DIY ribbon dipole tweeter, reductio ad minimum

Cheer up, Middle, I've tried your scheme with splitting the membrane, and it did not lower the distortion.

In fact, the sound was pretty bad. A better way is to increase the membrane thickness in the middle third of the width of the membrane, increasing the flow of current where it's needed.

But the best sounding membrane I've produced is an ordinary 5 uM one with Gerrits cable corrugator.

A peculiar fact I've found in my experiments, is that a 10 uM membrane sounds almost as good as a 5 uM membrane if you're using a Metglas transformer, and that 5 uM membrane have a tendency to get tiresome in the highest octave when Metglas driven!


Will you show that experiment?
I would like to see if scheme is complete or only cuts or stripes

a better way is....

It is a tale or what?

increase the current is a good one...increase were the tickness is greater is a new one!

Do you mean the current has two different velocity going trought ribbon?
 
What's the reason for making the ribbon 30cm long? what is the intended crossover?

regards,

Gerrit

Hi Gerrit
reason why ribbon is 30cm long is that it sound much better then small one.
As you can see on picture I have Ravens 2. membrane is about 8mm wide but it doesn't sound so good as my 30x2,5 cm. I use double magnets on each side to get homogenous magnetic field. Lundahl transformer used initially but later changed with my own amorphous.
 
Hi Gerrit
reason why ribbon is 30cm long is that it sound much better then small one.
As you can see on picture I have Ravens 2. membrane is about 8mm wide but it doesn't sound so good as my 30x2,5 cm. I use double magnets on each side to get homogenous magnetic field. Lundahl transformer used initially but later changed with my own amorphous.

Any measurements?

My first ribbon was 40cm. I found the narrow vertical directivity to be unacceptable and the ribbon was also very sensitive to draft.
I used purely objective (my Clio system is level calibrated) criteria for evaluation:
- frequency response
- harmonic distortion (k2-k5)
- cumulative spectral decay
- multitone (distortion) spectrum
- directivity (5 degree steps)
Using these measurements I arrived at the current design. I still have some ideas for improvement I'll have to try but it's not the season for driver experiments:)
Here's a multitone spectrum for my ribbon compared to the Seas Millennium tweeter as used in the Linkwitz Orion (my previous system):

ribbon-seas-millennium-multitone.png

Although the levels are not exactly the same the difference is obvious. Guess which one sounds harsher?

regards,

Gerrit
 
Any measurements?

Hi Gerrit
during my experiments with ribbon I only measured frequency response. I remeber that curve was flat but I never saved file.
I've used double magnets 1cm wide with air gap 2mm between and 5micron foil. All this should ensure homogen magnetic field and less bending during movement. And also less distortion. I remember when someone asked Linkwitz why is not using ribbon in their system his answer was distortion. Did you measure and compare distortion between ribbon and Seas?
Does many long peaks in your measurement means harsher? What is your subject opinion which one sound better? You said that your previous system was Orion. Which one is now?

BR//
Tesla
 
If I recall correctly Linkwitz does not use ribbon because of the limited vertical dispersion and the fact that there's no dipole ribbon available. The Raal dipole ribbon is not a true dipole in the sense that the level is 6dB down at 60 degrees, because of the width of the frame the dispersion will be much greater.

Here's the harmonics measurement for the Linkwitz Orion measured at 50cm halfway between the dome and the mid:

orion-50cm-harmonics.png

The harmonics for my ribbon with the Neo10 at 50cm measured mid ribbon (because the distance is so small and it is measured mid ribbon the response is not flat)

Di3-right-4th-ribbon-50cm-mid-ribbon-harmonics.png

As you can see there are a few peaks where the ribbon/Neo10 combination performs worse than the Orion but when it comes to the higher harmonics the ribbon clearly is the better performer.

The multitone spectrum shows both harmonic and intermodulation distortion. The test signal consists of 14 sine waves starting at 1kHz (actually rounded to the nearest FFT bin), these are the peaks above 60dBSPL. Everything below 30dBSPL is distortion, harmonic distortion of the sine waves and also all kinds of intermodulation distortion. If you look at the range between 10kHz and 20kHz you see that the baseline for the ribbon is still visible, with the Seas tweeter there are so many distortion peaks packed close together that the baseline is no longer visible and seem raised by about 20dB at 15kHz.
The 2nd harmonic of the first tone is 2kHz thus everything below 2kHz is intermodulation distortion only. It's a shame that the levels do not match in this area but it certainly looks like the ribbon seriously outperforms the dome when it comes to intermodulation distortion.
All this intermodulation distortion will add harshness to the sound and my guess would be that the closely packed high frequency peaks will result in increased sibilance.

My current system is described here: The Totem of Tone, an active 3-way dipole and active subwoofer
It's a significant improvement over the Orion, especially with respect to directivity, and it is even a bit smaller.

regards,

Gerrit
 
Gerrit
your The Totem project is very nice with many details. Where did you find NEO10? In Europe is impossible to find it. For bass driver I think better choice is speaker with higher Qts (less equalisation). AE is best but also impossible to find in Europe.
Similar to your NEO10 I've started in parallel another project as you can see it here:

http://www.diyaudio.com/forums/planars-exotics/285331-planar-midrange-ribbon.html

How did you compensate bass vibrations in your Totem. I found that plate which hold bass driver vibrate and produce own sound. Therefore my front plate plexiglass is physically isolated from plate which hold bass speakers.
BR//
Tesla
 
Good luck with you project!

The system is one and a half year old now, I got the Neo10 about two years ago from ProRaum but they do not have them anymore.

The baffle with the two woofers is a sandwich made of aircraft quality plywood and lead, four layers of plywood (each 6mm consisting of twelve layers of wood) and 3 layers of 3mm thick lead. With the drivers the baffle weighs 30kg, this in itself reduces the vibrations significantly. Because of the lead there's no resonance in the baffle itself. Mid and high a mounted on a separate frame, not directly coupled to the baffle.

regards,

Gerrit
 
In order to make the magnetic field more linear between poles in a ribbon speaker, i did some experiments with small help magnets with the help of a simulation program, see pictures.

The ordinary magnets are 10x10x40 mm, and the small help magnets are 3x3x40 mm. The help magnets are held in place by their own magnetic force, so no problems there....

Quite an improvement with the help magnets....

I'm going to test this in reality, I'll be back.....
 

Attachments

  • medmags.JPG
    medmags.JPG
    456.7 KB · Views: 577
  • With&Withouthelpmags.jpg
    With&Withouthelpmags.jpg
    98.4 KB · Views: 560
Greetings all

Hello Gerrit and others

Recently read the whole thread. Gerrit, I was very impressed with the symmetry and the waterfall of your creation. Good work. The thread has been quiet for a long time are there any developments ?

I was thinking of possibly using two tweeters one above the other arranged at 90 degrees to each other in a 'T' shape to possible improve the Dipole Dispersion. What do people think ?

James
 
Hello Gerrit and others

Recently read the whole thread. Gerrit, I was very impressed with the symmetry and the waterfall of your creation. Good work. The thread has been quiet for a long time are there any developments ?

I was thinking of possibly using two tweeters one above the other arranged at 90 degrees to each other in a 'T' shape to possible improve the Dipole Dispersion. What do people think ?

James

Thank you :)

The ribbons are still in use and work like a charm so I didn't feel the need to change anything. I've been thinking about the alternatives for the way the ribbon is mounted but haven't come up with an idea worth testing.

As for your suggestion to use two ribbons, I'm afraid you'd end up combining the worst aspects of the directivity in stead of the best and there's the issue of the interference of the two drivers at the lower end. Alas, a ribbon like this will never be a point source otherwise it would be a near perfect HF driver. One could create a line array to 'solve' the vertical directivity but array's have their own problems.

Regards,

Gerrit
 
.....The ribbons are still in use and work like a charm so I didn't feel the need to change anything .....

Hi Gerrit ....
Thank's for sharing your knowledge & craftsmanship .... & disclose your tech/music art.
I listened (yesterday) to your (actif) set up ... breathless.
Congrats ..... chapeau bas ... hoedje af ..... it's very fine music.
(& amazing tech stuff/concept/cabling)
Regards .... Allez, salukes.
Karel
 
I know part of your strategy to address the challenge has been to create magnet configurations that improve the uniformity of the field strength across the gap. I have been meaning to ask, how did you decide on 20mm for the width of your ribbon? When I built my test ribbon from Neo8 magnets I used 3/4"(19mm) gap just because I had some leftover spacers of that dimension already cut out from some ESL experiments.

The reason I ask about your chosen gap width is that the JonasKarud ribbon calculator makes it really easy to see that reducing gap width below 20mm can dramatically improve field uniformity. Not having experimented much with ribbons, my original thought was that there would probably be some optimum gap width for best efficiency since increasing gap width reduced field strength but increased radiating area.

However, this may not be the case. According to the ribbon SPL equations, the far-field SPL is proportional to average field strength and Length, but is not a function of width. So if two ribbons have the same length, but one has half the width of the other what would happen? (Assume for a moment that average field strength was the same.) If driven by the same current, the total force on the ribbon is the same, and the far-field SPL would be the same. This would also mean that the half width ribbon would be moving twice as far as the full width ribbon. Now we know that the field strength will be greater for the half width ribbon so it should be louder than the full width ribbon for the same current…likely by 2dB – 3dB.

So, more uniform magnetic field and no loss in sensitivity? What is not to like? Why not keep reducing the gap width and keep gaining improvements? Well, there is the matter of the doubling of ribbon motion for each halving of ribbon width. Also, there will be doubling of heat dissipated in the ribbon for each halving of width. I’m not sure which is would be the more limiting factor…the heat or the increased ribbon motion.

Ok…that was me just thinking while I was typing.
Has anybody tested ribbon of different width but same length driven with the same current?
If not, I may just have to give it a try to compare SPL and distortion for the same drive current.

Ha I know this is a little late coming but...

I have done this experiment a number of years ago. Basically made a magnet assembly with adjustable gap. It was 100mm long and used neo magnets 12mm deep. Built a number of ribbons from 8mm wide to 20 mm wide. Magnet assembly was set to give 1/2 mm space between ribbon and magnets each time. BTW it is extremely important to get exactly the same space between ribbon and magnets in each test or your data is worthless.
To get same power through each ribbon a 10 ohm resistor was in series with each. No transformer.

In the end you get the same sensitivity from all of them, BUT lower distortion from the larger surface area units. This was done with 4 foils. 4 micron, 9 micron, 12 micron, 18 micron. Also a number of ribbon configurations..flat,corrugated, embossed, pure foil, foil/film laminates etc etc etc
 
Last edited:
BTW more on above...

I do not see a dropoff in HF response to be related to ribbon width as some have mentioned. This drop off was strongly dependent on ribbon mass.

Also its interesting to me that we all seem to believe that distortion should be more of an issue as the magnets are spaced farther apart as in a wider ribbon. The assumption is that the field is less uniform so distortion should increase. However I have not seen this to be so in practice. Perhaps distortion does increase a bit?? BUT maybe is offset by the lower ribbon movements?? I consistently get lower distortion with wider ribbons reguarless of magnet geometry's. Within reason of course.
 
…I have done this experiment a number of years ago…To get same power through each ribbon a 10 ohm resistor was in series with each…it is extremely important to get exactly the same space between ribbon and magnets
Thanks for sharing your experience.
Based on your website and patent, would it be a good assumption that your experiments were for a closed back ribbon not a dipole? If so, I need to dig up formulas for SPL of closed back ribbons.

Were you really measuring at same power? Or did you mean same current thru the ribbon.
Yes, the gaps between ribbon and magnet can have a pronounced effect on SPL . Did you consider comparing output for gaps that were sized proportional to the ribbon width so the gap effect would be similar for each ribbon? A 0.5mm gap has more effect on the SPL of an 8mm width ribbon that a 20mm.

I do not see a dropoff in HF response to be related to ribbon width as some have mentioned. This drop off was strongly dependent on ribbon mass.
Agreed. For ribbon microphones width affects HF response, but not for ribbon loudspeakers.
It is the mass of the ribbon relative to the airload resistance that causes the HF roll-off.

The assumption is that the field is less uniform so distortion should increase. However I have not seen this to be so in practice. Perhaps distortion does increase a bit?? BUT maybe is offset by the lower ribbon movements?? I consistently get lower distortion with wider ribbons reguarless of magnet geometry's. Within reason of course.
This makes reasonable sense…minimizing motion being more important than field linearity in a transducer that doesn’t really have much displacement. Your mention of the importance of gap sizes reminded me I had read in some ribbon patents that gap size affected 3rd harmonic distortion as well. Most likely that has to do with differing airload and resulting motion on the edges of the ribbon vs the middle of the ribbon. Is this something you have experimented with? The patent verbage implied there was an optimum gap size.
 
Last edited:
Thanks for sharing your experience.
Based on your website and patent, would it be a good assumption that your experiments were for a closed back ribbon not a dipole? If so, I need to dig up formulas for SPL of closed back ribbons.

Good question. Actually those measurement were dipole. I did do some quick and dirty measurements closed box, however I cannot be sure of these as they were on the fly and were not set up with care ;)

Were you really measuring at same power? Or did you mean same current thru the ribbon.Yes, the gaps between ribbon and magnet can have a pronounced effect on SPL . Did you consider comparing output for gaps that were sized proportional to the ribbon width so the gap effect would be similar for each ribbon? A 0.5mm gap has more effect on the SPL of an 8mm width ribbon that a 20mm.

My intention was to have same power and current. Input same voltage into same resistance. While there are variations in ribbon resistance I assume this was swamped out by the rather large in comparison 10 ohm series resistor.The proportional gap thing is a good question. I did not do a careful examination of this but was wondering the same thing.


Agreed. For ribbon microphones width affects HF response, but not for ribbon loudspeakers.
It is the mass of the ribbon relative to the airload resistance that causes the HF roll-off.

Yes. BTW what do you think of the effect of the magnet slot depth on that airload and HF response?


This makes reasonable sense…minimizing motion being more important than field linearity in a transducer that doesn’t really have much displacement. Your mention of the importance of gap sizes reminded me I had read in some ribbon patents that gap size affected 3rd harmonic distortion as well. Most likely that has to do with differing airload and resulting motion on the edges of the ribbon vs the middle of the ribbon. Is this something you have experimented with? The patent verbage implied there was an optimum gap size.


Yes 3rd harmonic is the one effected by gap. This one is presently keeping me up at night as I am now developing a smaller ribbon capable of use down to 1 Khz. I loose sleep over it because at the moment I have two versions of the same ribbon. One with 90db sens BUT that damn 3rd starts to rise below 3khz. The other is 86 db sens but keeps 2nd - 5th below .5% across the window. I suppose 86 is still quite useful in many systems but would rather go to market with 90.
I have experimented a bit with this but theres much to learn and it looks complex. At first I suspected the distortion was more to do with the sharply rising magnet field strength as the ribbon edge was taken closer to magnet face. I assumed the sharp rise in field strength there was causing the ribbon to flex hard at the edge. Again the assumption that less than uniform field was to blame. However I made a version where the exact same ribbon was installed in a magnet assembly with the exact same geometry BUT there were plastic spacers between the magnets and the ribbon 1/4 inch thick. Now we have a much more uniform field across the ribbon and the same exact gap between the ribbon edge and the plastic placer face as we had before between ribbon and magnet face. The results were the same as with no spacers. Big gap between ribbon and surrounding structure gave low distortion. Small gap showed rise in 3rd harmonic.
If there is an "optimum" gap I am not sure. In my experiments so far it boils down to a decision between lower distortion/lower sensitivity, and higher distortion/ higher sensitivity.

Now heres what Im loosing sleep over. Neather I nor anyone else I have used in listening tests can hear the difference between the low distortion version and the "higher" distortion version. Im almost embarrassed to admit it. We all can easily hear the differences when changing to different diaphragm designs, for example the difference between a foil only design and a laminated foil/ film design and some others, BUT we cannot hear a difference between the low and "high" distortion versions within each diaphragm design. The coloration of each type is easily heard, BUT we cannot hear a difference between the low and higher distortion versions within each type.
There seems to be more to the distortion argument. Perhaps its the specific distortion profile that is responsible ? 2nd 4th and 5th are very low and 3rd is not a sharp bump but rather a smooth steady rise starting at about 3khz.
 
Last edited: