If you don't have a coupling cap you really ought to have a DC-offset protection circuit to prevent the amp destroying the speakers if a DC offset appears at the amp's input. An input cap is a way better decision unless you already plan to have DC-offset protection circuit and are irrationally allergic to coupling caps.
For an amp with input impedance around 10k, then 22µF is an appropriate value, with 50k input impedance 3.3µF or so would do. If electrolytic you can use larger values to reduce any residual LF distortion, if film caps you might want to reduce the value for practical reasons and take a small hit on 20Hz roll-off... I'm chosing an input time-constant around 1s, which has minimum effect on 20Hz roll off while quickly discharging when the input is unplugged.
Very important is a high value grounding resistor on the input to discharge the input cap (it can be 10x the amp's input impedance), other wise when you plug in a new source you may get a big click/thump from whatever charge was stored on the input cap.
A typical input network needs 5 or so components to block DC, discharge stored charge, attenuate RF, and provide a defined and linear impedance to the source.
What was the year of these tests?
Most probably materials and manufacturing processes of capacitors have changed since then so I would render Bateman's findings to "historical records" and would not take those for granted for today's capacitors.
I disagree - he tested the standard types we use today from PKT, PP, PS, PPS, PTFE film, Al electrolytic to ceramic types I and II. The dielectrics haven't changed, these are all still available (pretty much, PS is harder to find, PTFE is rare). There are a few more dielectrics, which can be tested to add some new entries using the same protocol if wanted. It would be wise to collect some new data, but don't expect the physics of dielectrics to change, solid state physics of the dielectric is a property of the material and its bonds and especially moisture absorption (for film types).
I'd like to see a graph of dielectric linearity v. moisture absorption v. dielectric absorption, I feel it would be interesting. Certainly PTFE and PE and PP are very moisture-repelling. Nylon would be extremely poor as it absorbs a lot of warer, and guess what, no one makes nylon film caps...
I'd like to see a graph of dielectric linearity v. moisture absorption v. dielectric absorption, I feel it would be interesting. Certainly PTFE and PE and PP are very moisture-repelling. Nylon would be extremely poor as it absorbs a lot of warer, and guess what, no one makes nylon film caps...
Sorry to disagree, everything in manufacturing processes changes - material electrical properties, thicknesses, physical dimensions etc
For me Bateman's results stay in their own era, it would be nice if someone would repeat the measurements with the capacitors on sale today.
Too bad Bateman chose to publish in a publication that has since stopped selling individual articles. Its a problem in science and technology in general that potentially useful knowledge gets lost when no longer available to most of interested parties, or eventually maybe not available to anyone.
Thanks to wealthy influencers, copyrights have been extended in time far beyond the original intent of their supposed purpose.
Thanks to wealthy influencers, copyrights have been extended in time far beyond the original intent of their supposed purpose.
Last edited:
Gee, I don't know Markw4. I purchased the Linear Audio series. Jan did a lot of work assembling and publishing that series. You could have learned a lot reading the articles.
Anyway, the fact remains pretty simply this. As long as you don't develop signal voltage across the capacitor and it isn't leaky, you can't hear the darned thing. You don't want it physically too large of course.
Absolutely true. Materials and construction change over time. Also, I would never omit an input capacitor to an amplifier. I have seen too many times what can happen if the source goes DC, and they can.
Anyway, the fact remains pretty simply this. As long as you don't develop signal voltage across the capacitor and it isn't leaky, you can't hear the darned thing. You don't want it physically too large of course.
Absolutely true. Materials and construction change over time. Also, I would never omit an input capacitor to an amplifier. I have seen too many times what can happen if the source goes DC, and they can.
I purchased a few articles from Linear Audio, and Jan has kindly chosen to share a few. What happens after Jan is gone and you are gone? We may as well start over from scratch?
Its not just Linear Audio. Many academics around the world are resorting to sci.xxx because they don't have funding to buy every article of possible interest for supporting future research. Most grant funding these days is pretty stingy. One can't find out if an article is of any use without reading it first; the abstract isn't usually enough to tell if you need to buy it.
That's the overall problem I was talking about. Its not just about Linear Audio.
Its not just Linear Audio. Many academics around the world are resorting to sci.xxx because they don't have funding to buy every article of possible interest for supporting future research. Most grant funding these days is pretty stingy. One can't find out if an article is of any use without reading it first; the abstract isn't usually enough to tell if you need to buy it.
That's the overall problem I was talking about. Its not just about Linear Audio.
I would agree, but don't know your particular rationale for saying that. Possibly we would not entirely agree on why, don't know.
Hi Markw4,
When I pass on, hopefully my bookcases full of technical books will be picked up by some really smart person. Or maybe someone will throw them all out. I can't control what happens when I pass on, I can will them on of course. The real shame is this. Every single experienced person that passes has even more information stored in their head. Instantly and completely gone. That's sad. The last tech I trained was in 1996. There is also more than dry knowledge. There is the rationel and care / attention - the mindset that also goes away. That is how we learned and helped others. This is one of the most important things to pass on in addition to the desire to learn why. As my fellow techs pass, this has been on my mind. The guys that mentored me have mostly passed now.
Why don't you want a large part? Well, you don't want it so small that the quality suffers. Too large and it becomes an antenna. It either transmits or receives unless you shield it. In addition, you have to pay attention to ventilation. So like everything else in life, there is a happy compromise. There is zero sense in a part being larger than it has to be. If the part is large with mass, you have to mechanically restrain it from vibrating or moving. Vibration or movement may break solder joints, the component leads or the traces. Happened a lot with amplifiers bolted to car sub boxes. Kinda funny actually!
When I pass on, hopefully my bookcases full of technical books will be picked up by some really smart person. Or maybe someone will throw them all out. I can't control what happens when I pass on, I can will them on of course. The real shame is this. Every single experienced person that passes has even more information stored in their head. Instantly and completely gone. That's sad. The last tech I trained was in 1996. There is also more than dry knowledge. There is the rationel and care / attention - the mindset that also goes away. That is how we learned and helped others. This is one of the most important things to pass on in addition to the desire to learn why. As my fellow techs pass, this has been on my mind. The guys that mentored me have mostly passed now.
Why don't you want a large part? Well, you don't want it so small that the quality suffers. Too large and it becomes an antenna. It either transmits or receives unless you shield it. In addition, you have to pay attention to ventilation. So like everything else in life, there is a happy compromise. There is zero sense in a part being larger than it has to be. If the part is large with mass, you have to mechanically restrain it from vibrating or moving. Vibration or movement may break solder joints, the component leads or the traces. Happened a lot with amplifiers bolted to car sub boxes. Kinda funny actually!
Member
Joined 2009
Paid Member
Those are pretty good, especially if given some time to settle in. However there have been better options in the past. Cap manufacturers have been dropping analog audio cap lines in favor of class-D audio and SMPS needs.
There were no real audio lines to begin with (with only a few exceptions). Audio is way too small a market to build special production lines for. Just different printing/labelling. The computer industry and today the e-car production need way higher numbers and they force new developments like better and higher current ultra low RDson MOSFETs, ultra low ESR caps, better/lighter batteries etc.
Class D = analog audio. The SMPS in it is also analog and used for analog audio. "Analog audio caps"...... How people come up with this pseudo scientific nonsense is remarkable.
Class D = analog audio. The SMPS in it is also analog and used for analog audio. "Analog audio caps"...... How people come up with this pseudo scientific nonsense is remarkable.
Last edited:
Some of you say that a capacitor should not be too big.
But what does "too big" mean for an amplifier input capacitor?
Give dimensions and justifications for these dimensions.
Nobody has the idea of putting at the input of an amplifier, a PP capacitor that supports 1500V
But what does "too big" mean for an amplifier input capacitor?
Give dimensions and justifications for these dimensions.
Nobody has the idea of putting at the input of an amplifier, a PP capacitor that supports 1500V
Not larger than necessary. Physical size not value. Why use a very large part with possible weight/antenna drawbacks because of its size only because (insert famous name here) says polyester is boo boo?
And there sure will be spectrum people using D cell sized 1500V polypropylene caps because (insert famous name here) says the dielectric is theoretically best in his lab with 250V 1 MHz square wave signal. Full circle. I have seen amplifiers standing on a side because polypropylene caps needed to be used according someone. Of course also melt glue/extension wire constructions and devices ruined with absurdly chosen caps and any transport of the item impossible.
A good medicine would be nice but what when the medicine is worse than the disease?
And there sure will be spectrum people using D cell sized 1500V polypropylene caps because (insert famous name here) says the dielectric is theoretically best in his lab with 250V 1 MHz square wave signal. Full circle. I have seen amplifiers standing on a side because polypropylene caps needed to be used according someone. Of course also melt glue/extension wire constructions and devices ruined with absurdly chosen caps and any transport of the item impossible.
A good medicine would be nice but what when the medicine is worse than the disease?
Last edited:
From the Linear Audio website "f you just want a single article, send me $ 2.90 through paypal@linearaudio.net and I will send you the article PDF"
How is that ambiguous?
Hi Alren,
I did outline the factors people need to consider. Anyone can get very silly and proclaim they know best because of what they think they hear. That is typically what is seen.
So, how do I select capacitors? I have a number of pieces of test equipment to test LCR components, like an HP 4263A, 4192A and an ESI Videobridge, plus others. I know what factors matter and what to test for. Then to confirm, we test audio performance with an audio analyser, and we listen (although the instruments have proved to make this step unnecessary).
So once the electrical characteristics pass, what next? Size, weight and cost come up. Larger parts cost more typically, if you have an equivalent part that is smaller and less expensive - you use it (unless you want the audiophile ad copy). You have to plan for board space, possible vibration, heat and other factors. If the part is large, you must attach the body to the PCB or chassis in some effective way. If you have to run wires to it, well it is probably a bad choice. Running low level signal around the chassis is a great way to pick up noise or cause instability.
When you design properly, you do end up with the very best performance at reasonable cost and higher reliability. Then there is always someone who comes in behind and replaces those selected parts with types that were rejected for some good reason, often ending up with poorer performance while they claim an improvement. And so the world goes around and around. Notice these folks never supply properly measured performance. CLUE!
You're the designer, you decide the value, type and voltage rating.
Okay, to do that we need to start with a value and voltage rating, case style and so forth. How about using common sense?
I did outline the factors people need to consider. Anyone can get very silly and proclaim they know best because of what they think they hear. That is typically what is seen.
So, how do I select capacitors? I have a number of pieces of test equipment to test LCR components, like an HP 4263A, 4192A and an ESI Videobridge, plus others. I know what factors matter and what to test for. Then to confirm, we test audio performance with an audio analyser, and we listen (although the instruments have proved to make this step unnecessary).
So once the electrical characteristics pass, what next? Size, weight and cost come up. Larger parts cost more typically, if you have an equivalent part that is smaller and less expensive - you use it (unless you want the audiophile ad copy). You have to plan for board space, possible vibration, heat and other factors. If the part is large, you must attach the body to the PCB or chassis in some effective way. If you have to run wires to it, well it is probably a bad choice. Running low level signal around the chassis is a great way to pick up noise or cause instability.
When you design properly, you do end up with the very best performance at reasonable cost and higher reliability. Then there is always someone who comes in behind and replaces those selected parts with types that were rejected for some good reason, often ending up with poorer performance while they claim an improvement. And so the world goes around and around. Notice these folks never supply properly measured performance. CLUE!
Polarized capacitor - 470uF- low ESR.
High values are better when it comes to ensuring stable phase behavior in the low frequency range.
High values are better when it comes to ensuring stable phase behavior in the low frequency range.
Okay, I just have to ask. Think on this please. Then do some research.
What are the highest values of ESR for a decent capacitor? Now, what is the circuit impedance? Please, compare the two and explain why even a high very ESR makes any difference at all. D/A might, that's depending on where it is and how much signal voltage is developed across it.
How does phase become "unstable"? And since you're looking at phase, it is locked mathematically to frequency response. Both channels being the same, the phase between them will be the same and very small. Harmon Kardon tried to spin the constant phase deal across the entire audio band. You do not detect phase differences between widely spaced frequencies. You detect phase differences between your ears (between channels) to allow you to location the source of a sound. Survival. So phase only matters between channels (within reason of course). It changes the appearance of a waveform on an oscilloscope display, but because the frequency content is the same you don't hear any difference.
Advertising nonsense polluting facts.
What are the highest values of ESR for a decent capacitor? Now, what is the circuit impedance? Please, compare the two and explain why even a high very ESR makes any difference at all. D/A might, that's depending on where it is and how much signal voltage is developed across it.
How does phase become "unstable"? And since you're looking at phase, it is locked mathematically to frequency response. Both channels being the same, the phase between them will be the same and very small. Harmon Kardon tried to spin the constant phase deal across the entire audio band. You do not detect phase differences between widely spaced frequencies. You detect phase differences between your ears (between channels) to allow you to location the source of a sound. Survival. So phase only matters between channels (within reason of course). It changes the appearance of a waveform on an oscilloscope display, but because the frequency content is the same you don't hear any difference.
Advertising nonsense polluting facts.
That's a change. For awhile individual articles were no longer sold. Only the full collection on memory stick, which IIRC was more expensive then than it is now.