Before I discovered VFETs/SITs, I was a vacuum tube addict. Now I have seen the light, but some old memories are starting to resurface. More iron has been creeping into my builds, and my thoughts have been turning to transformers - interstage and output transformers.
I am very happy with my last inductor loaded VFET/SIT builds and I thoroughly enjoy listening to them, but the itch has started again. I am a serial amp builder.
I like trying new things, and luckily, a new idea came to me. I remembered that I have a pair of interstage transformers from my previous life. They were used in a 5842 tube / 45 tube amp. They are Magnequest RIT-5 (5k:5k) gapped for 50H 20mA DC if I remember correctly. I still have these monoblock amps but I haven't used them for years. I can't remember exactly when I built them, but it was 20 or 25 years ago. So, these interstage transformers will be part of my new amp design.
When I was young and into Directly Heated Triodes, I was greatly impressed with Susumu Sakuma and his Direct Heating web site. His designs with transformer coupling, power triodes as drivers, and chokes galore were eye candy to me. Sadly, he has passed away, but his ideas live on. This new VFET / SIT amp was inspired by his designs.
Having decided that I wanted to build a transformer coupled VFET / SIT amp, I researched what has been done before. I found works by Jean Hiraga, Nelson Pass, Michael Rothacher, and others. So, I'm not the first with this idea. However, my idea is a little different in that I will use a source follower output stage, and I haven't found any references to this mode - probably because engineering wise, it is wasteful. But, I'm not constrained by cost (within limits) or practicality so I can live with a voltage amp stage and an interstage transformer which would not be necessary with a common source output.
I ordered a pair of J&K Audio Design's version of the Tango FG-50S output transformer so I'm starting this new build journey. It'll take a month or two to get the transformers but meanwhile, I will build and test some prototypes with choke loaded output.
My first version will be 2SJ28 driving 2SJ28. I also have some 2SK180ES and THF-51S for outputs and some Russian 2P926B for voltage amplifier. The larger SITs will require a bigger driver voltage swing for full power; that will need some experimentation.
A couple of photos below show my breadboarded 5842 IT 45 tube amp monoblock with its LCLC power supply, interstage transformer, and output choke and parafeed transformer. The interstage transformer is shown in the last picture.
I am very happy with my last inductor loaded VFET/SIT builds and I thoroughly enjoy listening to them, but the itch has started again. I am a serial amp builder.
I like trying new things, and luckily, a new idea came to me. I remembered that I have a pair of interstage transformers from my previous life. They were used in a 5842 tube / 45 tube amp. They are Magnequest RIT-5 (5k:5k) gapped for 50H 20mA DC if I remember correctly. I still have these monoblock amps but I haven't used them for years. I can't remember exactly when I built them, but it was 20 or 25 years ago. So, these interstage transformers will be part of my new amp design.
When I was young and into Directly Heated Triodes, I was greatly impressed with Susumu Sakuma and his Direct Heating web site. His designs with transformer coupling, power triodes as drivers, and chokes galore were eye candy to me. Sadly, he has passed away, but his ideas live on. This new VFET / SIT amp was inspired by his designs.
Having decided that I wanted to build a transformer coupled VFET / SIT amp, I researched what has been done before. I found works by Jean Hiraga, Nelson Pass, Michael Rothacher, and others. So, I'm not the first with this idea. However, my idea is a little different in that I will use a source follower output stage, and I haven't found any references to this mode - probably because engineering wise, it is wasteful. But, I'm not constrained by cost (within limits) or practicality so I can live with a voltage amp stage and an interstage transformer which would not be necessary with a common source output.
I ordered a pair of J&K Audio Design's version of the Tango FG-50S output transformer so I'm starting this new build journey. It'll take a month or two to get the transformers but meanwhile, I will build and test some prototypes with choke loaded output.
My first version will be 2SJ28 driving 2SJ28. I also have some 2SK180ES and THF-51S for outputs and some Russian 2P926B for voltage amplifier. The larger SITs will require a bigger driver voltage swing for full power; that will need some experimentation.
A couple of photos below show my breadboarded 5842 IT 45 tube amp monoblock with its LCLC power supply, interstage transformer, and output choke and parafeed transformer. The interstage transformer is shown in the last picture.
Attachments
Hello, brother!
My order of events is a bit different as I discovered SITs in 90s but went for powerful DHT (SV811-10) after that to come back to SITs just recently.
That resulted in a SE SIT and a PP SIT amplifiers built using KP802A. I also designed a SIT phono amplifier in my home stereo with 2SK79 and a choke loaded KP802A. And all of them used transformer inputs, transformer coupling between stages, and (except the phono) of course transformer outputs.
That is why my salutation is well justified, in my view - no coupling caps indeed!
Not mentioning the fact that only two locations in Canada are producing wine using Marechal Foche grapes (my absolute favorite!) which is actually good, and these are local to me Niagara on the Lake in Ontario and Okanagan Valley in BC!!
My order of events is a bit different as I discovered SITs in 90s but went for powerful DHT (SV811-10) after that to come back to SITs just recently.
That resulted in a SE SIT and a PP SIT amplifiers built using KP802A. I also designed a SIT phono amplifier in my home stereo with 2SK79 and a choke loaded KP802A. And all of them used transformer inputs, transformer coupling between stages, and (except the phono) of course transformer outputs.
That is why my salutation is well justified, in my view - no coupling caps indeed!
Not mentioning the fact that only two locations in Canada are producing wine using Marechal Foche grapes (my absolute favorite!) which is actually good, and these are local to me Niagara on the Lake in Ontario and Okanagan Valley in BC!!
Last edited:
Do you mean, taking V- as earth, and adding an extra rail of [book=]%[/book]25 volt V-; then with a p-fet, you will get a Vs 0f e.g -4 volts, and with a variable tap to the source resistor towards -25 then you have enough room to bias the 2SJ28?
And not much attenuation?
Still good handling of the input?
And no caps needed?
And not much attenuation?
Still good handling of the input?
And no caps needed?
To accomplish the goal of "no coupling capacitors", bias voltage source V2 needs to be a circuit with very low output impedance, AND it must be capable of both sourcing and sinking the audio signal current (coupling transformer secondary current) which charges and discharges the gate capacitance of the output VFET.
_
_
Attachments
Oh, now I understand. When you said V2, for some reason I was thinking in vacuum tube terms and thinking about the second VFET in the amp. You mean Version 2. Think of me as the Karate Kid and you as Sensei Myagi. I'm a step behind you.
For Version 2, my thoughts are that I would replace the 2SJ28 source follower with a 2SK182ES follower and adjust Vds, Vgs, and transformer load and go from there. I would use a separate power supply for fixed bias. Now, I will look at the operating voltages and think about what you said.
For Version 2, my thoughts are that I would replace the 2SJ28 source follower with a 2SK182ES follower and adjust Vds, Vgs, and transformer load and go from there. I would use a separate power supply for fixed bias. Now, I will look at the operating voltages and think about what you said.
Can that even work...? I'd assume that the transformer would need to provide increased current values, as the frequency goes up in the audio range? I think something like RIAA/NAB correction-controlled current source may be required...
To see Mark Johnson's suggestion as old practice, it is also in the TOKIN SIT brochure in the application examples.
View attachment 2SK180 2SK182 2SK183 TOKIN_SIT overzicht.pdf
Thought also of the following.
If the earth is as said, add a V+ of say 25 volts.
Then with a strong JFET, that has a high Idss so to say, then you need a large negative bias on the JFET, something like a few volts too. Now you have 8 volts in the sim, but if it is 2-4 volts, having a source follower this way is great too. Connect the source load to the most negative rail.
If you have succees the secret element of Passium will be in the system.
Passium is an element in the periodic system that opens up the fourth dimension.
__________________
View attachment 2SK180 2SK182 2SK183 TOKIN_SIT overzicht.pdf
Thought also of the following.
If the earth is as said, add a V+ of say 25 volts.
Then with a strong JFET, that has a high Idss so to say, then you need a large negative bias on the JFET, something like a few volts too. Now you have 8 volts in the sim, but if it is 2-4 volts, having a source follower this way is great too. Connect the source load to the most negative rail.
If you have succees the secret element of Passium will be in the system.
Passium is an element in the periodic system that opens up the fourth dimension.
__________________
what I was saying, and Mark complemented is - your V2 ( SIT Bias voltage source) "must" have as final element some sort of voltage follower ( be it emiter or source follower), obtaining low output impedance in DC domain
not necessary, but handy to have - in both AC and DC domains
observe how Papa made big Schade mosfet amp , arrangement of xformer secondary and biasing of mosfet
there low impedance is not a must in DC domain ( see 10K ) but for AC domain there is big elco, as low impedance path to GND
all this - simply because SIT gate is somewhat less predictable beastie than mosfet gate
while I was making Singing Bush, I was thinking about these issues - and result is biasing of input JFet buffer, while SIT gate is DC connected to JFet buffer output and - with hefty 20mA of buffer Iq, SIT gate is pretty much taken care of
not necessary, but handy to have - in both AC and DC domains
observe how Papa made big Schade mosfet amp , arrangement of xformer secondary and biasing of mosfet
there low impedance is not a must in DC domain ( see 10K ) but for AC domain there is big elco, as low impedance path to GND
all this - simply because SIT gate is somewhat less predictable beastie than mosfet gate
while I was making Singing Bush, I was thinking about these issues - and result is biasing of input JFet buffer, while SIT gate is DC connected to JFet buffer output and - with hefty 20mA of buffer Iq, SIT gate is pretty much taken care of
Attachments
Last edited:
And joining Zen Mod on that statement that powerful SITs leakage current needs to be treated with a lot of care. I do not like having the DC path from the gate at too low of resistance but after I killed some precious selected SIT pairs in PP due to thermal runaway, I am staying with 3-5KOhm, which seems to be adequate for handling leakage current variations with temperature, etc.
you can just imagine my care, during development of Singing Bush , from moment when I started connecting 2SK77B in circuit
Then that capacitor is in the audio signal path as well as any series cap, isnt it?
How about an op-amp configured as a voltage source? That could sit at 8V and both sink and source current in its effort to keep the node from deviating.