So, the LPA5 features in more detail:
First the Naim vs Bootstrapped topology.
Look at the air guitariste LPA1 schematic and the LPA3 funk monster. So the redbox LPA5 is designed to switch easily between these topologies.
For Classic Naim topology, on the LPA5 remove R16 and cut track B.
For Bootstrapped topology, insert R16, Remove R29, ensure track B is intact, and cut the track at point A.
Simples, urch.
Obviously I'm not going to go around cutting tracks, so I designed what I termed 'link pads'. Essentially an smt component footprint, with the pads very close together. A blob of solder bridges the pads easily to connect the tracks. A quick dab with desoldering braid and its removed again. Easy peasy. Though ahem, the pad design has to be of a certain shape to work well as I found out. So lets say very easy to middling easy, doh. Ive learnt. The point of this is I wanted to be very clear in my head about the differences between both topologies- i've been using bootstrapped for over 10 years but in truth can't remember the differences that clearly anymore and the games is now at such a different level that I'd like to revisit it. This whole redbox preamp project was really about consolidating my learning and understanding of a lot of the modding and reading I've done in the last few years, so its a development board as well as an astonishingly good preamp, so far.
1) Input transistor Q11 is a jfet not the usual BC550c. Its a very nice little part, the J304, with an Input capacitance of 2.2pf which is VERY low and can handle to 30V. In practice it is a refined, detailed, smooth and 'beautiful' sounding part compared to a BC550c which sounds a bit coarse in comparison. But... I felt the BC part was more ballsy. A tradeoff- do you want to go dancing with a ballerina or one of the fine ladies from Legs & Co? In the grand scheme of things its a trivial difference; I'd rate it as a 1/10 bang for buck, ie just cognisable.
2) BUT the point of Q11=jfet is to allow C12 the input coupling capacitor to drop to 0.22uf instead of 3-10uf without rolling off the freq response early and killing bass response. And THAT means you can try Teflon Vcaps for example.... which are still insane at around £70 each but better than £500 for a 3.3uf part or Duelund stacked copper foils at £1000+. If you want.
At present I'm using an 0.22 FKP1 film and foil from wima which I'm very happy with. Idly looking for an FKP3 0.22, but not found one in a couple of years of browsing.
**** On coupling caps:
I've used tants, oscons, BCwhatevers, blackgate, polyesters, polystyrenes, MKP4s, mkp10s, SMRs etc etc and really don't worry about this much anymore. They all have a slightly different character but the real win is moving up from tants / electrolytics to film. I do avoid wound film capacitors though and stick to Wima/Evox as I've found the wound caps can often have a weird sound, especially the audiophile ones, which I attribute to high inductance or tailoring for audiophile sensibilities. Yuchhh.
In general folks make a big drama about differences in coupling caps but I suspect more because its an easy mod that builds confidence. In the grand scheme of things, when you really start sorting out the power and grounding and other big stuff, its trivial. A polystyrene/foil Multicap RTXs failed to really impress compared to a wima for example. At some point I want to try a Vcap to see if THAT is actually significant but I very much doubt it from all the other caps I've tried.. The cost puts me off so it will be drunken impulse no doubt more fuelled by curiousity than expectation. I feel I have a good sense of the size of the ballpark and it aint that big. And don't diss tants, They're not actually that bad, certainly not a deal breaker ***
Anyway I designed space on the LPA5 board to try anything from a 5mm to a 50mm pitch component in all the standard capacitor pitches e.g 22.5mm, 37.5mm etc. should I ever want to.
3) the input transistor biasing string VR1, R14, R20, R28.
R20 is traditionally 1meg with jfets and this works fine.
R28 is another 50k to form a filter with C15 and trim off any lingering very high frequency noise getting through the regulators. C15 is a small polypropylene wima, ground planed for high frequency performance. The decoupling is right next to the lead of R20- ie the part it is supposed to be decoupling, not 6 inches away or on the regulator board. If you're going to decouple, do it properly.
C8 is another small polypropylene decoupling cap that combines with VR1 as a high frequency filter again.
R14 +VR1 form the top half of the biasing string. Why VR1 at all? Well because then you can trim the output DC offset to zero. And THAT means you can get rid of the output decoupling capacitor C13 all together. (C13 does need to be at least 10uf and big caps tend to perform worse than small uns in terms of most performance parameters. They're also more expensive. I use a polyprop for this again). In practice VR1 needs to be adjusted to around 600k to zero off output dc offset with the J304. Other parts will have other values. Note these values do not work with BC550s where Naims 330k values are appropriate.
R19 is a gatestopper resistor for the jfet just in case. Its almost 100% not needed and can be bypassed, but its there in case until ive confirmed the fact. Easy to bypass, not easy to build in if needed.
4) The input cascode we've already looked at above.
First the Naim vs Bootstrapped topology.
Look at the air guitariste LPA1 schematic and the LPA3 funk monster. So the redbox LPA5 is designed to switch easily between these topologies.
For Classic Naim topology, on the LPA5 remove R16 and cut track B.
For Bootstrapped topology, insert R16, Remove R29, ensure track B is intact, and cut the track at point A.
Simples, urch.
Obviously I'm not going to go around cutting tracks, so I designed what I termed 'link pads'. Essentially an smt component footprint, with the pads very close together. A blob of solder bridges the pads easily to connect the tracks. A quick dab with desoldering braid and its removed again. Easy peasy. Though ahem, the pad design has to be of a certain shape to work well as I found out. So lets say very easy to middling easy, doh. Ive learnt. The point of this is I wanted to be very clear in my head about the differences between both topologies- i've been using bootstrapped for over 10 years but in truth can't remember the differences that clearly anymore and the games is now at such a different level that I'd like to revisit it. This whole redbox preamp project was really about consolidating my learning and understanding of a lot of the modding and reading I've done in the last few years, so its a development board as well as an astonishingly good preamp, so far.
1) Input transistor Q11 is a jfet not the usual BC550c. Its a very nice little part, the J304, with an Input capacitance of 2.2pf which is VERY low and can handle to 30V. In practice it is a refined, detailed, smooth and 'beautiful' sounding part compared to a BC550c which sounds a bit coarse in comparison. But... I felt the BC part was more ballsy. A tradeoff- do you want to go dancing with a ballerina or one of the fine ladies from Legs & Co? In the grand scheme of things its a trivial difference; I'd rate it as a 1/10 bang for buck, ie just cognisable.
2) BUT the point of Q11=jfet is to allow C12 the input coupling capacitor to drop to 0.22uf instead of 3-10uf without rolling off the freq response early and killing bass response. And THAT means you can try Teflon Vcaps for example.... which are still insane at around £70 each but better than £500 for a 3.3uf part or Duelund stacked copper foils at £1000+. If you want.
At present I'm using an 0.22 FKP1 film and foil from wima which I'm very happy with. Idly looking for an FKP3 0.22, but not found one in a couple of years of browsing.
**** On coupling caps:
I've used tants, oscons, BCwhatevers, blackgate, polyesters, polystyrenes, MKP4s, mkp10s, SMRs etc etc and really don't worry about this much anymore. They all have a slightly different character but the real win is moving up from tants / electrolytics to film. I do avoid wound film capacitors though and stick to Wima/Evox as I've found the wound caps can often have a weird sound, especially the audiophile ones, which I attribute to high inductance or tailoring for audiophile sensibilities. Yuchhh.
In general folks make a big drama about differences in coupling caps but I suspect more because its an easy mod that builds confidence. In the grand scheme of things, when you really start sorting out the power and grounding and other big stuff, its trivial. A polystyrene/foil Multicap RTXs failed to really impress compared to a wima for example. At some point I want to try a Vcap to see if THAT is actually significant but I very much doubt it from all the other caps I've tried.. The cost puts me off so it will be drunken impulse no doubt more fuelled by curiousity than expectation. I feel I have a good sense of the size of the ballpark and it aint that big. And don't diss tants, They're not actually that bad, certainly not a deal breaker ***
Anyway I designed space on the LPA5 board to try anything from a 5mm to a 50mm pitch component in all the standard capacitor pitches e.g 22.5mm, 37.5mm etc. should I ever want to.
3) the input transistor biasing string VR1, R14, R20, R28.
R20 is traditionally 1meg with jfets and this works fine.
R28 is another 50k to form a filter with C15 and trim off any lingering very high frequency noise getting through the regulators. C15 is a small polypropylene wima, ground planed for high frequency performance. The decoupling is right next to the lead of R20- ie the part it is supposed to be decoupling, not 6 inches away or on the regulator board. If you're going to decouple, do it properly.
C8 is another small polypropylene decoupling cap that combines with VR1 as a high frequency filter again.
R14 +VR1 form the top half of the biasing string. Why VR1 at all? Well because then you can trim the output DC offset to zero. And THAT means you can get rid of the output decoupling capacitor C13 all together. (C13 does need to be at least 10uf and big caps tend to perform worse than small uns in terms of most performance parameters. They're also more expensive. I use a polyprop for this again). In practice VR1 needs to be adjusted to around 600k to zero off output dc offset with the J304. Other parts will have other values. Note these values do not work with BC550s where Naims 330k values are appropriate.
R19 is a gatestopper resistor for the jfet just in case. Its almost 100% not needed and can be bypassed, but its there in case until ive confirmed the fact. Easy to bypass, not easy to build in if needed.
4) The input cascode we've already looked at above.