Effect of Power Supply on Soundstage Width

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Affect of Power Supply on Soundstage Width

Apart from some interesting information on Aspen's and Mark Levinson's website, I have not found much information that talks about the effect of the power supply on stereo separation in power amplifiers.

Are there any general improvements I should make to improve the soundtage width? (The depth of soundstage seems to be fine).

What improvements would you expect on imaging/ soundstage and minimising crosstalk from the following changes made to the JLH 80w Mosfet amp:

1) Having a rectifier per channel for the voltage amplication stage, but the high current stage still will use a high quality shared supply (fast diodes, soft recovery witha full bridge per voltage line, and Elna Cerafine 10,000 muF pre line)
2)Having a shared transformer, bridge, capacitor, but both channels having a seperate rectifiiers for both the voltage and current gain stages (cheaper than new transformers, caps and diodes);
3) The whole hog: seperate power supplies per channel.

Additionally, if I go for option 3) - what is the recommended replacement for the 80V, 10,000muF Elna Cerafines in the other channel?

Many thanks for any information you may be able to provide.

Regards,

John
 
You know a lot of turntables image better than CDs?

Crosstalk on a typical phono cartridge is 20dB at 1Khz, dropping to 15dB or less at the frequency extremes.

Are we barking up the wrong tree?

I have seen amplifiers with a common high voltage supply for the front end have better than 80dB crosstalk at 1Khz.

If I was building a no feedback amplifier with poor PSRR I would do up as good a supply as possible.
 
Hi,

The effect on the sound stage is huge, depending which capacitors you use. Elna Cerafines are good, but soundstage can be improved dramatically if you cut the noise floor even further! The best way to do this is by using bypass capacitors. Many people use film capacitors at DIFERENT points. These points are important and can be defined only by experiment for the given power supply. I have tried many bypass caps, and the best one for me is Black Gate 0.1 uF / 50V non polar NX. If you need higher voltage, use two 0.47uF / 50 V non-polar NX caps in series "super E" configuration. Now, where to put them:

Try straight after the bridge rectifier (use only soft recovery, fast switching diodes, TWO IN PARALEL if you could afford it - but it's worth it!).
The other point to install these caps would be as close to the pre-amp section of your amp as possible. Avoid placing them in parallel with Elna Cerafines!!! - experiment a little and you'll notice a huge difference.

Let me know what you found!

Regards,
Nick
 
"The effect on the sound stage is huge, depending which capacitors you use."

Total agreement here.

Be careful when using values less than 1µF very close to the output stages, oscillation can occur. You may need to add a small series resistor to the cap, say 5 ohms ~ 10 ohms. Ideally the bypass caps should be grounded at where the high current returns. This is many times just not possible. In these cases you just have to find a ground where you can, and the oscillation caused by the inductance of the extra trace length is damped by the resistor in series with the cap.

Many times I take what improvement I get from adding 22µF~47µF to the main supply caps and just move on from there (look at coupling/feedback caps and diode snubbers).
 
larification Please

Extreme_Boky said:
Try straight after the bridge rectifier (use only soft recovery, fast switching diodes, TWO IN PARALEL if you could afford it - but it's worth it!).

Nick,

Could you clafify if "TWO IN PARALEL if you could afford it - but it's worth it!" is two bridge rectifiers or two capacitors?


Regards,

John L. Males
Willowdale, Ontario
Canada
10 December 2005 06:33
 
Affect of Power Supply on Soundstage Width

Apart from some interesting information on Aspen's and Mark Levinson's website, I have not found much information that talks about the effect of the power supply on stereo separation in power amplifiers.

Are there any general improvements I should make to improve the soundtage width? (The depth of soundstage seems to be fine).
I think dept and width of sound should depend on the size of your room.
If too wide and/or deep, the whole stage will not fit inside your listening space.

For minimum power supply impact on stereo separation,
wouldnt totally separate left/right amplifiers as well as loudspeakers
be good?
Preferable both amplifiers should be placed as far as possible from eachother.
You also should use two different AC mains wall outlets for minimum ground connection interference.


Just my simple thoughts, but I am no expert, so I am not sure how this would work compared to traditional setup.
:)
 
Could you clarify if "TWO IN PARALEL if you could afford it - but it's worth it!" is two bridge rectifiers or two capacitors?

Two diodes in parallel making "one diode". That's 8 diodes in total for a full wave bridge rectifier.
I saw this approach in some Japanese high end gear and tried it myself. It is definitely better, more relaxed and natural with better details retrieval.


Extreme_Boky
 
Hi foxyb,

Firstly, if ML are spruiking this on their website you can bet, London to a brick, that it's because they use elaborately regulated power supplies on their amplifiers - approaching the complexity of the amplifier module itself and there's the #1 commercial reason. Many manufacturers don't do this so they're trying to make a + out of it.

What they don't tell you is why they do this. I can tell you because I spent 4 months as a consultant to this company. The answer is they have found the only way THEY KNOW to improve their amps sound is to do this. So they're flogging it as necessary.

However, if they knew more about the finer aspects of amplifier design, they would design an intrinsically high PSRR topology AND they would design their PCB's and wiring looms and chassis layouts to minimise interactions of modulated Vs tracks and wiring with sensitive early/high impedance stages. It takes a very enlightened designer to do this and they invariably like to be paid for their work, accordingly.

That's too hard for many to get their heads around let alone do, so the easy, very expensive approach is to have lots of seperately regulated high and low power supply lines, to achieve an end result that could be achieved at very low cost of intrinsic design. And they've lost the advantage of power supply headroom and efficiency.

Not only that, it doesn't solve the problem of interactions as there are still track to track and wiring loom interactions not related to PS artefacts, like the naive running of high current driver tracks in proximity of the input stage causing high distortion (0.5% cf 0.03%), or the loud em-radiated traffo hum that had resulted in my re-working of one existing model after complaints by their Japanese agents that the hum was LOUD!

A more wholistic approach could use unregulated supplies for full dynamic headroom, better PCB and lead dress for lower THD and noise and a far better and cheaper amp.

But of course designing amps is a Black Art they're privvy to.

And here's an example of the cr@%^ that's on supply lines in AB amps at only 5W output. Imagine that all around your PCB near sensitive current sources.

Cheers,
Greg:D
 

Attachments

  • 5wsupplyhash.png
    5wsupplyhash.png
    11.6 KB · Views: 762
Now,
With soundstage size (height,width,depth) the information is contained in low level mostly HF directional detail so anything that reduces the miasma of artefacts from power supplies, grounding irregularities or bad PCB layout or lead dress is going to allow through more detail.
Soundstage size in particular relates to what spatial information comes out of well seperated speakers - that shouldn't. This is often measured as crosstalk and can be due to a shared power supply or grounding or proximity interactions. Often crosstalk is considered only as 'pure tone' while measurement techniques read 'residue + noise" or everything that comes out of one channel when the other is stimulated with a swept pure tone. This will include the miasma of harmonics,IM products and noise.

Where to put a capacitor to make it right?

Cheers,
Greg:D
 
One or Two rectifer to a +- PS ?

Extreme_Boky said:

Two diodes in parallel making "one diode". That's 8 diodes in total for a full wave bridge rectifier.
I saw this approach in some Japanese high end gear and tried it myself.
It is definitely better, more relaxed and natural with better details retrieval.

And whats your experience, opinion in the single vs. dual question about rectifiers ?
 
Re: One or Two rectifer to a +- PS ?

by Nelson Pass
Rectifiers.
Yeah, sure, rectifiers are important, after all, the AC has to get converted to DC,
but I don't like the fast recovery types that some audiophiles have raved about.
Fast recovery means that they withstand many amps and volts in a tenth of a few nano-seconds,
something we don't see very often on the old 60 Hz AC line.
They are essential element in switching power supplies,
but for regular "linear" power supplies, I much prefer SLOW diodes,
and we create them by placing small capacitor circuits across the diodes,
which greatly reduces radiated noise.
Four adequately current rated silicon diodes
in a bridge configuration
with capacitor across each diode
is a good rectifier for audio power amps.

Maybe we should have a look at how Nelson Pass did it.
:smash:
 
How important is crosstalk?

I have always wondered if there really is a need for extreme stereo separation? For most music, at least if there is an acoustic source like a symphony orchestra or a jazz band, I don't think there is, or should be, any need for it, at least for recordings that are mastered to have a realistic stereo image. I do realize, however, there may be a need for it for certain recordings of modern music or film music.

What other reasons could there be?

With a single power supply, a transient that makes one channel clip will affect also the other channel, but so what? If one channel clips, we are playing to loud anyway to get good sound.

A third reason, that maybe makes more sense, is if the crosstalk is high enough to cause non-negligible IM products between signals in the two channels. But if the crosstalk is reasonably good and the IM is already low, it is hard to see this should be a problem either.

Am I missing something? Or is this one of those things that some people hear, but that has no technical explanation?
 
Hi Amp Guru,
I know you like to protect your intellectual savvy.
But I would like you to teach us how to make our amps better at rejecting unwanted artifacts. How to minimise the transfer of the unwanted into other stages and maybe not generating the artifacts in the first place.

Even just a few pointers would be better than nought.
 
Thx, lineup !

> Fast recovery means that they withstand many amps and volts in a tenth of a
> few nano-seconds, something we don't see very often on the old 60 Hz AC line.
Well I am not sure about this. I think there is not only 50-60Hz load to the
rectifier, but also higher freqs. And hence, faster diodes can have a reason.

> placing small capacitor circuits across the diodes
Yea, that could be already reasonable, and should be tried definitely.

> which greatly reduces radiated noise
But maybe increases the LC effect on the (trafo sec - cap) circuit too.
 
Hi Christer,
maybe it's not the absolute level of crosstalk that's important.
The ability to resolve the low level detail above the noise floor could be more important.
This applies to both the recording and playback noise floors.
This is why many say that CD and other digital recording systems with a 16bit or less resolution ruins musical/stereo realism/effect.
The 20bit and beyond systems stand a chance of getting us closer to what analogue could achieve but DVDaudio is not taking off and the compressed audio tracks on DVD video are not there yet. Too much compression? or too low bit resolution to allow it to fit in?

If we design for overkill in all respects then when good stereo comes along our DIY systems should be able to resolve the data.
 
Andrew,

I get your point, I think. Yes, the noise floor matters, but I think it is only distorsion and other extramusical signals that matter. A bad separation of the musical content in the channesl hardly matters. As sombody pointed out, a cartridge has very poor stereo separation, but that is probably mostly a "leakage" of signal content between the channels, not addition of of new signals like distorsion.

So what would contribute to the noise floor in an amp, if it is not a "leakage" of the actual music signal that is the problem? Maybe the problem then is what our guru always persists in telling us, that a class AB OP generates a lot of high-level harmonics on the rails? Yes, I guss that could be a reasonable explanation.

Note that I am not trying to claim that we don't need good stereo separation, but just questioning if we need it. I am currently listening to a very nice dual mono amp I have borrowed from a friend, but I have no idea how important the dual mono is for the good sound of it.
 
Upupa Epops said:
Hi Christer, is better to make two separated supplies for each channel, than worry about distortion, which is " running " on the rails...

I don't mind overkill just in case, but do you mean distorsion caused by class AB output stage now, or do you mean that also other distorsion products could cause problem via the PSU (which seems less likely)?
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.