How Important is Power Supply Quality?

0 Members and 1 Guest are viewing this topic. Read 2612 times.

Steve

How Important is Power Supply Quality?
« on: 18 Sep 2018, 03:24 am »
Below is a link to a white paper I initially wrote in October of 1997, and recently updated. Although basic with little math, my paper deals with multiple purposes and considerations when designing the highest quality power supply for preamplifiers, and Class A, AB amplifiers.

Not only does the power supply have to provide pure DC, but also minimize any artificial artifacts from the AC line such as noise, RFI, but also deal with the musical signal at the other end. A power supply is much more difficult to design than most think. It is not just a couple of equations, but also includes parts quality.

http://www.sasaudiolabs.com/theory7a.htm

For those who need a little more help, here is some simplified information, some from my old Engineering textbook, Semiconductor and Tube Electronics by James Brazee, and RCA Radiotron Designers Handbook, written by 26 electrical engineers. I am an electronics engineer.


Let me first state that it is amazing how good the sound can be when a truly great power supply design is used with a great design and highest quality parts. Anyone can get pretty good sound from an ok power supply design, and pretty good is all you will get. Getting the best, most natural reproduction is another matter. One has to consider the entire design.

For some, my white paper is a bit too complicated, so I will start at the beginning and go step by step and prove the power supply decoupling capacitor not only handles musical signal current, but also alters the frequency response of an audio stage.
And there are usually more than one stage to design.

I will even do most of the mathematics, but I will also provide the equations in case one is interested. I will use a series of figures to help, which are located at the bottom of this post. As a suggestion, please print out those figures.

For those who wish to skip the points below, and get a quick summation, the points below prove that power supply filter capacitor (C1) is in series with RL, so actually RL + C1 (AC resistance/capacitive reactance Xc) is frequency dependent. Thus the gain of the stage is frequency dependent, so the frequency response is not ruler flat over several octaves.

Since RL has AC musical signal current flow, C1 does as well; in fact some 15 or more times Ccoupling capacitor and Rg in my example.

C1 capacitor also has AC voltage, musical signal voltage on its positive terminal. All these can be easily measured so they follow the laws of electronics.
Thus my white paper is correct.

For the rest of us, relax, take your time, and let’s get started. At the bottom of this post are Figures A, B, C etc. I also provided them in .pdf format at the very bottom in case some prefer .pdf.

Here we go.

1.   Figure A is simply a 10 volt DC battery with a 10k ohm (10,000 ohm) load resistor, resistor placed across it’s terminals. 1 ma (1 thousandths of an ampere) current flows through the resistor. The equation is: Current = voltage divided by resistance.
I = E / R

2.   Figure B is a 10 volt AC source with a 10k ohm load resistor. Again, 1 ma. of AC current is also flowing through the resistor. The same equation as point 1. This applies for whatever audio signal; 60hz, 1,000hz, 10,000hz are examples of AC, or a musical signal.)                     

3.   Next, Figure C is the same 10 volt AC source with a 10k ohm load resistor. However, I divided the 10k ohm resistor into a 4k ohm resistor Rp connected in series with a 6k ohm resistor RL. Again 1 ma. of current flow. I = V / R. The output AC voltage is calculated: Output = 10 volts x (6k / 6k +4k). 6 volts.

Relax, feel free to refer back because we will be coming back to these points.

4.   Next, Figure D is a typical gain stage triode circuit, and Figure E is the AC Thevenin Equivalent Circuit of Figure D. No DC voltage is applied. This will be covered later. (Engineering textbook, Semiconductor and Tube Electronics, by James G. Brazee).

5.   In Figure E, X is our AC source, Rp is the tube plate resistance from the specification sheet, and RL is our plate resistor to B+.  (I made up my own tube with specs sheet of Mu = 10, Rp is 4k ohms.)

6.   Figure E, RL is grounded at  point H.

7.   Notice Figure E, we have AC voltage source with Rp in series with RL to ground, just like in Figure C, because it is the same circuit. (The equation again is 10 volts x (RL / RL + Rp), or 10 x (6k/6k + 4K) = 6 volts AC.) All this can be easily measured to verify in our basic tube circuit.

8.   Our tube circuit will not work without DC voltage applied. But RL at H point to ground would short out the DC power supply.

9.   So let’s add the DC battery voltage to our Thevenin Equivalent Circuit Figure E? The result is Thevenin Equivalent Circuit Figure F, which deals with both AC and DC voltages.

10.   What is the difference between Figure E and F? One difference is power supply filter capacitor C1 is added so our B+ does not short to ground.

11.   Secondly, RL is not directly connected to ground anymore at H, but through C1 to ground.

12.   So we have a voltage divider, like Figure C, but with an additional part, C1. So we have our AC source, Rp, RL, and C1.

13.   Now, we need to understand that C1 has AC resistance (Xc). (Actually reactance but let’s keep it simple with no phase angles.)

14.   So now the AC output voltage = 10 x (RL + Xc / RL + Xc + Rp

15.   But what value is the AC resistance (Xc)? That depends upon the frequency.

16.   (The AC resistance (Xc) is calculated:   1 / 2PI x F x C. That is 1 divided by 6.28 times Frequency in hz times Capacitance in Farads.) As one can see, Xc is frequency dependent.

17.   So in our voltage divider, the output voltage will depend upon the frequency since RL + Xc changes with frequency..

18.   I have thus proven that the output voltage varies with frequency, thus the frequency response is not flat. (Thevenin Equivalent Circuits have been used in engineering circuits for decades and decades.)

The AC source we used can be simply musical voltage.

19.   Figure K is what you would see as the actual schematic circuit that we discussed.

20.   If we calculate or measure the AC musical signal current through C1, it will be at least 15 times the musical signal current through the coupling capacitor/Rg combination. I am lenient due to Y (a resistor or choke) siphoning off some AC musical current.

21.   We can calculate or measure the AC voltage at C1 but we need to know the frequency. At 20hz, a 20uf capacitor has 400 ohms Xc. The equation is: 6 AC volts x (Xc \ Xc + RL) The answer is approximately 0,2 volts AC. The C1 AC voltage will vary with circuit gain, C1’s value, Rp, RL, AC frequency.

The white paper could go further, because capacitors have ESR, internal inductance, and self resonance problems. Even very small capacitors have problems since tonal balance is so easily perceived.

For such information, please see an article by Walter Jung and Richard Marsh “Picking Capacitors” that includes measurements, resonance problems, ESR, DA etc.

https://www.yumpu.com/xx/document/view/23313452/picking-capacitors-part-1-walt-jung
http://www.reliablecapacitors.com/pickcap.htm







Cheers

steve

« Last Edit: 18 Sep 2018, 06:01 pm by Steve »

Cheytak.408

  • Restricted
  • Posts: 152
Re: How Important is Power Supply Quality?
« Reply #1 on: 18 Sep 2018, 03:33 am »
Send food during our hashing out something that allows for some breathing.  Go first responders!

Just happy they are all around people that understands the true effort in great food.

<><

Davej

FullRangeMan

  • Volunteer
  • Posts: 19853
  • To whom more was given more will be required.
    • Never go to a psychiatrist, adopt a straycat or dog. On the street they live only two years average.
Re: How Important is Power Supply Quality?
« Reply #2 on: 18 Sep 2018, 04:41 am »
deleted.
Excuse me the wrong post Steve.
« Last Edit: 19 Sep 2018, 02:12 am by FullRangeMan »

Steve

Re: How Important is Power Supply Quality?
« Reply #3 on: 18 Sep 2018, 12:51 pm »
Thanks FRM.

Cheers

steve
« Last Edit: 1 Nov 2018, 09:15 pm by Steve »

Mike B.

Re: How Important is Power Supply Quality?
« Reply #4 on: 18 Sep 2018, 02:52 pm »
Interesting stuff Steve. Have you ever discussed this with Curl?

RDavidson

  • Full Member
  • Posts: 2863
Re: How Important is Power Supply Quality?
« Reply #5 on: 18 Sep 2018, 03:06 pm »
Cheytak.408 appears to be a spam bot. Admin please boot them out.

mresseguie

  • Full Member
  • Posts: 4697
  • SW1X DAC+ D Sachs 300b + Daedalus Apollos = Heaven
Re: How Important is Power Supply Quality?
« Reply #6 on: 18 Sep 2018, 03:47 pm »
Rdavidson,

If you check Cheytak's posts, you will discover he isn't a bot. However, his response has absolutely nothing to do with the OP's post.  :scratch:

Steve,

Thank you for posting this information. I'm going to have to reread it after I've had my first coffee.

Michael

RDavidson

  • Full Member
  • Posts: 2863
Re: How Important is Power Supply Quality?
« Reply #7 on: 18 Sep 2018, 06:40 pm »
Hmmmm.....yeah you're right. Very odd. His post reads almost exactly like a bot made it. Perhaps his account has been compromised?

Steve

Re: How Important is Power Supply Quality?
« Reply #8 on: 18 Sep 2018, 08:02 pm »
Interesting stuff Steve. Have you ever discussed this with Curl?

No need to as he and other EEs should understand the basic 1st semester electronics. Interestingly, nobody seems to mention this information though, so maybe it is meant to keep it under wraps, an advantage to us engineers. I may have given away a secret, although only a bit. With all the different engineers offering different sounding components, most have to be off in terms of accuracy/naturalness, so it makes one wonder.

I thought I would bring up the subject here as I am retired from amps and preamp designing/building for some 6 years.

cheers

steve
« Last Edit: 13 Nov 2018, 04:27 am by Steve »

mresseguie

  • Full Member
  • Posts: 4697
  • SW1X DAC+ D Sachs 300b + Daedalus Apollos = Heaven
Re: How Important is Power Supply Quality?
« Reply #9 on: 18 Sep 2018, 08:15 pm »
Steve,

How do I (a non-engineer) discern whether any given power supply is good, bad, or excellent?

Thank you,

Michael


Steve

Re: How Important is Power Supply Quality?
« Reply #10 on: 18 Sep 2018, 10:17 pm »
Steve,

How do I (a non-engineer) discern whether any given power supply is good, bad, or excellent?

Thank you,

Michael

A few things to look for.
 
For a preamplifier:

1. If you see a schematic and the last filter capacitor (decoupling capacitor) for each and every stage is a 10uf, 20uf, 40uf,  then completely questionable design. One might get away with it using high plate resistance tubes like a 12AX7 or 12AT7 type tubes, but not with 12AU7, 6DJ8 types.

2. I would check to see if there is a voltage regulator for each stage. If there is none or only one for several stages, again questionable. We are concerned with both, between channels and between stages. The impedance/ac resistance of the regulator needs to be as close to constant vs frequency as possible.

3. This is not in the White Paper. If possible, stay away from power supplies with chokes/inductors. Oh this will add weight, but cause sonic problems. Chokes are usually between two filter capacitors (except power supply input chokes which we won't cover). A choke's AC resistance (impedance) varies very significantly with frequency. (I will attempt to keep this very simple for ease of understanding, so forget phase shifting etc.)

At high frequencies, a choke's AC resistance is extremely high, maybe 100,000 ohms, so the two filter capacitors are quite isolated from each other. However, at low frequencies, the choke's AC resistance might be a few hundred ohms, so isolation is minimal the two adjoining filter capacitors, their capacitance tends to "blend", almost like adding the capacitances. This causes problems.

Auditioning is very important. Remember, with a specification of +/- 0,1db from 20 to 20khz, the change in FR, the tonal balance change is only in the -54db area, easily within human perception. Rane, and Olson's work verified such.

A tube amplifier is somewhat different since we have an output transformer (OPT). It is extremely tough to judge any amplifier without special listening tests to detect what sort of problems the amplifier is causing. SS, that is another can of worms due mainly to capacitor size in uf, and multiple number of stages with a common power supply.

With tube amplifiers.

1. The problem is largely OPTs almost never have a flat FR, so they tend to be either a little bright sounding, a little bit dark sounding, or have resonance problems. The coupling capacitors and power supply capacitors have to be adjusted properly for the entire design. Special listening testing can only determine its quality. However, if one sees a .47uf, .68uf, .82uf, even 1uf, one has to really question the sonic accuracy/quality. Ask the size of the coupling capacitors.

By the way, properly designed tube amplifiers are incredibly good with very tight, punchy bass, due in part to the power supply design. So speaker design is not all there is to tight bass.
 
2. Best to use a totally separate power supply, and power transformers for each stage.

3. Same points for the input and driver stages of the amplifier as the preamplifier points.

Of course one must consider the whole design, parts quality etc.

I hope this helps some Michael. I am sick so back to bed I go.

cheers

steve


« Last Edit: 1 Nov 2018, 09:28 pm by Steve »

HsvHeelFan

  • Full Member
  • Posts: 452
Re: How Important is Power Supply Quality?
« Reply #11 on: 19 Sep 2018, 12:32 am »
Since the power supply is providing the voltage and current for the rest of your system, it's very important.  Don't cut corners in the area of the power supply.

You want a power supply that is quiet and can supply the instantaneous voltage and current needed, when needed.

Printed circuit board component placement and copper pours and signal routing, with attention to return paths is also important.

If you're interested in the Engineering side,  most component Datasheets have decent application notes on how to get the best out of that component.  National Semiconductor and Linear Technology (now part of Analog Devices) generally have good datasheets.   Digikey.com has a ton of datasheet links for parts.  Search by Manufacturer and part number (or type of part) can provide a lot of information.

HsvHeelFan

Steve

Re: How Important is Power Supply Quality?
« Reply #12 on: 4 Oct 2018, 03:44 am »
The article emphasizes what is never mentioned; what most people don't know, or understand is that the "output" end of the power supply, the "decoupling" capacitor/regulator/circuitry has to properly address the musical signal voltage and current.

As such, the impedance needs to be as flat as possible across the entire audio band and above. This is not easily accomplished and requires a knowledge of the entire design, plus special setup listening tests.

cheers

steve
« Last Edit: 1 Nov 2018, 09:30 pm by Steve »

Russell Dawkins

Re: How Important is Power Supply Quality?
« Reply #13 on: 4 Oct 2018, 06:03 am »
I seem to remember Martin Colloms making the point that in any amplifier you are essentially listening to the power supply, the output of which is simply modulated by the rest of the amplifier. That was a revelatory insight to me, steering all my subsequent musings on amplifier design (few as they have been!).
Here are a few of his other observations:
http://www.colloms.com/pages/exerpts.aspx

mresseguie

  • Full Member
  • Posts: 4697
  • SW1X DAC+ D Sachs 300b + Daedalus Apollos = Heaven
Re: How Important is Power Supply Quality?
« Reply #14 on: 4 Oct 2018, 04:12 pm »
A few years ago, I was given a personal tour of a factory in Taiwan. The owner of the factory is an audiophile who decided to produce vacuum tube amps, tube preamps, and tube CD players in addition to the rest of the equipment the company produces. His decision to produce audio gear didn't just come out of nowhere; the company had already been producing transformers for decades. I think they showed me 15 or 20 transformers. The question I posted above came out of that tour. While I was gazing at the transformers, I had asked myself, "How can I determine their quality?" No one offered to show me schematics - not that I can read schematics anyway:scratch:

I managed to persuade the business manager to loan me their best integrated tube amp and the tubed CD player for a one-week audition. This was my first real exposure to tube gear and I was very impressed by how good the music sounded. I would have loved to buy either component, but they were priced far above my ability (at the time) to afford them. As far as I know, this company's audio gear is primarily sold to the Chinese market. My impression is that it is intended to compete with Line Magnetic gear.

I'll provide a link to the products. The website's first language is Mandarin, but there is an option for an English version. Unfortunately, the English version lacks the Mandarin version's details, but you may find it useful nonetheless.

Links:

This page is in Chinese, but there are graphs at the bottom: http://www.kcr.com.tw/edcontent.php?lang=en&tb=2&id=148

The A8025 integrated amp is the one I auditioned (Chinese only): http://www.kcr.com.tw/archive/image/edcontent2/editor/Spec7_A7025_8025.jpg

Story behind their audio gear:  http://www.kcr.com.tw/edcontent.php?lang=en&tb=2&id=144

Thoughts? Impressions?

Michael

Steve

Re: How Important is Power Supply Quality?
« Reply #15 on: 7 Oct 2018, 01:47 am »
..............
Links:

This page is in Chinese, but there are graphs at the bottom: http://www.kcr.com.tw/edcontent.php?lang=en&tb=2&id=148

The A8025 integrated amp is the one I auditioned (Chinese only): http://www.kcr.com.tw/archive/image/edcontent2/editor/Spec7_A7025_8025.jpg

Story behind their audio gear:  http://www.kcr.com.tw/edcontent.php?lang=en&tb=2&id=144

Thoughts? Impressions?

Michael

I checked out the specs for both models and there are two areas I would like to address.

25 watts output at 10%, I would guestimate that at ~20 watts output, the distortion is probably somewhere around 2.5%. Not bad at all. The unknown is if the distortion is due to overload of the OPT or from clipping the output tubes, or both.

Normally, the power output at distortion measurement is somewhat arbitrary. They could easily have spec the amp at 10 watts at
~2.5% distortion. This assumes the driver/gain stages have low distortion.

The frequency response is a different matter though.
 
The FR is given 10hz to 30khz +/- 1db at 1 watt output. That means the frequency response actually varies by 2db from 10hz to 30khz. I would guess probably rolled off at both extremes. That means there is a 2db difference between mid-band and both low and high frequency extremes.

However, it is possible that +1db could occur at 30khz due to resonance, while -1db occurs at 10hz. As higher signal levels, the bass response will probably improve slightly, but still. This is just one example.

The problem lies in that the FR varies over a wide bandwidth, at least several octaves. Rane Corporation concluded that any deviation of 1/3 octave or less, will probably not be noticed. But over 1/3 octave, the perception increases rapidly, until at several octaves, the ear is very very sensitive to FR deviations.

As far as the power supply, difficult to tell. The questions become, how many stages of filtering, and how many ufds of filtering are used,,,, and the quality of the filter capacitors themselves.

Almost all tube amplifiers use chokes/inductors in the power supply, bad, but most designers, manufacturers won't design without them. Just adds cost while degrading the sound, yet is used as marketing, adds weight so must be good.

A further, tell tell sign is the size of the coupling capacitors. If the coupling capacitor size is less than 1.5uf, then there is a problem in the design. With that said, some don't care about bass, so small coupling capacitors are used.

Wish I could tell you more Mike. 

steve

« Last Edit: 1 Nov 2018, 09:35 pm by Steve »