@gootee
You may be interested to know that Dan d'Agostino uses a LOT of 47 uF capacitors to decouple his rails everywhere. I haven't counted them, but off hand, I'd say no less than 10 or more per rail, in addition to smaller value caps. The main PSU uses 6,800 uF caps in parallel, if memory serves, three in parallel per each rail. But, given the fact that EVERYTHING is electronically regulated, and that his supply lines are rather high, even those seemingly not-too-big caps in fact provide a LOT of energy storage when added up.
You may be interested to know that Dan d'Agostino uses a LOT of 47 uF capacitors to decouple his rails everywhere. I haven't counted them, but off hand, I'd say no less than 10 or more per rail, in addition to smaller value caps. The main PSU uses 6,800 uF caps in parallel, if memory serves, three in parallel per each rail. But, given the fact that EVERYTHING is electronically regulated, and that his supply lines are rather high, even those seemingly not-too-big caps in fact provide a LOT of energy storage when added up.
In a design note from Motorola, I think the one in which they propose two power amps, there's a good sideline on how to calculate your capacitor requirements.
Not scientific, more a rule of the thumb, but most useful nevertheless.
Essentially, they say that experience has taught us that we need 1...2 Joules of energy for every 10 Watts of dissipated power, depending on how complex and demanding the load is.
So when Wayne
D) sits down to work it out, he can easily "double down" as he pleases. Since one knows what the rails will be, and assuming we use plus and minus rails, PER RAIL capacitance can be worked out using a simple formula:
Joules = (V+ x V+) Farads
If my rails are +/- 50V, and I use 10,000 // 10,000 uF // 3x2,200uF for each device, I can have:
(50 x 50) 0.0266 = 66.5 Joules
good enough for 332.5 Watts into an evil load dmanind the maximum from the amp, to 665 Watts into a pristine clean and easy load.
Over time, this formula has shown itself to be very true indeed.
Obviously, it applies to the power supply only, for this to REALLY happen in REAL life, you also need beefy power tranformers, powerful rectifiers and an amp output stage capable of actually delivering this kind of power.
For power amp transformers, again there's a rule of the thumb - use as many VA in the transformer as you want watts out. If you want a 2x150W amp, minimum transformer value should be (2x150)2 = 600 VA for a stereo amp.
Not scientific, but hey, it works.
Not scientific, more a rule of the thumb, but most useful nevertheless.
Essentially, they say that experience has taught us that we need 1...2 Joules of energy for every 10 Watts of dissipated power, depending on how complex and demanding the load is.
So when Wayne
D) sits down to work it out, he can easily "double down" as he pleases. Since one knows what the rails will be, and assuming we use plus and minus rails, PER RAIL capacitance can be worked out using a simple formula:Joules = (V+ x V+) Farads
If my rails are +/- 50V, and I use 10,000 // 10,000 uF // 3x2,200uF for each device, I can have:
(50 x 50) 0.0266 = 66.5 Joules
good enough for 332.5 Watts into an evil load dmanind the maximum from the amp, to 665 Watts into a pristine clean and easy load.
Over time, this formula has shown itself to be very true indeed.
Obviously, it applies to the power supply only, for this to REALLY happen in REAL life, you also need beefy power tranformers, powerful rectifiers and an amp output stage capable of actually delivering this kind of power.
For power amp transformers, again there's a rule of the thumb - use as many VA in the transformer as you want watts out. If you want a 2x150W amp, minimum transformer value should be (2x150)2 = 600 VA for a stereo amp.
Not scientific, but hey, it works.
Yet it has also not yet been demonstrated that amps with large THD sound or low or no feedback sound better, in fact there are more hi end amps demonstrating very low THD and higher feedback winning awards than the other method........ makes one think
Dejan,
There are two issues here.
One is the evidence that low value power supply capacitors limit power delivery (they do not, undersized mains transformers do however).
The other issue what I personally, as supposed High End Guru and merchant of questionable (according to some) audio ideas, concepts and devices plan to do for myself. I mean they'd revoke my High End Guru licence if I made the Amp with a 250VA Torroid and 4 * 4,700uF, even if it did deliver 180W into 8 ohm (one channel) and was adequate with music...
Less tongue in cheek, as a rule (I have yet to find exceptions), there are no limits up to which one may increase PSU capacitance AND power transformer size (other than practical/monetary), with some benefit. Though at the levels I intend to use we are already pretty far past the point of diminishing returns.
Then again, there is something to an amp that runs of eight stacked 12V SLA Batteries (for +/-48V) that is hard to get with mains powered amp's...
Ciao T
There are two issues here.
One is the evidence that low value power supply capacitors limit power delivery (they do not, undersized mains transformers do however).
The other issue what I personally, as supposed High End Guru and merchant of questionable (according to some) audio ideas, concepts and devices plan to do for myself. I mean they'd revoke my High End Guru licence
if I made the Amp with a 250VA Torroid and 4 * 4,700uF, even if it did deliver 180W into 8 ohm (one channel) and was adequate with music... Less tongue in cheek, as a rule (I have yet to find exceptions), there are no limits up to which one may increase PSU capacitance AND power transformer size (other than practical/monetary), with some benefit. Though at the levels I intend to use we are already pretty far past the point of diminishing returns.
Then again, there is something to an amp that runs of eight stacked 12V SLA Batteries (for +/-48V) that is hard to get with mains powered amp's...
Ciao T
Hi,
As I happen to have 36,000uF per rail and 56V rails it would seem I'm well past that (twice actually). So I should be fine for 665W into a nasty load, which is probably quite nice for an Amp I'd call notionally 150W... Maybe my mains transformer is a wee bit undersized.
Ciao T
As I happen to have 36,000uF per rail and 56V rails it would seem I'm well past that (twice actually). So I should be fine for 665W into a nasty load, which is probably quite nice for an Amp I'd call notionally 150W... Maybe my mains transformer is a wee bit undersized.
Ciao T
Hi,
Well, first, there are by far fewer Amplifiers of (for example) the SE DHT no-NFB variety and Sturgeons rule applies. So the statistics you quote do not help us much.
As to: " it has also not yet been demonstrated that amps with large THD sound or low or no feedback sound better", I think the solution for everyone is to go and listen. We must also realise that we do not listen amplifiers, but to whole systems.
Ciao T
Yet it has also not yet been demonstrated that amps with large THD sound or low or no feedback sound better, in fact there are more hi end amps demonstrating very low THD and higher feedback winning awards than the other method........ makes one think
Well, first, there are by far fewer Amplifiers of (for example) the SE DHT no-NFB variety and Sturgeons rule applies. So the statistics you quote do not help us much.
As to: " it has also not yet been demonstrated that amps with large THD sound or low or no feedback sound better", I think the solution for everyone is to go and listen. We must also realise that we do not listen amplifiers, but to whole systems.
Ciao T
Music waveforms are non-periodic. Fourier transform, strictly speaking, can not be applied to non-periodic functions. But it is nevertheless applied, with definite degree of accuracy. A devil is in the limitations of its applicability. For some cases limitations are not essential, but they are essential for exploring audio perception.
I take my stats from reading about 15 audio publications from tubes to solid state, a lot of the facts about the circuits are unknown to the writers of these so they just state the little the manufacturer tells them.
The 2 best amps I have listened to are on opposite sides of the fence. One a french amp with very high feedback Lavardin (more so then usual) and one with fairly low NFB a electrocompaniet monster I own. Each one has its stronger and weaker points but I cannot say that one or the other sounds better because of the amout of feedback used or their THD figures.
not true as engineers know - yes the math is a little more complex to satisfy pure mathematicians but can still be made to work for bandwidth and time limited continuous signal representation
the simple gedankenexperiment version: for recorded music we can just loop the recording - instant "infinitely periodic" signal to analyze
Windowing functions are well studied - you can calculate the time domain "flitering" loss - can be way below real signal thermal noise, beyond recording microphone bandwidths
the time series and the full complex Fourier repesentaion of a Digital Audio signal are exact Duals - no information is lost in the conversion from one ot the other (finite rounding errors can be made small by working with long enough wordlength)
sorry for your "conspiracy theory" - but engineers do test, verify, use tools that really do work with real world signals, electronics - just how do think the 3 Mbaud DSL signal is made to work over voice telephone twisted pair??
by a million mathematically illiterate monkeys with soldering irons?? – that’s more the DiyAudio “just try it” version
(a little overlapping editing - SY has given the example before - its still a good example to "debug" your "nonperiodic" signal objection without heavy math)
the simple gedankenexperiment version: for recorded music we can just loop the recording - instant "infinitely periodic" signal to analyze
Windowing functions are well studied - you can calculate the time domain "flitering" loss - can be way below real signal thermal noise, beyond recording microphone bandwidths
the time series and the full complex Fourier repesentaion of a Digital Audio signal are exact Duals - no information is lost in the conversion from one ot the other (finite rounding errors can be made small by working with long enough wordlength)
sorry for your "conspiracy theory" - but engineers do test, verify, use tools that really do work with real world signals, electronics - just how do think the 3 Mbaud DSL signal is made to work over voice telephone twisted pair??
by a million mathematically illiterate monkeys with soldering irons?? – that’s more the DiyAudio “just try it” version
(a little overlapping editing - SY has given the example before - its still a good example to "debug" your "nonperiodic" signal objection without heavy math)
Last edited:
Actually, it can. Any function of finite duration can be made periodic without changing the physics. For example, let's say I want the frequency spectrum of Roland Kirk's performance of Seranade to a Cuckoo. It's 4:32 long. Call the time domain function SC(t). The boundary conditions are SC(0) = SC(4:32) = 0, since the song starts and ends. Then put SC(t + 4:32) = SC(t - 4:32) = SC(t). Voila, a periodic function!
Hi,
Well, this we can agree upon.
One of the best solid state Amp's I know is from Dennis Morecroft (DNM) and it is a high feedback design. One of the worst 'solid state Amp's I ever listened to is also a high feedback design. Either one has ephemeral THD.
Equally, some of the Tube Amp's I liked best contrast between zero NFB and one of my design which had ton's of Shunt NFB (excluding the output transformer).
Ciao T
Well, this we can agree upon.
One of the best solid state Amp's I know is from Dennis Morecroft (DNM) and it is a high feedback design. One of the worst 'solid state Amp's I ever listened to is also a high feedback design. Either one has ephemeral THD.
Equally, some of the Tube Amp's I liked best contrast between zero NFB and one of my design which had ton's of Shunt NFB (excluding the output transformer).
Ciao T
Dejan,
There are two issues here.
One is the evidence that low value power supply capacitors limit power delivery (they do not, undersized mains transformers do however).
The other issue what I personally, as supposed High End Guru and merchant of questionable (according to some) audio ideas, concepts and devices plan to do for myself. I mean they'd revoke my High End Guru licence
Less tongue in cheek, as a rule (I have yet to find exceptions), there are no limits up to which one may increase PSU capacitance AND power transformer size (other than practical/monetary), with some benefit. Though at the levels I intend to use we are already pretty far past the point of diminishing returns.
Then again, there is something to an amp that runs of eight stacked 12V SLA Batteries (for +/-48V) that is hard to get with mains powered amp's...
Ciao T
Oh, don't you worry, T., you ARE very questionable. For every post you make, at least several questions come to mind.
Heck Thorsen, we all know for our own experiences that there is good and there is better, small and smaller, big and too big, etc. But above all that stands tha fact that each and every one of us hears things (pun intended) his own way.
At the end of the day, there is but one question left (in DIY) - do I like it?
My personal liking of whatever is, I admit it, linked rather closely to how it was made. If I see some serious work built into it, even if I'd do it differently, I will tend to like it.
If I see one of those "take out everything until it stops working, then go back one step" jobs, I will tend not to like it simply because I would know that there was a price to pay somewhere.
As a young man, I did some drumming with some friends, so I am very well acquainted with how a set of drums sounds - this has left lasting consequences on me, because if a pair of speakers, an amp or whatever can't do the drums right, it's off the table for me no matter how good elsewhere it may be. Mind you, this does not mean it has to do the drums right and never mind the rest, no, doing the drums right also means keeping them in proper perspective with the rest. Nothing must push or pull.
Ultimately, watching out for the drum section means both lows and highs - hit that brass and it goes way over 20 kHz with harmonics.
So, at the end of the day, if I like it, frankly I don't care much what anybody else says about it. I also don't give a hoot if it uses bipolars, tubes, FETs, or horse dung. Or whether it uses grobal NFB or not, 0 dB or 100 dB. The last, er, word belongs to my ears only, non-negotiable.
As for power and capacitors, I do agree with your view on power transformers, but I disagree regarding electrolytics. Make the same circuit in two versions, both with an oversized transformer, but one with say 10,000 uF caps and the other with say 22,000 uF caps, in my experience, 9 out of 10 times the one with larger caps will actually make better music.
And thus it's impossible to get distortion figures comparable to those of a low distortion amplifier design. That doesn't give us a "free-to-ignore-distortion" card that we can use whenever our microphone, mic preamp, AD/DA or amplifier gear fails horribly in that aspect.
I do wish we could improve upon loudspeaker performance in that respect (instead of arguing that we don't need to) because, IME, low distortion drivers/loudspeakers are the ones that truly "stay out of the way" (at least compared to their lower performance counterparts) and let you enjoy (or run away from) the recording.
[Quads, ATCs, Lipinskis come to mind out of the ones I've heard. Probably Bruno's Grimm LS-something will perform on par or even better.]
I do wish we could improve upon loudspeaker performance in that respect (instead of arguing that we don't need to) because, IME, low distortion drivers/loudspeakers are the ones that truly "stay out of the way" (at least compared to their lower performance counterparts) and let you enjoy (or run away from) the recording.
[Quads, ATCs, Lipinskis come to mind out of the ones I've heard. Probably Bruno's Grimm LS-something will perform on par or even better.]
Sure.
When I used tube amps on campground for festivals of Russian bards I saw that power in outlet goes down to 90 and even 80 volts instead of 120. This year I am going to use class D amps powered from marine batteries, charged constantly from outlet by SM chargers.
Hi,
So, high distortion in a Speaker is Fine, but high distortion in an amplifier is bad, even if the actual air modulations arriving at the listener have high distortion?
There is something in this logic that I cannot reconcile. That is not even the "two legs bad, four legs good" logic of Orwell's "Animal Farm", it is more "two legs bad when I say it is bad" and "two legs good when I say it is good".
This is not logic or science, this is religion or politics, moral relativism all included.
That may does not. But if we can demonstrate the the distortion is reliably inaudible (regardless of amount), it does give us such card. And the evidence in published sources points to that.
I do not use the high distortion of speakers as an excuse, I use it as ILLUSTRATION that a high THD does not preclude high fidelity.
I do know either drivers used or the actual speakers you mention.
I would not call any of them "low distortion" at rated input power.
If we lower the power, we may find that even a "high THD" Amp has very little actual distortion (only noise) at such low power, so we are back at square one.
I am only asking the proponents of low THD in electronics for one things, that is proof that it reliably produces good sound or that reductions in THD reliably produce improvements in sound quality beyond some basic limits, if they wish to have their position seriously considered...
Until such proof is presented I suggest that any who claim "Low THD is required" or "THD is a measure of quality where lower THD means a device is better" are in the company of many religeons with grand pronouncements and no evidence, reliant strictly on blind belief.
Do not get me wrong, if you want to worship the god of low THD, be my guest.
And when I am in the chapel of this good I will be sure to pay lip service to his worship so as to not offend.
However do not expect to profess or agree with irrational beliefs in public or to join your cult, UNLESS you present evidence...
Ciao T
So, high distortion in a Speaker is Fine, but high distortion in an amplifier is bad, even if the actual air modulations arriving at the listener have high distortion?
There is something in this logic that I cannot reconcile. That is not even the "two legs bad, four legs good" logic of Orwell's "Animal Farm", it is more "two legs bad when I say it is bad" and "two legs good when I say it is good".
This is not logic or science, this is religion or politics, moral relativism all included.
That may does not. But if we can demonstrate the the distortion is reliably inaudible (regardless of amount), it does give us such card. And the evidence in published sources points to that.
I do not use the high distortion of speakers as an excuse, I use it as ILLUSTRATION that a high THD does not preclude high fidelity.
I do know either drivers used or the actual speakers you mention.
I would not call any of them "low distortion" at rated input power.
If we lower the power, we may find that even a "high THD" Amp has very little actual distortion (only noise) at such low power, so we are back at square one.
I am only asking the proponents of low THD in electronics for one things, that is proof that it reliably produces good sound or that reductions in THD reliably produce improvements in sound quality beyond some basic limits, if they wish to have their position seriously considered...
Until such proof is presented I suggest that any who claim "Low THD is required" or "THD is a measure of quality where lower THD means a device is better" are in the company of many religeons with grand pronouncements and no evidence, reliant strictly on blind belief.
Do not get me wrong, if you want to worship the god of low THD, be my guest.
And when I am in the chapel of this good I will be sure to pay lip service to his worship so as to not offend.
However do not expect to profess or agree with irrational beliefs in public or to join your cult, UNLESS you present evidence...
Ciao T
Last edited:
Exactly. The keyword is, "In a useful way". Temperature of patients, it's change, may be useful indicator of their health. But when we sum all temperatures of patients in the clinic and divide by their number we get some average number that can be no more useful than THD, while it can signal that some patients in the clinic have fever or inflammations.
Thanks for clarification, but do you believe that Fourier transform of such function will be useful enough, in order to characterize low-level signal constituents, like motion of fingers along the strings? Could they be visible over "calculation rounding noise"? Practical calculation are limited in accuracy, not any task can be solved numerically. Instruments like spectra analyzers also have their limitations.
Why not to directly compare just signal waveforms, without spectral tricks?
I feel that many are bounded to spectra, because they are the ground of convenient common practice, nothing more.
- Status
- Not open for further replies.