X-Can v3 - modded vs non-modded - blind tests

[szoze]

Hi!

I promised I would perform a more extensive listening test of the x-can V3 modified with Nichicon Muse ES 1000 uF output capacitors after a couple of days burn-in. Today, I and my two friends did it. My two friends are musicians and use their ears daily in their profession, exactly like I do.

We used one x-can v3 completely unmodded with all stock capacitors inside. The other x-can v3 we used had Panasonic FC in the psu part, polypropylene film nonpolar 10 uF capacitor in the input stage and Nichicon Muse ES 1000 uF in the output stage.

Music used:
Nirvana - "Nevermind"
Verdi - Requiem (Telarc, Shaw, Atlanta)
Beethoven - String Quartet op.59, No1, Takacs quartet, DECCA
Miles Davis - "Scetches of Spain"

Equipment:
cd-player: Harman/Kardon HD990
signal cables: Van den Hul
headphones: AKG k701

Every person did a blind test choosing the amp that sounded "best" after hearing three tracks from each of the above cds on each amp. A choice was made for every track, i.e. 12 choices per person (the persons didn't know if the headphones were moved or not between the amps after every track (sometimes it was done). Every person could chose to listen how much he wanted to every track on a certain amp, no time restrictions.

The result is as following:
Person 1: modded amp: 4 votes, nonmodded amp: 8 votes
Person 2: modded amp: 6 votes, nonmodded amp: 6 votes
Person 3: modded amp: 7 votes, nonmodded amp: 5 votes

So the nonmodded amp wins with 2 votes . Statistically significant? No. Chance? Probably .

I can only talk for myself and can say that it was very hard to choose between the amps.

I am not sure if it is possible to do a more objective testing.

Let this be the first in a series of tests.

[szoze]

Our testing continued yesterday. We used the same set-up. This time one of the amps had 10uF input capacitors bypased with a OFC wire (let's call it "the modded amp"). In the other one we used polypropylene film 10 uF capacitor.

This time we first did sighted testing and thought that we could hear slight "improvement" in the treble of the modded amp and even certain improvement in the bass region (definition). Otherwise the difference was far from night and day, the character of the sound was the same, the soundstage, transparency etc. we thought was the same.

The blind AB testing however gave us the following result:

Person 1: modded 5 votes non-modded 7 votes
Person 2: modded 8 votes non-modded 4 votes
Person 3: modded 7 votes non-modded 5 votes

This time the modded amp wins by 4 votes. Can we call this a night and day difference? No, I don't think so. Once again it was very hard to hear the difference between the amplifiers.

It would be interesting to repeat this test this time without the pre-listening sighted session and with more tries.

SZ

[Deleted]

If the PSU area is using 1N400x type diodes, you could also try the faster UF400x type.

[szoze]

Apr 23, 2012 6:23:59 GMT @sandyk said:

If the PSU area is using 1N400x type diodes, you could also try the faster UF400x type.

Thanks for the tip. It will make a nice set-up for our next AB test.

[PinkFloyd]

Stock diodes are 1N4007. You can replace with UF4007 or UF4001

UF4001 are even faster (50ns as opposed to 75ns) and perfectly good to use with the V3.

[Deleted]

Stock diodes are 1N4007. You can replace with UF4007 or UF4001UF4001 are even faster (50ns as opposed to 75ns) and perfectly good to use with the V3.

Hi Mike
The SF11 to 14 are even faster at 35nS max. , but harder to obtain.They also have a lower forward voltage drop than the 1N400x series.
The UF 4001 you suggested has a lower forward voltage drop than the UF4007 too.
Regards
Alex

[Deleted]

For comparison... some numbers so we have a reference.

turn-on time of 1N4007 = 500ns (2MHz), turn off time = 400ns a whopping 10 times slower than the ultra fast ones that are intended for high frequency switch mode power supplies !

We are rectifying 100Hz signal so every 10ms (10,000,000 ns) the diodes are 'on' during about (500,000ns) to replenish the reservoir caps.
During the remainder of the time 9,500,000ns (about 19 times that of the charging time) the caps are discharged by the circuit.
The difference between the top voltage and lowest voltage just before recharging is called ripple voltage.

With the fastest diodes in town (SF11) compared to the 'sluggish' (2MHz = 20,000x faster than full bridge rectified 50Hz) 1N4007 the difference in charging time is roughly 0.3% (shorter)
So when normalised the caps are charged with a SF11 is set at 100% 'on time' the 1N4007 is 'on' 99.7% of the time.

The rectified voltage using SF11 opposite 1N4007 is 0.1V higher (0.5% higher in this case)
when using UF4001 the rectified voltage will be 0.2V higher (+1%)
When using the UF4007 the rectified voltage will be 0.2V lower (-1%)

[Deleted]

Apr 23, 2012 9:40:07 GMT @solderdude said:

For comparison... some numbers so we have a reference.

turn-on time of 1N4007 = 500ns (2MHz), turn off time = 400ns a whopping 10 times slower than the ultra fast ones that are intended for high frequency switch mode power supplies !

We are rectifying 100Hz signal so every 10ms (10,000,000 ns) the diodes are 'on' during about (500,000ns) to replenish the reservoir caps.
During the remainder of the time 9,500,000ns (about 19 times that of the charging time) the caps are discharged by the circuit.
The difference between the top voltage and lowest voltage just before recharging is called ripple voltage.

With the fastest diodes in town (SF11) compared to the 'sluggish' (2MHz = 20,000x faster than full bridge rectified 50Hz) 1N4007 the difference in charging time is roughly 0.3% (shorter)
So when normalised the caps are charged with a SF11 is set at 100% 'on time' the 1N4007 is 'on' 99.7% of the time.

The rectified voltage using SF11 opposite 1N4007 is 0.1V higher (0.5% higher in this case)
when using UF4001 the rectified voltage will be 0.2V higher (+1%)
When using the UF4007 the rectified voltage will be 0.2V lower (-1%)

Frans
It's not all just about the specs you quoted, as Mike will attest.
Constructors of the Class A HA/preamp were also able to hear the improvement when the faster SF11 and SF12 were used instead of the normal 1N4004 supplied with the SC PSU kits. It was like a fine veil had been lifted.

RF Emissions from Bridge Rectifiers
Bridge rectifiers, even the massive ones intended solely for 100 Hz power rectification, generate
surprising quantities of RF. This happens when the bridge diodes turn off; the charge carriers
are swept rapidly from the junction and the current flw stops with a sudden jolt that generates
harmonics well into the RF bands. The greater the current, the more RF produced, though it is
not generally possible to predict how steep this increase will be. The effect can often be heard by
placing a transistor radio (long or medium wave) near the amplifier mains cable. It is the only area
in a conventional power amplifi er likely to give trouble in EMC emissions testing [3] .
Even if the amplifier is built into a solidly grounded metal case, and the mains transformer has a
grounded electrostatic screen, RF will be emitted via the live and neutral mains connections. The
first line of defense against this is usually four snubbing capacitors of approximately 100 nF across
each diode of the bridge, to reduce the abruptness of the turn-off. If these are to do any good, it is
vital that they are all as close as possible to the bridge rectifier connections. (Never forget that such
capacitors must be of a type intended to withstand continuous AC stress.)

Audio Power Amplifier Design 5th edition- D.Self

[Deleted]

Funnily enough the EMC emissions with fast diodes are WORSE than with slower diodes (such as the 1N4007).
Doing EMC measurements myself (work related) I can verify that.
The garbage that is emitted is way above the audible range by the way and is above the (several 100 kHz) region and is going up higher in frequency when fast diodes are used.
Normally these emitted RF components are filtered by mains filters IF it is above the minimum required levels.
For DIY of course there are no minimal requirements and emitted RF can be higher.

effectively what one does with snubbers caps in parallel to the fast diodes is slow the fast diodes down even MUCH (factors 1000 or more) slower than the original 1N4007 diodes were, completely degrading all the extra speed the fast diodes give (0.3%)

I know about the subjectively perceived differences though.
Technical explanations and subjectively found differences often do not seem to go hand in hand.

[Deleted]

The garbage that is emitted is way above the audible range by the way and is above the (several 100 kHz) region and is going up higher in frequency when fast diodes are used.

In other words, an area where the amplifier is less susceptible perhaps?

Technical explanations and subjectively found differences often do not seem to go hand in hand.

Perhaps you are not looking at the more appropiate technical explanations ?

[Deleted]

Alas no... it starts at the same frequency band but has more harmonics higher up top (>100MHz).

Since it is common practice in audio to snubber the diodes there is NO difference anymore between emitted RF between a snubbered 'fast' and 'slow' diode as these are only a factor 10 apart unsnubbered.
When snubbered they are a factor 100 to several 1000 times slower (depending on used snubber Cap) effectively eliminating all the 'speed' differences.

These high frequencies are not present at the amp side of the rectifier (reservoir caps) but exist in the secondary winding circuit (and through capacative coupling of windings) also on the primary side of the amp as that part is switched on and off rapidly.
The frequency spectrum depends on wire capacitance, inductance and impedance of the circuit.
a DIY'er can lower emissions by twisting the wiring from a trafo to the rectifiers and using a mains filter (properly applied) and the correct shielding methods.

The more appropriate (technical) explanations (if you have them) I will be glad to hear.

[szoze]

Apr 23, 2012 10:05:13 GMT @solderdude said:

I know about the subjectively perceived differences though.
Technical explanations and subjectively found differences often do not seem to go hand in hand.

Subjectively perceived differences are hard to measure especielly when we rely on our "sound memory". We simply have too much confidence in our senses (and too little knowledge about our sensory perception perhaps) when doing this.

Direct "blind" AB testing is in my experience the easiest and the most reliable approach to detecting possible SQ differences. It is the only way to eliminate our preformed beliefs about something we have done to our equippment.

It is fascinating how I for example "heard" some difference in bass and treble definition in my latest sighted test. In the blind test nothing was obvious and chosing the amp that sounds "better" was not at all that easy.

[Deleted]

Since it is common practice in audio to snubber the diodes there is NO difference anymore between emitted RF between a snubbered 'fast' and 'slow' diode as these are only a factor 10 apart unsnubbered

That is not common practice in Headphone Amplifiers, although Mike has done that. It is more reserved for the likes of big bolt on bridge rectifiers in Power Amplifiers.Even then it wasn't found necessary in the case of most Silicon Chip designed amplifiers, where extensive use is made of their Audio Precision test equipment. Many recent design amplifiers use big TO220 style Schottky diodes, Hexfred,(some are 17nS ) and Stealth etc.diodes , so why on earth would they fit snubbers to them ? Obviously they must find an improvement when using them, as they are normally considerably more expensive .

[Deleted]

That is not common practice in Headphone Amplifiers, although Mike has done that. It is more reserved for the likes of big bolt on bridge rectifiers in Power Amplifiers.Even then it wasn't found necessary in the case of most Silicon Chip designed amplifiers, where extensive use is made of their Audio Precision test equipment. Many recent design amplifiers use big TO220 style Schottky diodes, Hexfred,(some are 17nS ) and Stealth etc.diodes , so why on earth would they fit snubbers to them ?

You are correct about the snubbering which is not common in low current applications (HA) and more for high current (power) applications or when low emission is required (medical or industrial).

Why they would fit fast diodes (and subsequently snubber them ?) is more or less known.

I have no objections to fitting fast diodes and or snubbering if that enhances the subjective experiences you have (i.e. improved SQ) and it doesn't cost that much.
It's when speed numbers are used to state this is what is causing this is when I start questioning things.
Simply look at the posted nanosecond numbers to get a feel of the importance of it.
Of course I do not know everything and my electronics knowledge is not even 0.01% of what's there to know.
This, most likely, explains my ignorance in sonic matters.

The strange part is how these diodes can change SQ while measurably, the power supply voltage (at the reservoir caps) is not changed.

All music 'power' is drawn from the reservoir caps (also true for power amps) and only a few 100 us per 10ms directly from the mains.
The rest of the 9.99ms the amp (or whatever is being fed) is virtually DISCONNECTED from the mains all of that time.

Whether the charging time is 0.3% shorter (could just as well be 0.1 or 1%) doesn't seem to make much of a difference.
At least not to me.

but as said... IF you hear an improvement, no matter how small it may be and it's worth it to you to increase sonic bliss levels... please mount them.

It is a bit strange to me though that each and every equipment, no matter low or how high end it is can often be improved sonically with various modifications.
How stupid/ignorant can those pesky designers be ?

[Deleted]

but as said... IF you hear an improvement, no matter how small it may be and it's worth it to you to increase sonic bliss levels... please mount them.It is a bit strange to me though that each and every equipment, no matter low or how high end it is can often be improved sonically with various modifications.How stupid/ignorant can those pesky designers be ?

Why don't you try questioning the use of these types of diodes in a forum with >100,000 members such as DIY Audio, where MANY members, including some highly qualified members use them ?
Fortunately. not all designers, including the engineers who design these components, are as closed minded as some, and recognise their advantages in more than just SMPS. !

It is a bit strange to me though that each and every equipment, no matter low or how high end it is can often be improved sonically with various modifications.

Hmmm. I thought that was the rationale behind your improvements to the Indeed Bravo, and your latest variations on a theme ? ;D

I am now out of this stupid "tit for tat" crap.

[szoze]

Apr 23, 2012 11:32:03 GMT @solderdude said:

It is a bit strange to me though that each and every equipment, no matter low or how high end it is can often be improved sonically with various modifications.
How stupid/ignorant can those pesky designers be ?

It has been puzzling me too. Moreover, almost every replacement/modification of the stock configuration improves the sound quality.

[szoze]

Apr 23, 2012 11:43:58 GMT @sandyk said:

Hmmm. I thought that was the rationale behind your improvements to the Indeed Bravo, and your latest variations on a theme ? ;D.

Indeed Bravo costs 50-60 dollars and is no high end equipment at all and can surely be improved. I beleive that Frans is referring to all those close-mindend people spending multum of cash on their equipment and then modifying it. Of course if they hear the difference then what can anybody else say to discourage them?

Apr 23, 2012 11:43:58 GMT @sandyk said:

I am now out of this stupid "tit for tat" crap.

Why give up so easily?

[Deleted]

Why give up so easily?

Simple.
Originally, Frans and myself agreed on the majority of purely technical issues.
These days, I get the feeling that if Mike or myself said something was white, Frans would insist it was black.
P.S.
It's my bed time, so as far as I am concerned, Frans can continue on telling you and others why your modifications have no technical basis, and that you must be imagining the differences that you report. ;D
Good Night from DownUnder.

[szoze]

Apr 23, 2012 8:18:17 GMT PinkFloyd said:

Stock diodes are 1N4007. You can replace with UF4007 or UF4001

UF4001 are even faster (50ns as opposed to 75ns) and perfectly good to use with the V3.

Mike, what capacitor values do you use för snubbering the diodes in v3?

[PinkFloyd]

Apr 23, 2012 15:24:17 GMT szoze said:

Apr 23, 2012 8:18:17 GMT PinkFloyd said:

Stock diodes are 1N4007. You can replace with UF4007 or UF4001

UF4001 are even faster (50ns as opposed to 75ns) and perfectly good to use with the V3.

Mike, what capacitor values do you use för snubbering the diodes in v3?

10uF polypropylene.... don't worry about the lead inductance, it actually improves the performance of the diode.... smoother mids, ripe bass, shimmering treble etc. You will have to modify the board / enclosure though as the 10uF polyprops are quite large

[Deleted]

;D

[szoze]

Apr 23, 2012 19:57:52 GMT PinkFloyd said:

10uF polypropylene.... don't worry about the lead inductance, it actually improves the performance of the diode.... smoother mids, ripe bass, shimmering treble etc. You will have to modify the board / enclosure though as the 10uF polyprops are quite large

But isn't 10 uF too small value? I have some 100 uF polyprops I think I will use to make the diodes even slower. Maybe the sound parameters will improve even further with bigger caps as we all know: the bigger the better.

The 100 uF caps are quite big physically but hey I will need a new enclosure anyway. One problem will be to mount the caps onto the diods but maybe this time I shall do it the other way around. The other problem will be to find a shielded enclosure cuz I am really scared of those RFs. They are my real enemy right now, don't you think??

[szoze]

Joking aside, I think I will fit 1 nF capacitors onto the diodes and make an AB test. We need some kind of evidence that these mods really work.

[Deleted]

Coin flippin'....

I flipped some coins to get a better statistical view on these test where the 'boundaries' lie of what can be attributed by chance.
If the differences are obvious/night and day one would expect 11 or 12 out of 12 tries, perhaps when less audible 10 out of 12 would be correct.

We also don't know the personal preference nor competence of the listeners involved.
Differences however are differences.

O.K. flipped some coins.. 12 coins head=modded, tails=nonmodded.

3 tries and later on again 3 tries, 3 persons each trying 12 times in 2 sessions.

person 1: modded amp: 8 votes, nonmodded amp: 4 votes
person 2: modded amp: 7 votes, nonmodded amp: 5 votes
person 3: modded amp: 7 votes, nonmodded amp: 5 votes

In this case by simple coin flipping there seems to be a 'convincing' evidence to support the modded amp would be preffered by 8votes (of a total of 36)

next attempt.
person 1: modded amp: 10 votes, nonmodded amp: 2 votes
person 2: modded amp: 5 votes, nonmodded amp: 7 votes
person 3: modded amp: 6 votes, nonmodded amp: 6 votes

again the modded amp has preference for 5 votes.

So unless you reach 11 or 12 out of 12 the differences might well be due to chance.. or was the difference small yet perceivable in a couple of cases ?

When we add the results of both attempts the difference in the real tests are the modded amp wins by 2 votes.
This can be seen as... there is a difference but in the light of a total number of attemps of 72 it seems more like chance.

This is why blind tests are not accepted by many.
In most cases (unless very measurable differences are found) it seems impossible to pass a test blind.
Sighted differences appear to be quite obvious and beyond all doubt.

an interesting phenomenon.

[PinkFloyd]

Apr 23, 2012 15:24:17 GMT szoze said:

Apr 23, 2012 8:18:17 GMT PinkFloyd said:

Stock diodes are 1N4007. You can replace with UF4007 or UF4001

UF4001 are even faster (50ns as opposed to 75ns) and perfectly good to use with the V3.

Mike, what capacitor values do you use för snubbering the diodes in v3?

33pf polystyrene.