What are those values? Is it a 1k drain resistor and a 3.3k source resistor?

Hard to tell..

But if Source resistor is 3k3 to 3k9, then working current goes down to 4mA to 5mA..
(my previous estimation for 15mA was based on 1k resistor..)

1k Drain resistor sets Q7 working current at about 700μA (Q6 V_BE / 1k)

Quote from: antonis on October 17, 2024, 07:08:19 AMHard to tell..

But if Source resistor is 3k3 to 3k9, then working current goes down to 4mA to 5mA..
(my previous estimation for 15mA was based on 1k resistor..)

1k Drain resistor sets Q7 working current at about 700μA (Q6 V_BE / 1k)

If source resistor > drain resistor, isn't the JFET misbiased? I did something similar for the soft clipping on the negative voltage swing this causes, mind you.

Quote from: fryingpan on October 17, 2024, 08:32:21 AMIf source resistor > drain resistor, isn't the JFET misbiased?

If so, then ALL JFET Source followers are misbiased..

With 1k Drain resistor and 3k3 Source resistor, JFT's Drain sits at about 14,3V and Source at 1.7V..

Pretty good bias for a +/- 15V Sziklai pair Source follower..

Quote from: fryingpan on October 17, 2024, 08:32:21 AMI did something similar for the soft clipping on the negative voltage swing this causes, mind you.

Are we talking about CS or CD amp..??

Quote from: antonis on October 17, 2024, 09:28:57 AM

Quote from: fryingpan on October 17, 2024, 08:32:21 AMIf source resistor > drain resistor, isn't the JFET misbiased?

If so, then ALL JFET Source followers are misbiased..

With 1k Drain resistor and 3k3 Source resistor, JFT's Drain sits at about 14,3V and Source at 1.7V..

Pretty good bias for a +/- 15V Sziklai pair Source follower..

Quote from: fryingpan on October 17, 2024, 08:32:21 AMI did something similar for the soft clipping on the negative voltage swing this causes, mind you.

Are we talking about CS or CD amp..??

Sziklai pair.

https://postlmg.cc/xNQ2nZ3g

It's with both bipolar transistors, but results are similar with a JFET at Q2 (and gate tied to ground).

I'd say that if the power supply is dual +/-15V, source should ideally sit at a negative value, like -13.5V.

Quote from: fryingpan on October 17, 2024, 09:53:49 AMI'd say that if the power supply is dual +/-15V, source should ideally sit at a negative value, like -13.5V.

And do exactly what there..??

Going (ideally) to 28.5V positive and -1.5V negative..??

Quote from: antonis on October 17, 2024, 11:12:59 AM

Quote from: fryingpan on October 17, 2024, 09:53:49 AMI'd say that if the power supply is dual +/-15V, source should ideally sit at a negative value, like -13.5V.

And do exactly what there..??

Going (ideally) to 28.5V positive and -1.5V negative..??

Why? Vgs(off) isn't necessarily referenced to ground. In fact it is Vgs, not Vg.

EDIT: I actually simulated it and yes, Vg tends to sit slightly above ground. And that appears to be because of the large current flowing through it (coming mostly from the BJT transistor).

EDIT 2: of course I had a brainfart. Gate in a JFET has to be more negative than source, not more positive.

Mouser has a number of dual JFETs in stock. A low cost one with a biggish Vds (50V) is here:
https://www.mouser.com/datasheet/2/408/2SK2145_datasheet_en_20140301-3177240.pdf
As the maker Toshiba notes it's intended or "Audio Frequency Low Noise Amplifier Applications".

The typical noise curves versus source resistance are surprising. With Rsource over 2K (and up to 1M), the NF is below 1db. The other NF charts are similarly amazing. Mouser stocks the Y and BL selections - 3-6ma and 6-14ma Idss, both costing $0.76 in ones and down to $0.36 in 100s.

Toshiba has consistently been amazing in audio-focused discrete semiconductors. I guess I shouldn't be surprised.

There are some others listed, but I was quite taken with this device. Should be a good start for low noise, wide source impedance buffer or amplifier for audio.

Quote from: fryingpan on October 17, 2024, 11:26:30 AMof course I had a brainfart.

Brainfarting is the distinctive feature of active brains..

Quote from: fryingpan on October 17, 2024, 06:55:28 AMWhat are those values? Is it a 1k drain resistor and a 3.3k source resistor?

The schematic has a few issues.   What's weird is the mixed use of decimal points and EU notation: 3k9 and 3.9k!!!

The value of R87 is 3k9.



There's some scope to change the value although when you start getting to down to 600uA through R87 the JFET will operate at a small current.

QuoteI'd say that if the power supply is dual +/-15V, source should ideally sit at a negative value, like -13.5V.

Yes dual supply and a source voltage something around +1.5V.   Gate at zero, output a small positive voltage, at just under the JFET's |VP|.

FWIW,

Like all class A amp the output swing depends on the ratio of the bias resistor (in this case the source resistor RS) and the load resistance.  When RL=RS we might expect the negative swing to limit to -7.5V.

The load resistance also affects THD.  (As seen on most ckts in Self's book.)

The THD can be improved by making the drain more than 1k, say upto 3k9.   The JFET operating current would decrease and noise would need to be evaluate; probably not changing much.   Alternative the JFET current can be left as is and the source resistance reduced.    RD=RS = 2k2 to 3k3 is a reasonable compromise.

The THD is less than 0.001% at 1V swing and RL = 10*RS, and increases as the input level increases.

Not bad for two transistors.

If using this as an output buffer things like stability with capacitive loads would need to be addressed.  As least a resistor in series with the output lead.

For the base design I got about 3nV/rtHz input noise but that from spice, which doesn't model the JFET noise that well.   Nonetheless, a 1k gate resistor will introduce 4nV/rtHz input noise.    However a guitar pickup is going to introduce a lot more than that.   The advantage of a JFET input over a BJT is the noise current is lower and that means the noise doesn't go up as much with high impedance sources.

A +/-15V supply pushes a lot of common JFETs.  +/-12V would help.

Quote from: merlinb on October 16, 2024, 10:07:30 AMThis is for guitar, right? You seem very concerned about triple-zero THD which makes me wonder if this is for something else?

Aye. I may have gotten carried away with myself I won't lie. My first thought when I saw the design in the book was, I can make this cheaper than an opamp with similar performance. I do not think this is the case any more lol. Anyways, I wanted to pick the brains on this forum because for whatever reason, I am most likely to get a response form people who are willing to actually engage in the design questions, provide equations and link to external resources that can get me to where I need to go. Can't say I've had the same experience in other forms, even in the nominally electronic engineering ones, which is firmly the territory into which we have strayed in this thread. This certainly had been a learning experience. 

I will be examining the other options that have been posted after your response hahah. Checking the power dissipation at idle on LTSpice at a more sane +/-15V gives me roughly 200mW on both rails. That's probably more than I want on a simple buffer. Plus the requirement for kind of knowing how much current you need to sink and source is kind of pain. But that does mean I could probably reduce the power consumption even more as I really don't need to push things that far with a guitar buffer.

That being said, I do want to get into building studio effects so, to circle back to the original question on why do I want triple zero distortion figures, I guess that counts as an answer

Quote from: mzy12 on October 20, 2024, 08:55:52 AM

Quote from: merlinb on October 16, 2024, 10:07:30 AMThis is for guitar, right? You seem very concerned about triple-zero THD which makes me wonder if this is for something else?

Aye. I may have gotten carried away with myself I won't lie. My first thought when I saw the design in the book was, I can make this cheaper than an opamp with similar performance. I do not think this is the case any more lol. Anyways, I wanted to pick the brains on this forum because for whatever reason, I am most likely to get a response form people who are willing to actually engage in the design questions, provide equations and link to external resources that can get me to where I need to go. Can't say I've had the same experience in other forms, even in the nominally electronic engineering ones, which is firmly the territory into which we have strayed in this thread. This certainly had been a learning experience. 

I will be examining the other options that have been posted after your response hahah. Checking the power dissipation at idle on LTSpice at a more sane +/-15V gives me roughly 200mW on both rails. That's probably more than I want on a simple buffer. Plus the requirement for kind of knowing how much current you need to sink and source is kind of pain. But that does mean I could probably reduce the power consumption even more as I really don't need to push things that far with a guitar buffer.

That being said, I do want to get into building studio effects so, to circle back to the original question on why do I want triple zero distortion figures, I guess that counts as an answer

Studio effects? Keep in mind that for ultra pristine processing, heavily oversampled digital processing is probably the way to go these days. For character, on the other hand, ultralinear performance may not be needed. Actually, for single tracks, even 2-3% THD may be not only inconsequential but even desirable. If you have a look at the sound engineering forums (fora?) people are prepared to spend top dollar (pound? euro? kronen?) for '60s tech. For what it's worth, more than THD, what may be important is the transition into significant THD. Or even more important, the actual sound.

As an aside: sound "engineers" and producers may be a bit gullible these days (but not necessarily these days: read up on the silly '80s practice of buying NS10s and covering their tweeters with tissues because some producer with a few hits on his hands apparently used them and did so). Once upon a time, a sound engineer was expected to, well, not necessarily design their own equipment, but certainly find their way around a console or an outboard processor, not to mention tape machines and the like. To them distortion was an inescapable problem, and their mission was to be able to come up with technical or creative solutions that sounded good (or satisfied the requirements of the artists) while juggling the comparatively huge amount of noise and distortion even just recording on tape meant.

These days, audio (the purely electrical stuff) is basically a solved problem. (The next big step might be to find a way to actually transduce the electrical signal/power into sound much more efficiently and cleanly, as apart from neodymium magnets and better designs, current-day speakers are not all that different from the stuff from a century ago - the average efficiency of a woofer is around 2% and significant distortion already sets in at 10% of their rated excursion). Even the cheapest audio interface affords performance that would completely annihilate, from a technical point of view, high end gear from days past. Of course, it turns out that the distortion (maybe not the noise) was actually pleasing - or useful, at least - in pop (rock, whatever) production. (If you choose to target the classical music world, it's another story). Also, any dynamic processing, for instance, is going to introduce much more distortion and colouration than the surrounding circuitry. Anyway, a lot of sound engineers today pay through the nose for equipment that "colours tone in specific ways" - there is a lot of mystique and hype in the field, and people are happy to pay well into the thousands for things I find are more marketing than substance. I find that "cheap creation" - music, photography, whatever - is riddled with this kind of gullibility: cost of entry is and was relatively low, and some stuff that was borne out of expedience (or simply cost) has become highly coveted despite being technically "mediocre". I'm not saying that "lo-fi" is a fad, or whatever. It has its place. I'm not even saying that people should reject certain classic "tones" - an LA-2A, a Neve 1073 preamp, a Fender amp, a U47, whatever, they all work very well. Sometimes, they are relics of a distant past that can be replaced by other gear which can sound similar or even better, but they don't have the "name" cachet that some stuff has.

For instance, look at the film world. To them, technical excellence is paramount. The huge desks they use, the processors they employ, they are all genuinely "good stuff". Everything is tightly calibrated and standardised. It's a more "serious" world (much more money involved) and the people working in that field do not improvise, they do not (usually) do things out of expedience, they do not just forget to turn down a compressor (or badly calibrate a camera) and say "wow, this looks/sounds cool!".

Anyway, what's my point? There will be one in there somewhere.

Quote from: fryingpan on October 20, 2024, 09:40:40 AMefficiency of a woofer is around 2%

As long as we play large-chamber music in small rooms, the efficiency number won't go up. If the air-paddle is much smaller than a wavelength there is more slosh than power transfer. A half-wave of 41 Hz is like 161 inches or 13 feet (4m). Little 15-inch cones just flap futilely and we call that good.

But so what? A Watt of acoustic power in a domestic room is LOUD; 10 acoustic watts in a theater is loud. Taking 2% as an optimistic number, we need 50W or 500W of electric power. When we could hardly pass 15W through a type 300 vacuum tube, it was a challenge. But even 50 Watts is less than a TENTH of a horsepower. The cheapest e-Car (Nissan Leaf) has a "modest" 147 HP motor (13 more than my 2-ton Thunderbird). We Audio Men have been afraid of high power.

Quote from: PRR on October 20, 2024, 09:20:39 PM

Quote from: fryingpan on October 20, 2024, 09:40:40 AMefficiency of a woofer is around 2%

As long as we play large-chamber music in small rooms, the efficiency number won't go up. If the air-paddle is much smaller than a wavelength there is more slosh than power transfer. A half-wave of 41 Hz is like 161 inches or 13 feet (4m). Little 15-inch cones just flap futilely and we call that good.

But so what? A Watt of acoustic power in a domestic room is LOUD; 10 acoustic watts in a theater is loud. Taking 2% as an optimistic number, we need 50W or 500W of electric power. When we could hardly pass 15W through a type 300 vacuum tube, it was a challenge. But even 50 Watts is less than a TENTH of a horsepower. The cheapest e-Car (Nissan Leaf) has a "modest" 147 HP motor (13 more than my 2-ton Thunderbird). We Audio Men have been afraid of high power.

I would dispute that audio men are afraid of high power.  We are afraid of having our eardrums meet in the centre of our heads.

I used to work with vibration machines that could give a sinusoidal or random noise vibration for testing avionics equipment.  This was usually a voice coil about two feet in diameter, sometimes water-cooled, moving boxes that could weigh up to 20 pounds and using a 5000 watt audio stage with a flat response down to 5 Hz.  One thing I found from this is that my threshold of bass hearing started at 35 Hz.  Anything below that was silent.  Note that this was for sine waves.  Hi Fi people obsess over getting a flat response down to 20 Hz.  Note that a sine wave is not 20 pulses or clicks or other sounds that are not sine waves.  The pure sine wave below 35 Hz is meaningless for me.  Maybe it is different for other people and maybe someone should conduct that experiment.  We already have a lot of information about high-frequency hearing loss.  I have no idea of whether my bass limit is everybody's bass limit.

Quote from: mzy12 on October 20, 2024, 08:55:52 AMThat being said, I do want to get into building studio effects so, to circle back to the original question on why do I want triple zero distortion figures, I guess that counts as an answer

I design studio gear for a living and have met Doug Self a couple of times. I can tell you the one thing customers clamour for *constantly* is 'colour' and 'warmth' and 'character'. Which are all basically ways of saying 'imperfect fidelity'! Especially from Di inputs. They absolutely *love* transformer distortion. Guitars sound really quite boring with perfect fidelity.

Quote from: merlinb on October 21, 2024, 06:56:35 AMGuitars sound really quite boring with perfect fidelity.

    

Quote from: PRR on October 20, 2024, 09:20:39 PM

Quote from: fryingpan on October 20, 2024, 09:40:40 AMefficiency of a woofer is around 2%

As long as we play large-chamber music in small rooms, the efficiency number won't go up. If the air-paddle is much smaller than a wavelength there is more slosh than power transfer. A half-wave of 41 Hz is like 161 inches or 13 feet (4m). Little 15-inch cones just flap futilely and we call that good.

But so what? A Watt of acoustic power in a domestic room is LOUD; 10 acoustic watts in a theater is loud. Taking 2% as an optimistic number, we need 50W or 500W of electric power. When we could hardly pass 15W through a type 300 vacuum tube, it was a challenge. But even 50 Watts is less than a TENTH of a horsepower. The cheapest e-Car (Nissan Leaf) has a "modest" 147 HP motor (13 more than my 2-ton Thunderbird). We Audio Men have been afraid of high power.

If there were some novel way to convert electrical power into acoustic power that were as large as your hand and weighed half a kilo (a pound for you Unitedstatians) and were capable of, dunno, 20% efficiency, a loud band could probably cover a large-ish live club with something they could basically fit in a couple of backpacks.

Audio people are not afraid of power, the speakers are. Your average high-power woofer can only take up to 500W or so (I might be optimistic) before exhibiting significant power compression that makes any further increase in power futile. So, right now, for large enough venues, you either need large horn-loaded cabinets to maximise efficiency or a large number of speakers. Due to logistics, the latter is far more popular (although it is technically inferior on a number of counts, unless it is very well calibrated with lobing and positioning etc.). It still means a whole truck or more of heavy speakers and gear (AFAIK, the pro audio world hasn't yet fully embraced neodymium and class D amps, but they would only make it slightly better).

Quote from: merlinb on October 21, 2024, 06:56:35 AM

Quote from: mzy12 on October 20, 2024, 08:55:52 AMThat being said, I do want to get into building studio effects so, to circle back to the original question on why do I want triple zero distortion figures, I guess that counts as an answer

I design studio gear for a living and have met Doug Self a couple of times. I can tell you the one thing customers clamour for *constantly* is 'colour' and 'warmth' and 'character'. Which are all basically ways of saying 'imperfect fidelity'! Especially from Di inputs. They absolutely *love* transformer distortion. Guitars sound really quite boring with perfect fidelity.

I still maintain that the obsession with transformers, although it may be somewhat justified, is a very inefficient and expensive way to colour sound. Transformers are nice for other reasons, of course, and they can sound nice.

I did a test a few weeks ago. The same bass guitar signal going into my inexpensive passive DI (with a Monacor transformer you can buy for, like, €15, probably much less in bulk) and the Hi-Z input on my Focusrite Scarlett interface.

https://audio.com/adriano-petrosillo/collections/transformer-or-interface

I took the trouble to volume-match the tracks, upload the files in 24-bit .wav and all (not that you need more than 16 bits...) and the website still manages to do something to the tracks, so if you want to listen to them you gotta download the files. 

For what it's worth, the nicest bass guitar tone I've heard was from an Avalon preamp/DI (which should be active).