If this is from an empty stadium? Wont 70,000 "waterbags" make a huge difference.
Can you point to a chart of loss vs frequency vs distance? Most of the concert halls I get to enjoy I'm at leat 100' from the performers. I was at a smaller restaurant with brick walls listening to 3 operatic sopranos last night. in the higher registers they came off as a screech. Some I'm sure is due to the sheer level 10' vs the 200' to the seats I can afford at the opera house. However it would be interesting to know what kind of low pass filter I'm listening through in the affordable (cheap at $120) seats in the house.
Thank you for posting this.
If you put a serious voltage over a bad capacitor it will distort. 680 Ohms source resistance on a 68 Ohm load (1:10) through a non linear capacitor is exactly the opposite of what any sensible designer would venture to do. And showing measurements below the f3 point is below the belt.
Put no serious voltage over a bad capacitor, and it will not distort. Look at your own posting, this is the fun part of your reference. Size conquers all. There is no technical merrit to this story, just marketing. The article referred to proves my point.
Properly sized CoG, metal film, oversized elco's: please show me any measurement by any metric that demonstrates even the slightest badness.
Last edited:
Funky!
My next project works with AKM. No NDA's, just a company making good products and giving proper info on them hoping that other companies will use their product.
NS blue book was a good example.
Last edited:
Much harder to measure occupied! But as there is no reverberant field as almost all reflections go out of the stadium results don't change much. Also in many cases there are padded seats. They tend to absorb the same empty or full.
The echo that is a worry is the cross field one. It comes in quite late and is about the only worry. Secret stuff on how to handle it!
Last edited:
Not at the office so can't post the charts easily. It really depends on humidity. Rear row seats generally have more reverberation and sound "fuller."
A quick google will get you the IEC formula.
Outside wind also comes into play. Inside airconditioning has some effect.
Allow me to explain the cap measurements that I published over 40 years ago (1978) and their relevance both then and now.
First the ceramic. It was a typical ceramic cap, easily purchased and used at the time, and for the most part, even today.
It was to show how a ceramic cap can add distortion where you don't expect. This basic RC circuit was in a crossover for an electrostatic speaker sold at that time. The distortion that I measured (IM) was added to the audio path, because the ceramic cap was so non-linear. When I proposed an alternative, I was told, like so many here would like to tell me today, that it would make no difference. So I measured the cap to 'rub their noses in it' so to speak, because I refused to work for a company so backward.
Now, what about the other measurement with Tantalum caps?
This test was conducted in order to mimic using Tantalum caps in a FEEDBACK LOOP of either a power amp or preamp. You know: 680 ohms to ground with a series cap to force the output to unity gain at DC. Like so many of you out there still do, except perhaps you now use Aluminum caps instead. Back 40 years ago Tantalum was dominant, not Aluminum, in this application. My analyzer at the time had a fixed 600 ohm output, so if I were to add a 600 ohm load, I would have had 1200 ohms as the equivalent feedback, and this was too easy, so I selected about 10% of 600 ohms as a load, to get approximately 668 ohms as an equivalent feedback resistor.
Today with newer instrumentation, 25 ohms might be selected at the analyzer, instead of 600, and a 650 ohm load resistor might be more appropriate. The measured results should be essentially the same. Try it for yourself!
First the ceramic. It was a typical ceramic cap, easily purchased and used at the time, and for the most part, even today.
It was to show how a ceramic cap can add distortion where you don't expect. This basic RC circuit was in a crossover for an electrostatic speaker sold at that time. The distortion that I measured (IM) was added to the audio path, because the ceramic cap was so non-linear. When I proposed an alternative, I was told, like so many here would like to tell me today, that it would make no difference. So I measured the cap to 'rub their noses in it' so to speak, because I refused to work for a company so backward.
Now, what about the other measurement with Tantalum caps?
This test was conducted in order to mimic using Tantalum caps in a FEEDBACK LOOP of either a power amp or preamp. You know: 680 ohms to ground with a series cap to force the output to unity gain at DC. Like so many of you out there still do, except perhaps you now use Aluminum caps instead. Back 40 years ago Tantalum was dominant, not Aluminum, in this application. My analyzer at the time had a fixed 600 ohm output, so if I were to add a 600 ohm load, I would have had 1200 ohms as the equivalent feedback, and this was too easy, so I selected about 10% of 600 ohms as a load, to get approximately 668 ohms as an equivalent feedback resistor.
Today with newer instrumentation, 25 ohms might be selected at the analyzer, instead of 600, and a 650 ohm load resistor might be more appropriate. The measured results should be essentially the same. Try it for yourself!
Last edited:
John, the problem is more the 4.7uF driving a 68Ohm load, in general you would not want to have that pole at that location in your feedback loop, would you?
It is great that you and others have done the work and provided information on how bad capacitors can be. So, pitfalls to be avoided, nanos gigantum humeris insidentes, I give you that fair and square.
But my point really is: a properly dimensioned quality capacitor is as good as a piece of copper wire to connect two points for AC (but only if there is no DC potential between said two points, in that case typically a capacitor is the better option).
It is great that you and others have done the work and provided information on how bad capacitors can be. So, pitfalls to be avoided, nanos gigantum humeris insidentes, I give you that fair and square.
But my point really is: a properly dimensioned quality capacitor is as good as a piece of copper wire to connect two points for AC (but only if there is no DC potential between said two points, in that case typically a capacitor is the better option).
That's the problem with staying referenced in 1970's. Not that you will, but measurements of modern electrolytics are completely boring for all but the most specialized circuits. E.g., Nichicon Muse ES bipolar caps measured: <-120dB THD, <-140dB IMD
It's like nothing happened in the last 40-50 years in terms of electronic parts... wonder why Bonsai said what he did about your designs.
Last edited:
My ears are not sold on that yet. It seems like a legit idea but the sound that is often bad from a lot of servo gear seems to point at some sorta complex problem caps negate.
DPH, did you measure the DA or dielectric absorption of these caps? It is probably several %.
Unfortunately, folks, it is not easy to find, buy, or fit a nearly 'perfect' capacitor. They are really expensive, like $100 for a 1uF Teflon cap. Now you can find some 'close fits' that are not too bad, like metalized PP but you can get even better performance by using a simple servo with a 1uF Mylar cap, and getting as low a frequency response as you might like. It used to be hard to design a cost-effective servo, but it isn't today with cost effective fet input IC's like the OPA2134 (like I tend to use) or many others.
I realize, like DPH, some don't notice 1uF ceramic coupling caps in otherwise well designed audio equipment, but they are often there. After all, they are small, cost-effective, etc. It is just that they will have lots of distortion and DA, compared to a servo.
Now, I can't say that ALL servos sound great. Sometimes servos can be designed that have real problems, like too high a cutoff frequency. Never use a servo as a hi pass filter, it works the servo too hard, use a quality cap instead. But a servo has an additional advantage of buffering its 'imperfections' (all IC's and small hi value caps have 'imperfections' with an almost always necessary output attenuator to keep their contribution low in regards to the audio signal.
If any of you have had problems with servos, I suspect it is due to something wrong with the servo.
Unfortunately, folks, it is not easy to find, buy, or fit a nearly 'perfect' capacitor. They are really expensive, like $100 for a 1uF Teflon cap. Now you can find some 'close fits' that are not too bad, like metalized PP but you can get even better performance by using a simple servo with a 1uF Mylar cap, and getting as low a frequency response as you might like. It used to be hard to design a cost-effective servo, but it isn't today with cost effective fet input IC's like the OPA2134 (like I tend to use) or many others.
I realize, like DPH, some don't notice 1uF ceramic coupling caps in otherwise well designed audio equipment, but they are often there. After all, they are small, cost-effective, etc. It is just that they will have lots of distortion and DA, compared to a servo.
Now, I can't say that ALL servos sound great. Sometimes servos can be designed that have real problems, like too high a cutoff frequency. Never use a servo as a hi pass filter, it works the servo too hard, use a quality cap instead. But a servo has an additional advantage of buffering its 'imperfections' (all IC's and small hi value caps have 'imperfections' with an almost always necessary output attenuator to keep their contribution low in regards to the audio signal.
If any of you have had problems with servos, I suspect it is due to something wrong with the servo.
Last edited:
Derfy,
If you do the tests the way shown in that reference those are reasonable results. There are better methods to look at the distortion such as using a bridge as I do. What is ignored is what happens as capacitors age, behavior above room temperature and the influence of vibration.
If there is no DC across the capacitor why is it in a circuit? If it is being used in a crossover in a loudspeaker then vibration is a real big issue.
What is actually being proposed is to use coupling capacitors at least an order of magnitude greater in value than what would be required for any given low frequency cutoff.
So the actual trade off is noise vs low frequency distortion. My math says on a preamp with a nominal input level of 100 millivolts would have thermal noise at -75 dB with a 100,000 ohm input impedance. So the distortion of the capacitor would not be the limit! (A valid test would be to match the capacitor distortion at a load that would have noise matching distortion level.)
If you do the tests the way shown in that reference those are reasonable results. There are better methods to look at the distortion such as using a bridge as I do. What is ignored is what happens as capacitors age, behavior above room temperature and the influence of vibration.
If there is no DC across the capacitor why is it in a circuit? If it is being used in a crossover in a loudspeaker then vibration is a real big issue.
What is actually being proposed is to use coupling capacitors at least an order of magnitude greater in value than what would be required for any given low frequency cutoff.
So the actual trade off is noise vs low frequency distortion. My math says on a preamp with a nominal input level of 100 millivolts would have thermal noise at -75 dB with a 100,000 ohm input impedance. So the distortion of the capacitor would not be the limit! (A valid test would be to match the capacitor distortion at a load that would have noise matching distortion level.)
Last edited:
What is the source you are using with 100K Ohms impedance? Maybe a ceramic cartridge but even then it has a reactive Z lower than 100K. Thermal Noise Calculator comes up with -104 dBV with 100K 20 KHz bandwidth. Subtract 20 dB for the 100 mV signal and you get 84 dB SNR. Except for some really nasty ceramics the distortion products would mostly be lower if there were a voltage across that cap.
One aspect not mentioned is the cumulative effect of cascaded high pass filters. An audio recording will have at least the following high pass filters:a microphone (1), a mike preamp (2) a DAW (typically 3) a digital link and then a DAC (4) a preamp (5) an amp (6) and a speaker (7) so a seven pole high pass filter. Looking at part of it would be misleading. and on a seven pole filter even a small change in a part can have a big effect on the overall behavior. Just maybe some of the cap sound comes from small changes in value?
One aspect not mentioned is the cumulative effect of cascaded high pass filters. An audio recording will have at least the following high pass filters:a microphone (1), a mike preamp (2) a DAW (typically 3) a digital link and then a DAC (4) a preamp (5) an amp (6) and a speaker (7) so a seven pole high pass filter. Looking at part of it would be misleading. and on a seven pole filter even a small change in a part can have a big effect on the overall behavior. Just maybe some of the cap sound comes from small changes in value?
- Status
- Not open for further replies.