View Full Version : Woofer driver positioning on baffle - in front or behind baffle?
12-08-2010, 10:24 AM
First i have to thank you for making such great speakers. Previlosi i ovned some Chartwell PM 110 speakers who I loved, but the woofer made a noise after 20 years of use, so i was lokking for a replasement. I have looked for years, until i discovered Harbeth. They have the same "sound", but they have more air and detail. The Chartwell also used thin wall construktion(but in a vented disign), so there has to be something with it I like. I Think they have a fullness in voices(around 100 - 300Hz) that other speakers lak. It also creates the big holografic room of the speakers I think.
Then to my questions. In many designs you please the woofer on the back of the baffle. What is the advantages of doing that?
Also, I was wondering about some of the crossoverparts you seem to use.I recon you use them for their sound, but it looks like polyester capasitors and I have always heard they were infirior of polypropylens. So can polyestercaps sound as good as/better than polypropylens? Pardon for my bad english, it is not my first language.
Thank you for this great website and for all the information you give about speaker designing. Its great!
Best regards Mattism
Welcome to the HUG. I understand you perfectly. Good questions, thanks for the opportunity to do some research for you and to reply.
Yes, I think that there is a common origin for the Harbeth and Chartwell brand. There is also a connection with the Rogers brand. They are all intellectually rooted in the BBC 'school' of monitor speaker design. Harbeth was formed in 1977 and I believe that the Chartwell company was already in existance by then. I looked-up in my archive 'HiFi Choice, Loudspeakers' No.26/1881 and I see that by 1981, the Chartwell speakers were listed as Chartwell/Rogers and the factory address was that of the Rogers factory, Swisstone Electronics Ltd.. So I guess that by then, Chartwell had been absorbed into Rogers. I will see if I can find more information later.
Anyway, your question was about mounting the driver on or behind the baffle. First. let's show some pictures of this so we can be clear about the concept of flush or rear mounting.
TO BE CONTINUED;
Sorry for the delay. I thought I'd do a proper little CAD drawing first which I've now done. I need a day or two more on this please.
I've made another TechTalk item. This is just part of the explanation as to why the bass/mid driver is mounted behind the panel for some models and in front for other. It's about path length adjustment. There are other factors such as on/off axis response, but this is a major reason.
Hope this is of interest - I made five 'takes' to try and reduce the length, but I just couldn't without leaving out what I thought was valuable information.
New TechTalk here (http://www.harbeth.co.uk/uk/index.php?section=products&page=designersnotebookdetail&id=18).
17-08-2010, 08:13 PM
Thank you for a great answer, but I still wonder why you put the woofer on the inside. I understand that the time alignment has to be right, but the moving of the woofer into the cabinet do not compensate for the delay, it makes it worse. Would it not be easier to make the right delay in the tweeter and not move the woofer, you still have to make a delay in the crossover, dont you?
Ah! Remember - what matters is the relative delay between the woofer and tweeter as I tried to illustrate in my video. By placing the woofer behind (or flush on the front of) the baffle it may well assist aligning the relative delay with the tweeter. Every speaker system design will be different, and this aspect will have to be considered afresh. As with all things relating to quality loudspeakers there are no hard and fast rules.
Absolute alignment is impossible - relative alignment over a narrow frequency band is (sometimes) possible with a lot of design effort. And providing that you are not frightened by a complex crossover and have the measuring tools and techniques to attack the problem of alignment. It's one of the most difficult aspects of quality speaker design - and not one I can go much deeper into here.
Alan, enjoyed the TechTalk video. I was thinking of investing in a pair of heavy studio grade speaker stands that can both be tilted ( forwards and backwards at any angle) and adjusted in height. Do you think being able to adjust my speakers ( M30s) in this way could help with their alignment and relative delay in relation to my seating position. Thanks Philip.
... Do you think being able to adjust my speakers ( M30s) in this way could help with their alignment and relative delay in relation to my seating position...Thanks for the feedback which is extremely rare and I appreciate. Without feedback I have no idea if I'm pitching at too high, too low or at just the right technical level. I think I've had only two or three comments on these TTs in the past few years so I must assume that they're of zero interest, incomprehensible or boring. They may look shambolic but they do take time to self-produce!
Ok, thanks to your feedback I can see that maybe I didn't convey quite what I intended. Yes, from one perspective, if you have a flexible stand solution for height and angle you are surely going to give yourself the maximum performance possible from your speakers. You can trim the stand height and/or tilt to compensate for your own height, different speakers and different listening seats so you can 'gear' the speaker optimally to whatever your personal set-up requires.
On the other hand, at the design stage I have to take a view on all the height and angle factors impinging on listening at home and inevitably settle on what will be to many users, a compromise. The fact is - as I hope I showed in the TT video - that the poor old designer can sort-out the time alignment (in the crossover electrically or physically) but at only one vertical point in space. And that point is usually on or about level with the tweeter's axis. But how many users are willing to either put their speakers on appropriately tall stands to place the tweeter at or about ear level and/or tilt the speaker back? Few I think. But if they don't, then they are, inevitably, not getting the best from the speakers by a substantial margin. They will, inevitably, be listening off-axis and off-axis means, in the crossover region that the amount of energy at the top end of the woofer/bottom end of the tweeter is many dBs below my design intention. In short, the speaker is going to sound less detailed and somewhat 'dryer' in the all-critical presence region than it is capable of listended to on-axis. That's not by any means unique to a Harbeth design - all multi-driver speakers have the same issues of different path lengths from the drive units. Coaxial speakers don't have this issue but they have others which can be even more serious problems.
I could go into more detail about this with a follow-up TT, but I'm extremely busy and without more feedback as to whether man-days of my effort to make TechTalks are doing any good for the users (or for Harbeth) I'm reluctant to steal the time from other important activities. Sorry.
03-09-2010, 11:01 AM
I totally agree (but then again, why argue with Alan) that for best results you need the appropriate height stands which place the tweeter at ear level. It makes a big difference to the presence and focus of images
To my knowledge, all but a couple of pairs of Harbeth I have sold, have been supplied with tailor made stands. I get the customer to measure from ear to floor level, I calculate the required height and get the stands made to suit.
The correct height is well worth achieving to get the very best from your speakers, even if they don't quite fit in with the decor. As a second best, the speakers may be tilted slightly back to direct more HF energy towards the listener but I would alwas opt for the correct stands.
03-09-2010, 11:10 AM
On a slightly different note, we do appreciate your 'tech talk' but comments from us are not always necessary as you have usually explained the points with great clarity. We read and learn.
As to mounting the main driver behind or on the baffle, I have been reading my copy of 'Loudspeakers' by Gilbert Briggs for the umpteenth time and he describes the various mounting arrangements including having the driver standing off of the baffle to provide the appropriate venting without having a separate hole in the cabinet. This looks like a neat arrangement but I wonder how effective this might be ?
... we do appreciate your 'tech talk' but comments from us are not always necessary as you have usually explained the points with great clarity. We read and learn.Thanks for that. I'm not actually fishing for compliments, but I must know if I'm conveying the explanation at the right technical level. An explanation that fails to convey the key message is worthless. Less than worthless because it just further confuses the viewer. After planning, recording, editing, packaging for the internet, conversion to Flash and uploading all by myself late into the night I want to know how the next one can be better, simpler, more intelligible, smarter graphics, better sound or whatever.
Obviously, there is a limit to my own creative abilities - we're not video professionals and it's the content that matters more than the slickness - but there must be room for improvement. Or are they just about right? I'm acutely aware that English is not the first language of many - most - of our visitors. Do I have a good sense of what's needed? Is the audio clear enough?
03-09-2010, 03:22 PM
Just had a look at the TT which is very informative, just a couple of points though:
I was curious, at a very simple level, about the maths involved: what sort of time discrepancies are we talking about?
Moving a driver from the front to the back of the baffle would be a distance of 20mm at most I'd guess, or 0.02 of a metre. If sound travels at 340 m/s then 0.02m takes 340/0.02 of a second or a 1/17000th - is that right? Doesn't sound like much, but presumably in this context it is!
Secondly, the critical issue with respect to time alignment is the vertical axis isn't it? Yet as audiophiles we seem far more preoccupied with the horizontal - this is for different reasons but have we got the balance wrong?
... Moving a driver from the front to the back of the baffle would be a distance of 20mm at most I'd guess, or 0.02 of a metre. If sound travels at 340 m/s then 0.02m takes 340/0.02 of a second or a 1/17000th - is that right?Ummm. You've missed a trick here! Have another go-round on the dimensions involved. Have a really good look at my little CAD drawing. Your starting point is incorrect .......
03-09-2010, 03:58 PM
Do you mean where you move the bass/mid such that the magnet is flush with the baffle? ie the cone is actually outside the cabinet?
In your post #2 above you show a driver mounted on the face of the baffle and a driver mounted behind the baffle. I understand the point that the sound is actually coming from down inside the cone in the case of a base/mid, but relative to it's starting point and in the case of the M30 and C7 above, any point on the cone is only being shifted by 20mm or so isn't it?
The velocity of sound waves coming from the bass/mid and tweeter are the same regardless of frequency. That is, approx. 340mS (very, very, very slow compared with the velocity of light).
At 340Hz, the wavelength of each cycle of a sine wave is 1.0m. At 3400Hz (3.4kHz) the wavelength is 0.1m (10cm) .....
03-09-2010, 08:17 PM
The velocity of sound waves coming from the bass/mid and tweeter are the same regardless of frequency.
Yes, so if I were to assume that an ideal crossover had managed to exactly time align tweeter and bass/mid and then the bass/mid was moved back by 20mm would that mean that sound from the bass/mid now lagged behind the sound from the tweeter by the fraction I suggested?
I am confused as to whether you're point is that the statement is incorrect or that it is simply not the right way to view this issue.
That is, approx. 340mS (very, very, very slow compared with the velocity of light).
At 340Hz, the wavelength of each cycle of a sine wave is 1.0m. At 3400Hz (3.4kHz) the wavelength is 0.1m (10cm) .....
340 Hz is around F above middle C isn't it; 2cm as a fraction of 100cm (the wavelength) now seems relatively large (1/50th). 3400 Hz is close to the top of the piano range; 2cm as a fraction of a 10cm wavelength is significant (but my assumption is that the bass/mid isn't doing anything at this frequency.)
The relationship between wavelength and frequency (wavelength = velocity/frequency) is one that I had long since forgotten, thanks for the reminder.
Yes, so if I were to assume that an ideal crossover had managed to exactly time align tweeter and bass/mid and then the bass/mid was moved back by 20mm would that mean that sound from the bass/mid now lagged behind the sound from the tweeter by the fraction I suggested?Let's be honest, we're not really "time-aligning" the two drive units. We can't do that in a passive crossover. We could (theoretically) do that in a digital crossover because we'd hold the data that would otherwise feed the tweeter directly in a frame-buffer and release it at exactly the right time. But analogue passive crossovers don't have frame stores do they. So we have to say "time alignment" in quotes because that's impossible to achieve without digital technology, when, should you wish, it would be a trivial matter to delay the signal to the tweeter by a second, a day, a week .... just depending upon how big the frame store is. But "time aligmnent" of a compromised type is definitely what we are aiming at.
340 Hz is around F above middle C isn't it; 2cm as a fraction of 100cm (the wavelength) now seems relatively large (1/50th). 3400 Hz is close to the top of the piano range; 2cm as a fraction of a 10cm wavelength is significant (but my assumption is that the bass/mid isn't doing anything at this frequency.)Right to say that 2cms as a fraction of 10cms is a huge percentage. Percentage of what? Percentage of the cycle. And we know that a cycle is another way of saying 360 degrees ......
And most definitely, the bass/midrange is working nearly at full loudness to about 3kHz, is at about 1/3 loudness to 4kHz, tenth loudness to perhaps 6kHz. Again, crossover filters gradually fade out the woofer and fade up the tweeter. IT IS THIS LARGE OVERLAP BETWEEN DRIVE UNITS WHERE THE "TIME ALIGMENT PROBLEM" LIES. If the bass/mid's contribution stopped dead and handed over sharply to the tweeter the problem of mis-alignment would be inaudible.
We've covered crossover slopes in this TechTalk (http://www.harbeth.co.uk/uk/index.php?section=products&page=designersnotebookdetail&id=5) issue.
04-09-2010, 09:03 AM
Many thanks Alan, that starts to become much clearer.
Back up at the top of the thread you said:
It's about path length adjustment.which appears to be the better way of thinking about it; as you say above, actual "time alignment" would be something else.
Is the (possible) reduction in energy in the crossover region due to "time alignment" issues therefore a consequence of signals being slightly out of phase and cancelling each other?
To come back to your query about the technical level at which you pitch the TT videos. In general I think you do very well at keeping things in 'everyday' language. In common with most industries though there are terms which can appear to mean one thing if taken literally but their accepted usage within that industry actually means something else -"time alignment" being the example here.
That said, I believe your explanations are clearest when you describe things directly and don't try to over-simplify, perhaps it is just a case of making clear when a term being used isn't being used literally.
Lastly, a point which crops up frequently is the difficulties audiophiles can get into when we believe we understand something, when we believe we know that such and such component is making such and such contribution - when in fact the science says something completely different. But if we don't understand the science, if the terminology doesn't actually mean what we think it means, it's easy to see how these issues are perpetuated.
05-09-2010, 08:11 AM
Just crossing over with the comments from Paul G Smith in the mic thread (http://www.harbeth.co.uk/usergroup/showthread.php?971-Microphones-the-critical-input-to-the-audio-chain-....&p=10939#post10939) regarding spot mics and so on got me wondering about delays in sound reaching us - or more specifically microphones - in an orchestral context.
A thread on recording.org (http://recording.org/acoustic-music-forum/16746-time-alignment.html) answers much of what I had been thinking about.
It also talks about 'time alignment' in the way that I had originally assumed was meant in this thread.