I have googled a bit and found some info why its possible the LM4780 can function with low capacitance and still sound good.
Gainclone - Wikipedia, the free encyclopedia
where its states that it is because of the "excellent PSRR" what means it has its own voltage regulator inside the chip.
Amplifier
The original site of the gain card: "If energy supply depends on the capacity of filter/condensers, you can easily lose the freshness of sound. The high capacity transformer of Model 4700 (170 VA) regulates enough energy to support the extremely small filter/condenser (1000 uF) of Model 4706,enabling it to trace avalanches of fff."
Well, anyway, With my Mission M52 speakers I'm very pleased the way my amp is now, if I change to the Vifa Basis speakers I will build another LM4780 parallel set with high capacitance to bi-amp the speaker and use for the amp for the low drivers high capacity caps to prevent clipping and if this amp looses its "freshness" in high and mid it doesn't matter.
Greetz,
Joery
Gainclone - Wikipedia, the free encyclopedia
where its states that it is because of the "excellent PSRR" what means it has its own voltage regulator inside the chip.
Amplifier
The original site of the gain card: "If energy supply depends on the capacity of filter/condensers, you can easily lose the freshness of sound. The high capacity transformer of Model 4700 (170 VA) regulates enough energy to support the extremely small filter/condenser (1000 uF) of Model 4706,enabling it to trace avalanches of fff."
Well, anyway, With my Mission M52 speakers I'm very pleased the way my amp is now, if I change to the Vifa Basis speakers I will build another LM4780 parallel set with high capacitance to bi-amp the speaker and use for the amp for the low drivers high capacity caps to prevent clipping and if this amp looses its "freshness" in high and mid it doesn't matter.
Greetz,
Joery
Last edited:
I have no problem believing it could sound excellent on the highs and mids with only 1000 uF. And the bass, too, could be perfect, at lower output power levels.
But all of the music signal current comes from the capacitors, except during a charging pulse. So at some higher output power level, and some lower frequency, the capacitors will discharge too far, before the next charging pulse, and the amp will clip.
You can have nearly-perfect reproduction with almost any value of reservoir capacitance. It's just that with lower values, the rated maximum output power level (where clipping starts) must be lower, for any given "lowest frequency the amp can handle" spec.
But a lot of music signals are rather sparse, compared to those that are assumed when calculating the max rated output power. So the max power rating is usually a very conservative estimate and the actual performance with real music is usually better than predicted.
The "solution", if one is even needed, is very simple: If you ever get to the point where you want more output power but then some particular music makes bad things happen, just add a little more capacitance until it sounds OK again.
However, be aware that when clipping happens, the sharp edges that are created could send a spray of very high frequencies to the output, which has been known to sometimes destroy tweeters.
My preference would be to design a system so that nothing bad could possibly happen, to either the music or the hardware, even with the volume turned all the way up.
But all of the music signal current comes from the capacitors, except during a charging pulse. So at some higher output power level, and some lower frequency, the capacitors will discharge too far, before the next charging pulse, and the amp will clip.
You can have nearly-perfect reproduction with almost any value of reservoir capacitance. It's just that with lower values, the rated maximum output power level (where clipping starts) must be lower, for any given "lowest frequency the amp can handle" spec.
But a lot of music signals are rather sparse, compared to those that are assumed when calculating the max rated output power. So the max power rating is usually a very conservative estimate and the actual performance with real music is usually better than predicted.
The "solution", if one is even needed, is very simple: If you ever get to the point where you want more output power but then some particular music makes bad things happen, just add a little more capacitance until it sounds OK again.
However, be aware that when clipping happens, the sharp edges that are created could send a spray of very high frequencies to the output, which has been known to sometimes destroy tweeters.
My preference would be to design a system so that nothing bad could possibly happen, to either the music or the hardware, even with the volume turned all the way up.
Last edited:
To be honest, I actually have a problem to believe that it could sound well with low capacitance but I found out it for sure does not sound bad.
The whole theory that bigger caps should always sound better is something that I would have agreed with if I would not have build the amp I have build now.
To bad I don't have a scope here because then I could have measured input and output signals with different volume levels to see when things go wrong.
Sims for me are maybe a lead to how things maybe will be but I know in my profession that theoretical assumptions often do not present the reality because easily underestimated parameters that in the first place seem to have no or little influence can have much more influence than expected.
In this thread until now I have seen many reasons why bigger caps are always better but no explanation why small caps seem to sound better with gain card or gain clone amps....
When I see the gain card site with all the references how good this amp sounds (with low capacitance) its for me hard to believe that so many people think this amp sounds SO great only because some hype has spread.:
http://www.sakurasystems.com/reviews.html
I think there must be a reason for this and I still don't see this explained.
Last edited:
Did you see this one already? Mick Feuerbacher Audio Projects
The prototype of the gaincard. It gives you some insight into the design philosophies used.
It is really easy to build an unstable gainclone, add some long PCB tracks to the feedback network, long leads to the decoupling caps, a nice groundplane, create some good ground loops, then you will have an oscillating, humming and overheating gainclone.
Ill try to think up something positive about low capacitance.
One thing relevant about capacitance on the gaincard site is that 1000uF gives you 25 Watt per channel and 2200uF gives you 50 watt per channel.
I was also sceptical about using simulations, but its quick, cheap and safe.
The prototype of the gaincard. It gives you some insight into the design philosophies used.
It is really easy to build an unstable gainclone, add some long PCB tracks to the feedback network, long leads to the decoupling caps, a nice groundplane, create some good ground loops, then you will have an oscillating, humming and overheating gainclone.
Ill try to think up something positive about low capacitance.
One thing relevant about capacitance on the gaincard site is that 1000uF gives you 25 Watt per channel and 2200uF gives you 50 watt per channel.
I was also sceptical about using simulations, but its quick, cheap and safe.
My simulations and my theoretical calculations and my testing have all strongly suggested to me that even small reservoir capacitance values can result in accurate audio at a power amplifier's output. As I have said, the only difference is that the RATED maximum output power level must be lowered, for lower capacitance values. The equations that I derived clearly show that. So I am not surprised at all by your good results.
However, I can NOT agree that smaller capacitance values should or would "sound better". And the reason given in the quote from the website that you posted is nonsensical, especially since ALL of the output current comes from ONLY the capacitors, a majority of the time.
Larger capacitance values are more robust, in the sense that they can produce more-accurate music output at higher output power levels, where the smaller capacitance would allow clipping to occur. So that's one major point in their favor.
And at an output level where clipping would not occur with either a given smaller C value or a larger C value, the only major difference between the two should be that the larger C value would not allow the rail voltage to fluctuate as much, which should always result in a more-linear-performing amplifier (since PSRR is never perfect). The difference in the accuracy of reproduction might be small, between the two C values, in that case. But the larger C value would result in more-accurate sound reproduction.
And again, MY over-riding goal, in all of this, is the most-accurate reproduction of the input signal, whether I "like the sound" or not. Therefore, a claim that something "sounds better" carries almost no weight, in making my judgments or decisions about audio reproduction, and is weighted even less when it's not based on my personal listening. It's simply unreliable as a measurement tool for judging audio reproduction, which has been absolutely proven, many, many times.
(Not to put too fine a point on it, but...) There are those for whom "personal preference" is a valid method of judging audio reproduction and choosing audio equipment. I believe that attitude to be an abhorrent anti-science evil.
So I will have to vote for "larger" reservoir capacitance values providing more-accurate sound reproduction, in general, while holding that "sounds better" must usually be considered to be mostly meaningless and irrelevant.
For anyone who might be interested, I will attach the results of a little bit of the work that I did in the past that is related to the reservoir capacitance and power supply issues for class AB audio power amplifiers. Note that the ".txt" will need to be removed from the xls file's name. Also, below is a link to a post about choosing the capacitance and transformer values/ratings, which has a link to one version of my other spreadsheet, which simulates the power supply, and contains the scalable transformer model's equations.
http://www.diyaudio.com/forums/soli...n-attempt-number-2-simpler-9.html#post3532280
Cheers,
Tom Gootee
Last edited:
I forgot to include the paper describing my other spreadsheet, i.e. the one that actually simulates the power supply, which includes the scalable transformer model. Anyone who thinks it should be easy to find a closed-form solution for the equations of the simple transformer-rectifier-capacitor power supply should read this. It is attached.
NOTE: Remove the ".asc" from the filename. It is a PDF file but .asc files have a larger file size limit, on diyaudio.
NOTE: Remove the ".asc" from the filename. It is a PDF file but .asc files have a larger file size limit, on diyaudio.
Here to me you touch an interesting note.
What actually is the signal of a music file?
Of course it starts with the recording of it, its often done with an analog microphone but it can also be produced with a digital source like a synthesizer but the evaluation of the sound is always done by the mixer/producer listening to it through an amplifier and speakers.
To hear what the mixer/producer intended you actually should listen to the signal with exactly the same amplifier, speakers, room etc.
If you would be able to exactly reproduce the sound signal, the music would not sound like the mixer/producer intended it to be.
This all makes the whole theory of sound reproduction not as simple as you describe and it means that "it sounds good" is a phrase that can not simply be considered to be mostly meaningless and irrelevant
I don't think that I can disagree with that. But I think that my view is consistent with it.
I believe that if the recorded music "sounded good" when it was made, and we do a good job when implementing our system, then it should also sound good to us. But I don't believe that "sounds good" can be relied upon as a measurement or evaluation tool, especially when it's "hearsay", even though sometimes it might work OK.
We have no way of knowing what the producer/mixer heard. So the best we can do is to try to reproduce the recorded signal as accurately as possible.
One of my sub-goals here is to develop design and analysis methods and criteria that can be proven to be valid, and that are repeatable, by anyone. I don't know of any other good way to go about it. And the ear-brain systems of people do not fit well into that model, unfortunately.
I am sorry if it seemed like I was trying to make it sound simpler than it is. My intent was actually more like the opposite of that.
Last edited:
Correct me if I'm wrong but I think you try to replicate the audio input audio signal as best as possible and in theory that is the best you can do being the one on the other side of the audio replication circle (listener).
I can see from the work you have put into that its very complicated to do so.
What I mean with ""simple" is that even you could accomplish that, it doesn't mean you make the ideal system for replicating the sound as it was intended to be.
Don't get me wrong, People like you are very much needed to get the sound experience as it was intended.
I don't have much experience with amp building but in the past I have been busy with speaker design for a couple of years and found out that speakers that came close to theoretical best performance really did not sound well.
Not only me but many have scratched there head a lots figuring out why this happened and I think its still not really clear why this is the case.
If there would be a single way how to reproduce the sound as it was intended by the mixer/producer of the music there would be only one amplifier and one speaker system on the market now.
Last edited:
I concur, completely, I think.
With currently-available recordings, we can never even really know how to "correctly" reproduce what was recorded, with perfect accuracy, and also would not be able to know that we had, if we did.
It seems that the best we can possibly do is to try to reproduce the recorded material from our source media, as accurately as possible.
But, of course, when we talk only about the amplifier, we are not including the entire reproduction system. The feedback is from before the speakers.
But that is more practical, I guess, than taking feedback from our brain's audio perceptions, or even just from microphones that are placed at or in our ears, although I have thought about doing that.
On the other hand, that producer/mixer person was recording from somewhere before his power amplifier, and not from microphones at his ears. So maybe it would not be a theoretically correct approach to getting the best reproduction accuracy, after all. But maybe it would at least help correct for the room we were in. I don't know. I am pretty sure that it has been tried, or at least theoretically analyzed. But I can't remember seeing any details about it.
There are threads right here on diyaudio about taking feedback from speaker cone motion (with laser or radar-like reflective sensors), but I think that most of those were only for bass. I wonder if anyone has tried it with microphones. But from what I know about field theory, there would probably be severe problems due to the exact microphone placement, because of the constructive and destructive interference patterns that are so variable, which is why placing the feedback mics at a listener's ears seemed like it might be a good idea. But then, of course, it would almost certainly only work well when doing that for one listener.
Anyway, it can be as complicated as we want it to be. But there is no perfect solution. So we just try to do the best that we can do, with whatever time and money constraints we have.
I am happy-enough with either my Magnepan planar magnetic speakers or my Vandersteen 2Ce speakers (both of which I purchased used, for very reasonable prices), with almost any good power amplifier. I think that they are both extremely accurate, as are my power amplifiers (which are either hand-made or older used ones like Adcom 545 II or 585). The sound is exquisite, with a very detailed and realistic audio image ("soundstage"). But, for all I know, the hair-raising sound quality might be mostly because of my room layout and speaker positions. I do know that if I do not position the speakers (and myself) EXACTLY right, some of the best quality gets lost.
I'd love to have a friend who were nuts about audio while not being an electronics educated person. Somehow to use him to test hearing my amplifiers. It's like my brain tend to prefer a Class-A amplifier rather than a chinese small class D just because all the effort and money invested, like I know I have to prefer the heavier one
I'd love to have a friend who were nuts about audio while not being an electronics educated person. Somehow to use him to test hearing my amplifiers. It's like my brain tend to prefer a Class-A amplifier rather than a chinese small class D just because all the effort and money invested, like I know I have to prefer the heavier one
I can imagine that people around you find you to be a real hoot.....
Hello and first of all thank you very much for this excellent 3D.
PS capacitors are an old obsession of mine.
I have a question
Has anyone tried to see how the voltage at the PS caps terminal varies during most powerful passages ?
My feeling is that in most commercial amps it will sag quite clearly, and i think this is not good.
Thanks a lot,
gino
PS capacitors are an old obsession of mine.
I have a question
Has anyone tried to see how the voltage at the PS caps terminal varies during most powerful passages ?
An externally hosted image should be here but it was not working when we last tested it.
My feeling is that in most commercial amps it will sag quite clearly, and i think this is not good.
Thanks a lot,
gino
What I'm interested in is a scoop reading (not a sim) of input vs output signal from a GC with small and big caps below the clipping level.
This would give some insight of the quality of the sound produced with big and small caps on a gain-clone.
Hi, thanks for the very interesting reply.
I would try very big uF anyway ... like 30,000 uF/rail/channel
Just to see the effect on this fluctuation.
In the ideal case the PS voltage should not sag i suppose.
Thanks and regards,
gino
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.