John Curl's Blowtorch preamplifier part II

[simon7000]

If there is anyone interested, here is the complete measurement of opamps + schematics.

Very Nice!

 

[SY]

I haven't seen preamps where line stage has 40dB of gain lately.
What happens when you reduce the gain to 20 or 10dB? Still no 7th?

Best ask Pavel- he did the experiments. I use a unity gain line stage, and no discernable 7th.

 

[PMA]

Thanks for this post, John:
http://www.diyaudio.com/forums/anal...ch-preamplifier-part-ii-3329.html#post3337331

What I thought is that here we could exchange our experience and results, and we may leave marketing to marketing people and market places.

As we are getting older, our experience and results are the most valuable things we may share. Everything else is passing.

 

[PMA]

Best ask Pavel- he did the experiments. I use a unity gain line stage, and no discernable 7th.

We may quite well predict the distortion much less for gain 0dB compared to 40dB. It is almost 40dB lower then, for many opamps. But difficult to measure. Yes, no 7th for 0dB. The 7th is easier to see at higher gain - like OP27 in my measurements.

 

[jneutron]

George
At least for the current crowding effect being discussed wouldnt the co-ax symmetry prevent it all together?

Thanks
-Antonio

No. Details below.

Coax symmetry is not the problem. The issue with coax is that the spacing between inner and outer is large enough that normal skin effect dominates. Now I believe that skin effect and proximity effect are merely auto and mutual versions (respectively) of the same thing so will behave in the same way, but I am not sure everyone agrees so demonstrating nothing strange with skin effect might not convince those looking for strange proximity effects.

Coax exhibits two effects.

First, as frequency rises, the core wire current will skin to the outer surface of the core, and will skin to the inner surface of the shield.
Second. If the core wire is not centered (offset) within the coax, or the shield is asymmetric (oval for example), the shield current will redistribute on the shield so that it's centroid will coincide with the centroid of the core current. This is proximity based.

You would probably need to setup something where conductor spacing is smaller than skin depth so proximity effect dominates, as may sometimes be the case for audio. Maybe an air-cored inductor or transformer, but with well-encapsulated windings so no physical movement is possible?

Done it all, measured same. Measured air coils of the same physical dimensions all wound identically using single conductor magnet wire, two conductor magnet wire, and now some awful litz stuff with 133 individual filaments, 15awg IIRC..and all potted in a two part alumina filled epoxy. The litz is difficult to connect to.

That all the irony of all this passionating discussion: The distortion levels we are talking about are so low that we can doubt of their exixtance, and there is hundred places in all our systems, where we can improve distortions of 100X this levels.

Ed did not try the resistors I mentioned...low value ones. The magfield drive currents in his test were rather small..

Comments -- skin affect and proximity have the same affect... but are caused by different means. The lowering of inductance is what causes the currents at higher freqs to flow in that region of the 'skin'. But what causes that to happen? The wire(s) geometry causes electromagnetic fields to be concentrated more and less across the conductor. In a round conducting wire the field is concentrated -denser- at the center(actually, that is backwards...the self induced magnetic field intensity within a round conductor is zero at center and increases linearly out to the surface of the conductor.) ... thus the center has the higher inductance. Where ever the fields get concentrated, the inductance is increased. Proximity can reduce inductance but it is not evenly distributed or symetrical anymore..

It's best to think of it as the current is trying to take the path of least reactance. As frequency increases, the path becomes the lower inductance path. For a wire pair next to each other, that path is as close to the other conductor as is possible given the fields. for a coax, it is the surfaces closest to one another.

Thsi also occurs on PCboards with a full ground plane, and also within chassis when ground current is flowing through the chassis, and when equipment is connected via more than one ground.

Because the inductances of the wires differ ?

The physical arrangement of the wires changes the inductance of the system. Also of concern would be if the entire foil construct is antiparallel current or parallel.

jn

 

[Esperado]

It entertained the mass, and did no harm in epoxy it MAGIC . The buyer saw it as value no harm to them you will be forgiven.

I hope Mr Dolby will benefit of the same indulgence, with his "surround" process he imposed to the movie (tv) market, while he really harmed us with this poor and stupid standard, production side, for its own profit.

mylife.com:
We were obliged to get an official 'approval' of the studio equipment and tunings, to be "Dolby certified". The day the Dolby guy had finished the inspection of our brand new "Surround" movie mixing studio, he was writing the official certification paper.
When he handed me the certificate, I fell on my knees, in the position of the peasant about to be promoted to Knight. Everyone laughed (but he).

I remember too the story that happened to one of my technical manager friends. They just finished a fantastic movie mixing facility in a big theater in Paris. Sumptuous. The speakers where included in a very thick concrete wall across all the width ot the room, behind the screen. The guy from THX refused the certification, because, in the official requisites, the wall was described as to be "plaster panels" !!!!
They had to exchange letters with George Lucas, for this shame was arranged.

 

[Telstar]

And finally, OPA637. Low distortion, low noise and low sensitivity to interference. My vote

No 7th!!

Or under 130dB which is your setup noisefloor. Edit: ok you raised the gain to show the harmonics above the noise floor, good work. So we can assume all values 10-20db lower than that.

BYW why dont you try with a 50 or 100hz tone and show us all the harmonic spectra? There may not be enough dissonant 7th, but there can be anything much higher than that.

 

[Telstar]

Many thanks, PMA, for sharing your measurements. Once again, it prove all those evil little black boxes works damn well !

'fraid it doesn't prove anything.
1k TDH and noise spectra is nowhere near a sufficient measurement of sound quality.

 

[PMA]

Or under 130dB which is your setup noisefloor.

Why dont you try with a 50 or 100hz tone and show us all the harmonic spectra? There may not be enough dissonant 7th, but there can be anything much higher than that.

1) -130dBFS is NOT the measurement noise floor. The lowest noise floor had the AD797 - you can check the measurements, it was -135dBFS. The noise floor of my measurement setup is below -140dBFS.

2) Why not to try 50Hz or 100Hz - for it shows nothing more and nothing new. The test condition would be softer compared to 1kHz (higher feedback factor at 50 or 100Hz). With 1kHz, you might see harmonics up to 47th, if there were any.

I made the measurements for the reason I want to stay accurate - I do not like speculations and unsupported claims.

 

[PMA]

'fraid it doesn't prove anything.
1k TDH and noise spectra is nowhere near a sufficient measurement of sound quality.

It proves that speculations about the 7th were fairy tales. But it takes much less time to figure out and tell the fairy tale then to prove it false by measurements.

If you could read the plots you might see the difference in sensitivity to interference. It clearly shows that JFETs are superior to BJTs in this.

 

[Max Headroom]

Subjective Evaluations ?...

None of the good opamps (OPA627, AD797, OPA134 ...) has any residual 7th above -120dB. They do not sound same, but the 7th is not the reason.
Hello again Pavel, I respect your measurements data, that said it reads that you are making subjective comment here. Could you please elaborate on your subjective findings of these opamps, or at least your subjective findings of differences between this collection of opamps.

Thanks, Dan.

 

[PMA]

Dan,

it is pretty difficult and subjective.

From the set here, I like most OPA637 and AD844.

AD825 sounds "nice", but it wipes off differences in quality of different recordings, equalizing them a bit - to me.

AD797 is a very good opamp but I prefer AD844 and OPA627/637 for line level, they sound more cultivated to me. IMO this is because of the worse immunity of AD797 to interference.

 

[john curl]

Please give it a break, PMA. 7th harmonic is STILL very important. I challenge YOU to make something better than my designs, rather than 'bad rap' me.

 

[SY]

No one said it wasn't. What was challenged was your apparently incorrect claim that it is what distinguishes ICs from discrete, relegating the former to "mid fi."

 

[Esperado]

None of the good opamps (OPA627, AD797, OPA134 ...) has any residual 7th above -120dB. They do not sound same, but the 7th is not the reason.

Right, Dan. The question remains, witch is the only interesting (and mysterious) one: WHY ?

 

[31697B]

It's best to think of it as the current is trying to take the path of least reactance. As frequency increases, the path becomes the lower inductance path. For a wire pair next to each other, that path is as close to the other conductor as is possible given the fields. for a coax, it is the surfaces closest to one another.

Thsi also occurs on PCboards with a full ground plane, and also within chassis when ground current is flowing through the chassis, and when equipment is connected via more than one ground.

I agree with all this part. It is all documented in any book on the subject. Nothing new here. -RNM

jn

I agree also with the inductance vs freq....goes to path of least resistance... towards the surface of the conductor where the L is lowest.
Do not agree on your explaination regarding the fields. Perhaps just a different view of the same thing? -RNM

 

[Max Headroom]

English Can be A Powerful Language...

Cultivated
Synonyms: accomplished, civilized, educated, enlightened, learned, polished, refined, sophisticated, urbane.

Ok, not a term I have heard or used to describe audio equipment, but I think I get the picture...maybe.
Interesting that these opamps that all measure close to perfect (low ppm distortions) should elicit subtly different subjective opinions.
Nonetheless, you have provided a shortlist and give me the inspiration to do some ordering and soldering in order to take a closer listen for myself.

Thanks and regards, Dan.

PS - Anybody here have same or different subjective opinions of this collection of opamps when used in line level applications such as volume control recovery amp or line output driver ?.

 

[PMA]

Please give it a break, PMA. 7th harmonic is STILL very important.

Yes it is, but it is non-existent in good opamps up to 40dB gain in any audible level. I support the answer that SY gave to you. Please stay technical and at engineering level.

 

[john curl]

Let me explain the situation as I see it.
A number of comparisons of THD (the easiest measurement possible) at 1KHz,(the easiest frequency possible) with MAXIMUM feedback applied, to the point of absurdity, where something else would be a better solution, just to MASK the inherent non-linearity in the AD797, and hide its intrinsic properties.
For the record, I seldom MEASURE IC op amps, anymore. I just LISTEN to them, as I did with the AD797. I would have included it in my JC-3 phono design, but I found another IC that sounded just as good, at less than 1/2 the price, with the same basic specs. I avoided using the AD825 for the second stage, for other practical reasons. I suspect, all else being equal, it would sound better than the OPA134 that I selected, but I needed a DUAL op amp, not a single.
Harmonic distortion is useful ONLY when a frequency sweep is used as well as a low residual. Heck, you might as well use SMPTE IM. It is more sensitive to many nonlinearities, so why not? Where are the SMPTE IM analyzers with a resolution to .0000001? I sure could use one.

 

[Max Headroom]

The Ultimate (Audio) Question...

Right, Dan. The question remains, witch is the only interesting (and mysterious) one: WHY ?

With all the brainpower collected here, we still don't have the answer it seems.
I do know that the answer is not 42 .

Dan.