Dynamic displacement vs resistance vs temperature measurement for ribbon transducer

[thadman]

Hello DIYAudio members,

I am an engineering student at Purdue University and am interested in doing an experimental analysis of ribbon transducers to contrast with simulations. This data could be made available to the community to aid in transducer development.

A ribbon transducer consists of a rectangular element of metal foil suspended within a magnetic gap and clamped at its ends (ie a clamped-clamped membrane). Current is applied to the foil, which produces an electromagnetic field that interacts with the magnetic field generated by the permanent magnets. The foil bends in response to this force.

Assuming it is used as a loudspeaker, current is applied to the foil to create sound pressure. The loudspeaker is <10% efficient due to the poor membrane-fluid coupling, which results in a majority of the energy being converted to heat. Resistance increases with temperature, which results in clipping of the waveform. I'd like to measure this effect.

A modal analysis could be conducted with the use of a scanning laser vibrometer system. This would appear sufficient for the task.
PSV-400-3D Polytec Scanning Vibrometer - Features

A dynamic displacement vs resistance vs temperature measurement of the foil would appear to be valuable. However, I'm not sure what other instrumentation would be required or how to set up such an experiment.

Any thoughts?

Best Regards,
Thadman

 

[Steve Dunlap]

Honestly, if you have to explain to someone what a ribbon speaker is, they are not likely to have answers to your questions.

Addressing your points of study:

Dynamic displacement will only be caused by the input frequency, a DC offset or the movement of air around the ribbon. If we assume the second two can be eliminated or minimized to an acceptable level, that leaves only frequency. The lower the frequency, the more displacement.

The temperature of the foil element (assuming it is Al, not nichrome) is very unlikely to change to any significant degree with normal music input. The reasons for this are 1. There is very little actual power delivered to the driver at higher frequencies. 2. The foil is in free air and moving very rapidly, enhancing heat dissipation. 3. Thin Al foil cools very rapidly when the heat source is removed. There is simply not enough mass to store the heat energy.

Points 1 and 3 will not necessarily apply for high power steady state testing, or at very low frequencies.

My thoughts. Is that a good place to start?

 

[thadman]

Honestly, if you have to explain to someone what a ribbon speaker is, they are not likely to have answers to your questions.

Addressing your points of study:

Dynamic displacement will only be caused by the input frequency, a DC offset or the movement of air around the ribbon. If we assume the second two can be eliminated or minimized to an acceptable level, that leaves only frequency. The lower the frequency, the more displacement.

The temperature of the foil element (assuming it is Al, not nichrome) is very unlikely to change to any significant degree with normal music input. The reasons for this are 1. There is very little actual power delivered to the driver at higher frequencies. 2. The foil is in free air and moving very rapidly, enhancing heat dissipation. 3. Thin Al foil cools very rapidly when the heat source is removed. There is simply not enough mass to store the heat energy.

Points 1 and 3 will not necessarily apply for high power steady state testing, or at very low frequencies.

My thoughts. Is that a good place to start?

Excellent response

I do not expect the transducer to be stressed with normal listening levels. However, I would like to differentiate between the different types of foil as well as different motor structures. I believe a high power transient as well as high power steady state analysis would offer the opportunity for significant contrast.

A thicker membrane will be more invariant to temperature changes due to its higher thermal mass, however a thinner membrane will dissipate heat more quickly. Beryllium also offers potential as a substitute for the membrane element. It's coefficient of thermal expansion, specific heat, wear resistance, and internal damping are much superior to those inherent to aluminum.

I have contacted Brush Wellman's electrofusion division and they have expressed interest in testing a variety of their IF-1 and PF-60 beryllium foils.

In addition, I'd like to test 4 motor structures with 2 different gaps.

Samarium Cobalt 32H w/ 8mm gap, Neodymium 48H w/ 8mm gap, Samarium Cobalt 32H w/ 16mm gap, Neodymium 48H w/ 16mm gap, Samarium Cobalt 32H/Hiperco50 w/ 8mm gap, Neodymium 48H/Hiperco50 w/ 8mm gap, Samarium Cobalt 32H/Hiperco50 w/ 16mm gap, and Neodymium 48H/Hiperco50 w/ 16mm gap.

^The bessel function defining radiation suggests an 8mm gap width to provide a uniform distribution of energy wrt angle to 20khz and a 16mm gap width to provide a uniform distribution of energy wrt angle to 10khz. The gap will generate lobes if ka>3.83 (ie the gap width is greater than 3.83 times the wavelength).

The properties of the magnetic structure will vary with temperature, specifically permeability. Samarium Cobalt is more linear than Neodymium with regards to temperature, however the higher field strength imparted by the neodymium will require less current passed through the foil for a comparable displacement.

The addition of the ferrous materials within the structure may also present a reactive load to the amplifier.

A FLIR high speed infrared camera may suffice for thermal measurements. For transients, I assume the temperature will change very rapidly. However, I'm not sure what refresh rate would be required to capture this event.

What waveforms would you recommend for transient/steady state analysis? dirac pulse? 100khz square wave? white noise?

I would like to apply a signal which produces the maximum repeatable displacement (thermal/mechanical). I expect a filter must be applied to restrict over-excursion. However I am finding the maximum thermal limit to be non-trivial.

For each experiment, 25 samples would seem to be the lower limit for statistical correlation. If 8 motor structures were used, 200 membrane samples would have to be measured for each experiment

The cost of the aforementioned motor structures will exceed >$1000. I may have to apply for grant money and/or persuade a magnet supplier. Any thoughts on suppliers?

 

[Steve Dunlap]

^The bessel function defining radiation suggests an 8mm gap width to provide a uniform distribution of energy wrt angle to 20khz and a 16mm gap width to provide a uniform distribution of energy wrt angle to 10khz. The gap will generate lobes if ka>3.83 (ie the gap width is greater than 3.83 times the wavelength).

I would agree with this. My ribbons are 0.25" wide and provide a uniform distribution of energy to 20KHz.

The properties of the magnetic structure will vary with temperature, specifically permeability. Samarium Cobalt is more linear than Neodymium with regards to temperature, however the higher field strength imparted by the neodymium will require less current passed through the foil for a comparable displacement.

Why do you think the magnet structure will change temperature? I would not expect it to change much from room temperature.

The addition of the ferrous materials within the structure may also present a reactive load to the amplifier.

I don't think so.

FLIR high speed infrared camera may suffice for thermal measurements. For transients, I assume the temperature will change very rapidly. However, I'm not sure what refresh rate would be required to capture this event.

What waveforms would you recommend for transient/steady state analysis? dirac pulse? 100khz square wave? white noise?

I don't think I can give a qualified answer to the use of the camera. 100KHz sounds a bit high for an audio transducer. White noise should work to generate the desired thermal rise. You might find this interesting also.

http://www.diyaudio.com/forums/full-range/156466-one-only-full-range-amt-2.html#post2010628

I would like to apply a signal which produces the maximum repeatable displacement (thermal/mechanical). I expect a filter must be applied to restrict over-excursion. However I am finding the maximum thermal limit to be non-trivial.

A square wave should do this. I don't believe a filter will be necessary if you are using a single frequency.

It has been several years since I worked on making my own ribbons. I had to give up doing any physical work. Here is a link to one magnet supplier. The others I had bookmarked seem to no longer be there. A Google search should turn up plenty more.

https://unitednuclear.com/index.php?main_page=index&cPath=70_79

Good luck with all that testing.

 

[thadman]

Why do you think the magnet structure will change temperature? I would not expect it to change much from room temperature.

Ribbons are inherently displacement limited transducers, thus the applied signal must be restricted in amplitude at low frequencies. However, massive power handling (>1kw) may be realized if the bandwidth is restricted to higher frequencies (>1khz). I do not expect a dipole linesource to be stressed with levels exceeding 130dB above 1khz.

Due to the surrounding fluids (air) very low thermal conductivity (.0243 W/m-K), I believe the primary conduit for heat transfer may exist at the membrane/structure interface. Ribbons are inherently inefficient due to the poor membrane-fluid coupling. If we assume <10% efficiency, >900 watts may result as heat if a 1kw signal was applied. Assuming a high power steady state analysis, a significant amount of thermal energy may pass through the membrane/structure interface and permeate the structure. I expect this to effect the magnetic field, which is temperature dependent.

The transducers could potentially be tested at a variety of levels to establish power handling.

>100hz: 90dB,100dB
>200hz: 90dB,100dB,110dB
>300hz: 90dB,100dB,110dB,120dB
>500hz: 90dB,100dB,110dB,120dB,130dB
>1khz: 90dB,100dB,110dB,120dB,130dB,140dB

For a comparative analysis, would it be appropriate to apply a constant signal (ex. 100 watts) or provide a signal which produces a constant displacement (ex. 1mm)?

For constant power handling (steady-state), would white noise be appropriate? How long would it take to achieve a steady state (thermal/mechanical)?

I don't think so.

I assumed the ferrous components would inductively couple to the membrane and function as a transformer, which may include a reactive component. Is that assumption erroneous?

I don't think I can give a qualified answer to the use of the camera. 100KHz sounds a bit high for an audio transducer. White noise should work to generate the desired thermal rise. You might find this interesting also.

http://www.diyaudio.com/forums/full-range/156466-one-only-full-range-amt-2.html#post2010628

Most interesting. AMTs are a very interesting concept.

A square wave should do this. I don't believe a filter will be necessary if you are using a single frequency.

Good luck with all that testing.

How can a square wave be a single frequency? I was under the impression it was composed of a fundamental + odd order components.

Thanks for the link, your input is very appreciated

 

[Steve Dunlap]

Due to the surrounding fluids (air) very low thermal conductivity (.0243 W/m-K), I believe the primary conduit for heat transfer may exist at the membrane/structure interface. Ribbons are inherently inefficient due to the poor membrane-fluid coupling. If we assume <10% efficiency, >900 watts may result as heat if a 1kw signal was applied. Assuming a high power steady state analysis, a significant amount of thermal energy may pass through the membrane/structure interface and permeate the structure. I expect this to effect the magnetic field, which is temperature dependent.

Remember that the magnets are on either side of the foil, and there can not be a magnetic material holding them in place (connecting one to the other). Also, the foil must be insulated from electrically conductive material on at least one end. You could use an Al heat sink at each end of the ribbon, and insulate the foil with something like SilPad or Kapton film.

For a comparative analysis, would it be appropriate to apply a constant signal (ex. 100 watts) or provide a signal which produces a constant displacement (ex. 1mm)?

For constant power handling (steady-state), would white noise be appropriate? How long would it take to achieve a steady state (thermal/mechanical)?

With the same input power and signal, the displace should be the same unless you are comparing different lengths or tensions. Even then, I think the constant signal would be more desirable.

I think white noise would be appropriate. A steady state should be reached very quickly under these conditions. Well under 0.01 second.

I assumed the ferrous components would inductively couple to the membrane and function as a transformer, which may include a reactive component. Is that assumption erroneous?

I don't think so. Every ribbon I know of has an almost purely resistive impedance. You only have one "turn" on the transformer primary or secondary, you have an air gap, and you don't have to have a closed circuit.

How can a square wave be a single frequency? I was under the impression it was composed of a fundamental + odd order components.

You are correct about that, but the low frequency component will be very small. If you use a filter anywhere near the square wave frequency, you no longer have a square wave.

 

[thadman]

In my "Direct Drive amplifier for DIY ribbon" thread, you suggested the Agilent AT-34420A as a suitable tool for measuring the resistance of the membrane.

Electro Rent Corporation - Product Detail Page

For transients, I assume the temperature will change very rapidly. Since resistance is linearly related to temperature, a high speed measurement device may be required to capture the resistance shift.

^Would this be able to measure dynamic resistance?

---

I'm having difficulty calculating the natural frequencies of the membrane.

I have a copy of Den Hartog's text, "Mechanical Vibrations". Membranes are discussed very briefly on page 205. The author suggested that their calculation is significantly more complex than most of the problems in the book and "Vibration problems in Engineering" by Timoshenko would be a good source for an in-depth theoretical analysis.

I have acquired access to Timoshenkos "Vibration problems in Engineering" through my Universities online library. On page 411 "Vibration of membranes", membranes are discussed as per the title.

-Timoshenko assumes that the membrane is uniformly stretched in all directions, however this does not apply to my particular problem. Is there anyway to get around this assumption? If not, are there any other texts anyone could recommend?

-The provided equations do not include a materials property element, only weight/area and tension. Would this not suggest that two materials with equal density, but dissimilar material properties, have the same natural frequencies?

Such an analytical solution would also ignore fluid damping and the resistive/reactive components of the duct, which may result in a gross error. A numerical solution would probably be required for a high degree of accuracy.

Would it be possible to calculate the minimum tension applied to the membrane which does not result in sagging without FEM? My laptop runs OSX and I do not currently have enough funds for a linux workstation.

If the natural frequency of the system was placed below 10hz, I believe some potentially interesting results could arise. I believe the response coefficients of the eigenmodes will decrease at a rate of 1/J^2, where J=mode number. Above 200hz, significant mode dispersion, insignificant mass, and significant fluid damping may result in a displacement primarily defined by the mode shape J=1.

 

[thadman]

YouTube - Crowley and Tripp Roswellite (tm) Ribbon Microphone Material

Bob Crowley - Soundwave Research: Roswellite FAQ- Frequently Asked Questions About Crowley and Tripp Ribbon Microphones

What is Roswellite?

---

Roswellite is the trademarked name of a new nano-enabled ribbon material invented, patented and produced by Soundwave Research Laboratories, Inc. and used in certain Crowley and Tripp microphones. Roswellite is also known as "acoustic nanofilm". It is an extremely strong, low mass, superelastic, paramagnetic composite with high inherent conductivity and shape memory properties. Roswellite can be employed to replace the "foils" such as those used in ribbon microphones. Due to its high strength, toughness and shape retaining properties, Roswellite can withstand windblast, plosives, phantom power applications, and high sound pressure levels, even at low frequencies.

This could be a potentially very interesting material.

 

[tvi]

Methods for forming and using thin film ribbon microphone elements and the like]
"A geometrically shaped acoustic polymer ribbon with shape memory, high conductivity, high toughness. A method of manufacturing the ribbon comprises: forming a sized, elongated, coated or coatable polymeric substrate film between a pair of opposed, geometrically shaped dies, pinching the dies about the polymeric substrate film to form an assembly, heating the dies and the pinched die and polymeric film assembly to a temperature of at about 300 degrees F. for a period of about 15 minutes to set the elongated film into a predetermined geometric pattern, cooling the assembly, removing the film from the dies; and if not already coated, coating the geometrically formed, set, elongated film with a conductive coating."

The patent mentions PET as the polymer layer and Nickel Titanium as a possible metal layer, Nitinol is a shape memory alloy.
"acoustic nanofilm" may have come from the advertising department?

Regards
James

 

[thadman]

Methods for forming and using thin film ribbon microphone elements and the like]
"A geometrically shaped acoustic polymer ribbon with shape memory, high conductivity, high toughness. A method of manufacturing the ribbon comprises: forming a sized, elongated, coated or coatable polymeric substrate film between a pair of opposed, geometrically shaped dies, pinching the dies about the polymeric substrate film to form an assembly, heating the dies and the pinched die and polymeric film assembly to a temperature of at about 300 degrees F. for a period of about 15 minutes to set the elongated film into a predetermined geometric pattern, cooling the assembly, removing the film from the dies; and if not already coated, coating the geometrically formed, set, elongated film with a conductive coating."

The patent mentions PET as the polymer layer and Nickel Titanium as a possible metal layer, Nitinol is a shape memory alloy.
"acoustic nanofilm" may have come from the advertising department?

Regards
James

Interesting.

It's too bad Nitinol's (Nickel-Titanium) thermal (conductivity=10 W/m-K) and electrical (resistivity=.0000820 ohm-cm) properties are so poor. With that said, I would still like to see it tested.

Perhaps another superelastic alloy would be more appropriate. CuAlNi or CuAlBe may offer superior thermal and electrical properties. Buckypaper could be VERY interesting

 

[GusK]

United States Patent Application: 0080152186

is the patent that everybody is talking about on the prodigy-pro board

Gus

 

[Steve Dunlap]

In my "Direct Drive amplifier for DIY ribbon" thread, you suggested the Agilent AT-34420A as a suitable tool for measuring the resistance of the membrane.

Electro Rent Corporation - Product Detail Page

For transients, I assume the temperature will change very rapidly. Since resistance is linearly related to temperature, a high speed measurement device may be required to capture the resistance shift.

^Would this be able to measure dynamic resistance?

That will measure the resistance change with temperature, but not the way you probably have in mind. The signal source will affect the readings. You will most likely need to generate an AC frequency well away from the test frequency, very low in amplitude so it doesn't contribute to heating. Use a filter to remove the test signal, and calculate resistance change from the amplitude change in the AC signal.

I'm having difficulty calculating the natural frequencies of the membrane.

I can't help you with that as I have no background in that field.

Would it be possible to calculate the minimum tension applied to the membrane which does not result in sagging without FEM? My laptop runs OSX and I do not currently have enough funds for a linux workstation.

The magnetic field will have no effect on the sag of the membrane when no signal is present.

 

[1audio]

Before you get too deep into the theory of operation of a ribbon transducer you should get one or two to play with. I learned a lot from reworking and modifying a Decca ribbon.

First, there are real ribbon transducers (Like the Decca) and planar magnetic transducers often called ribbons. We are discussing the genuine version here. In the examples I have seen (including the Apogee and the Magneplanar tweeter) the ribbon is unstressed, in fact loose. If it has a resonance it is well below its operating frequency. Many examples use tricks to keep the ribbon in the magnetic field (the Decca had some grease at strategic locations). Most have corrugated ribbons to provide the elasticity.

A large part of the performance comes from the very light "diaphragm" reacting against air. In many cases the air load is higher in mass than the diaphragm.

The resistance of the ribbon isn't that big a deal. In many cases the matching transformer has a higher internal resistance than the diaphragm. Making an amplifier that will work directly into a ribbon is difficult and very inefficient so transformers are usually used.

Be foil is a great idea, however the stuff is extremely toxic and should be handled very carefully. BW should be able to give you guidelines. Its also astonishingly expensive.

Measuring the resistance of the ribbon in operation is a difficult task. You can't use DC with the low resistance and thermals everywhere. I have the earlier version of the HP meter you mentioned and it works pretty well. But it may be possible to build a version that could go inline with a ribbon, but difficult. The heating may be pretty small and the effect on the DCR of the ribbon will be small. Its usually only an issue on commercial sound reinforcement drivers. In normal audio the average power is rarely even a watt when playing loud.

I'm really interested in this and would like to support your efforts. Let me know how.

 

[thadman]

Before you get too deep into the theory of operation of a ribbon transducer you should get one or two to play with. I learned a lot from reworking and modifying a Decca ribbon.

I have built a pre-prototype, which includes an apparatus for tensioning the membrane. I used standard aluminum foil. The aforementioned system is VERY fragile.

Some thoughts:

The fragility of the element is significant. Mechanical shielding is probably required to prevent foreign objects from entering the gap. However, placing obstructions in front of the element may lead to turbulence at higher sound pressure levels and energy storage (similar to the effects of a phase plug in a compression driver). I have some ideas, however CFD will be required for resolution.

First, there are real ribbon transducers (Like the Decca) and planar magnetic transducers often called ribbons. We are discussing the genuine version here. In the examples I have seen (including the Apogee and the Magneplanar tweeter) the ribbon is unstressed, in fact loose. If it has a resonance it is well below its operating frequency. Many examples use tricks to keep the ribbon in the magnetic field (the Decca had some grease at strategic locations). Most have corrugated ribbons to provide the elasticity.

I'd like to stay away from non-newtonian fluids (grease?), as I believe they may be a source of inherent non-linearity.

Corrugating the element appears interesting. However, the solution space is incredibly large. I'm not sure how to quantify the length of time the membrane will hold a particular shape. I'll have to schedule further meetings with faculty in the Materials Science department for resolution.

A large part of the performance comes from the very light "diaphragm" reacting against air. In many cases the air load is higher in mass than the diaphragm.

Yes However, the membrane-fluid-duct interaction is EXTREMELY complex. I believe an optimized membrane may approach functioning as a massless, oscillating barrier between the fluids.

Be foil is a great idea, however the stuff is extremely toxic and should be handled very carefully. BW should be able to give you guidelines. Its also astonishingly expensive.

This is causing me significant trouble. Brush Wellman has expressed interest in providing a quantity of foil for my research. With regards to Aluminum, I may simply purchase an extremely large quantity. However, I'm not sure how much Beryllium to request due to the incredible cost.

I'd like 25 samples for each experiment to achieve sufficient statistical correlation. The material will be incredibly fragile and I expect a percentage of the material may be destroyed during the research process (tensioning/handling/testing). How much extra should I request to account for this? I don't think there is any way to avoid this problem.

Measuring the resistance of the ribbon in operation is a difficult task. You can't use DC with the low resistance and thermals everywhere. I have the earlier version of the HP meter you mentioned and it works pretty well. But it may be possible to build a version that could go inline with a ribbon, but difficult. The heating may be pretty small and the effect on the DCR of the ribbon will be small. Its usually only an issue on commercial sound reinforcement drivers. In normal audio the average power is rarely even a watt when playing loud.

I agree. Unless a user was interested in acquiring Tinnitus, <.1 watts should be sufficient for normal playback levels. However, operating the transducer at its limits should offer opportunity for significant contrast.

Thank you for your interest in this project thus far With a little bit of effort, I believe some very interesting conclusions could be reached.

 

[1audio]

I had to replace a diaphragm for the Decca once. I used the foil separated from its paper backing from some chewing gum. It worked very well.

BW has some alloys that may also be useful and a lot cheaper. Both a magnesium-aluminum and a BE-aluminum that have good properties are available from BW.

I suspect the sensitivity of the tensioned ribbon will be lower but I may be wrong.

Most of the existing samples I know of have corrugated ribbons, The MG has a 5' corrugated ribbon. They seems to be quite stable for a long time.

The grease is used to keep the ribbon in the gap at rest. There is 4 or 6 small blobs in critical locations. However Ferrofluid could be used in some effective way as well. That would need some thought.

 

[thadman]

BW has some alloys that may also be useful and a lot cheaper. Both a magnesium-aluminum and a BE-aluminum that have good properties are available from BW.

Does Brush Wellman produce these alloys as foils?

I suspect the sensitivity of the tensioned ribbon will be lower but I may be wrong.

I would tend to agree. The reactive force (tension) will counter the input force. A tensioned membrane may also store more potential energy (ie higher mode energy).

However, I believe a tensioning (whether minimum or maximum) mechanism may be required for repeatable experimental results as it may reduce human assembly error.

 

[1audio]

I believe BW makes all of them as foils and others to explore as well. I recently discussed making domes for tweeters with them. They do know their materials.

While playing with a FLIR would be fun you could also start by modifying a Raytek IR thermometer by adding an output and coupling it to a scope to see how fast it responds. If its quick enough you could use it and see how the different parts of the ribbon respond. To keep the ribbon in a consistent place you would need an AC excitation, I would use 100 KHz to keep the motion small.

If I had to do this with the limited budget and resources of a commercial enterprise this would be my approach:
First measure the resistance change with heat by heating the foil indirectly. Check to see if the results are consistent with published tables.
Then set up the transducer and drive it with a bridge that would show the effective impedance and see what happens as the drive level changes. It should be pretty straightforward (Ohm's Law), some scopes can do the necessary math. The ribbon should be pretty resistive in impedance above its resonance (or so I remember) so the change should be related to the temperature of the ribbon. A really thin ribbon should have a very short time-constant so it would be pretty clear if the resistance changed with signal and how much.