John Curl's Blowtorch preamplifier part II

[Esperado]

I recommend Jan to anyone who needs a straight man.

The other fashionable insult, during my freak period, was "square".

 

[bogdan_borko]

This thread became the WHITE NOISE again. Someone, please stop the agony and start other meaningful discussion.
Ego fighting is infantile...

 

[SY]

Sorry, that was an Americanism. A "straight man" is one half of a comedy duo, the guy who sets up the joke for the other guy to deliver the punch line. Bergen to McCarthy, Abbott to Costello, Crosby to Hope.

 

[31697B]

My 'theory' (in other words, IMO) about some parts replaced which had low 'distortion'.... ... there seems to be some forgetfulness that we dont replace one single component part in a system. Going from all cc (like I showed of a Dynaco tube preamp) to all mf resistors is what we typically would do and dozens changed thru out the entire system on both channels..... THAT is when even small amounts begin to make a larger difference.

Same might apply to one single cap or one opamp somewhere. The differences show more easily when a lot of them are changed. And, generally, that is what we have done over time..... all are upgraded. Does it measure or sound different? The total affect is always greater.

Typically, without servo in audio circuits, we have 8 polar caps per channel (LP). Still that way in many low-mid priced receivers and amps and add in Cd players.... one on each in/out/and gain.

There is a necessary place for individual component testing, of course, and there is also a place to see what accumulated affect using several (dozens?) gizmo's will have as well. What would be the best dut topology for testing several cap at same time to reasonably represent the total circuit behavior of all the cap??


THx-RNMarsh

Vda is simply this...... the value of Cda and Rda compared to the value of the load. The load could be a transistor input Z.


View attachment DA.pdf



As the TC is usually longer than the main C and Rl it is both absorbed and released at a different rate and time. The Rda can be high like in film dielectrics and Cda small or it can be larger Cda and lower Rda as in polar types. And, the values of subsequent Cda and Rda shown in the DA model are different values from one another. iirc more log in vlaue but dont hold me that exactly.



THx-RNMarsh

 

[Waly]

Waly, you are asleep at the wheel. Wake up! The amount of DA changes with the RC time constant. We actually measured it.

Of course, 40 years ago. What was I thinking?

 

[john curl]

To note differences with DA with loading. Look at fig.1 and fig.6. on page 2 of the paper Walt and I wrote.

Same CAP, same test signal, different load ONLY.

 

[fas42]

Yes, but they are pretty good at HF, no ? So why not use them in // with other brands as power shunts? If i'm looking for near perfect, I use a 1/10 value for each. Like 1000µ lityc,, 100µ lityc, 10µ film, 1µ film, 0.1µ ceramic.
PMA, thanks for all this very instructive measurements.

Christophe, this not a good strategy! You'll likely have resonant impedance peaks, unless you're very lucky - either have the values very close in value, and/or significantly further apart, a factor of 50 or more is a good rule of thumb.

 

[john curl]

For the record, Waly, (and SY) this is a test that I did with a 1uf ceramic, today with a 10K load and 10Hz. Kind of lousy, isn't it SY? '-)

 

[Esperado]

Sorry, that was an Americanism.

Sorry too, that was French humor. (pince sans rire)

 

[Esperado]

Christophe, this not a good strategy! You'll likely have resonant impedance peaks, unless you're very lucky - either have the values very close in value, and/or significantly further apart, a factor of 50 or more is a good rule of thumb.

If i remember well (i used-it long time ago), 1/10 is the "very close" value. If the values are "further apart", yes, you have resonances between the inductance of the big cap and the higher capacitance. But, yes, it is good, depending of the wiring, to dump the LCs with little Rs.

 

[fas42]

You remember poorly! 1/10 is bad, something like just a couple of steps in the normal E12 range is what's needed. Forget about adding R, you're just throwing away the value of the capacitors in doing their job! Getting exactly the right combination can make a big difference, and the higher in frequency you're looking at the more difficult it becomes, Diminishing Returns kicks in, and placement in the physical circuit becomes everything.

 

[31697B]

John..... I found a lot of Wireless World and others in boxes in my attic in California where recorded and two phono carts that had capability..... a LTE from A.J.Watts: he quoted from a book Hi-Fi System by G.King a ref to a LP signal peaks of 80cm/sec. From that, two cart gave 200mV peak output -- Ortophon SL15 and Decca.

-RNM

 

[SY]

For the record, Waly, (and SY) this is a test that I did with a 1uf ceramic, today with a 10K load and 10Hz. Kind of lousy, isn't it SY? '-)

Yes, a high K ceramic is nonlinear, that's not what we're discussing. That shows nothing about DA's effect on audio signals.

 

[31697B]

For the record, Waly, (and SY) this is a test that I did with a 1uf ceramic, today with a 10K load and 10Hz. Kind of lousy, isn't it SY? '-)

What is the % DA rating of it?





-RNM

 

[john curl]

Ceramics have BOTH DA and non-linear distortion, and hysteresis. Use with care, SY. You were giving PMA a bad jibe, and I want to correct you.

 

[31697B]

Vda is simply this...... the value of Cda and Rda compared to the value of the load. The load could be a transistor input Z.

View attachment Vda.pdf



As the TC is usually longer than the main C and Rl it is both absorbed and released at a different rate and time. The Rda can be high like in film dielectrics and Cda small or it can be larger Cda and lower Rda as in polar types. And, the values of subsequent Cda and Rda shown in the DA model are different values from one another. iirc more log in vlaue but dont hold me that exactly.


THx-RNMarsh

Lets try this picture above ------ Vda

 

[fas42]

IMO:
Easiest to hear: frequency response changes (or "linear distortions" if you like). Low-Q resonances are especially audible, high-Q ones require signal energy close in frequency to the peak or dip, with the peak usually more audible.

Fairly easy to hear with low-level program material: crossover distortion near the origin, amounting to a severe loss of gain at those low levels. Hard to hear with normal level program material.

Fairly easy to hear if there is a lot of it: clipping distortion.

More difficult: IM distortion, unless it is terrible, or resulting from a pathological pair of adjacent tones and absence of masking (example, the 1kHz difference from 19kHz and 20kHz).

Much more difficult: single-frequency-signal harmonic distortions from smooth nonlinearities. They sound like timbral shifts and small level differences.

The other thing: when B follows A, all other things equal, B is preferred---unless B is just awful. If B is louder than A, B is preferred---as SY described a while back, from his experience, even when the small difference is not perceived as a level difference.

POOGE may have been a noble effort, and undoubtedly lots of fun, but I think suffered significantly from the creator effect. How can one not like one's progeny---unless they are just awful?

My experiences are totally different ...

Distortion of low level information, which may be considered noise modulation or IM depending upon your point of view is the killer - this is what always immediately identifies an audio system as being a "fake", not convincing. It's also the factor that makes turning up conventional systems in volume beyond a certain point unpleasant - it becomes easy to hear that all the low level detail is screwed, and therefore it's no longer nice to listen to.

 

[john curl]

Richard, the tested cap (the Wurcer test) showed 6-12% deviation. DA, follows proportionally with the same test with other caps such as Mylar.

 

[31697B]

Richard, the tested cap (the Wurcer test) showed 6-12% deviation. DA, follows proportionally with the same test with other caps such as Mylar.

High Volt coeff and high DA, both. The worst of all worlds.


THx-RNMarsh

 

[SY]

Ceramics have BOTH DA and non-linear distortion, and hysteresis.

Yes, but your data do not show any nonlinearity due to DA. It may be there, who knows, but I can't see anything in any of the data you and Richard have presented which shows that. Sure, move an RC zero around the test signal frequency and you can see lots of issues with nonlinear dielectrics- and changes in pulse shape (assuming the polarity reversal was the photo!). No-one disagrees with that. But your attribution of the nonlinearity to DA is the question.