I've been contemplating for a long while now about building ribbon speakers. I am of the opinion that ribbons are the most superior method of audio reproduction at the current time although I have not actually heard a ribbon yet 
Ribbons are among the lightest possible diaphragms and so logically must be among the best methods of reproduction. They also seem to have the least amount of issues that prevent them from reaching their potential compared to other technologies.
When I design amplifiers I try to consider every possible deficiency in the design and attempt to solve it. This leaves my designs to be both pure and effective without being overly complex. I would take the same approach for speaker design. Not only do I want to design a full range ribbon speaker but I want it to be as close to 0 ohms as possible.
This removes the need for voltage amplification and vastly reduces the effects of capacitance and distortion within the electrical systems. The remaining unwanted parasitic traits that plague practical systems are inductance and resistance. For this reason I will be using a current source amplifier instead of a voltage amplifier. It would be very difficult to get a voltage amplifier to function into a close-to-zero-ohm load anyway. The current source amp can direct couple to the ribbon without issue. I'll post the current source amp design here soon once I finish tweaking it in spice.
In order to get as close to zero ohms as possible I would need to have a parallel ribbon configuration. This combined with the fact that smaller ribbons should have less distortion (correct me if I'm wrong) leads me to believe that a line array type system is in order. I'm thinking something along the lines of this
but with much much thinner magnets in between the ribbons to prevent comb filter distortion. Thinner magnets means reduced sensitivity but I only really care about performance and sound.
I've read that ribbons have a problem with vertical dispersion so I am thinking about using a configuration like this to combat the issue
The issue I have with this type of design is mainly that I am a bit stuck on the design and construction of the physical driver, I have never done this before. I have attempted to purchase a 3d printed model of an array structure as shown above but the materials I've used are not stiff enough I think I would have to order a metal printed structure and that is expensive when I am shooting in the dark with design.
I have some cool ideas for a (much) lighter than air suspension for a non corrugated ribbon design but it involves expensive materials so I think I will save for that another thread since my budget is only about $2000 for this build.
I am limited in knowledge about the sciences involved in speaker design so let me know if there is some reason these ideas are not good.
I'm going to leave this thread open as my official thread for this project so discussion is welcome.
Ribbons are among the lightest possible diaphragms and so logically must be among the best methods of reproduction. They also seem to have the least amount of issues that prevent them from reaching their potential compared to other technologies.
When I design amplifiers I try to consider every possible deficiency in the design and attempt to solve it. This leaves my designs to be both pure and effective without being overly complex. I would take the same approach for speaker design. Not only do I want to design a full range ribbon speaker but I want it to be as close to 0 ohms as possible.
This removes the need for voltage amplification and vastly reduces the effects of capacitance and distortion within the electrical systems. The remaining unwanted parasitic traits that plague practical systems are inductance and resistance. For this reason I will be using a current source amplifier instead of a voltage amplifier. It would be very difficult to get a voltage amplifier to function into a close-to-zero-ohm load anyway. The current source amp can direct couple to the ribbon without issue. I'll post the current source amp design here soon once I finish tweaking it in spice.
In order to get as close to zero ohms as possible I would need to have a parallel ribbon configuration. This combined with the fact that smaller ribbons should have less distortion (correct me if I'm wrong) leads me to believe that a line array type system is in order. I'm thinking something along the lines of this
but with much much thinner magnets in between the ribbons to prevent comb filter distortion. Thinner magnets means reduced sensitivity but I only really care about performance and sound.
I've read that ribbons have a problem with vertical dispersion so I am thinking about using a configuration like this to combat the issue
The issue I have with this type of design is mainly that I am a bit stuck on the design and construction of the physical driver, I have never done this before. I have attempted to purchase a 3d printed model of an array structure as shown above but the materials I've used are not stiff enough I think I would have to order a metal printed structure and that is expensive when I am shooting in the dark with design.
I have some cool ideas for a (much) lighter than air suspension for a non corrugated ribbon design but it involves expensive materials so I think I will save for that another thread since my budget is only about $2000 for this build.
I am limited in knowledge about the sciences involved in speaker design so let me know if there is some reason these ideas are not good.
I'm going to leave this thread open as my official thread for this project so discussion is welcome.
Last edited:
You mean this one? Ribbon question This is a more specific thread for this proposed design, the other thread was not really about this. I was just asking broad questions in the other one, I also forgot it even existed. I'm pretty dead set on making a low impedance full range ribbon so this thread is meant to document and discuss this specific design.
Last edited:
I've been thinking about how to build "modules" of small ribbons that can be stacked and expanded and paralleled.
I'm thinking maybe I can clamp a longer ribbon at certain points along its length in order to mechanically separate it into smaller ribbons.
In order to make the bezel as small as possible to prevent comb filter distortion I would probably use the magnets as the structure itself using them to clamp the ribbon between two slices of magnets.
Like this
A complete "driver" might look like this
The problem is that when I attempt to sandwich the magnets together they will be repelled by like-magnetic fields. I need to find a way to lock them onto each other.
What do you think? I cannot think of another way to make a stackable array of small ribbons.
Also how do I calculate at what width ribbons must be from each other to prevent comb filtering?
I'm thinking maybe I can clamp a longer ribbon at certain points along its length in order to mechanically separate it into smaller ribbons.
In order to make the bezel as small as possible to prevent comb filter distortion I would probably use the magnets as the structure itself using them to clamp the ribbon between two slices of magnets.
Like this
A complete "driver" might look like this
The problem is that when I attempt to sandwich the magnets together they will be repelled by like-magnetic fields. I need to find a way to lock them onto each other.
What do you think? I cannot think of another way to make a stackable array of small ribbons.
Also how do I calculate at what width ribbons must be from each other to prevent comb filtering?
well i never have problems getting my ribbons to low in resistance so i dont see the problem whats the question ? hehe , jokes asside most people cant find 3 micron alu foil ribbons of 1 meter.. but using thicker foil you can get your wanted resistance with ease, and still light enough to have a good response to 20khz. i think you just have to invest so much in magnets it is no fun at all since ribbons are efficient with all those neo's it still becomes a nasty expensive business if you want to create surface area needed for low end without letting ribbons flap all over the place.
i once thought of making an array with rather weak magnets, just to be able to go low, so every ribbon only moves a tiny bit. that insane efficiency is not needed if you are gone use 40+ anyways
so i dont see the problem whats the question ? hehe , jokes asside most people cant find 3 micron alu foil ribbons of 1 meter.. but using thicker foil you can get your wanted resistance with ease, and still light enough to have a good response to 20khz. i think you just have to invest so much in magnets it is no fun at all since ribbons are efficient with all those neo's it still becomes a nasty expensive business if you want to create surface area needed for low end without letting ribbons flap all over the place.i once thought of making an array with rather weak magnets, just to be able to go low, so every ribbon only moves a tiny bit. that insane efficiency is not needed if you are gone use 40+ anyways
I know there are many people here are are knowledgeable on the physics of driver design. Currently I'm wondering if my magnet array idea has merit. I also need to know the spacing requirements to avoid comb filtering.well i never have problems getting my ribbons to low in resistance
Haha good point.that insane efficiency is not needed if you are gone use 40+ anyways
I will still need a build methodology in order to create an array though. Keeping the ribbons as close as possible seems to be key. The only thing I can think of in order to achieve this is the above magnet array configuration I posted.
i would just use strips of magnet. meters long rubber magnet that is, and then the thickest i could find. its not strong but thats not key, key is having low excursion and allot of it.
i use 3 mm thick rubber magnet allot, and as well 4mm thick. and i can get it up to 6mm thick. still weak though. but its all about surface area. pahsing is not an isue since you dont want to play fulrange on 4000 ribbons since it acts as a huge planar, so dispersion will be terible if you play it fullrange
i use 3 mm thick rubber magnet allot, and as well 4mm thick. and i can get it up to 6mm thick. still weak though. but its all about surface area. pahsing is not an isue since you dont want to play fulrange on 4000 ribbons since it acts as a huge planar, so dispersion will be terible if you play it fullrange
Could you elaborate on how you would build this array? What is the advantage of your suggested configuration over mine?
It seems like mine would be superior because custom magnets can be made less than 1 mm thick which means many ribbons can be arrayed right next each other. But then again I don't completely understand your suggestion.
The dream would be for me to create a modular expandable array.
It seems like mine would be superior because custom magnets can be made less than 1 mm thick which means many ribbons can be arrayed right next each other. But then again I don't completely understand your suggestion.
The dream would be for me to create a modular expandable array.
well with the idea you wont combat the dispersion. it is the size that emits high frequency that is the problem so no mater how you put them horizontal or vertical it does not mater.
if the size is small dispersion will be better the bigger you make the combined emiting surface (without using angles) the worse it gets. thats the reason why they are so small in width to have at least good horizontal dispersion. so you will end up with a ribbon tweeter, and allot of ribbons that do the rest. but to do the rest there are beter and more efficient option to do so
if the size is small dispersion will be better the bigger you make the combined emiting surface (without using angles) the worse it gets. thats the reason why they are so small in width to have at least good horizontal dispersion. so you will end up with a ribbon tweeter, and allot of ribbons that do the rest. but to do the rest there are beter and more efficient option to do so
So you're saying a bigger tweeter array will give worse dispersion than a smaller one? When you say size do you mean width, length, or total area? If I were to do this it would be a multi way array using active crossovers.
You said there are better options for "the rest"? You're saying ribbons are inferior at lower frequencies even with lots of surface area? I think I must misunderstand, one consistency I've seen is that people claim full range ribbons are among the best sounds things they've ever heard.
With such low mass they have greater control and lower distortion without the sound degrading issues of planar. Surely a massive full range ribbon is superior to other technologies?
You said there are better options for "the rest"? You're saying ribbons are inferior at lower frequencies even with lots of surface area? I think I must misunderstand, one consistency I've seen is that people claim full range ribbons are among the best sounds things they've ever heard.
With such low mass they have greater control and lower distortion without the sound degrading issues of planar. Surely a massive full range ribbon is superior to other technologies?
Last edited:
I mean indeed width length. which gives you the surface area BIG is bad for dispersion up high. i think max one wave length is often used, as for tweeters max 1.7cm up to 20khz. of course you can use bigger and have slightly different dispersion. but going fullrange on a big panel is not pleasant. yeah about the options i think it is almost impossible to make a true ribbon go low. and using allot of small ones to have any magnetic field is an option but costly and a hell of a work to make allllllll those ribbons, and even then question remains if you will blow them out of the gap with each low note. in theory yes best way. practical absolutely not. why not go for ESL ? you can make an esl with lighter and thiner membrane then a ribbon.
i wonder by the way how many people have heard a full range ribbon? since apogee is not one of them for instance. also just a planar , bit more fancy looking but still no ribbon in the low end, same goes for the carvers. also woofers or planar. as far as i know
BIG is bad for dispersion up high. i think max one wave length is often used, as for tweeters max 1.7cm up to 20khz. of course you can use bigger and have slightly different dispersion. but going fullrange on a big panel is not pleasant. yeah about the options i think it is almost impossible to make a true ribbon go low. and using allot of small ones to have any magnetic field is an option but costly and a hell of a work to make allllllll those ribbons, and even then question remains if you will blow them out of the gap with each low note. in theory yes best way. practical absolutely not. why not go for ESL ? you can make an esl with lighter and thiner membrane then a ribbon.i wonder by the way how many people have heard a full range ribbon? since apogee is not one of them for instance. also just a planar , bit more fancy looking but still no ribbon in the low end, same goes for the carvers. also woofers or planar. as far as i know
Last edited:
So maybe a more traditional ribbon tweeter and a larger mid/bass?I mean indeed width length. which gives you the surface area
I suppose I can use a cone woofer with servo feedback without it being a huge deal. It's mostly the treble and midrange I'm concerned with.
Esl is expensive and difficult to make. There are a lot ways it can go wrong and you need to have just the right tension using just right materials, ect, ect. Ribbons just "work". Plus ESLs require extremely high voltages and are a reactive load which comes with its own non linearities and issues. I'm sure this is only touching the surface on their issues. I recall doing some reading on ESL a while back and it is a deep subject to get involved in. Which just means there are a lot of compromises. I don't like compromises.
Part of the reason I want to make this ribbon array is so will not require voltage amplification at all. In a pure parallel configuration with thick interconnects there would be barely any resistance and thus require almost no voltage. I would aim for milivolts. This means the amplifier and DAC will have almost no voltage variation which also means better performance.
I haven't done the math but a huge array such as this in parallel would make the ribbons almost 0 ohms so the interconnects will make up the majority of the resistance. The impedance will be flat as ruler and the sheer surface area will make each individual ribbon produce vanishingly low distortion upon their already extremely low distortion.
I can't even see a comparison between ESL and ribbons, ribbons seem to win every time, assuming a proper array can be made.
Not many people I'm sure. I only know of a few people I've talked to that have heard such a thing and I'm pretty sure it was DIY.
Last edited:
Hi hellokitty123,
Of course you realise you're just trading one set of problems for another by trying to go all current without voltage gain. Now inductance is your problem instead of capacitance. That and the extreme energy loss that high current causes. That's why current technology is a mid range impedance where everything is controlled a little easier. With your plan you will find that connections will become a major problem instead of being just one small concern.
At the end of the day, executing your plan will require voltage gain to overcome connection and wire impedance. All of a sudden speaker wire lead length becomes a variable that matters a great deal. On top of that, if your ribbons heat up, the resistance will also increase.
All I can say is that there is a reason why the state of the art is what it is. To each his own I guess.
-Chris
Of course you realise you're just trading one set of problems for another by trying to go all current without voltage gain. Now inductance is your problem instead of capacitance. That and the extreme energy loss that high current causes. That's why current technology is a mid range impedance where everything is controlled a little easier. With your plan you will find that connections will become a major problem instead of being just one small concern.
At the end of the day, executing your plan will require voltage gain to overcome connection and wire impedance. All of a sudden speaker wire lead length becomes a variable that matters a great deal. On top of that, if your ribbons heat up, the resistance will also increase.
All I can say is that there is a reason why the state of the art is what it is. To each his own I guess.
-Chris
In a huge parallel ribbon array using thick interconnects and a current source amplifier the inductance issue should be negligible at best, I would imagine. Normal amplifiers have both capacitance and inductance to deal with as well as voltage and current device lienarities. A current source amplifier ignores back EMF and parasitics in general and into a near short the device linearities are effectively current dependent only. Wire impedance is a relatively straightforward problem to solve. I think the bigger issue is ensuring the current gets shared equally across the ribbons. A tiny mismatch in wire length can mean a big difference in current sharing when resistances get so low.
If my understanding is correct, SPL in a ribbon is not effected by power but current only. The magnetic field across the ribbon for a given current will be the same regardless of the resistance. Meaning the ribbons should require almost no effective power as resistance becomes close to zero as current requirements do not increase in a parallel configured ribbon over a series configured ribbon but the voltage requirements decrease greatly.
Instead of trading a set of problems, I just reduced them.
If my understanding is correct, SPL in a ribbon is not effected by power but current only. The magnetic field across the ribbon for a given current will be the same regardless of the resistance. Meaning the ribbons should require almost no effective power as resistance becomes close to zero as current requirements do not increase in a parallel configured ribbon over a series configured ribbon but the voltage requirements decrease greatly.
Instead of trading a set of problems, I just reduced them.
Last edited:
With very low impedance any inductance will be a serious problem as it will act as a low pass filter. With a traditional ribbon tweeter the transformer has to be as close as possible to the ribbon to reduce the inductance of the circuit. Here's a picture of a transformer driven ribbon I made:
What looks like a single strip of copper foil are actually two strips separated by very thin polyester foil. This is done to keep the inductance as low as possible. Here's a picture of the test jig:
The ribbon is simulated by a 50mΩ Caddock film resistor. If the two foil strip are separated by 1mm thick plastic in stead of the thin foil the HF response drops by 30% !!! For my minimal ribbon design I decided to drop the transformer and just use a series resistor and a powerful amp.
The series resistor is 2Ω consisting of 4 0.5Ω 100W film resistors. You might think this is a space heater but it isn't because the high frequency audio (>2kHz) is mostly peaks (up to 16 amps). In normal use the resistors don't even get lukewarm. With the series resistor the ribbon is essentially current driven as the series resistor is 2Ω and the ribbon just 30mΩ. The non-inductive film resistors are also a very benign load for the amp, near perfectly resistive. The minimal ribbon is described in this thread: DIY ribbon dipole tweeter, reductio ad minimum
What looks like a single strip of copper foil are actually two strips separated by very thin polyester foil. This is done to keep the inductance as low as possible. Here's a picture of the test jig:
The ribbon is simulated by a 50mΩ Caddock film resistor. If the two foil strip are separated by 1mm thick plastic in stead of the thin foil the HF response drops by 30% !!! For my minimal ribbon design I decided to drop the transformer and just use a series resistor and a powerful amp.
The series resistor is 2Ω consisting of 4 0.5Ω 100W film resistors. You might think this is a space heater but it isn't because the high frequency audio (>2kHz) is mostly peaks (up to 16 amps). In normal use the resistors don't even get lukewarm. With the series resistor the ribbon is essentially current driven as the series resistor is 2Ω and the ribbon just 30mΩ. The non-inductive film resistors are also a very benign load for the amp, near perfectly resistive. The minimal ribbon is described in this thread: DIY ribbon dipole tweeter, reductio ad minimum
Ah yes I forgot about the low pass filter factor. A current source amplifier will brute force through the RL filter but at the expense of increased voltage swing which I don't want.
How much inductance could thick interconnects in twisted pair config possibly have?
Doing some quick internet math, 7 gauge wire @ 1FT is sufficient to pass 20khz into a 0.1 ohm load. This doesn't count using smaller gauge wires in a twisted pair configuration that equate to a larger gauge. Twisted pair config should significantly reduce gauge required.
How much inductance could thick interconnects in twisted pair config possibly have?
Doing some quick internet math, 7 gauge wire @ 1FT is sufficient to pass 20khz into a 0.1 ohm load. This doesn't count using smaller gauge wires in a twisted pair configuration that equate to a larger gauge. Twisted pair config should significantly reduce gauge required.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.