I'm looking to build a current source amp to direct drive ribbons. I have scanned all of the previous threads from over the years on the topic and haven't found the info I'm looking for.
I want the amp to be powerful enough to drive theoretical "Full range" ribbon speakers. I.E. very big ribbons.
I am pulling my hair out trying to find out how much power I need, I have no clue how power translates into SPL for ribbons and I can't seem to find the answer anywhere.
I can't begin to design such an amp without knowing the required power output.
I have a few design thoughts in my head but I first need the answer to my above question.
Can anyone help me out?
I want the amp to be powerful enough to drive theoretical "Full range" ribbon speakers. I.E. very big ribbons.
I am pulling my hair out trying to find out how much power I need, I have no clue how power translates into SPL for ribbons and I can't seem to find the answer anywhere.
I can't begin to design such an amp without knowing the required power output.
I have a few design thoughts in my head but I first need the answer to my above question.
Can anyone help me out?
I'd like to keep the ribbon resistance as low as possible, Ideally 0 ohms if it was possible. I may end up paralleling ribbons or something if I have to.
I think rather than asking how much power I need I should ask how many amps do I need?
How does current translate into ribbon SPL?
I was originally going to make it put out 2 amps of swing but I realized this build will be far too expensive to run off of guesswork.
I think rather than asking how much power I need I should ask how many amps do I need?
How does current translate into ribbon SPL?
I was originally going to make it put out 2 amps of swing but I realized this build will be far too expensive to run off of guesswork.
Last edited:
Hi,
Have a look here, Woofertester is creating some beast to drive a sub-ohm ribbon speaker.
http://www.diyaudio.com/forums/pass-labs/300233-f4-beast-builders-120.html#post5338916
Have a look here, Woofertester is creating some beast to drive a sub-ohm ribbon speaker.
http://www.diyaudio.com/forums/pass-labs/300233-f4-beast-builders-120.html#post5338916
Hi
If there are speaker products out on the market past or present that use ribbons, those will give a clue as to power and SPL.
If ribbon efficiency is similar to other types of drivers like dynamic or ESL, then you would expect 80-100dB for 1W at 1m. Keep in mind that 100dB is stupidly loud, so you won't be playing the system that loud even though many people will parrot arguments stating you need a zillion watts to achieve the 130dB peak of an orchestra or a live rock, pop or country performance. That is all nonsense.
However, you can arbitrariy assume a low efficiency like 80dB/1W and a high peak capability just for headroom. Say you decide 120dB is ample. That puts peak power at 40dB above 1W, or 10kW - yikes!. So, maybe the guesstimates are too wild and you settle for just a couple or few hundred watts.
As moschfet pointed out, the ribbon is almost purely resistive and extremely low resistance. This means the currents will be very high and the supply voltages quite low. This is actually an ideal situation for BJTs as there will be no worries about secondary breakdown and you can set the current limiting quite simply as a clamp. Then just make the output stage large enough to dissipate the short current continuously.
If Rz=100mR, and peak power is 200W, then Ipk is 45A at 4V5. This is just a 100Wavg am.
Even if your ribbon is low-z, the cable from the amp will introduce loss. For this reason you might want to use a distributed-amplification scheme where each ribbon section is driven by its own amplifier. This will reduce the effective losses of the interconnects.
Were you to go with everything in parallel, it is no problem to make an output stage with 100s of amperes of capability - you wouldn't need that much current unless the ribbons are super-low-z.
When you know the load impedance, it is dead easy to design an amplifier to drive it.
If there are speaker products out on the market past or present that use ribbons, those will give a clue as to power and SPL.
If ribbon efficiency is similar to other types of drivers like dynamic or ESL, then you would expect 80-100dB for 1W at 1m. Keep in mind that 100dB is stupidly loud, so you won't be playing the system that loud even though many people will parrot arguments stating you need a zillion watts to achieve the 130dB peak of an orchestra or a live rock, pop or country performance. That is all nonsense.
However, you can arbitrariy assume a low efficiency like 80dB/1W and a high peak capability just for headroom. Say you decide 120dB is ample. That puts peak power at 40dB above 1W, or 10kW - yikes!. So, maybe the guesstimates are too wild and you settle for just a couple or few hundred watts.
As moschfet pointed out, the ribbon is almost purely resistive and extremely low resistance. This means the currents will be very high and the supply voltages quite low. This is actually an ideal situation for BJTs as there will be no worries about secondary breakdown and you can set the current limiting quite simply as a clamp. Then just make the output stage large enough to dissipate the short current continuously.
If Rz=100mR, and peak power is 200W, then Ipk is 45A at 4V5. This is just a 100Wavg am.
Even if your ribbon is low-z, the cable from the amp will introduce loss. For this reason you might want to use a distributed-amplification scheme where each ribbon section is driven by its own amplifier. This will reduce the effective losses of the interconnects.
Were you to go with everything in parallel, it is no problem to make an output stage with 100s of amperes of capability - you wouldn't need that much current unless the ribbons are super-low-z.
When you know the load impedance, it is dead easy to design an amplifier to drive it.
Whew that's a lot of current. I was going to go for something like this
I was going to use https://datasheets.maximintegrated.com/en/ds/MAX4238-MAX4239.pdf
for the DC offset.
I like the circuit's elegance and performance but the idle currents in the range you speak of would be enormous.
I was thinking maybe 5 amps idle with 2 amp swings.
It wouldn't be so bad if it weren't for minimum Vdrop requirements across the pass devices.
Does there exist pass devices that can pass that much current at 5 or less volts that would fit the source follower duty of the above circuit?
If so I could give the circuit more gain to reduce the required V-swing into the sense resistor and use very low voltage supplies. For some reason the simulation is showing that distortion increases as the overall voltage swings decrease in amplitude even though the RMS current stays the same. I question whether I can trust that result.
Lower Z means lower voltage swing which means less power into the load and less influence from parasitics in the amp, etc.
I was going to use https://datasheets.maximintegrated.com/en/ds/MAX4238-MAX4239.pdf
for the DC offset.
I like the circuit's elegance and performance but the idle currents in the range you speak of would be enormous.
I was thinking maybe 5 amps idle with 2 amp swings.
It wouldn't be so bad if it weren't for minimum Vdrop requirements across the pass devices.
Does there exist pass devices that can pass that much current at 5 or less volts that would fit the source follower duty of the above circuit?
If so I could give the circuit more gain to reduce the required V-swing into the sense resistor and use very low voltage supplies. For some reason the simulation is showing that distortion increases as the overall voltage swings decrease in amplitude even though the RMS current stays the same. I question whether I can trust that result.
I would prefer to run it at as low z as possible but I don't know the practical limitations of that.
Lower Z means lower voltage swing which means less power into the load and less influence from parasitics in the amp, etc.
Maybe build the amp sections right next to the ribbon sections on the back side of the speaker?
Last edited:
I can recall an Elektor project "The Current Amp", dedicated for ribbons. Designed by Ton Giesberts:
Output amplifier for ribbon loudspeakers (1):
Output amplifier for ribbon loudspeakers (2):
Output amplifier for ribbon loudspeakers (1):
Output amplifier for ribbon loudspeakers (2):
Hi Guys
Wherever there is current through a resistance there is heat and that heat is measured in watts.
Although class-A nominally eliminates crossover distortion, you can easily attain performance that is below the measurement capability of an AP27xx without resorting to class-A. This is a power amp afterall, and in this app there is a requirement for high current drive. However, this does not mean the amplifier should behave like a current source; rather, it should be a voltage source as usual and have a current capability appropriate to the load.
For example, an 8R load driven to 200Wpk needs 5Apk. As a pure resistive load, having no more current available than 5A is acceptable. Real speakers are reactive, so current can peak much higher but not when the voltage peaks, so the output stage has to be able to deliver these higher currents at phase angles up to 60-degrees without stressing the output stage.
The ribbon driver might look like 100mR and the peak current and voltage iat 200Wpk is 44.7A and 4.47V, respectively. Again, if this is purely resistive, then no current reserve is required. One would assume a ribbon should have some inductance but it is very low, so the application is almost like building a spot welder - haha.
It is a common mistake to assume that the circuit can be "excessively simple" if only unity voltage gain is required. Einstein said that things should be as simple as possible to achieve the goal but no simpler.
Regarding the 20-30A output of the article mentioned: That is not difficult at all to achieve even for parallel dynamic drivers. Built a few amps for such apps using standard circuits and excellent THD20.
Wherever there is current through a resistance there is heat and that heat is measured in watts.
Although class-A nominally eliminates crossover distortion, you can easily attain performance that is below the measurement capability of an AP27xx without resorting to class-A. This is a power amp afterall, and in this app there is a requirement for high current drive. However, this does not mean the amplifier should behave like a current source; rather, it should be a voltage source as usual and have a current capability appropriate to the load.
For example, an 8R load driven to 200Wpk needs 5Apk. As a pure resistive load, having no more current available than 5A is acceptable. Real speakers are reactive, so current can peak much higher but not when the voltage peaks, so the output stage has to be able to deliver these higher currents at phase angles up to 60-degrees without stressing the output stage.
The ribbon driver might look like 100mR and the peak current and voltage iat 200Wpk is 44.7A and 4.47V, respectively. Again, if this is purely resistive, then no current reserve is required. One would assume a ribbon should have some inductance but it is very low, so the application is almost like building a spot welder - haha.
It is a common mistake to assume that the circuit can be "excessively simple" if only unity voltage gain is required. Einstein said that things should be as simple as possible to achieve the goal but no simpler.
Regarding the 20-30A output of the article mentioned: That is not difficult at all to achieve even for parallel dynamic drivers. Built a few amps for such apps using standard circuits and excellent THD20.
Last edited:
I still don't understand. The lower the resistance of the ribbon the lower the wattage is going to be through it at the same current. Yet it is the current that causes the ribbon to vibrate for through the magnetic field.
If the magnet is X tesla and the ribbon is X ohms then 1A of current should move the ribbon the same amount no matter what the value of X is right? I don't understand how ribbon wattage matters at all.
If the magnet is X tesla and the ribbon is X ohms then 1A of current should move the ribbon the same amount no matter what the value of X is right? I don't understand how ribbon wattage matters at all.
We do not have super-conductors for the living room. Therefore a LOW resistance implies a THICK HEAVY ribbon.
While you are looking for zero resistance, for decent ribbon-to-air coupling efficiency we want ribbon to be "lighter than air", or at least not a lot heavier than the air it couples to.
And all this is low-efficiency approximation. What you really want is Power IN the air. A high-efficiency speaker (a huge thin ribbon may come close) will "feel air resistance" and reflect that back to the electric side as electrical resistance. A JBL E-130 shows several Ohms of acoustic reflected impedance on top of its dead coil resistance (below 500Hz). Big horn drivers have working impedance over twice their unloaded electrical resistance.
There's always resistance. Get used to it.
Anyway for most home-size diaphragms, if we did have superconductors, we would face all the reactances of the mass and diaphragm. A basic voice-coil driver would still be 50 Ohms at 50Hz, but 5r at 500hz and 1r at 2500hz. No straight linear amplifier can put good power into such a wide range of impedances.
How big? Big as a box? Shoebox? Blanket box? Big as a billboard? In general, if you wish to go into audio bass, a bigger speaker allows a more efficient design and *less* audio power. (However there are big drivers which just suck.)
Oh wow I didn't know that.
Why not just make many small ribbons instead of one large one? Would that change anything?
How do you calculate acoustic reflected impedance?
I'm still confused on the question of how much 'current' I need to swing for a given SPL into a ribbon.
Why not just make many small ribbons instead of one large one? Would that change anything?
How do you calculate acoustic reflected impedance?
I'm still confused on the question of how much 'current' I need to swing for a given SPL into a ribbon.
I was told at least 7 feet of ribbon would make a good full range ribbon.
I submit that the speaker is much more the unknown problem than the amplifier.
You can easily stack enough amp-parts to power a locomotive. Ribbon impedance is low but smooth; I-drive or E-drive does not make any primary difference on Frequency Response (as it does for a small cone in a box). I think you can test experimental speakers with a car-audio chip (good for low-Z loads), just to see if the speaker works and how much bigger the amp needs to be to be "satisfying". Scaling-up is simple.
The last ribbon speaker I saw had 5' by 2' radiating area made of 1" ribbons, almost certainly Aluminum, apparently dead-soft annealed (Al foil is hardened by the rolling and usually sold part-annealed.) And a ton (almost) of magnet! Heavy! And the ones I saw had been damaged beyond repair. Reviews said they were excellent on female voice but a bit shy on full-bass, and a 100W amp would strain (perhaps more due to the very low resistance than actual power).
This will be a major project. Build a couple-foot of ribbon speaker and see how it carries the mids, how much Volts/Amps it eats to be loud-enough, how more magnet helps or hurts (poles block sound), if thinner ribbon adds efficiency or the ribbon's stray resonances just stand out more.
You can easily stack enough amp-parts to power a locomotive. Ribbon impedance is low but smooth; I-drive or E-drive does not make any primary difference on Frequency Response (as it does for a small cone in a box). I think you can test experimental speakers with a car-audio chip (good for low-Z loads), just to see if the speaker works and how much bigger the amp needs to be to be "satisfying". Scaling-up is simple.
The last ribbon speaker I saw had 5' by 2' radiating area made of 1" ribbons, almost certainly Aluminum, apparently dead-soft annealed (Al foil is hardened by the rolling and usually sold part-annealed.) And a ton (almost) of magnet! Heavy! And the ones I saw had been damaged beyond repair. Reviews said they were excellent on female voice but a bit shy on full-bass, and a 100W amp would strain (perhaps more due to the very low resistance than actual power).
This will be a major project. Build a couple-foot of ribbon speaker and see how it carries the mids, how much Volts/Amps it eats to be loud-enough, how more magnet helps or hurts (poles block sound), if thinner ribbon adds efficiency or the ribbon's stray resonances just stand out more.
Have you actually built a ribbon? Before you worry about the amplifier, you should actually build one or more ribbons and see what the design trade-offs are.
Once you decide on a ribbon design, that will dictate what current you need to drive it to a listening loudness you want.
After that, the current required to produce the sound you want multiplied by the resistance of the ribbon (yes your ribbon will have a resistance) yields the voltage needed across the ribbon.
For large ribbons, wiring the ribbon to the amp is a design challenge. You will not appreciate the challenges until you build some different versions and actually integrate the ribbons with conventional midranges and woofers.
Once you decide on a ribbon design, that will dictate what current you need to drive it to a listening loudness you want.
After that, the current required to produce the sound you want multiplied by the resistance of the ribbon (yes your ribbon will have a resistance) yields the voltage needed across the ribbon.
For large ribbons, wiring the ribbon to the amp is a design challenge. You will not appreciate the challenges until you build some different versions and actually integrate the ribbons with conventional midranges and woofers.
Here is the thread where I brought up the topic
Citation 12 non-complimentary to drive ribbon tweeter
Citation 12 non-complimentary to drive ribbon tweeter