DIY ribbon dipole tweeter, reductio ad minimum

@JonasKarud Titanium is heavier than aluminium, so probably not as good for ribbons.

@matejsirk Yeah, that's what i was looking for !!! :) The one i unwrapped looked almost identical, maybe was also Russian from Soviet era. Do they still make that one?

Edit: Found them online. Only problem is that dielectric is paper in oil, so they should be cleaned well. But otherwise great find, thanks matejsirk!
 
Last edited:
The minimal ribbon produces 75dBSPL at 2kHz at 50cm with 1A current, but remember there's a significant dipole loss. If you look at the response with the amp output you'll see that at 4kHz 15dBV is required for 88dBSPL at 50cm. 15dBV is 5.62V, the series resistor is 2Ω so the dissipation is 15.8W. Now you know why I use a 400W amp for the tweeter :D
The steel frame ribbon designs do not have any dipole loss so much less power is required and because of the reduced excursion requirements they can also be crossed lower (1.5kHz). This is why I also want to experiment with a cardioid, although it's not a dipole anymore a constant directivity could be achieved and the ribbon could be crossed lower.

I'm sitting here doing some contemplating; wonder if it's possible
to use 2-4uM Titanium foil instead of aluminium?
That was the reason I contacted Salomons metalen, I had the same idea :) I don't know if it is stiff enough at 2µm.

When I've finished my new steel frame ribbon with double magnets a proper comparison can be made between the single and dual magnet construction. I hope to be able to show that the double magnet construction requires less corrugation and will work with thinner ribbons while also reducing distortion. IIRC Apogee ribbons are build like this.

regards,

Gerrit
 
Thanks Gerrit, that is very detailed analysis. 400W amp makes sense now :) , but with low prices and great sound of digital amps these days, maybe that route is better than making transformers, which will always affect the sound.
Probably the biggest problem for distortion is uniformity of magnetic field at the ends of the ribbon. From side to side, even the simplest designs have very uniform magnetic field, but making that field the strongest at the points that ribbon attaches to the frame is counter intuitive. I think there should be gradual weakening of magnetic field from some point near the end of the ribbon to the end where it attaches to the frame. On the other hand, maybe i am wrong, ends are the points where ribbon moves the least, so they affect the sound in smallest way.
 
...The signal was measured at the NCore output as I did not dare to put alligator clips on the brass machine screws on the ribbon, if the clips snap off they will be pulled in the gap and the ribbon will be destroyed.
It's possible that the signal at the ribbon clamps is different, I will have to test this with my prototype ribbon when I have my test amp ready again.

You don't need to measure the voltage directly at the ribbon clamps, you can calculate it.
With 2ohm precision resistors in series with the ribbon, just measure voltage drop across this series resistance and divide by 2 to get the current flowing through the ribbon element. This is what the force on the ribbon will be proportional to. If interested, you can calculate the voltage across the ribbon by multiplying the current by the ribbon resistance.
 
...I did manage to locate a screenshot...
Here is another measurement showing comparison of the 11µm and a 2µm ribbon element.
You can see that the 2µm is starting to approach the continuous 6dB/oct slope of a massless ribbon shown back in post #23

I also found a pic of the ribbon structure used for these measurements, which was made from magnets pulled from a burned up Neo8, stacked and glued to a thin plywood base. The corrugation pattern on the ribbon was one that worked well at reducing distortion and response anomalies.
 

Attachments

  • ribbon_mass1-2.png
    ribbon_mass1-2.png
    38.2 KB · Views: 750
  • Neo8-to-Ribbon.png
    Neo8-to-Ribbon.png
    638.3 KB · Views: 736
Thanks Gerrit, that is very detailed analysis. 400W amp makes sense now :) , but with low prices and great sound of digital amps these days, maybe that route is better than making transformers, which will always affect the sound.
Probably the biggest problem for distortion is uniformity of magnetic field at the ends of the ribbon. From side to side, even the simplest designs have very uniform magnetic field, but making that field the strongest at the points that ribbon attaches to the frame is counter intuitive. I think there should be gradual weakening of magnetic field from some point near the end of the ribbon to the end where it attaches to the frame. On the other hand, maybe i am wrong, ends are the points where ribbon moves the least, so they affect the sound in smallest way.
The field from side to side is not constant! Look at the first post in this thread.
FEMM is a 2D simulation tool!

ribbon-motor-new.png

The flux lines crossing the gap are NOT correct as the depth dimension is not taken into account. A separate top view simulation shows what going on:

ribbon-motor-top-single-flux.png

This is not the same model but a crude way to simulate the top view while still closing the magnetic circuit so that the flux at the edge of the gap is about the same as in the previous simulation. It is this variation of the flux density that cause flexing of the ribbon which has to be prevented by corrugating the ribbon. Reduce the variation and you can use thinner foil and/or less corrugation.
The top view for the double magnet construction:

ribbon-motor-top-dual-flux.png

The flux lines are much more parallel in this case. Here's a detailed comparison of the flux density across the gap:

flux_compare.png

This for a 2mm gap between the magnets, the flux density variation is even less than with the 1mm gap in de minimal ribbon.
I use 15cm magnets for a 10cm gap length to keep the flux density constant (more or less) lengthwise.

You don't need to measure the voltage directly at the ribbon clamps, you can calculate it.
With 2ohm precision resistors in series with the ribbon, just measure voltage drop across this series resistance and divide by 2 to get the current flowing through the ribbon element. This is what the force on the ribbon will be proportional to. If interested, you can calculate the voltage across the ribbon by multiplying the current by the ribbon resistance.
I was thinking of the possible induction of the wiring and the effect this could have on the response at the ribbon. Working with transformers the inductance in the secondary circuit is a serious issue but that's with less than 50mΩ impedance perhaps it is not an issue at all with 2Ω.

I made some new one meter long, 23mm wide membranes, etched them down from 0.71 grams to 0.30, applied 3 mm strips ordinary Scotch office tape at the long edges, fiddly work I must say!

Big improvement! well not so big, but it sounds better...cleaner Ssss:s, How do you say, not so sibilant?

I have to do some measurements!

Indeed, less sibilance, same here. Great that you've got some positive results already:)

regards,

Gerrit
 
Here is another measurement showing comparison of the 11µm and a 2µm ribbon element.
You can see that the 2µm is starting to approach the continuous 6dB/oct slope of a massless ribbon shown back in post #23

I also found a pic of the ribbon structure used for these measurements, which was made from magnets pulled from a burned up Neo8, stacked and glued to a thin plywood base. The corrugation pattern on the ribbon was one that worked well at reducing distortion and response anomalies.
Now that's what I call recycling. :)
Very interesting pattern and results. How was the pattern made?

regards,

Gerrit
 
Yes Gerrit, i also used FEMM a lot 10 years ago when I was making magnetic structure for 1m tall ribbons. They had much more complicated steel/magnets design that allows for only 0.004T difference in magnetic field across volume where ribbon is suspended. That took me several months to simulate and make, but at around 16kg, the magnetic structure is heavy and pretty big. That was dipole quasi ribbon design. There was thread here on forum where I published some picture when they were completed. Making quasi ribbon membranes was painfull, but they do tend to sound much less tamed with less sibilance as Jonas said, than pure alu foil.
As I remember, Mr. Bengt Dahlberg first published that double magnets structure on his excellent site.

@JonasKarud I didn't understand where did you put the scotch tape. But I am glad that you also found lighter foil to be less hard on the ears, 2um foil that Bolserst made now intrigues me...

@Bolserst That is awesome pattern on ribbon, very well made! I also tried patterned designs, but they had much denser grooves. This pattern looks like it makes alu foil very rigid, similar to old titanium JBL tweeters. The response is also excellent. Can you tell us how hard is to work with such thin foil?
 
…I was thinking of the possible induction of the wiring and the effect this could have on the response at the ribbon. Working with transformers the inductance in the secondary circuit is a serious issue but that's with less than 50mΩ impedance perhaps it is not an issue at all with 2Ω.
There may, or may not, be significant inductance in the wiring to the ribbon. However, since it is a series circuit, the effect of any inductance will show up in the current measured through the 2Ω resistor.


Very interesting pattern and results. How was the pattern made?
The foil was place on top of some perforated sheet metal and then rolled over with a small paint roller made of dense foam. I tested the stiffening ability of the pattern by gently waving/wiggling the ribbon around in the air from one end and listening to the metallic rustling noises it makes. I kept rolling with more pressure to deepen the pattern until the test produced silence.

… Can you tell us how hard is to work with such thin foil?
I found it extremely difficult. It took several tries to get a 2µm foil mounted without tearing it. Adding a pattern to it was out of the question. I should have mentioned that the ribbon pattern shown in the image was on 11µm foil.

My reason for pursuing 2µm foil was for performing the test to compare with theoretical massless ribbon.
I know 2µm foil is used for microphones, but personally I think it is too fragile to use in a tweeter unless it was backed with mylar or paper.
 

Attachments

  • ribbon_pattern.png
    ribbon_pattern.png
    138.1 KB · Views: 697
Thx for reply. Taking everybody's opinion, i think that all of us agree that optimal thickness for pure aluminium foil ribbons would be in 4-6um range, with some type of corrugation. In the meantime i unrolled some small foil cap found in old CFL lamp, and it also has mylar/alu foil construction, but foil is too thin to unroll it in longer pieces. I can only imagine how fragile is 2um.
Btw, 1.8-2.5um foil can be found on ebay in small pieces.

I must correct myself, your pattern reminds me more of the Pioneer PT-R9 ribbon tweeter. http://img.fmhifi.com/old/sp/201309/1378508341144520586.jpg ;)
 
There may, or may not, be significant inductance in the wiring to the ribbon. However, since it is a series circuit, the effect of any inductance will show up in the current measured through the 2Ω resistor.

Of course, silly me :headbash:


I had made some 2µm foil but as you said it's extremely fragile and I fully agree that it's unusable as such in a ribbon tweeter.

Nice trick with the paint roller :cool:


regards,

Gerrit
 
I thought that you said that, but somehow it seemed too complicated to do. That's something between pure ribbon and quasi ones, i really like your idea! Did you made comparison with same etched down foil to same thickness with and without that tape at the ends? I assume that would make ribbon slower because of bigger mass, but like quasi ribbons it has good sides when it comes to better midrange reproduction and eliminating sibilants as you said.
There are probably many more variations that we didn't try with corrugation and embossing of pure alu foil or quasi ribbons, but probably they all get the similar effect. I assume that larger differences are accomplished by lower mass of ribbon (thinner foil for example) and stronger magnetic field, but shape of the ribbons can only improve sound (distortion) by some small margin (and add rigidity to thin membranes).

"@JonasKarud I didn't understand where did you put the scotch tape."

View attachment 504193

The black strips are the tape strips....
 
That's similar to some 3M spray I used long time ago for gluing mylar and alu foil on quasi ribbon. But I don't know if there is point to doing this, you're adding too much weight to membrane. I' am curious to hear your thoughts after listening to that ribbon. It's probably better to make old fashioned quasi ribbon in the first place, and get added bonus of wanted resistance without transformers, very rigid membrane and higher efficiency compared to pure ribbon with resistors.
I tried some embossing patterns today, and found some perforated aluminium mesh that was painted and had smooth holes. IMG_20150915_214759.jpg Basically, it's the same pattern as Bolserst kindly posted few posts before, and it's done on thick kitchen foil. Tried ones with irregular patterns, very small corrugation, small bumps further apart, but the one on the picture seems most rigid. Need to make ribbon and at least hear how it sounds.
 
" you're adding too much weight to membrane"

Na, that's no problem, You can etch it with the extra strip attached down to a weight much less than a 10 um ordinary membrane....

Which gives me an idea: Mask off a central strip with photo resist on both sides before etching.

This will give a 10uM thick central strip, and thinner foil on the sides, which will higher
current density where its needed.
 
" you're adding too much weight to membrane"

Na, that's no problem, You can etch it with the extra strip attached down to a weight much less than a 10 um ordinary membrane....

Which gives me an idea: Mask off a central strip with photo resist on both sides before etching.

This will give a 10uM thick central strip, and thinner foil on the sides, which will higher
current density where its needed.
Wouldn't the additional force be cancelled by the additional weight?

regards,

Gerrit