Yeah, input and output caps are a notorious subject. Here again, George shows us the way. Somewhere in the diyaudio pages, he had recommended Auricaps for coupling purposes in tube amps. When I was trying out different boutique caps, I took George's suggestion and threw Auricaps in the mix. They are simply more transparent than the others. Every other cap adds its own flavor. But the Auricaps are truly invisible. So, I have stopped worrying about the input and output coupling caps. Currently, the preamp is using 1-2 uF input and output caps. With that, I still get decent bandwidth on the bottom end. Smaller Auricaps are much cheaper 
Agree ZM! I am also lazy and want to get away without adding more parts, including a second supply. I built the neg supply to try but still playing with the single supply variations… so many to try.
SCG Pre -- LSK Special
I have been looking at various datasheets of MOSFETs in the Mouser collection to look at their curves. Here’s a look at the STQ1N60 that I have been trying in the top position.
I have been looking for FETs with more true-pentode-like curves. The FQPs have decent curves. Then I got an idea. Is there a better, current production, solid state small signal pentode than the lsk170? Maybe, but I don’t know about it, and the lsk170, and the Toshiba original, the venerable 2sk170, have really nice curves to begin with. So why not pop it into the SCG configuration and see what we get? The part is not unobtanium, which is one condition for entry, and it is available from our very own diyaudiostore.
Well, it sounds really good. Currently paired with a VP0106, whose source is at about 15 V and the drain of the lsk170 at 45 V. CCS adjusted for 8-9 mA through the FETs. The feedback network draws an additional 3 mA. Here’s a quick spice DC operating point analysis, which is quite spot on. Compared to the others, the LSK Special sounds blisteringly direct. It reminded me of the B1 pre's sound.
I tried measuring it but it was moving all over the place. My layout is on a breadboard and these are fast parts. Wish I could post the full juicy update, but alas the measurements will have to wait. I’ll draw up a schematic and post here.
Feel like this is a good time for me to start developing a PCB in KiCAD. One that accommodates both the TO-220 and TO-92 parts. The basic topology doesn’t really change and with the bias circuits, you can use a wide range of MOSFETS and now a JFET.
I have been looking at various datasheets of MOSFETs in the Mouser collection to look at their curves. Here’s a look at the STQ1N60 that I have been trying in the top position.
I have been looking for FETs with more true-pentode-like curves. The FQPs have decent curves. Then I got an idea. Is there a better, current production, solid state small signal pentode than the lsk170? Maybe, but I don’t know about it, and the lsk170, and the Toshiba original, the venerable 2sk170, have really nice curves to begin with. So why not pop it into the SCG configuration and see what we get? The part is not unobtanium, which is one condition for entry, and it is available from our very own diyaudiostore.
Well, it sounds really good. Currently paired with a VP0106, whose source is at about 15 V and the drain of the lsk170 at 45 V. CCS adjusted for 8-9 mA through the FETs. The feedback network draws an additional 3 mA. Here’s a quick spice DC operating point analysis, which is quite spot on. Compared to the others, the LSK Special sounds blisteringly direct. It reminded me of the B1 pre's sound.
I tried measuring it but it was moving all over the place. My layout is on a breadboard and these are fast parts. Wish I could post the full juicy update, but alas the measurements will have to wait. I’ll draw up a schematic and post here.
Feel like this is a good time for me to start developing a PCB in KiCAD. One that accommodates both the TO-220 and TO-92 parts. The basic topology doesn’t really change and with the bias circuits, you can use a wide range of MOSFETS and now a JFET.
Thanks Geroge! That’s a good one! I also found some other JFETs on Mouser that look pretty good. Will grab that one and some others for testing.
Some of the biasing techniques you posted in the Spice schematics earlier are very interesting. I haven’t tried the depletion mode ones yet but that is another round of fun.
Some of the biasing techniques you posted in the Spice schematics earlier are very interesting. I haven’t tried the depletion mode ones yet but that is another round of fun.
Here's a look at the LSK Special's measurements. All measurements were at 1.6 Vrms output.
1 kHz FFT. Nice spread of 2nd, 3rd, and 4th. Nothing above that. The other channel is a little worse at the same voltage on the drain of the JFET. If I adjust the drain voltage lower, I can get it to match up with this channel, but I wanted at least 25 volts between the drain and source of the JFET.
Distortion versus frequency. Nothing to be said here. Ruler flat.
Same graph but with percent distortion.
Lastly, distortion v amplitude. It is doing 5 Volts RMS at about 0.5% distortion. The ones with bigger FETs did better than this, but not by much. Though, we are limited by how much voltage we can swing by the device limits.
1 kHz FFT. Nice spread of 2nd, 3rd, and 4th. Nothing above that. The other channel is a little worse at the same voltage on the drain of the JFET. If I adjust the drain voltage lower, I can get it to match up with this channel, but I wanted at least 25 volts between the drain and source of the JFET.
Distortion versus frequency. Nothing to be said here. Ruler flat.
Same graph but with percent distortion.
Lastly, distortion v amplitude. It is doing 5 Volts RMS at about 0.5% distortion. The ones with bigger FETs did better than this, but not by much. Though, we are limited by how much voltage we can swing by the device limits.
Oh, one more thing: the instability I encountered earlier was because the gate stopper resistor on the VP0106 was too small. I mistakenly picked up 32R4 instead of 332R. After putting in 332R, the instability was gone as well as the slight buzz that was coming through the speakers.
Note to potential builders: don't undersize the gate stopper.
Note to potential builders: don't undersize the gate stopper.
Hi Everyone,
I've been testing a lot of FETs in the gain device position. It is becoming like tube rolling, and it really is a bit like that because each device seems to have its own sound, even if the distortion numbers and profile is similar.
Here's a list of what I've listened to so far, with the favorites at the top:
STQ1NK60ZR-AP: Wow, this one is stunning. Currently out of stock everywhere, but the 800V version is available at Digikey.
J109
2N3819
STF3LN80
SIHP10N40D
LSK170
FQP19N20
More in the queue.
I've been testing a lot of FETs in the gain device position. It is becoming like tube rolling, and it really is a bit like that because each device seems to have its own sound, even if the distortion numbers and profile is similar.
Here's a list of what I've listened to so far, with the favorites at the top:
STQ1NK60ZR-AP: Wow, this one is stunning. Currently out of stock everywhere, but the 800V version is available at Digikey.
J109
2N3819
STF3LN80
SIHP10N40D
LSK170
FQP19N20
More in the queue.
Yes, I do have it. The crazy thing is that the SCG is competitive with the LuminAria, which sets a very high bar. And the STQ1NK60ZR version is really quite something else. It has that tube-like ease, with extreme transparency and almost no distortion. They claim in the STQ1NK60ZR that it is a high dV/dt device. I don't know if that is the property that sets it apart.
My goal with this project was to see if I can live without uber-expensive and unobtanium devices. The answer is an overwhelming yes. The VFETs are just so special, but there is hope beyond VFETs.
Some other changes to the schematic is a higher impedance Schade network and lower drain voltage. Will post schematic soon.
Have started working up a PCB for it.
My goal with this project was to see if I can live without uber-expensive and unobtanium devices. The answer is an overwhelming yes. The VFETs are just so special, but there is hope beyond VFETs.
Some other changes to the schematic is a higher impedance Schade network and lower drain voltage. Will post schematic soon.
Have started working up a PCB for it.
The SOT-223 SMD version is in stock at Digikey. It's easy to solder wires onto an SOT-223. The 800 volt STN1N80Z may be somewhat different since the data sheet is from 2006 and the capacitance numbers are quite a bit higher. I have a few of both on order.
I have a crude perf board version built, but it has not seen any power yet. I have a small selection of old j-fets, some dating back to the 70's, and I have ordered a few that were mentioned in this thread previously including the LSK189's. Yesterday's post convinced me to order some J109's, STN1N60Z, STN1N80Z from Digikey, but it can take a week to get them here. I have LOTS of miscellaneous mosfets to try as well, again some are quite old.
I will test all at a voltage level suitable for the lower voltage j-fets, and the suitable mosfets at a vacuum tube friendly voltage in the 400 volt range for duty as a driver for a BIG cathode follower once all parts are here.
I have been tinkering with a cascode implementation using a j-fet and a high voltage mosfet for the gain device in LT Spice, but so fat I have not created anything worth breadboarding.