Voltage before output capacitor is less then 10mV. Bias through 2sj79 and 2sk216 is 40mA. Sound is not clean, sounding like cheap commercial preamplifier. Signal wire is silver upocc 28awg, resistor in signal path is Vishay Z-foil and tx2575, output capacitor Ampohm with MKP1837 in parallel. Potentiometer is Alps RK27 in shunt with 15k vishay z-foil in series.
https://www.dropbox.com/s/pptq1cpv5fhojiz/BA3_Z_foil.jpg?dl=0
https://www.dropbox.com/s/pptq1cpv5fhojiz/BA3_Z_foil.jpg?dl=0
Hi impuls,
I think you chose wrong parts for this topology. k216/j79 have very low gm (typ. 40mS). BA-3 is essentially a transconductance amplifier - input voltage is converted to output current and that current is converted to voltage by means of R13. So it's essential to use parts with higher gm (k2013/j313 are typ. 700 mS). It's no surprise that k216/j79 sound bad n BA-3. Even irf610/9620 will sound dramatically better here (although they are less linaer devices).
BTW, what's the value of your R13 and how much gain you got in your version of BA-3 ?
I think you chose wrong parts for this topology. k216/j79 have very low gm (typ. 40mS). BA-3 is essentially a transconductance amplifier - input voltage is converted to output current and that current is converted to voltage by means of R13. So it's essential to use parts with higher gm (k2013/j313 are typ. 700 mS). It's no surprise that k216/j79 sound bad n BA-3. Even irf610/9620 will sound dramatically better here (although they are less linaer devices).
BTW, what's the value of your R13 and how much gain you got in your version of BA-3 ?
65R is too much degeneration and R13=170R is too much of a load for k216/j79 here.
Leave the input JFETs (k170/j74) alone, they have nothing to do with problem at hand - take the k216/j79 out and save them for appropriate project (they will shine in LSK-BAF preamp).
k216/j79 are not good for BA-3. Just for a quick experiment, get back to original resistor values and use IRF9610/9620 instead of k216/j79 and see what you get.
EDIT:
we were typing at the same time, good that you have k2013/j313
Yes, change the resistor to original values and don't worry about gain - you can reduce it by changing the Value of R5 (to 220R for example) and/or R6/R7 (to 47R). It will linearize the input stage without messing with the transconductance stage (Q3/Q4). High K2 means unmatched parts and P3 is very effectively dealing with that.
Of course, after changing R5,6,7 you'll have to adjust Q3/Q4 Id to 50mA again.
Of course, after changing R5,6,7 you'll have to adjust Q3/Q4 Id to 50mA again.
Hallo Herr juma! Now I am getting confused 
I thought the gain was given by: "The voltage appearing at the Gate of Q3 is amplified by something less than the ratio of R13 divided by R10, and with the same happening at Q4(R11)..."
And reading through this thread, it seems many have succesfully adjusted gain to suit their needs, simply by reducing R13 (often to around 200r).
Now you are suggesting to adjust R5, R6 and R7 instead. Do you consider this a better method to reduce gain? If so, because?? Or is it just an alternative method for same result?..
I ask for two reasons: 1. I would like to build an BA-3 with variable gain (dual knobs on front panel like Aleph P). 2. I value your opinion, as I have read many of your posts - and you seem to know quite a bit about this stuff
I thought the gain was given by: "The voltage appearing at the Gate of Q3 is amplified by something less than the ratio of R13 divided by R10, and with the same happening at Q4(R11)..."
And reading through this thread, it seems many have succesfully adjusted gain to suit their needs, simply by reducing R13 (often to around 200r).
Now you are suggesting to adjust R5, R6 and R7 instead. Do you consider this a better method to reduce gain? If so, because?? Or is it just an alternative method for same result?..
I ask for two reasons: 1. I would like to build an BA-3 with variable gain (dual knobs on front panel like Aleph P). 2. I value your opinion, as I have read many of your posts - and you seem to know quite a bit about this stuff
Yes it is. But Q3/Q4 form an transconductance stage and those like to "breath" in order to sound good. You can't press it too much and expect wonders from it...
OTOH Q1/Q2 form a regular voltage gain stage (Common Source) and with the values given in original schematic that stage delivers about -6dB gain (with 1V on the gates of JFETs you'll get less than 0.5V on their Drains). So the input stage is practically attenuating the input signal (that was causing the conundrum about BA-3's gain) so that output stage can work optimally. If we further reduce the gain of the input stage we will get lower total gain and more linear input stage without compromising the output stage...
I recently had a major problem adjusting P3 in a F5T. I don't want to repeat it. Will adjusting P3 in BA3-pre affect K2 and K3 or will just work to reduce distortion over all? Will playing with P3 potentially fry the mosfets in the stage after the jfets? Should I watch for bias change and DC off-set?
Thank you for the explanation. I think I understand it now. But... Just for clarification: Does that mean, that it isn't such a good idea to reduce gain by reducing R13 value? Not even a little bit, as many have done?
If so, I guess I will have to drop my idea of pot variable gain.. :\
The best way to think of P3 is a "K2 nulling adjustment". Proper adjustment of P3 will reduce or increase K2.
K3 will also follow and get bigger and smaller as the total distortion changes.
If you reduce K2 as much as possible, K3 is also smaller, and the total is obviously smaller.
Potentially, yes, but it's a pretty small risk.
Yes. But it shouldn't be particularly big.
Place a resistor on the output and measure from that. 10K-47K, Whatever is closest to your amp input. The output needs a load of some sort.
Most important questions - How are you measuring the distortion?
Thank you for the explanation. I think I understand it now. But... Just for clarification: Does that mean, that it isn't such a good idea to reduce gain by reducing R13 value? Not even a little bit, as many have done?
If so, I guess I will have to drop my idea of pot variable gain.. :\
Nothing bad will happen - it's easy to try and see if you like it.
Pot variable gain is possible, just make sure you use part with adequate Pd (2W or so)