I read that back when you posted it; the reread was a good idea, as I had already forgotten about the IM.
Also, it's important to remember that even in applications like passive crossovers where there is no DC bias applied, instrument waveforms are often asymmetrical; obviously they integrate to zero over the long term, (depending on the LF characteristics of the chain) but can be quite asymmetrical over the short term.
And with respect to point 5, I believe those were all of the same type; that does not cover the issue of a fancy small cap bypassing more mundane larger capacitors. (Which I have not had good luck with. )
Also, it's important to remember that even in applications like passive crossovers where there is no DC bias applied, instrument waveforms are often asymmetrical; obviously they integrate to zero over the long term, (depending on the LF characteristics of the chain) but can be quite asymmetrical over the short term.
And with respect to point 5, I believe those were all of the same type; that does not cover the issue of a fancy small cap bypassing more mundane larger capacitors. (Which I have not had good luck with. )
Capacitor bypassing
Although I have little personal experience with bypassing large caps with high quality smaller caps, the examples I've seen posted are of fairly tiny and probably questionable ratios.
I see electrolytic power supply capacitors running 50,000 uF bypassed with a 0.1uF or two. A ratio of 1/2 million to one or to put it another way 0.0002%.
Looking at the problem from a current flow standpoint, I'd guess that a ratio of 1% to 5% made more sense. So in the example above a bypass of 500uF would be needed for 1% of the capacitance value to be a high grade low ESR path. Clearly a very costly answer.
Does anyone have experience with actual improvement in sound, and if so at what ratio?
Although I have little personal experience with bypassing large caps with high quality smaller caps, the examples I've seen posted are of fairly tiny and probably questionable ratios.
I see electrolytic power supply capacitors running 50,000 uF bypassed with a 0.1uF or two. A ratio of 1/2 million to one or to put it another way 0.0002%.
Looking at the problem from a current flow standpoint, I'd guess that a ratio of 1% to 5% made more sense. So in the example above a bypass of 500uF would be needed for 1% of the capacitance value to be a high grade low ESR path. Clearly a very costly answer.
Does anyone have experience with actual improvement in sound, and if so at what ratio?
"I see electrolytic power supply capacitors running 50,000 uF bypassed with a 0.1uF or two."
IMO it's pointless to "bypass" PS caps, whose job is to supply pure DC.
IOW, there's no AC signal to bypass.
IMO it's pointless to "bypass" PS caps, whose job is to supply pure DC.
IOW, there's no AC signal to bypass.
noah katz said:
An "ideal" power supply presents a zero Ohm source impedance to all loads AC or DC. Power amplifiers have substantial AC load currents and these currents flow into and out of power supply bypass capacitors. (Actually if you follow the load current all the way around the loop, the power supply capacitors are usually in series with the load)
Having a very low AC impedance prevents a signal voltage from being developed across the power supply output and therefore prevents any intermodulation of other stages.
Needing a low AC impedance means that electrolytic capacitors might not be the best choice, bypassing them is one possible solution. Other solutions include regulating the supply rails (usually results in a lower Z out) or capacitor multipliers where the gain of an active device helps to lower the power supply output impedance (but not its actual stored energy).
"Power amplifiers have substantial AC load currents and these currents flow into and out of power supply bypass capacitors."
If you count pulsating DC as AC, then I guess that's true.
Still not sure it applies for PS caps; what's needed is low internal impedance so that when current is delivered the voltage doesn't fall.
If you count pulsating DC as AC, then I guess that's true.
Still not sure it applies for PS caps; what's needed is low internal impedance so that when current is delivered the voltage doesn't fall.
As HermanV says, the capacitors are in series with the load. It's easier to see if you draw the PS as two batteries, with the center point grounded (or not, as is the case in some full bridge configurations), and the load returned to the center point.
Looking at the cap's job as minimizing AC rail voltage is actually another way of looking at the same thing, and is certainly valid.
Unfortunately really high quality high voltage capacitors in the 100 uF (less than 0.2 Ohms Z at 1 kHz) range are not inexpensive.
Looking at the cap's job as minimizing AC rail voltage is actually another way of looking at the same thing, and is certainly valid.
Unfortunately really high quality high voltage capacitors in the 100 uF (less than 0.2 Ohms Z at 1 kHz) range are not inexpensive.
OK, I see what you guys are saying now.
But I'm still having trouble equating what (I think) bypass caps are supposed to do, which is transmit AC faithfully, to what a PS is supposed to do, which is supply DC.
But I'm still having trouble equating what (I think) bypass caps are supposed to do, which is transmit AC faithfully, to what a PS is supposed to do, which is supply DC.
There are 2 substantial AC quantities to bypass: 1, the current from the power supply; 2, the current required by the load.
The capacitors "supply pure DC" by storing and releasing charge... i.e. AC current. Pure DC current flows past the capacitors; the AC currents flow through the capacitors.
Electrolytic capacitors really behave their best at low frequencies. At higher frequencies, the effects ESL (equivalent series inductance) and ESR (blah blah resistance) begin to dominate. In other words, electrolytics do not offer a "low impedance" at high frequencies. Paralleling caps of different technology is usefull because it can provide reduce impedance at higher frequencies. There are, however, no golden ratios. Proper selection is based on the frequency spectrum in need of low impedance. Capacitor roulette with the aid of a scope is a popular and sound method.
poobah said:
Use the load characteristics that the PS will encounter and try and pass square waves..or... conversely..attach scope leads to the PS cap terminals..or at the point where the raw DC enters the outputs. Both are vaild, as long as you practice logic with your interpretations of the resultant data.
Measure.
The capacitor choice/mix that creates the best 'looking' square wave..is the most valid combination.
Looking at the output of the amplifier is not the issue here. Fundmentally correcting the power supply in isolation..or..fundmentally correcting the PS In Situ is valid as well. As I said, all depends on how you interpret and act on the data that is observed.
The human ear hears (the ear as a diode) by way of transients and timing..so peak points..and the differences in time and level of the positive edges is the key component.
This is why the old 'square wave' test works so well..a thing the younger set doesn't tend to know much about...and the older set does not know why the dang test works so well. There's your explaination. Same reason Horns sound so good. They get the leading transient right..even though they destroy everything else with gross levels of distortion. Why do they work? Cuz the ear simply does not hear the rest of the signal.
I'm in this thread because I'm looking for the best sounding speaker crossover Electrolytics. I might even have to make them! But..such is life. Electrolytics basically have NO place in ANY speaker crossover, but they -are- used quite a bit. Sadly enough. The main reason..being their price. To produce a "competitive" speaker, one tends to have to use them. Especially if one likes to compensate a tweeter's fundamental Impedance issues.
I'm likely going to ahve to find the HIGHEST ripple capacity 200uf or so, 200 to 160VDC electrolytics I can find and start matching them, to get in the 60 to 100uf bipolar range. It should be fun, but in the end, FAR cheaper than any film cap. and smaller. But no current density..ug. To add..any thing added down there in the fundamental res of a tweeter is going to affect it's -entire- sound. I've tried photoflash before..they degrade too fast. About as fast as eggs in your fridge. Good for a couple of weeks.
Any contenders with the current density needed to maintain transient control?
Eg: Electrolytic 100uf handles about 2.5 amps of current.. with maximum frequency issues thrown in too boot.....and the 100uf film cap handles about 120 amps of current at, comparibly speaking ..astronomically high frequencies..
Film is nearly a full magnitude better, which is clearly and easily heard in the resultant sound.
Oh yeah, I agree that bypass does NOT work, it only muddies the transient, if one is not VERY careful. And remember the rest of the low distortion characteristic is literally MEANINGLESS (If you are thinking measurements mean something)..it's the transient shape, postion, and timing that is CRITICAL. Ya gotta remember to measure and weigh the results... in the same way the ear hears and interprets things. The key point that is lost on numbers/measurement fanatics. If you think about all that, sudddenly listening vs measurement (confusion!) begins to make sense. Ie, why digital sucks: critical micro-timing 'cues' are all messed up.
Switchig power supply designers often parallel many small electrolytics to get a low ESR. While not as cheap as a single big one this might work OK.
Bi-polar electrolytics are a special case, most are designed for high current but not necessarily low ESL. Putting two polar electrolytics back to back just doubles the ESR problem.
Me, I'd pay the money for a film. In production quantities the price can be much more reasonable, if you're building only one (two?) then get over the cost, buy a decent film.
Bi-polar electrolytics are a special case, most are designed for high current but not necessarily low ESL. Putting two polar electrolytics back to back just doubles the ESR problem.
Me, I'd pay the money for a film. In production quantities the price can be much more reasonable, if you're building only one (two?) then get over the cost, buy a decent film.
At the quantities and values I'm looking at..even a 4x price reduction..which is highly unlikely, is still too much-in terms of film prices. I dunno. I'll have to look at it a bit more closely. The economies of scale don't kick in for quite a while, when one is just starting production.
The pricing gets so high, it starts to make a case for suppying active speakers to the customer.
The pricing gets so high, it starts to make a case for suppying active speakers to the customer.
Have you considered?:
http://cgi.ebay.com/100-uF-Oil-Run-...oryZ4662QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
The Shizukis are excellent caps, brass as opposed to the steel terminals of domestic ASC caps. Brass or steel, still way preferable to electros for many.
http://cgi.ebay.com/100-uF-Oil-Run-...oryZ4662QQssPageNameZWDVWQQrdZ1QQcmdZViewItem
The Shizukis are excellent caps, brass as opposed to the steel terminals of domestic ASC caps. Brass or steel, still way preferable to electros for many.
Member
Joined 2004
Hi
ClarityCap just finished their extensive research on capacitors. More exactly, it looks like they can relate measurements to the sound of their capacitors.
Research paper
Kind regards
Roland Breteler
www.capsandcoils.com
ClarityCap just finished their extensive research on capacitors. More exactly, it looks like they can relate measurements to the sound of their capacitors.
Research paper
Kind regards
Roland Breteler
www.capsandcoils.com
Do you call that a research paper ? Is that a joke ?😀
I'm not saying they DIDN'T figure out things right or that ClarityCaps are bad. It's just the text in your link is a rather weird mix of pseudo-science and marketing. It's like a research paper written by the marketing department. Full of generally known things, lacking ANY real information or measurement (the graphs and picture are only there to look cool).
Parameter 1 vs parameter 2 ?? Yeah sure, I can also write a similar text in which I "show" how I reach perfection.
I'm not saying they DIDN'T figure out things right or that ClarityCaps are bad. It's just the text in your link is a rather weird mix of pseudo-science and marketing. It's like a research paper written by the marketing department. Full of generally known things, lacking ANY real information or measurement (the graphs and picture are only there to look cool).
Parameter 1 vs parameter 2 ?? Yeah sure, I can also write a similar text in which I "show" how I reach perfection.
It would be interesting what kind of caps they come up with. Mechanical resonance is one aspect. I once evaluated a capacitor made up of multiple dielectric material. Due the difference in strink factor in the process of baking, thw layers could not fit tight and you can hear some sound when you press it hard with you fingers. This does create additional harmonics in a low pass filter below 100Hz and some other areas I cannot remember. This is however not the only issue though.roland bios said:
Member
Joined 2004
bzfcocon said:
The paper will be presented at the Institute of Acoustics (UK) this month.
The research has been done in cooperation with the University of Salford.
Taking 300.000 measurements just for a marketing paper is not very smart?
Kind regards
Roland
As I said before: I do not question the research/results ! They might be great and lead to "perfect" caps - actually I am sure they DID help a lot.
It's the written form that I find annoying, although I know there are reasons for it. It happens that I've also have worked in the academic/research area for a while, and I know that often a trade-off must be chosen as to how much information to reveal into a paper such as not to compromise commercial interests.
In this case, I find the balance clearly in the direction of not revealing much of the technical issues, but, on the other hand, there is a clear "we're the best" marketing thing that would have require more arguments, IMHO.
It's the written form that I find annoying, although I know there are reasons for it. It happens that I've also have worked in the academic/research area for a while, and I know that often a trade-off must be chosen as to how much information to reveal into a paper such as not to compromise commercial interests.
In this case, I find the balance clearly in the direction of not revealing much of the technical issues, but, on the other hand, there is a clear "we're the best" marketing thing that would have require more arguments, IMHO.
roland bios said:
Member
Joined 2004
Yes your are right about the "we" are the best part. That's not scientific.
I can only pass the information i got from ClarityCap. I believe that despite some marketing phrases in it, it is quite an interesting paper. The full report/ research/ paper will follow later this year or next year. And will probably satisfy your more scientific needs.
I know how long they worked on this. So i was happy i finaly got permission to make it public. It was and is not my intention to spam.
Kind regards
Roland
I can only pass the information i got from ClarityCap. I believe that despite some marketing phrases in it, it is quite an interesting paper. The full report/ research/ paper will follow later this year or next year. And will probably satisfy your more scientific needs.
I know how long they worked on this. So i was happy i finaly got permission to make it public. It was and is not my intention to spam.
Kind regards
Roland
- Status
- Not open for further replies.