LSK pre - BAF 2013

What appealed to me about this circuit is it's elegance, scalability (wide range of gain and output level settings, all without feedback loop), wide usability range and, most importantly, extraordinary good sound (I'd describe it as perfectly clear and detailed, strongly vivid, neutral and transparent but emotional and pleasant in presentation) so I've been playing with different versions of this preamp for the last two weeks and I think now I can share my impressions. I toyed mostly with:
  • CCS instead of resistor to supply the current into cascoding node
  • different input devices (j74/k170 V and BL grade, k246/j103 GR and BL grade),
  • different PS and cascode voltages,
  • different gain levels and different gain distribution between input and cascode,
  • different currents through cascode (for the input I always chose most linear region close to Idss),
  • different cascode devices (j74/k170, j103/k246, j313/k2013, BC550c/BC560c).

Since I need a high gain and high output (for low level sources and for driving a small/medium power followers), I opted for gain of 19dB (9V/V) and for about 12V_RMS of output.
The schematics shows the circuit I like the most - j74V/k170V (I managed to obtain 2 pairs) as input devices and BJTs as cascodes, followed by a very nice SE, cascoded, BF862 buffer stage that can happily swing more than 20V_peak into 10k load. Buffer enables a use of low gm JFETs at input which leads to high Zout (k246/j103 sounded really great too, I think I chose j74/k170 over them only because it would be a shame to left them unused after all the trouble I went through to obtain them).
Thermal drift is small - few mV in the preamp (which can be improved by thermal coupling of transistors) and less than 1mV in the buffer, but instead I added a nice small coupling cap between them (WIMA MKP10, 100nF) just to make sure that naughty source component won't make a mess.
R14 and C3 snubber I put for hygiene reasons - a lot of experimenting went on here and I wanted to limit a preamp's band a bit - with a snubber it's good up to almost 200kHz.
PSU is self-explanatory - good thing here is that circuit has a constant current draw so the three legged regs can be used with RC filter after them which solves their greatest problem, the HF noise.
I had the suitable case handy (with connectors, selector and attenuator already done) so I used it although it's very small (22x5x13 cm). That's why it looks messy - the wiring scheme had to compensate for the lack of space.

This post starts to look like blog so I'll stop now - feel free to ask things I forgot to write about...
Hi Juma,

Could you please help understand the point "- different gain levels and different gain distribution between input and cascode,"
what is mean by "gain distribution between input and cascode"
how do we get same gain with j74/k170 JFETs having high trans-conductance and with j103/k246 JFETs having low trans-conductance? should high transconductance result in higher gain , as I think , in cascode amplifier total gain would be the gain provided by input device in common source/emitter stage, and cascode devices only providing shielding the input devices and reducing its capacitance effects ?
 
Whatever matches with the population of J175s I have floating around. I have about 1k Siliconix J175s, and around 500 Fairchild devices, which probably aren't as consistent as the Siliconix fets. I'm still determined to eventually try and build a version of your original circuit using depletion mode current source biasing for the cascodes and a Darlington output buffer. The sims I did in the past for that particular variation looked promising, This is assuming I can get some complementary jfets that will play well together. Part of my frustration back then was finding less exalted jfets that would match reasonably well, as all my J113s were on the low side of their IDSS range, and all the J175s I had around were on the hot side. I may have better luck with J112 nd J175. The J112s came as a mixed batch of parts form Jameco, so I likely have several part vendors to play with.
 
  • Like
Reactions: 1 user
...please help understand ...
Transconductance (gm) is intrinsic property of given FET and you use it according to your design goal(s). Of course the higher gm parts can provide higher gain but the gain value we choose in our designs uses only part of it (degeneration is a form of reflective negative feedback that limits the gain of the active part we decide to use).
In this circuit it's like this:
Input pair's gain mostly depends on chosen level of degeneration (source resistors).
Second pair's gain/output is mostly determined by the value of I/V resistor's value at the output.
The ratio of those two chosen values sets the gain of our circuit and that's where the game is played - more degeneration means less gain but better linearity (but subjectively better sound is not guaranteed). Higher I/V resitance means higher output voltage but with it we get the higher output impedance. Linearity depends on the parts properties (transfer curve, loadline etc...).
So, how the parts work and their interaction is what one needs to understand in order to achieve design goals.

Please, before you ask more questions, try to carefully read and understand this:
https://audioxpress.com/article/JFETs-The-New-Frontier-Part-1
https://audioxpress.com/article/JFETs-The-New-Frontier-Part-2
 
Last edited:
  • Like
Reactions: 1 users