John Curl's Blowtorch preamplifier part II

[EUVL]

Many thanks, Scott.

Patrick

 

[Joao@AltheaMusica]

Hi John,

I certainly agree with you that most op amps are happier in many regards when operating in the inverting mode, and that sharp edges in the input signal don't get to the input differential pair as badly, due to the RC filter you describe in the inverting integrator configuration. There is also the matter of little or no high-frequency, sharp-edged common mode applied to the op amp.

This does not, however, make it impossible to get low TIM in an op amp of high quality using feedback in a non-inverting configuration.

BTW, these discussions bring back to mind the use of passive RIAA equalization that has often been preferred as rendering higher quality. In such a case, of course, the fast edges are LPF'd without stress on an amplifier. The use of a passive RC for the high-frequency turnover followed by an integrator-type low-frequency turnover is also a good approach that still keeps fast transients out of the following amplifier.

Of course, there still remains the first buffer stage, which must deal with the fast edges that may come from the cartridge. At least this stage then only has one job to do. It must be a chunk of gain with low noise that has very low distortion, even at high frequencies at fairly high amplitude, and must be very resistant to EMI from all sources. This is particularly important for the first stage in moving coil preamplifiers.

Cheers,
Bob

I agree and vote for the passive RIAA as well. Some thoughts regarding time delay with using active, or better, negative feedback loops.

Your music signal has / or is a texture. Assume that you have an active RIAA and therefore a negative feedback signal - so what happens:

Time 0:
the signal, or better musical texture, arrives at the input and passes the amplifier stage. The signal leaves the phono stage and arrives in the line stage. Same time the signal cycles the feedback loop - here RIAA EQ

Time 0+x:
the next signal, or better musical texture, arrives at the input of the phon stage.
The corrected feedback signal of the musical texture from "T=0" arives at the input of the phono stage. Here it will added to the next signal form or musical texture.

And so on ....

From measurement view the active RIAA-EQ can be better than the passive version, but looking on the above sight of view it seems logical that the passive RIAA-EQ is the better one because it lets the musical texture intact.

 

[Joao@AltheaMusica]

For my MC cartridge loading, I will use a remote controlled 10 turn wirewound triple pot of 5K ohms, that will switch from there. Already designed into my latest preamp. I want resolution to 1 ohm or so. The resistor value will come up on a video screen and can be stored for future setting. We shall see whether we can tell that much difference or not, especially with $5,000 and above phono cartridges, like Joachim uses.

That's what I was thinking to do.... do you not have problems with interferences from the microcontroller?

 

[vuki]

In case you really want to get surprised, this is the typical phono record distortion. Distortion of the preamp itself is about 0.001%.

It's not surprising Every well regarded tube SE amplifier has similar distortion spectrum. It's "natural sound" spectrum and we love it!

 

[PMA]

Thanks, that makes sense.

To clarify completely, there are 100 samples per division. We have 10us/div, i.e. 100ns per sample, i.e. sampling frequency is 10MHz in this case. But amplitude resolution is only 8-bit per full vertical scale, so amplitude accuracy is nothing to call home about. Anyway, the sampling is fast enough to show sudden bigger changes in plot slope and to follow the wave shape.

Regards,

 

[PMA]

It's not surprising Every well regarded tube SE amplifier has similar distortion spectrum. It's "natural sound" spectrum and we love it!

I agree. And the result I have shown is still quite good.

Yes, Phono distortion + tube distortion + speaker distortion together make unique, unforgettable blend. Like wines. Wines are much better analogy than cars for sound tastes

 

[Johnloudb]

In case you really want to get surprised, this is the typical phono record distortion. Distortion of the preamp itself is about 0.001%.

Pavel,

I see the fundamental frequency for that test is at 300 Hz. Doesn't the LP have lower distortion at higher frequencies due to pre-emphasis of the higher frequencies in the record?

That distortion spectrum also reminds me of a Krell integrated amp I saw reviewed awhile back to. It got really good reviews. Interesting.

John

 

[Johnloudb]

I'm not sure if that makes sense or not. Just thought that might have an effect on distortion.

 

[PMA]

Distortion gets worse with frequency and closer to the center of the record.

http://www.badenhausen.com/VSR_History.htm

 

[Bob Cordell]

It is no ultrasonic noise, BUT 8-bit A/D digitization of the scope (only 255 levels per full screen scale). You measure the same from generator only. Sometimes I am very surprised what I read here.

Darn, I miss those good old analog scopes! But, of course, one can do so much more with a DSO, especially for non-repetitive waveforms.

Cheers,
Bob

 

[Bob Cordell]

In case you really want to get surprised, this is the typical phono record distortion. Distortion of the preamp itself is about 0.001%.

Look at that 7th - only 70 dB down. It all looks like crossover distortion. I would have expected that LP distortion would be much more low-order dominated. What was the setup you had here, and are you really sure this is real?

Cheers,
Bob

 

[PMA]

Darn, I miss those good old analog scopes! But, of course, one can do so much more with a DSO, especially for non-repetitive waveforms.

Cheers,
Bob

I love the analog scopes as well, and I have it, but it takes a lot of time to make screen photos, and we have another sources of distortion like angle etc.

Regards,

 

[PMA]

Look at that 7th - only 70 dB down. It all looks like crossover distortion. I would have expected that LP distortion would be much more low-order dominated. What was the setup you had here, and are you really sure this is real?

Cheers,
Bob

I understand, but the preamp has distortion about 0.001% and only 2nd and 3rd visible in the spectrum (both below -100dB). Everything far below groove noise.

http://www.synthgear.com/2010/audio-gear/record-grooves-electron-microscope/

Regards,

 

[Bob Cordell]

I agree and vote for the passive RIAA as well. Some thoughts regarding time delay with using active, or better, negative feedback loops.

Your music signal has / or is a texture. Assume that you have an active RIAA and therefore a negative feedback signal - so what happens:

Time 0:
the signal, or better musical texture, arrives at the input and passes the amplifier stage. The signal leaves the phono stage and arrives in the line stage. Same time the signal cycles the feedback loop - here RIAA EQ

Time 0+x:
the next signal, or better musical texture, arrives at the input of the phon stage.
The corrected feedback signal of the musical texture from "T=0" arives at the input of the phono stage. Here it will added to the next signal form or musical texture.

And so on ....

From measurement view the active RIAA-EQ can be better than the passive version, but looking on the above sight of view it seems logical that the passive RIAA-EQ is the better one because it lets the musical texture intact.

Hi Joao,

I like passive RIAA, but not for the reason you cite.

Unfortunately, the "chronological" way of thinking about NFB is a classically flawed argument; one of the ones that leads to distrust and misunderstanding of NFB. Those time delays are microscopic compared to the period of the audio signal if the amplifier has a good high frequency gain crossover frequency and good stability.

The closest to truth that this way of looking at things comes is the Baxendall effect where Baxandall illustrated re-modulation in a feedback loop due to interactions between the fed-back signal and the input signal. I discuss this effect at length in my book. The effect is negligible in practice. It turns out that this effect actually pretty much disappears once the total amount of NFB around a nonlinearity exceeds about 15 dB. This amount of feedback INCLUDES any amount of feedback arizing from local degeneration. The Baxandall effect is a mathematical reality that has little or nothing to do with time delay.

Cheers,
Bob

 

[jan.didden]

[snip]Time 0:
the signal, or better musical texture, arrives at the input and passes the amplifier stage. The signal leaves the phono stage and arrives in the line stage. Same time the signal cycles the feedback loop - here RIAA EQ

Time 0+x:
the next signal, or better musical texture, arrives at the input of the phon stage.
The corrected feedback signal of the musical texture from "T=0" arives at the input of the phono stage. Here it will added to the next signal form or musical texture.

And so on ....

[snip].

I don't think this is a correct way to look at it. Even if you have phase shift in the feedback loop (which you will get if you get high enough in freq) there is no 'delay' as implied here.

For instance, it is known that the voltage across a cap is shifted 90 degrees wrt the current into the cap.
If you use a sine wave current, the voltage is also a sine wave, 90 degrees phase shifted. It *looks* like a delayed input wave but it isn't.

Think about it: every minuscule change in the input current gives an *immediate* minuscule change in cap voltage. There is NO delay!

I tried to explain here:
http://www.diyaudio.com/forums/blogs/janneman/454-feedback-how-late-time-same-time-all-time.html
Let me know if I succeeded

jan didden

 

[john curl]

Returning to my 32 year old IEEE paper, I have found that instead of triangle waves coming out of RIAA equalized phono stages, 10us rise-time SQUARE waves came out. This is REASONABLE WORST CASE, and this happens if you apply a TIM (30,30) pulse (figure 8) to the input of a phono stage or the mistracking pulse of the Ortofon MC phono cartridge. THIS is what I tested even the lowly 741 with, and the LF356, the results shown in graph 1.

 

[AndrewT]

Returning to my 32 year old IEEE paper, I have found that instead of triangle waves coming out of RIAA equalized phono stages, 10us rise-time SQUARE waves came out. This is REASONABLE WORST CASE,

yes, a constant velocity of the stylus resulting from a triangle wave gives rise to a constant voltage at the RIAA output.
Change the slope of the triangle waveshape and you change the voltage value.
Swap the slope from +ve to -ve and vice versa and the constant voltage changes from one side of zero volts to the other side.

Yes, triangle input gives squarewave output.
The difficult bit is how fast can the stylus follow the change in slope of the triangle wave. That determines the rise & fall times of the square wave.

 

[AndrewT]

I don't think this is a correct way to look at it..................
Think about it: every minuscule change in the input current gives an *immediate* minuscule change in cap voltage. ..................
Let me know if I succeeded

A flow of charge (current) into a capacitor changes (gradually) the voltage stored between the plates.
A constant current into (or out of) the cap gives rise to a straight line slope of voltage with time.

The current does not need to change for there to be a change in voltage.
The current being ON or OFF is the criteria for a change in voltage across the cap.
The current value changing gives rise to a change in voltage vs time slope. Note: slope of Voltage vs time, not absolute change in voltage.

 

[myhrrhleine]

Of course, there still remains the first buffer stage, which must deal with the fast edges that may come from the cartridge. At least this stage then only has one job to do. It must be a chunk of gain with low noise that has very low distortion, even at high frequencies at fairly high amplitude, and must be very resistant to EMI from all sources. This is particularly important for the first stage in moving coil preamplifiers.

Cheers,
Bob

just want to add low gain.

just enough to set overall noise levels.

 

[PMA]

Even if you have phase shift in the feedback loop (which you will get if you get high enough in freq) there is no 'delay' as implied here.

That's it.
But makes no sense to explain to people who do not have certain level of knowledge in circuit theory and feedback theory. Even real results would not convince them.

And, do not forget Jan, that marketing purposes require to protect and emphasize certain design philosophy. This is the one and only reason.