the smoothest roll-off rates, the smoothest response wins the day. Even if summed response is flat, I have found the sound of a filter having a sharp "shoulder" to be audible.
Good replies by all, my favourite being the one from Joe Carrow.
In my experience, appreciating the effects of different methods and practises requires a learning curve. In the end, one might discover that certain things affect others, and that others are at the opposite ends of the scale. On my learning journey, I am currently concerned about the following points:
I have not discovered a need to align acoustic centres, but if I were to do it, I would align them a long a vertical axis (physically) as opposed to electronic delay. Alignment by electronic time delay only works on-axis.
These four are interrelated.
3. Narrow baffle to increase imaging and reduce diffraction
I am not convinced that a narrow baffle decreases diffraction, or that it is responsible for better imaging.
6. Offset tweeter for smooth response
Offsetting the tweeter may yield a smooth on-axis response, but it does not solve the problem of diffraction.
9. Rounded baffle for smoother response
Yes, absolutely. This results in less harsh diffraction. But rounding (radius) must be large enough to be effective.
10. Flush mounted drivers
I always do this. I believe that some drivers need it more than others, though.
I believe that narrow baffles got the credit for producing the best imaging only because this forces the drivers to be centrally mounted, and therefore produce a symmetrical power response. It is the symmetrical power power response, I believe, that produces good imaging. With this reasoning, I think that wide baffles can also produce good imaging.
Again, offset mounting of the tweeter can achieve the smoothest on-axis response for that baffle, but would result in asymmetrical power response.
Power response should be smooth off-axis. No sudden excursions. The rule in audio is Nice and Easy does it. The same counts for group delay: the smoothest roll-off rates, the smoothest response wins the day. Even if summed response is flat, I have found the sound of a filter having a sharp "shoulder" to be audible.
Hi,
Coincidentially, I went to Joe Rasmussen's place (www.customanalogue.com) yesterday and had a listen to his Elsinore Loudspeakers. The speakers use exactly the same drivers you are considering. It is worth reading his pages.
Regards,
Bill
Coincidentially, I went to Joe Rasmussen's place (www.customanalogue.com) yesterday and had a listen to his Elsinore Loudspeakers. The speakers use exactly the same drivers you are considering. It is worth reading his pages.
Regards,
Bill
Rounder said:
You cannot include everything - but it is not just a matter of tradeoffs - it's how the sum of the parts contribute to a coherent whole. It is after all a loudspeaker system.
By all means get the best bits you can - within reason (or within sanity?). The best designers work for years to develop a methodology and when they sit down with a brief, they almost clearly have the approach down pat. It's here that the occasional DIY constructor is at a disadvantage. Where does he start, how does he proceed, what are the clear aims and does he have the confidence to know that he is going to get there?
For this reason it is not a bad idea to look for an established design where you are confident that somebody with the where-with-all have done the hard yakker (Oz speak). Get familiar with the elements, learn to understand how this works as a system and chances are you start out on the longer road and develop your own signature.
Bill says he has heard the Elsinores on www.customanalogue.com - it has not just a few hundred of hours of work behind it - but building it you will learn as the design is ambitious. The total philosophy behind the design is explained in exhaustive detail, including "The Renegade Tweeter Theory" explaining why the Tweeter correct acoustic phase can only be achieved when connecting the Tweeter electrically out of phase. Don't dismiss it out of hand, (as many will), there is a definite rationale behind it. The reproduction of genuine square waves at the listening position does not lie.
So it is guaranteed you will not just make a good speaker (an under statement if I may say so myself) but also take something else away in the end.
Joe R.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.