Quote from: Lino22 on January 19, 2025, 01:50:17 PMWhy do i need twice as much headroom for the negative swing?

Due to R12/R13 equal values..
Source can't go lower than half of negative swing due to 50-50% voltage divider..

e.g.
Considering Source biased at 4.5V (and voltage gain of unity) any input higher than 2.25V should be distorted at negative waveform..

Quote from: ElectricDruid on January 19, 2025, 02:46:30 PMGiven that no-one is really arguing that the choice of output buffer technology is a massive part of the sound of this circuit

Pardon me Sir but I do argue a lot..

Guys would it be a good idea to use a double opamp chip, and stick one buffer where it belongs and the other one after volume so the output impedance stays low no matter what volume level?

Quote from: Lino22 on January 20, 2025, 03:34:53 AMGuys would it be a good idea to use a double opamp chip, and stick one buffer where it belongs and the other one after volume so the output impedance stays low no matter what volume level?

Output impedance is considered 1/g_m or about 140R (according to datasheets) in parallel with R12,R13 & VR3 so lower output impedance should be beneficial only for succeeding effect input impedance lower than 1k5, say...

P.S.
You should be right in case of VR3 lug 1 disconnected..  
(pot wired as series variable resistor..)

Quote from: Lino22 on January 20, 2025, 03:34:53 AMGuys would it be a good idea to use a double opamp chip, and stick one buffer where it belongs and the other one after volume so the output impedance stays low no matter what volume level?

Another voice here in favour of an op amp buffer.  Easy to design and excellent power supply rejection with an output on the order of a few ohms.  Of course, the output would probably have a series resistor to avoid overcurrent if the output is grounded, but that can be a relatively small value like 1K.  In production, each unit would behave the same.

Hi

By bootstrapping I meant nulling the voltage across Rb by feeding one end from emitter but for PSRR I meant adding a lowpass filter in series with Drain or Collector and while these transistor buffers have a number of great specs PSRR is not one of them.

Regarding J113 it is also very typical you can if you place big enough order have transistors selected for say 2mA because you happen to need a stable 2mA current source in a product. What you don't buy somebody else will buy . Then you change the design and all your selected J113 ends up at local supplier.
Specifically for J fets this is very common with selection from transistor manufacturer where they anyway sort groups but also with BJT-s and if you buy enough you can get even your custom part number on your part


It is as I said my experience through the years and I have used a about one thousand all very consistent.
Actually I don't remember exactly what they measured in Idss because I don't use them at that working point but I also have J112 which are much higher current but that could possibly be worked around by allowing higher resting current and higher current draw.
 I use Idss more as an indicator of where working point is going to land and so a piece of information to go on when shopping for transistors but I jump forward to what is really interesting here.
 

However the Idss is not the point the point is that out of the J111 and J112 and J113 the J113 would be close to what OP is looking for which is why I suggested to use this part number as replacement for 2N5485.
In my experience a J113 would in the circuit bias  just fine for the signal it would handle
Of course if only J112 would be available then one can adjust the Rs to as outlined above and also differences in Oz would be negligible.
From the ECA books there are roughly maybe 1000 part numbers of J Fets and of small signal J Fets that could replace 2N5458 at 9V and current less than1mA in the working point there would be maybe about 100 part numbers which shrinks to about 30 excluding the exotic ones
But sometimes those could be found surplus and it used to be possible to get say at least 10 part numbers from an array of dealers and sadly selection at dealers is shrinking.
I mean the 2N5458 is a standard transistor which should be easy to find and then the application is as uncritical as it really gets in terms of getting  it to work. It is with sadness I see selection shrinking to J111 and J112.

Now there is another rabbit hole regarding the signal processing through this buffer.

With sincere love for J Fets

Have fun
Bjorn Juhl
BJF Electronics
Sweden

Quote from: antonis on January 19, 2025, 03:17:42 PM

Quote from: Lino22 on January 19, 2025, 01:50:17 PMWhy do i need twice as much headroom for the negative swing?

Due to R12/R13 equal values..
Source can't go lower than half of negative swing due to 50-50% voltage divider..

e.g.
Considering Source biased at 4.5V (and voltage gain of unity) any input higher than 2.25V should be distorted at negative waveform..

Antonis, i still don't comprehend. Dure to hard clipping the signal will swing something like 650mV up and 650mV down. Why do i need to reserve more room for the negative swing.

Hi there,

That is the rabbit hole

Think of the gate space of a J Fet like one half of a 2nd degree equation and place that in the fourth quadrant of coordinate system and you will see.

Now will the complete peak to peak signal pass through linearly and will it matter if it does or does not and in case it does matter was this intentional to design or not or is it just something that makes some sound more organic than others...........

Bjorn Juhl
BJF Electronics
Sweden

Well said Bjorn but I think OP is still confused..

@Lino22: A Source (or Emitter) follower has limited load-driving capability.. While the transistor can, within limits, source plenty of current (as much as required to the load) its current-sinking ability is limited by its Source (Emitter) resistor R12, which forms a voltage divider with the load (R13//VR3)..

Below is the waveform of a BJT buffer (didn't want to mess-up with JFET parameters spread) marginally biased at input signal amplitude..



P.S.
I should bias it at exactly 700mV to make more distinct the Emitter/Load equal values 50/50 dividing effect but you get the idea..

edit: Actually, current "source/sink" should be more precise terms for a bipolar supply buffer with grounded load..
(which I had in mind to demostrate for more ease..)

I see now, thank you Antonis.

Hi,

Yes actually I only glanced at the schematic  and good point why is R13 there at all?
Total load the J Fet buffer sees is then about 9K.
A very hands on method here you could increase current through the transistor by lowering the value of R12 to 4,7K or even 2,2K and drive capability will be higher.


Now without going down the rabbit hole and very hands on if voltage at Source is between 2V-5V all is fine
If you get a J fet that rests at lower voltage you can use a a 10M Ohm resistor from Gate to Drain and that will shift resting point at Source towards mid point. If you get J201 then you really need a voltage divider at the Gate and very hands on again a 1M Ohm to Drain will do the job just fine,

Also because the Power Supply Rejection Rate is very low for Q5 it would be better for Q 5 Drain to be returned to pin 7 of  U1
Actually the PSRR of the OP amp is very high so it doesn't need filtering but Q5 needs all filtering it can get

Have fun
Bjorn Juhl
BJF Electronics
Sweden

So you would put the DC filter section close to the buffer collector, right?

Quote from: Lino22 on January 23, 2025, 12:27:52 PMSo you would put the DC filter section close to the buffer collector, right?

Better make Q5 Drain own supply filter..
A copy of R1/C1 should be OK..

You can even make R1 bigger, according to affordable voltage drop in relation with Source bias level..

That's interesting. I reverse ingeneered an RF digital board and noticed that multiple DC filters were hanging on Dc rails at each node to a semiconductor component.
What are the rules for this for my curiosity?