|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
66.220.116.223
In Reply to: RE: Linear Power Supply or Battery on Mac Mini - pics posted by AbeCollins on December 21, 2014 at 18:27:01
Your last plot shows exactly what I have been talking about. The power is drooping with HF current transients. A faster responding regulator is needed to eliminate these droops.
Follow Ups:
Right. But all that is shown is the effct on the power supply.
It is what happens within the computer power and data rails that is important.
An expensive way to show scope shots which Tony Lauck would sure like, but without indications of what happens in the sub lf and high hf ranges in the MB power rails, so what?
But how do you know that these 'droops' are detrimental ?And remember, those measurements were taken before the additional filtering module installed within the Mac Mini. And another thing, decoupling caps right at the Vcc source pin to a component tend to help with those fast 'droops'. There are several such caps in most digital designs.
I'm not saying that a 'faster' regulator won't improve the screen shot but how important it might be is another question.
Edit: I just posted two more scope screen shots and it appears that the battery actually has better transient regulation, although it has more wide band noise under steady state DC load. Pick you poison. ;-)
Edits: 12/21/14
The only way I know is by anecdotal evidence, my ears.Decoupling caps should eliminate these HF droops, but they didn't in this case. If the decoupling is done right, the power supply only sees lower frequency current transients. The voltage droops on the power only get worse as the get to the loads, not better.
Your battery is better than your linear, but that just says that your linear is not very good. I have compared several LI battery supplies with ultracaps on the output to the linear I use. No contest. The fast linear wins every time.
Edits: 12/23/14
I have compared several LI battery supplies with ultracaps on the output to the linear I use. No contest. The fast linear wins every time.
I'm not sure how much help the ultracaps should be with higher frequency transients. Much smaller value caps in parallel might actually do better. I'm thinking of the parasitic inductance of the larger caps having a cancelling effect on it's effectiveness at higher frequencies. As you should know, a bypass capacitor becomes more inductive at higher freqs where even lead lengths, as short as they may be, become a factor.
Do you know of an extremely 'fast' 12V linear power supply that can source upwards of 6 to 8 Amps on peaks? I know that the Belleson local regulators are good for only 2-Amps or so, so they don't qualify for my application.
What I don't get is why the concern about ultra fast responding power supplies being good despite the inductance of the cabling vs. well bypassed stiff batteries being bad because of the inductance of the cabling.
It makes no sense as described. There has to be something else going on.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Nothing else going on. The speed makes a BIG difference. Once you hear it, you are a believer.
It comes down to the undesired inductive reactance in the capacitor design, capacitor leads, and the power supply DC cables having a cancelling effect on the desired capacitive reactance as the frequency goes up.
In a bypass capacitor, you want a low resistance (reactance actually because we're talking AC) path to ground for the high frequency noise.
In other words, a perfect capacitor will have capacitive reactance only, but that is not the case in the real world. The capacitor is more correctly represented as having series resistance and series inductance along with it's capacitance.
Capacitive reactance is inversely proportional to frequency so as frequency goes up, the reactance goes down, making for an easier path for bypassing high frequency noise (to ground).... but, inductive reactance goes up with frequency cancelling the effectiveness of the bypass capacitor at higher frequencies.
How do you accommodate for that? Inductance in parallel goes down in value so to reduce the effects of inductive reactance in the capacitors, you can wire them in parallel. That's why you often see smaller value capacitors wired in parallel with larger capacitors.... and depending on the circuit needs, you might see several very small value capacitors wired in parallel rather than using just one very large capacitor. As you wire the capacitors in parallel, their capacitance goes up AND the total series inductance goes down. The effects of inductive reactance is reduced.
It's also worth noting that when someone speaks of impedance, impedance is comprised of resistance, capacitive reactance, and inductive reactance.
Google Electrical Reactance and separately Google Impedance.
While it might be fun to 'look at' the output of various power supplies and compare their ripple, wide band noise, and transient response, but beyond being sufficient to power the computer, does it matter much?
Personally, I would be more concerned with the power supply in the DAC and more so for the DAC's analog section.
Since my computer is now set up to accept 12VDC, it will be pretty easy to try various 12V power sources on it and listen. I have a couple linear power supplies, a couple switchers, and some batteries.
Quite right. Real Capacitors are far from the ideal text-book models. Only the most expensive Teflon and paper-oils even come close, and these are usually too large to be practical for power decoupling.
Abe - putting small value decoupling caps at the power supply output does not have much effect, because of the inductance of the cabling to the load. I have some in there, but only for noise filtering of the power supply itself, not for load regulation. The ultracaps do help with medium frequency transients. They have very low ESR.
The best fast linear supply is the paulhynesdesign.com SR5-12 or SR7-12, 5 amps and 7 amps respectively.
"But how do you know that these 'droops' are detrimental ?"The circuits that the power rail is feeding will have some inherent noise rejecting character. The key to success is understanding this character, what the supply can deliver, and how the resulting performance due to interaction of power supply and the load circuit systems compare with required performance. It's a multivariable equation at least.
"And remember, those measurements were taken before the additional filtering module installed within the Mac Mini. And another thing, decoupling caps right at the Vcc source pin to a component tend to help with those fast 'droops'. There are several such caps in most digital designs."
It's true, for differential noise measurements like this you definitely have to measure right at the load to get a realistic idea of what is truly going on where it counts. Measuring upstream of the filtering will tell you very little about the power quality as seen by the load.
Though most audio gear these days, depending on application of course, is pretty awesome at rejecting differential noise but much less effective at rejecting common mode noise which is coincidentally cropping up more regularly.
Less talked about and less well spec'd on power supply equipment is the common mode noise performance on the outputs with respect to earth referencing as most common audio equipment uses. Getting common mode noise way down in earth referenced audio gear power supplies is critical. The humongous problem here is testing gear for doing proper evaluation at an audiophile level is going to cost a fortune. Your 8 bit scopes and high end fluke dmm's just wont have what it takes to see -120dB throughout the audio band and a preferably a reasonable bit beyond. A good audio adc could probably be helpful fur the lower frequencies but what about the 200kHz and up? the answer is $$$$$ gear I can't afford. Even my work won't buy that kind of stuff. It's specialty gear and gonna cost ya.
Edits: 12/23/14
I mostly agree with you. But as I mentioned to fmak (who can't seem to understand), my measurements are part of a preliminary functional test, i.e. does it appear within reason and does it work?
It's true, for differential noise measurements like this you definitely have to measure right at the load to get a realistic idea of what is truly going on where it counts. Measuring upstream of the filtering will tell you very little about the power quality as seen by the load.
Absolutely! With hundreds of individual 'load points' (Vcc pins) to test within the computer, and the thousands of variables which can affect the results, we can only assume that the power at each 'load point' is sufficient for proper operation.
On a larger scale, I don't own a high-end spectrum analyzer to view the end result at the output of the external DAC or power amp. There are very few here who do. At least one manufacturer has posted a screen shot from an Audio Precision analyzer.
"Absolutely! With hundreds of individual 'load points' (Vcc pins) to test within the computer, and the thousands of variables which can affect the results, we can only assume that the power at each 'load point' is sufficient for proper operation."
There are billions of load points in a computer, but most of them are inside the chips and hence inaccessible without very specialized equipment. The real issue is the definition "proper operation".
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
"The real issue is the definition "proper operation"."
Who else would define it besides the component manufacturer? If a manufacturer doesn't supply the info good designers need the good designers won't use their stuff.
"The real issue is the definition'proper operation'."Who else would define it besides the component manufacturer? If a manufacturer doesn't supply the info good designers need the good designers won't use their stuff.
Not correct. The proper operation of the chip according to the chip manufacturer does not necessarily imply proper operation of the motherboard according to the computer system manufacturer. The proper operation of the computer system according to the computer system manufacturer does not imply proper operation of the audio playback chain, according to the audiophile who assembled the playback chain.
If there were complete and accurate engineering specifications at each level and designs used subsystem components in accordance with these specifications, then these problems would not exist, or at the least, it would be possible to identify a particular organization or person who screwed up. However, this is not "real world". This is a particular problem when a digital computer system is used as part of an audio chain that has inadequate isolation from digital noise. Here, there are no specs whatsoever, e.g. there is no DAC on the marketplace that quotes isolation from power noise or digital interconnect noise. Indeed, there is no standard or even widely known method of measuring the impact of this noise on the audio output from the DAC. There are no known engineering methods of correlating measurements with sound quality.
Tony Lauck
"Diversity is the law of nature; no two entities in this universe are uniform." - P.R. Sarkar
Edits: 12/26/14
"Not correct. "
Sure it is. At each level of system complexity you get the choice of going integrated or piecing together discretes of your own choosing. Going integrated allows one the luxury of leaving the worrying up to the manufacturers. Limiting your choices to what has been integrated and made commercially available obviously has it's limitations as well. pick your poison.
But for practical reasons we can't get to the 'load points' within a chip so that's why I reduced the count down to hundreds. ;-)
Since we're talking about a computer, and I was talking about preliminary functional tests.... The fact that the computer boots up on the various power supplies and is working as a computer should and like it did before, that would be my definition for 'proper operation'.
You don't need an 'audiophile' power supply to make a computer operate properly.
there is no point in any tests, preliminary or otherwise, if you don't carry out the preliminary setup correctly. This is the first mandatory step towards any credible test or meaurement.
You appear to be repeating yourself at every opportunity that my setup was not correct for the preliminary functional tests that I was performing. So please explain what was not correct for these specific tests.
This shows that you don't know what you are saying or doing.
Go away Fred, you're making an ass of yourself.
the ass that doesn't know what he is doing?
.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: