As usual, a super quick action in building. Well done! Congratulations.
The ultra long bolts for the woofer worry me somewhat, though. They are long cantilevers. The force at the connection point on the frame should be very large. The woofer looks very heavy, doesn't it?
Is it possible to hold the 'neck' between magnet assembly and frame as the main supporting point (ring)? And then add another one or two points on the frame for balancing. Still fully suspended, just an easier mechanical load on the driver. Just my 2c.
The ultra long bolts for the woofer worry me somewhat, though. They are long cantilevers. The force at the connection point on the frame should be very large. The woofer looks very heavy, doesn't it?
Is it possible to hold the 'neck' between magnet assembly and frame as the main supporting point (ring)? And then add another one or two points on the frame for balancing. Still fully suspended, just an easier mechanical load on the driver. Just my 2c.
Yes, the mids and tweeter are in perfect time-align. The woofer is still somewhat behind the others.... oh well, I have digital time alignment.It looks as if you got physical time alignment (at least almost) på kjöpet !
/Erling
As usual, a super quick action in building. Well done! Congratulations.
The ultra long bolts for the woofer worry me somewhat, though. They are long cantilevers. The force at the connection point on the frame should be very large. The woofer looks very heavy, doesn't it?
Is it possible to hold the 'neck' between magnet assembly and frame as the main supporting point (ring)? And then add another one or two points on the frame for balancing. Still fully suspended, just an easier mechanical load on the driver. Just my 2c.
Good point. The bolts are 10 mm in diameter, and 200 mm long so the stress at the connection points must be large. The woofer weights 15 kg... quite heavy. I was already thinking of drilling holes in the driver frame at the balance points, and attach the chains there. Lets see....
Some more changes here again!
When I converted to baffle-less, I noticed that the tweeter suddenly got a dip around 4.5 - 5 kHz that wasnt there with the baffles. I EQ'it flat on axis, but off-axis it was BAD (a large peak because of the EQ).
I just had to find out what was causing this. I guessed edge diffraction... and I was right. I modeled it the EDGE software, and the model was pretty spot-on, althought the dip was larger in real life than predicted.
It was easy to control this diffraction with foam on the edges, as the pictures show. The foam damping didn't disturb the off-axis response. On 90 degrees the tweeter is almost flat, and 25 dB lower than on axis!
Listening tests showed that there was a slight harshness before, which is cured now. Saxophone could get a little hard sounding, that problem is gone. It also sounds better off-axis.
The next thing to attack is that the dipole peak of the mid's is higher off-axis than on-axis. Even if I EQ it flat on-axis, there is still a peak off-axis. Any ideas of what to to? I dont think foam will work here...
EDGE model:
Measurement with and without foam:
Pictures... (pretty ugly, or what? )
When I converted to baffle-less, I noticed that the tweeter suddenly got a dip around 4.5 - 5 kHz that wasnt there with the baffles. I EQ'it flat on axis, but off-axis it was BAD (a large peak because of the EQ).
I just had to find out what was causing this. I guessed edge diffraction... and I was right. I modeled it the EDGE software, and the model was pretty spot-on, althought the dip was larger in real life than predicted.
It was easy to control this diffraction with foam on the edges, as the pictures show. The foam damping didn't disturb the off-axis response. On 90 degrees the tweeter is almost flat, and 25 dB lower than on axis!
Listening tests showed that there was a slight harshness before, which is cured now. Saxophone could get a little hard sounding, that problem is gone. It also sounds better off-axis.
The next thing to attack is that the dipole peak of the mid's is higher off-axis than on-axis. Even if I EQ it flat on-axis, there is still a peak off-axis. Any ideas of what to to? I dont think foam will work here...
EDGE model:
An externally hosted image should be here but it was not working when we last tested it.
Measurement with and without foam:
An externally hosted image should be here but it was not working when we last tested it.
Pictures... (pretty ugly, or what?
)
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
Nice work Stig!
What you need to do is go to a smaller mid driver (sorry!). An 8" driver, with no baffle, has its 'baffle' peak within its passband - and so above that (approx 700Hz), you get non-ideal dipole response off axis. A 6" driver might work, but smaller is better. Of course, then you get into excursion problems at the low end of the driver.
I've been trying to sort this out. Here is a thread where I ask for help finding appropriate drivers: http://www.diyaudio.com/forums/multi-way/150836-missing-link-600-4khz.html There turned out to be some good ideas, my favorites being the HiVi DMB-A, Galaxy S5N-8 (needs to be measured first), or the SB Acoustics SB15NRXC30-8. But we are almost in 4way territory then, because the low end is limited, esp with the HiVi. I'd do Xmax calculation to see what excursion is needed depending on the crossover frequency.
Today actually, I'm planning on replacing my 8" driver (Dayton RS225) with an Eminence Alpha 6a. It should have enough excursion to do 300-350Hz okay (I hope). Off axis response should be much more regular, and my initial distortion measurement say it can go low enough.
How does that sound?
(EDIT: sorry again, I was going too quick - the driver beaming has a lot to do with it too, perhaps more than the baffle peak)
Last edited:
Okay, I guessed so... the dipole peak of the 8" is in the 12-1300 Hz area, and I CANT cross over the Beyma TPL any lower than 1500 Hz.
I have four Seas Excel 5" ... so I could try the 4-way route. Actually, it will be 5-way, since I have sub's as well. Will measure the 5" to see where the peak comes, but it should be around 2000 Hz - above XO.
I have four Seas Excel 5" ... so I could try the 4-way route. Actually, it will be 5-way, since I have sub's as well. Will measure the 5" to see where the peak comes, but it should be around 2000 Hz - above XO.
Some more changes here again!
When I converted to baffle-less, I noticed that the tweeter suddenly got a dip around 4.5 - 5 kHz that wasnt there with the baffles. I EQ'it flat on axis, but off-axis it was BAD (a large peak because of the EQ).
I just had to find out what was causing this. I guessed edge diffraction... and I was right. I modeled it the EDGE software, and the model was pretty spot-on, althought the dip was larger in real life than predicted.
It was easy to control this diffraction with foam on the edges, as the pictures show. The foam damping didn't disturb the off-axis response. On 90 degrees the tweeter is almost flat, and 25 dB lower than on axis!
Listening tests showed that there was a slight harshness before, which is cured now. Saxophone could get a little hard sounding, that problem is gone. It also sounds better off-axis.
The next thing to attack is that the dipole peak of the mid's is higher off-axis than on-axis. Even if I EQ it flat on-axis, there is still a peak off-axis. Any ideas of what to to? I dont think foam will work here...
Interesting foam idea Stig, thanks!
Yes unfortunately the baffle peak (for the mid) is not at the same place for off axis listening positions – there is not much one can do here - have a look up at John Kreskovky's page for good explanation – he also offers precise modelling tools for OB
Above pix is an overlay from on axis towards 45deg off axis
for my 8" Jantzen in minimalistic OB - I cross around 1500Hz
###########
Additionally you possibly can get some inspiration from what my findings were on the whole issue of integration an AMT with an OB mid
http://www.diyaudio.com/forums/multi-way/100392-beyond-ariel-573.html#post1835857
hop over to the follow ups of mine or have a look at my paper that's basically a compressed write up from all the my postings in Lynn's and other threads on the subject.
Though focusing more on the tweeter where you already found a solution...
http://www.kinotechnik.edis.at/pages/diyaudio/DDCD/DDCD_dipole_horn.html
#######
Cuibono
Interesting – I actually went exactly the other route – upscaling from two 6" (B&C 6PEV13) to two 8" as mids (JA8008) – to my own surprise it was a considerable improvement regarding sonics and seamless merging.
I don't find my soundstage or imaging lacking – quite in contrary – this too got better
Michael
Last edited:
Those sound like good drivers - and my system is evolving exactly the same way - with subs, its getting complicated... but its fun.
Measurements showed that the dipole peak of the 5" Excel is at 1750 Hz. That makes it possible to cross over below the peak. The peak of the 8" happens around 1250 Hz. The response from the 5" is also much smoother 90 degrees off axis. This might be beause of its much smaller neodymium magnet. This is getting really interesting! I will look into building a 4-way dipole using two 5" and two 8". I have all the drivers and amps I need, the only thing missing is a taller frame to hang the thing!
EDIT: Oh by the way, I can try 4-way using the current frame, just replace one of the 8" with a 5".
EDIT: Oh by the way, I can try 4-way using the current frame, just replace one of the 8" with a 5".
Last edited:
Stig Erik, you are amazing - would be great you could do a 0 15 30 45 polar measurement for each setup for us to compare.
By the way, I let my second mid come in some lower (split power configuration) but this should not have any influence on the horizontal directivity you are investigating - I just wanted to mention, as it seems to have some influence on integration and seamless blending to the tweeter AFAICT.
As can be seen from the Edge simu FR polars are still rather parallel till the first dip (~2kHz) - meaning - as a compromise we can use it there as well - my current setup - crossing slightly above the peak - does fine for me.
For nude drivers this may not be true as FR polars go rather wild in that region.
Better anyway - I agree - to leave some "headroom" for the XO if you can..
Michael
By the way, I let my second mid come in some lower (split power configuration) but this should not have any influence on the horizontal directivity you are investigating - I just wanted to mention, as it seems to have some influence on integration and seamless blending to the tweeter AFAICT.
As can be seen from the Edge simu FR polars are still rather parallel till the first dip (~2kHz) - meaning - as a compromise we can use it there as well - my current setup - crossing slightly above the peak - does fine for me.
For nude drivers this may not be true as FR polars go rather wild in that region.
Better anyway - I agree - to leave some "headroom" for the XO if you can..
Michael
Last edited:
Don´t take that magnet size myth for granted. My own experiments showed, that making the magnet more bulky changed almost nothing at 90deg. It was the shape of the basket - how it traps the air behind the cone - and the mounting of the driver (no issue here) which made the bigger difference.
Rudolf
In my case I didn't find "reflections" to conform to the measured peak frequencies. Could it be some sort of Helmholtz resonator instead?
This is what SL says: "The peak is caused by an acoustic filter formed by the basket openings and trapped air between cone and basket. This filter is the reason for the differences in high frequency response between front and rear."
Rudolf
Don't know but speculative I'd say to state a *filter* from the small basket rods (towards the sides as in OB) is rather "optimistic".
A cavity resonance (depth from spider rim to the voice coil) seems the most presumably reason for me - but just guessing.
It could be seen as sort of a axially symmetric - very open - single pleat in AMT operation. One pleat formed by spider with the cone and another one with the magent - not exactly the same so not canceling (spider could be seen as a divider - more simply)
Michael
A cavity resonance (depth from spider rim to the voice coil) seems the most presumably reason for me - but just guessing.
It could be seen as sort of a axially symmetric - very open - single pleat in AMT operation. One pleat formed by spider with the cone and another one with the magent - not exactly the same so not canceling (spider could be seen as a divider - more simply)
Michael
Last edited:
It wouldn't have to be "scientifically" - for quick I use a marker at the floor to roughly place the mic and ARTA's normalizing function - so no equalizing has to be done for excellent comparable plots.
Michael
Last edited:
Which makes it look like this:
An externally hosted image should be here but it was not working when we last tested it.
It sounds significantly more even off-axis than any of the 8" + AMT setups I've had. Will do off-axis measurements tomorrow!
XO frequencies are now 22, 350, 700 and 1700 Hz.
Thank you, but I think I will not take the effort, sorry... I did some off-axis measurements just to find that previous designs were not good enough. If this latest development measures well, I will post graphs.
PS. It sure sounds great!
Check this out - the product of today's work: Two drivers, one an 8" RS225, the other a 6" Alpha 6a, both measured with NO baffle.
I took outdoor measurements, 3m in the air, at a distance of 1.7m, giving a resolution of 85Hz or so. In the following graphs, IGNORE below 200Hz. Each graph shows 0, 30, 45 and 60 degs. What I am looking for is an 'ideal' dipole response, which I define as down 1dB at 30degs, -3dB at 45deg, and -6dB at 60deg. Also notice that each vertical division is 3dB in the graphs below.
It is common to see 8" drivers used for OB systems between 150 and 1500Hz. At issue is the off axis response, and the irregularity caused by the width of the driver. As will be shown, 6" drivers have a much more ideal off axis response compared to 8" drivers, when used in this range. Of course, the caveat with 6" drivers is their limited output. They will be excursion limited below 350Hz or so.
Here is the RS225, the driver I'm currently using:
Two things - first, it's figure of eight pattern is narrower - each off axis measurement is consistently more attenuated that one would expect. The other thing to notice it the drop in level with increasing frequency. I like to look at 45deg. At this angle things should be down -3dB: but instead, between 300-400Hz, the level is -4dB; at 600Hz, it is -5dB; at 900Hz, it is -8dB; and at 1500Hz it is -10dB. If you were trying to have a 'constant directivity' design and maintain -3dB at 45deg, you would have missed it by 7dB at the upper end of the driver's passband. I think it is safe to assume this is audible.
Now, lets look at the 6" driver.
Notice how tight the polar pattern is - up to 1000Hz, there is almost no deviation from the ideal dipole pattern. Then at 1500Hz, you can see there is some loss of regularity - at 45deg, the level is down 7dB, which is an error of 4dB from the ideal. I would say this is a substantial improvement (and tomorrow, when the rest of the drivers come, I'll tell you how it sounds).
So - the bottom line is, if you want to maintain a very regular off axis response, it is possible. My current design is a 3 way, but is still a little compromised - at the top of the midrange driver's passband, it loses a little directivity control, and at the bottom, it will either begin distorting because of over excursion, or will start lobing due to to large a distance between drivers for a higher XO point. Before too long, this system is likely to become a 4 way, using a 5" or smaller driver for an upper mid, and a 10" driver as a lower mid. This has the potential for very constant directivity for the whole system's bandpass, high sensitivity, low distortion and high output. The main compromise will be complexity and cost.
As a final note, the Alpha 6a was a good bit more sensitive in the lower midrange (where it matters a lot), despite its much smaller cone area. It does not have as advanced a motor design as the RS 225 though.
An externally hosted image should be here but it was not working when we last tested it.
I took outdoor measurements, 3m in the air, at a distance of 1.7m, giving a resolution of 85Hz or so. In the following graphs, IGNORE below 200Hz. Each graph shows 0, 30, 45 and 60 degs. What I am looking for is an 'ideal' dipole response, which I define as down 1dB at 30degs, -3dB at 45deg, and -6dB at 60deg. Also notice that each vertical division is 3dB in the graphs below.
It is common to see 8" drivers used for OB systems between 150 and 1500Hz. At issue is the off axis response, and the irregularity caused by the width of the driver. As will be shown, 6" drivers have a much more ideal off axis response compared to 8" drivers, when used in this range. Of course, the caveat with 6" drivers is their limited output. They will be excursion limited below 350Hz or so.
Here is the RS225, the driver I'm currently using:
An externally hosted image should be here but it was not working when we last tested it.
Two things - first, it's figure of eight pattern is narrower - each off axis measurement is consistently more attenuated that one would expect. The other thing to notice it the drop in level with increasing frequency. I like to look at 45deg. At this angle things should be down -3dB: but instead, between 300-400Hz, the level is -4dB; at 600Hz, it is -5dB; at 900Hz, it is -8dB; and at 1500Hz it is -10dB. If you were trying to have a 'constant directivity' design and maintain -3dB at 45deg, you would have missed it by 7dB at the upper end of the driver's passband. I think it is safe to assume this is audible.
Now, lets look at the 6" driver.
An externally hosted image should be here but it was not working when we last tested it.
Notice how tight the polar pattern is - up to 1000Hz, there is almost no deviation from the ideal dipole pattern. Then at 1500Hz, you can see there is some loss of regularity - at 45deg, the level is down 7dB, which is an error of 4dB from the ideal. I would say this is a substantial improvement (and tomorrow, when the rest of the drivers come, I'll tell you how it sounds).
So - the bottom line is, if you want to maintain a very regular off axis response, it is possible. My current design is a 3 way, but is still a little compromised - at the top of the midrange driver's passband, it loses a little directivity control, and at the bottom, it will either begin distorting because of over excursion, or will start lobing due to to large a distance between drivers for a higher XO point. Before too long, this system is likely to become a 4 way, using a 5" or smaller driver for an upper mid, and a 10" driver as a lower mid. This has the potential for very constant directivity for the whole system's bandpass, high sensitivity, low distortion and high output. The main compromise will be complexity and cost.
As a final note, the Alpha 6a was a good bit more sensitive in the lower midrange (where it matters a lot), despite its much smaller cone area. It does not have as advanced a motor design as the RS 225 though.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.