John Curl's Blowtorch preamplifier

[Justcallmedad]

... A schematic will be clearer to show what I have in mind:

Fet/BJT_PSU.

[Elso Kwak]

Originally posted by Justcallmedad
... A schematic will be clearer to show what I have in mind:

Fet/BJT_PSU.

The LEDs should have a constant current source. Your circuit has poor load regulation as there is no feedback from the output. There are better references than a few LEDs and also way more expensive.

[lineup]

Originally posted by Justcallmedad
... A schematic will be clearer to show what I have in mind:

Fet/BJT_PSU.

Voltage at output may change a tiny bit depending of load.

But I think your circuit will have a very clean supply.

[jbaudiophile]

Elso is right, Justcallmedad

But he should just have said you that another N-Channel Jfet with the gate attached to te source could just replace the resistor feeding the led chain...

Far better constant current through the leds... and not as expensive as some integrated ones. For such a purpose, long channel Jfets are the best (2N4338 and the like... Just have a look on some Siliconix data books).

Seems too that Richard Marsh used a very similar circuit in early or mid-80s... Will try to find it out and put it on the forum if the thread just not collapse before...

jbaudiophile

[Justcallmedad]

Hi Elso, Hi lineup,

“The LEDs should have a constant current source.”

Yes for PSRR a CCS will be better, but not sure for the sound, especially if I use a Pre Reg, but it's something that could be tested.
I just intended to draw a simplified last stage PSU.

“Your circuit has poor load regualtion as there is no feedback from the output”

I don’t like feedback even in PSU (at least for the last stage), in the case of this thread the PSU is intended for a complementary symmetrical balanced input balanced output class A preamplifier, so load regulation is not a problem.

“There are better references than a few LEDs and also way more expensive.”

I got best results with LEDs than with TL431 or LM336 but I have not tried other ones from AD LT etc.

[john curl]

Justcallmedad, thanks for showing a schematic. Personally I like your design, AND you avoided many oversights often made by others.
Your approach is truly low noise. The output transistor is driven by low impedance and the output is 'shorted' at audio by the large cap at the output. Your DC stability is OK, but could be improved, IF you felt it necessary by using a current source instead of a resistor, but it will still work well enough and it will be very quiet at audio frequencies.
Now, why don't I do this? Well, as I said in a previous comment, I can hear the 'sound contribution' of the output electrolytic cap when it powered my folded cascode circuit. As an engineer, I would not have thought it possible, but I heard it in a direct AB test with a large value film cap, that sounded better. This forced me to use another approach, still very similar to yours:
I just used the FET driver directly and removed the output transistor and the large output cap. Now what is the tradeoff? Well the output impedance WILL go up, BUT if I use a regulator for each individual circuit that is running class A at all times, then the 5-20 ohm output impedance might be OK, because the change in Gm will be very small, because the power supply current load will be essentially constant. Then I don't need the final large cap. There won't be any added X-talk either, because each channel uses separate regulators, in fact each gain stage has its own regulators. That is why I use so many regulators in my circuits. In the picture of the CTC there are 8 visible, and 12 total, in that box alone. With the power supply box added, there are a total of 28 active regulators to drive the audio circuits in this box.
Now, what about DC regulation? Well in the actual power supply box, I use conventional 3 terminal regulators, to remove hum and ultra low frequency (breathing) of the power line. I have found that 2 regulators in series is better than one 'super' regulator. I hope that this makes sense to you.

[serengetiplains]

Originally posted by john curl Well, as I said in a previous comment, I can hear the 'sound contribution' of the output electrolytic cap when it powered my folded cascode circuit. As an engineer, I would not have thought it possible, but I heard it in a direct AB test with a large value film cap, that sounded better.

Seems to me an output cap is ghosting any AC due to its dielectric absorption, DA being high for an electrolytic. Also seems to me uncontroversial that such ghosting could be audible (as IME it is), and that its audibility is an engineering consideration that may hard-limit a regulator's actual audio performance.

[Steve Eddy]

Originally posted by serengetiplains
Seems to me an output cap is ghosting any AC due to its dielectric absorption, DA being high for an electrolytic. Also seems to me uncontroversial that such ghosting could be audible (as IME it is), and that its audibility is an engineering consideration that may hard-limit a regulator's actual audio performance.

What do you mean by "ghosting"? It seems to imply some "second" signal that's basically a time-delayed version of the original signal. I've seen similar notions before, usually by those who are somewhat familiar with the DC-based test for DA and draw erroneous conclusions from it, i.e. that of a time-delayed "second" signal, or "ghost."

se

[serengetiplains]

Call it what you will, capacitors muck up AC signals in a way that seems tightly correlated with their DA rating. Beyond stating that correlation, who really knows how capacitors *actually* muck up AC signals? Steve, your use of the word "erroneous" suggests you're one who does know. Care to share?

[Darry]

Mr. Curl,

You say:

""""Well in the actual power supply box, I use conventional 3 terminal regulators.""""

You use them like voltage regulator or as current regulator (current regulator for to pilot your shunt regulator). Maybe the 2 in series, voltage regulator then current regulator.

Darry