swak,
I have no experience with any other TOKIN devices or the Schaded mosfets.
I would tend to think all TOKIN devices would have far more in common than differences. Zen Mod, who has extensive experience with both approaches, says the Schade Mosfet (which i assume is what you are calling a simulated SIT) is different more than less good.
I know I like the idea of the choke loading of TDV and the MOFO amplifiers. I have found, in my case, having the choke by itself in the circuit without any additional resistance makes for a better amplifier. I got lucky in my case - I also am not obsessed with all four amplifiers having the same dissipation.
I am using MEH speakers with efficient drivers but after active EQ much of that sensitivity is erased since no one wants a broad 10 dB rise in the midrange. Whatever the speaker I think having as much power as possible is better than trying to get by with a few watts. As you surmise no crossover components between the amplifier and the speakers - each driver has its own TDV.
Rahul's designs are easy to assemble but do require tweaking of components which has the advantage of one getting to know the amplifier on an intimate basis.
I have no experience with any other TOKIN devices or the Schaded mosfets.
I would tend to think all TOKIN devices would have far more in common than differences. Zen Mod, who has extensive experience with both approaches, says the Schade Mosfet (which i assume is what you are calling a simulated SIT) is different more than less good.
I know I like the idea of the choke loading of TDV and the MOFO amplifiers. I have found, in my case, having the choke by itself in the circuit without any additional resistance makes for a better amplifier. I got lucky in my case - I also am not obsessed with all four amplifiers having the same dissipation.
I am using MEH speakers with efficient drivers but after active EQ much of that sensitivity is erased since no one wants a broad 10 dB rise in the midrange. Whatever the speaker I think having as much power as possible is better than trying to get by with a few watts. As you surmise no crossover components between the amplifier and the speakers - each driver has its own TDV.
Rahul's designs are easy to assemble but do require tweaking of components which has the advantage of one getting to know the amplifier on an intimate basis.
Here's a new schematic with JFETs. This is an LTSpice schematic, not a PCB schematic. So there will likely be revisions to current and voltage setpoints and maybe some component values based on actual devices and parameter spread. I have ordered large quantities of JFETs to test what the spread is, especially for the J111. If you are interested in revising your existing preamp, it can be done. Read on.
The key revisions and rationale are as follows:
1. JFET input and gain device. I am showing a J74 here for the input, but J175 is likely to be the final choice.
2. J111 has a high pinch-off voltage. This has two benefits: first, it allows operation of the input device in its linear range and second, it allows a large swing while staying mostly linear. It can swing 50V with <0.5% THD with mostly second harmonic.
The upshot of all this is that it sounds relaxed with a feeling of headroom.
3. The cap and pot below R2 are gone. This can be implemented in the existing PCBs by taking out the bypass coupling cap and shorting across the cap pads (C102byp).
4. The JFETs are rated at about 400 mW. So, the J111 drain voltage should be limited to about 40V to keep dissipation under control. That means we can bring down the supply voltage to 80V. Okay, this is lower than earlier, but the sound is so much cleaner at higher swings due to the devices and new operating points.
5. The output impedance is somewhat high and so it won't drive really low input impedances well, something like 10k in the next stage is okay. For the PCB, I am working on a JFET buffer.
Why does this sound better? I spent a LOT of time (lot for an amateur, anyway) listening to various device combinations. JFETs are simply more linear in the current range we want compared to other devices. They generate lower amounts of higher order harmonics and provide more headroom (when correctly designed). There are some combination of devices in this design that produce lower THD, but when subjected to large swings, they don't sound as good as JFETs.
Next steps is to test device spread for J111s and J175s and then commit to a PCB. I am not making any promises of when I will be done with the PCB. But I am done with the circuit and device testing. If you have an existing PCB, you can replace the gain and input devices, short C102byp, and adjust setpoints to get most of the way there.
The key revisions and rationale are as follows:
1. JFET input and gain device. I am showing a J74 here for the input, but J175 is likely to be the final choice.
2. J111 has a high pinch-off voltage. This has two benefits: first, it allows operation of the input device in its linear range and second, it allows a large swing while staying mostly linear. It can swing 50V with <0.5% THD with mostly second harmonic.
The upshot of all this is that it sounds relaxed with a feeling of headroom.
3. The cap and pot below R2 are gone. This can be implemented in the existing PCBs by taking out the bypass coupling cap and shorting across the cap pads (C102byp).
4. The JFETs are rated at about 400 mW. So, the J111 drain voltage should be limited to about 40V to keep dissipation under control. That means we can bring down the supply voltage to 80V. Okay, this is lower than earlier, but the sound is so much cleaner at higher swings due to the devices and new operating points.
5. The output impedance is somewhat high and so it won't drive really low input impedances well, something like 10k in the next stage is okay. For the PCB, I am working on a JFET buffer.
Why does this sound better? I spent a LOT of time (lot for an amateur, anyway) listening to various device combinations. JFETs are simply more linear in the current range we want compared to other devices. They generate lower amounts of higher order harmonics and provide more headroom (when correctly designed). There are some combination of devices in this design that produce lower THD, but when subjected to large swings, they don't sound as good as JFETs.
Next steps is to test device spread for J111s and J175s and then commit to a PCB. I am not making any promises of when I will be done with the PCB. But I am done with the circuit and device testing. If you have an existing PCB, you can replace the gain and input devices, short C102byp, and adjust setpoints to get most of the way there.
THF51 can simply do much higher power than the 2SJ28, but if you have horn speakers, then 2SJ28 is fine in terms of power. But hey, what's stopping you from building two amplifiers and listen for yourself?
I have tried that and the sound was... flat. It looks good in the simulation but did not sound better to me. Did you do any listening with it? What were your impressions?
Another option would be to not triodize the gain device, which could be done with JFETs coz they so linear. And then apply feedback from the output to the input to control gain. I tried that but unfortunately it was not stable on my build with wires going long distances. But it could be tested on a PCB with short distances.
A third idea, which I also tried, is to put the LND150 in the input position and use it like a traditional cascode but with the triode top device. LND150 is a gem of a device because it has a very wide Vgs base. It results in not much gain but a wide input capability (the all important headroom). It sounds quite good with the J111 on top and produces >40 db gain (with 10X from the top device). It can swing huge amounts of voltage and you could have the voltage gain for the entire system sorted from this one gain stage. Amazing. But, JFET follower input just sounded better to me in my system where 20 db gain from this stage is enough. Still, this is worth a try and I really want to use LND150 somewhere in the future.
I preferred the resistor loaded for its warmth or the ccs with the feedback taken from the output as mentioned above. The buffered output version per se is not good sounding, but was the driver for a dc coupled, inductor loaded SE amp I’ve never built.
I had a PP tube amp with this configuration for the phase splitter, but I didn’t like it so much.