• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

The Truth About HiFi Amplifier Power Supplies

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
11,585
Likes
12,785
Location
Central Fl

egellings

Major Contributor
Joined
Feb 6, 2020
Messages
2,404
Likes
1,849
62dB of gain? That must be for use with MC cartridges. An MM would likely overdrive it. Also, why so much supply current for a small signal amplifier?
 

Ken Tajalli

Major Contributor
Forum Donor
Joined
Sep 8, 2021
Messages
1,099
Likes
817
Location
London UK
It is a MC stage but it has enough input voltage headroom that one can use HO MC carts up to around 1.5mV.

Because its not just one opamp and a 3 pin reg... ;)


There's 108 transistors in a stereo build and a fair bit of class A psu.
Nice one, but honestly , wouldn't it have been easier, to replace all that regulation with a bank of LiPo's and a simple SMPS charger?
You could charge when not listening, run on pure battery power for use, while the charger is off.
 

atmasphere

Active Member
Manufacturer
Joined
May 26, 2021
Messages
284
Likes
408
Mostly positive responses but not many. Nobody tried to argue with my assertion about the mains cable being an unimportant component :)
OK I'll bite...

For most solid state amps it probably isn't. But for tube amps (or larger class A solid state amps) a power cord can be a deal due to voltage drops. I've seen a power cord gobble up significant power, causing the amp to lose 40 Watts of power at its output (about 30% of total output power). The voltage drop across the cord was something just north of 2 Volts. Not surprisingly the power cord was warm to the touch not just at the connectors but along its length. Its a series resistance after all and Ohm's Law still applies.

In a tube amp you can have a significant filament circuit. The one in the example just above had a 20 Amp filament circuit- it had a lot of power tubes. In addition its output section was composed of triodes which are more sensitive to B+ than pentodes. So the result was the filaments cooling off a bit in tandem with there being less B+ available, which is dropped across the now cooler power tube.

If the power cord has lousy high frequency bandwidth you might get some current limiting on the peaks of the 'sine' wave since that is the only time filter caps charge; this could lead to the power supply being a bit low. But I see that as a minor issue, if there is one at all.

If you have enough feedback in the amplifier design it will be able to reject effects like this (a self-oscillating class D amp likely being the ultimate in this regard as they tend to have the most feedback and the least idle current, often also run by SMPSs). Lower feedback (or zero feedback) designs will be more susceptible, especially if they are biased heavily (class A, or 'enriched' class AB.... I hate that 'enriched' word in this case since its a marketing term...). I suspect that with the equipment most people use on this forum the difference in performance will be negligible since idle currents will be low. But anyway, the things to measure are output power, distortion and output impedance at power. The differences will be subtle with most amps but they can be seen. The more feedback the design employs the less distortion and output impedance will be affected.

So the bottom line here is unless you're running a larger tube amplifier or a very large class A solid state amp, the power cord will probably not be an influence. But if it is, if you run that kind of equipment and found that a power cord does make a difference, you may be disappointed to learn that you can get materials at Menard's (heavy power cord used for stoves or clothes dryers) and with some nice looking connectors off of Amazon you can make a power cord that will work just as well and for under $100 including shipping. ....Bad news unless you've not already sprung for that $5,000 cord, in which case this news might make you quite happy ;)
 

Sal1950

Grand Contributor
The Chicago Crusher
Forum Donor
Joined
Mar 1, 2016
Messages
11,585
Likes
12,785
Location
Central Fl
So the bottom line here is unless you're running a larger tube amplifier or a very large class A solid state amp, the power cord will probably not be an influence. But if it is, if you run that kind of equipment and found that a power cord does make a difference, you may be disappointed to learn that you can get materials at Menard's (heavy power cord used for stoves or clothes dryers) and with some nice looking connectors off of Amazon you can make a power cord that will work just as well and for under $100 including shipping. ...
Monoprice caries some heavy duty power cords, 14AWG, 15A/1875W, 3-Prong, Black, 6ft for $5.99.
Also available in different colors and lengths to suite your needs/desires.
I bought a couple a few years back to upgrade the cords on my triggered power distribution blocks that handle the current of 6 stereo power amps at turn-on. Nice cords.
 

Ken Tajalli

Major Contributor
Forum Donor
Joined
Sep 8, 2021
Messages
1,099
Likes
817
Location
London UK
For most solid state amps it probably isn't. But for tube amps (or larger class A solid state amps) a power cord can be a deal due to voltage drops. I've seen a power cord gobble up significant power, causing the amp to lose 40 Watts of power at its output (about 30% of total output power). The voltage drop across the cord was something just north of 2 Volts. Not surprisingly the power cord was warm to the touch not just at the connectors but along its length. Its a series resistance after all and Ohm's Law still applies.
To start with, I am a fan of Atma-Sphere, you guys make very interesting amps, I have been tempted to get one.
A 2V drop on the power cord is insignificant to normal voltage dropouts that can occur for reasons beyond our control. In UK (240V) I have monitored voltages as low as 230V and as high as 245V. Does that affect some amps to the tone of 40%?? I find that hard to believe.
If a small voltage drop causes so much problem, shouldn't a reputable manufacturer account for that, or at least use a chunky power cord, bolted to internal terminals, therefore not be replaceable?
In a tube amp you can have a significant filament circuit. The one in the example just above had a 20 Amp filament circuit- it had a lot of power tubes. In addition its output section was composed of triodes which are more sensitive to B+ than pentodes. So the result was the filaments cooling off a bit in tandem with there being less B+ available, which is dropped across the now cooler power tube.
Again, 20 amps is in 12V or 6V rail, on a 110V mains input it translates to 2A or 1A ! and a cable of 1 ohm resistance, it means 2V drop max? again significant?
If the power cord has lousy high frequency bandwidth you might get some current limiting on the peaks of the 'sine' wave since that is the only time filter caps charge; this could lead to the power supply being a bit low. But I see that as a minor issue, if there is one at all.
This is where you lose me completely!
High frequency bandwidth? on a power cord? The only frequency I wish to pass through my power cords is 50/60Hz and nothing beyond - Indeed, I use elaborate L-C networks to make sure. Any high frequency through a power cord is RF noise!
 

egellings

Major Contributor
Joined
Feb 6, 2020
Messages
2,404
Likes
1,849

sq225917

Major Contributor
Joined
Jun 23, 2019
Messages
1,292
Likes
1,437
There's not a hint of over design in it, it uses absolutely the least amount of parts possible to achieve the desired result. ;)
 

atmasphere

Active Member
Manufacturer
Joined
May 26, 2021
Messages
284
Likes
408
To start with, I am a fan of Atma-Sphere, you guys make very interesting amps, I have been tempted to get one.
A 2V drop on the power cord is insignificant to normal voltage dropouts that can occur for reasons beyond our control. In UK (240V) I have monitored voltages as low as 230V and as high as 245V. Does that affect some amps to the tone of 40%?? I find that hard to believe.
If a small voltage drop causes so much problem, shouldn't a reputable manufacturer account for that, or at least use a chunky power cord, bolted to internal terminals, therefore not be replaceable?

Again, 20 amps is in 12V or 6V rail, on a 110V mains input it translates to 2A or 1A ! and a cable of 1 ohm resistance, it means 2V drop max? again significant?

This is where you lose me completely!
High frequency bandwidth? on a power cord? The only frequency I wish to pass through my power cords is 50/60Hz and nothing beyond - Indeed, I use elaborate L-C networks to make sure. Any high frequency through a power cord is RF noise!
If the filament in a tube runs at a lower voltage, the tube will have less emissions. So you can see how this will lead to lower power in tandem with lower B+.

It is true that the mains frequency is only 50 or 60 Hz. But the area of conduction is a very short period of perhaps a few milliseconds or less. The conducting current is thus limited to very short duration so you need bandwidth to conduct it properly, else the current will be limited by the cord. IOW if there isn't bandwidth, the power cord represents a higher impedance a those frequencies.

If you build the power cord into the product as some manufacturers do, you run the risk that some audiophiles will tamper with the equipment to splice in their own special sauce. This could lead to shock hazards and the like as too many audiophiles are not as good at soldering as they like to think!
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
13,033
Likes
29,357
Location
The Neitherlands
Are you implying that a BW of 1kHz is not enough for a power cord and that there are mains cables that cannot even reach 1kHz BW ?
All frequencies above 1kHz do not contribute to the reservoir caps charging.

Below measured current spectrum in the rectifier path (24A peak)
fft-single-phase-dual-winding-dual-diode1.png
 
Last edited:

sq225917

Major Contributor
Joined
Jun 23, 2019
Messages
1,292
Likes
1,437
Power chords have impedance in the miliohm range, that would have to be some serious current draw to cause a 2v drop.
 

atmasphere

Active Member
Manufacturer
Joined
May 26, 2021
Messages
284
Likes
408
Are you implying that a BW of 1kHz is not enough for a power cord and that there are mains cables that cannot even reach 1kHz BW ?
All frequencies above 1kHz do not contribute to the reservoir caps charging.

Below measured current spectrum in the rectifier path (24A peak)
fft-single-phase-dual-winding-dual-diode1.png
Ever seen what makes up a square wave? With each succeeding harmonic the leading edge gets steeper. You need bandwidth in order to pass that. Fortunately this really isn't an issue with most power cords. That is why I mentioned that you can get decent materials to build a good power cord at Menard's.
Power chords have impedance in the miliohm range, that would have to be some serious current draw to cause a 2v drop.
IME most of the resistance is actually at the cable terminations; the connectors heat up before the cordage does. And yes that takes some current. Big class A power amps draw lots of current.
 

theREALdotnet

Addicted to Fun and Learning
Joined
Mar 11, 2022
Messages
532
Likes
782
One potential differences between tube and solid state power amps is how the mains transformer affects the mains supply’s source impedance (from the street transformer, through your house wiring, the power cord, all the way to the transformer terminals). With tube power amps, the main transformer usually transforms up. Given a supply voltage of 120V, this could well mean by a factor of three or four. In a scenario like that, the mains supply’s impedance is magnified by a factor of 9 to 16 (transformer ratio squared). That said, given the substantial impedances found on the secondary side (filter chokes and resistors), I can’t see how that matters much.

Mains transformers in solid state amps typically transform down, and are thus lowering the mains supply’s impedance, as seen by the amp.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
13,033
Likes
29,357
Location
The Neitherlands
Ever seen what makes up a square wave? With each succeeding harmonic the leading edge gets steeper. You need bandwidth in order to pass that. Fortunately this really isn't an issue with most power cords.
Yep, seen plenty and in no way the current draw from any mains supply (is does not matter which diodes are used in the rectifier) looks anything like a square-wave.
A BW of 1kHz is more than sufficient. No special cable geometries or materials are needed. For high power amps all that is needed is a low resistance (thicker wire).

I have not seen a single cable (even ones with low pass filters in it) that have problems conducting frequencies below 1kHz.
Do you happen to have any shots from current draw of high power speaker amp power supplies that somehow have worthwhile currents above 1kHz that would indicate an amp could suffer from power shortage because of frequencies > 1kHz not being present ?
 

atmasphere

Active Member
Manufacturer
Joined
May 26, 2021
Messages
284
Likes
408
Yep, seen plenty and in no way the current draw from any mains supply (is does not matter which diodes are used in the rectifier) looks anything like a square-wave.
It looks to me as if you missed my point. I'm not talking about a square wave. Filter caps can only charge when their voltage is lower than that of the output of the rectifiers; the rectifiers can only turn on when the power transformer voltage is higher than that of the filter caps. That will be at the peak of the AC waveform. At this time the rectifiers commutate; current flows with a very steep leading and falling edge. This has nothing to do with harmonics of the AC waveform as it does simply switching from off to on (commutation). The current flow will have a lot to do with how depleted the filter caps got between peaks of the AC waveform. All this was covered in opening posts of this thread.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
13,033
Likes
29,357
Location
The Neitherlands
This has nothing to do with harmonics of the AC waveform

Not with the AC voltage waveform and that's not what my plot showed.
The FFT I showed is from the diode current waveform on the secondary side of a transformer.


Diodes do not really 'switch' in a rectifier, it is not as 'steep' because the AC voltage rises relative slowly near the tops and the charging current is mostly limited by the transformer. When nothing is really 'stressed' we see a 1.2ms rise-time which is roughly equal to what a 1kHz sine would do.

3300uf-schottky-33ohm-big-trafo.png


Below a similar condition but with the transformer limiting the charge current.
single-phase-dual-winding-dual-diode1.png


Below is the spectrum of this particular current 'peak' not of the AC input voltage.

fft-single-phase-dual-winding-dual-diode1.png
 
Last edited:

atmasphere

Active Member
Manufacturer
Joined
May 26, 2021
Messages
284
Likes
408
-And there you go. As I mentioned earlier this is a minor influence; most AC cords have plenty of bandwidth to handle this issue.
 

Ken Tajalli

Major Contributor
Forum Donor
Joined
Sep 8, 2021
Messages
1,099
Likes
817
Location
London UK
-And there you go. As I mentioned earlier this is a minor influence; most AC cords have plenty of bandwidth to handle this issue.
And again, as you mentioned earlier, a chord bolted to internal posts, negates the termination resistance (almost!), and as I mentioned earlier a chunky cable, with factory plug and bolted to the inside terminals, should give ideal arrangement for high current situations, probably why we see them in fridges, washing machines, and we should see them on high current amplifiers!
Those who may decide to finger them and splice things together without enough knowledge, deserve what comes to them!
At any rate, grid voltage swings far more than any cable's influence.
 
Top Bottom