Thank you for the praise Mighty One! 
Someone asked me about BOM - The only change I made was to use SMD equivalents for the BJTs - namely FJV1845 and FJV992 and in the case of the 1-layer board, I swapped to an smd led as well. I have updated the BOM I made for my own use so be warned, use at your own peril!
All resistors are 805 footprints, even though there are 1206 footprints in some places (for track routing). 805 fits just fine between the pads or order 1206 resistors for those (R1 and R9) if it helps you sleep better. I like Murata caps but there are cheaper options. Just note that you are looking for ceramic caps and at least in the case of C2 and C3, they are power supply decoupling caps so MLCC if I understand correctly. I have no idea what C1 and C4 are doing but I just used MLCC for those as well.
And while I'm disclaiming things, I used GR jfets (2SK209). I don't know if GR or Y grade is preferred or if it's not of concern. Like I said earlier, you've been warned! Now go have fun
Cheers,
Stephen
Someone asked me about BOM - The only change I made was to use SMD equivalents for the BJTs - namely FJV1845 and FJV992 and in the case of the 1-layer board, I swapped to an smd led as well. I have updated the BOM I made for my own use so be warned, use at your own peril!
All resistors are 805 footprints, even though there are 1206 footprints in some places (for track routing). 805 fits just fine between the pads or order 1206 resistors for those (R1 and R9) if it helps you sleep better. I like Murata caps but there are cheaper options. Just note that you are looking for ceramic caps and at least in the case of C2 and C3, they are power supply decoupling caps so MLCC if I understand correctly. I have no idea what C1 and C4 are doing but I just used MLCC for those as well.
And while I'm disclaiming things, I used GR jfets (2SK209). I don't know if GR or Y grade is preferred or if it's not of concern. Like I said earlier, you've been warned! Now go have fun
Cheers,
Stephen
Last edited:
As far as I know, the BA2018 kit sold on the DIYAudio Store are also the GR bin. These bin designations just indicate a binning based on Idss (I = current, dss = drain-source, second s is for funsies). The Y bin is the lowest max current bin, the GR is the middle one, and BL is the highest one. As far as I'm aware, all the BL ones have already sold out because they are the rarest of them all, followed by Y and then GR. GR is the middle ground, and that's in line with the yield of the silicon production process.
You want the GR.
C1 (and probably C4 too then?) are, to my knowledge, there to shunt high frequency content so it doesn't pass through the resistor. Since the 27K R16 resistor combined with the 10K R17 resistor form a voltage divider, this basically increases the intensity of high frequency signals by effectively significantly reducing the apparent value of R16 at those frequencies.
Since the Q2 JFET is in a differential pair with the Q1 JFET, this forms a negative feedback path where the significantly higher intensity of the high frequency signals on Q2 (JFET) simultaneously reduces the current in Q1 (JFET). It is negative feedback because it, for the lack of better words, steals current from Q1 given they can only draw so much current at once combined.
I do have to acknowledge my terminology used here is focussed on what I'm used to, which is BJT based long tail pairs, aka differential input stage pairs.
You want the GR.
C1 (and probably C4 too then?) are, to my knowledge, there to shunt high frequency content so it doesn't pass through the resistor. Since the 27K R16 resistor combined with the 10K R17 resistor form a voltage divider, this basically increases the intensity of high frequency signals by effectively significantly reducing the apparent value of R16 at those frequencies.
Since the Q2 JFET is in a differential pair with the Q1 JFET, this forms a negative feedback path where the significantly higher intensity of the high frequency signals on Q2 (JFET) simultaneously reduces the current in Q1 (JFET). It is negative feedback because it, for the lack of better words, steals current from Q1 given they can only draw so much current at once combined.
I do have to acknowledge my terminology used here is focussed on what I'm used to, which is BJT based long tail pairs, aka differential input stage pairs.
Last edited:
Thank you HumbleDeer for the clear and detailed explanation!
If C1and C4 are low pass filters, my googling says to use C0G capacitors for those as well
If C1and C4 are low pass filters, my googling says to use C0G capacitors for those as well
They aren't low pass filters; they're actually acting as high pass filters. But they're acting as a high pass filter that bypasses an otherwise all-attenuating resistor. And that signal feeds into an amplifying circuit that actually reduces the output.
So the high pass filter allows full high frequency stuff, while leaving low frequency stuff attenuated, and then feeds that to the "backwards-amplifier" that attenuates everything that's boosted.
Mind you, high frequency here refers to well above the audible range.
I'd recommend you go for something else than just ceramic capacitors. The microphonics and DC-bias induced degradation of specifications really aren't desirable. For your MLCC's on the PSU decoupling: double or triple up the value and double or quadruple the voltage tolerance. Both are severely affected.
Perhaps it's not a bad idea to keep some surface mount pads & space reserved in case you need to surface-tack on a film capacitor instead? I don't know exactly what the effects will be.
So the high pass filter allows full high frequency stuff, while leaving low frequency stuff attenuated, and then feeds that to the "backwards-amplifier" that attenuates everything that's boosted.
Mind you, high frequency here refers to well above the audible range.
I'd recommend you go for something else than just ceramic capacitors. The microphonics and DC-bias induced degradation of specifications really aren't desirable. For your MLCC's on the PSU decoupling: double or triple up the value and double or quadruple the voltage tolerance. Both are severely affected.
Perhaps it's not a bad idea to keep some surface mount pads & space reserved in case you need to surface-tack on a film capacitor instead? I don't know exactly what the effects will be.
Last edited: