back biasing - in principle that is it
can't remember where Pa posted it (somewhere in some SIT thread) and schm or two I have with that trick, I'm not authorized to post
but you got it
as I said - I prefer other way of doing it
there is a saying in my neck of wood - roughly translated - who was bitten by snake, he fears even of small lizard (I'm sure there is equivalent one in everyone's neck of wood)
so, someone to whom I'm related had problems with that approach, thus I'm Chicken; besides always preferring to reinvent Da Wheel
can't remember where Pa posted it (somewhere in some SIT thread) and schm or two I have with that trick, I'm not authorized to post
but you got it
as I said - I prefer other way of doing it
there is a saying in my neck of wood - roughly translated - who was bitten by snake, he fears even of small lizard (I'm sure there is equivalent one in everyone's neck of wood)
so, someone to whom I'm related had problems with that approach, thus I'm Chicken; besides always preferring to reinvent Da Wheel
It depends on how you want to look at it. In some ways, it is the same, in others it is not the same. See the two circuits below.
Left circuit: back bias, with -1.4V on the gate and +1V or so on the source for a combined Vgs of, say, -2.4V.
Right circuit: 0V on the gate, with same total Vgs. The device does not care where we put the 0V.
The full 1.4 amps of current will flow through the 1R resistor in both cases and so it should be sized appropriately given the power dissipation. The only real difference is the speaker out, which does not have the resistor in the load. Does it matter sonically? I'm not hearing it. But it is a neat way to generate a negative voltage off the same positive supply.
Left circuit: back bias, with -1.4V on the gate and +1V or so on the source for a combined Vgs of, say, -2.4V.
Right circuit: 0V on the gate, with same total Vgs. The device does not care where we put the 0V.
The full 1.4 amps of current will flow through the 1R resistor in both cases and so it should be sized appropriately given the power dissipation. The only real difference is the speaker out, which does not have the resistor in the load. Does it matter sonically? I'm not hearing it. But it is a neat way to generate a negative voltage off the same positive supply.
In your earlier drawing, the speaker was always connected across the choke only, so there really was no difference. Does not matter if you ask me.
I'd look more into how you drive the gate of the SIT. A proper buffer with a low-impedance output and a solid control on the bias of the SIT gate might be a good thing.
I'd look more into how you drive the gate of the SIT. A proper buffer with a low-impedance output and a solid control on the bias of the SIT gate might be a good thing.
ZM already has a great way of buffering and driving the gate. What is being explored here is a no buffer solution, just because I can and because it is simple and fun. There is no problem driving this amp with a preamp that has some balls. I am getting quite good sound with a setup like that. I also enjoy the simplicity and the astonishing sound that this circuit provides. And I'm starting to like the simple versions better than even my own version in post #1.
Last weekend, I spent some time listening to the version in post #1, the one with the LM regulators and the bias supply. I had gone through so many iterations and short listening sessions that I hadn't really sat down and given #1 a good listen. After spending some time, I found that while the bass was good and all of it was there, some of the fun and musicality was missing and the sound was sticking to the speakers more, instead of the bloom that I had gotten with the earlier simple versions. So, I started trying various things like flipping polarity, changing sources, etc. Ultimately, I concluded that it is the regulators. They just do something to the sound that makes it not likeable.
After switching back to the "Simple #1" solution illustrated in post #18, I got that feeling of sheer musicality, of being there, and the bloom. This particular VFET has a unique ability to unmask bass notes and make them audible amongst everything else. It gets the complex resonances and decay of percussion instruments so right. It also has all of the other goodness that we get from VFETs. Besides sounding nicer, it is much simpler by getting rid of a number of parts and two separate power supplies. What's not to like? So, I am calling it good and going to live with this for a while.
After switching back to the "Simple #1" solution illustrated in post #18, I got that feeling of sheer musicality, of being there, and the bloom. This particular VFET has a unique ability to unmask bass notes and make them audible amongst everything else. It gets the complex resonances and decay of percussion instruments so right. It also has all of the other goodness that we get from VFETs. Besides sounding nicer, it is much simpler by getting rid of a number of parts and two separate power supplies. What's not to like? So, I am calling it good and going to live with this for a while.
Haha... yeah, I tried that as you can see. But it becomes complicated. At a minimum, two extra supplies are needed, plus some form of regulation. I'm sure Papa is watching and sipping wine as he listens to his super simple trimpot adjustable version 
It won't take long to switch out resistors. I do it using the Wago snap connector thingies. Start at 1.0R and see what you get. If you are below 1.4 amps then reduce the resistance, if more than 1.4 amps then increase the resistance. I would go in 0.25 ohm increments. So, if you get 0.47 ohm 3W resistors, you can make parallel-series combinations and get in the range of 0.75, 1.0, 1.25, and 1.5R. That would be plenty. I wouldn't worry too much about matching channels exactly.
I looked up rheostats at Mouser, but they are pricey. I remember getting cheap ones at Radioshack many years ago, guess they don't make cheap ones anymore. Maybe on ebay.
It won't take long to switch out resistors. I do it using the Wago snap connector thingies. Start at 1.0R and see what you get. If you are below 1.4 amps then reduce the resistance, if more than 1.4 amps then increase the resistance. I would go in 0.25 ohm increments. So, if you get 0.47 ohm 3W resistors, you can make parallel-series combinations and get in the range of 0.75, 1.0, 1.25, and 1.5R. That would be plenty. I wouldn't worry too much about matching channels exactly.
I looked up rheostats at Mouser, but they are pricey. I remember getting cheap ones at Radioshack many years ago, guess they don't make cheap ones anymore. Maybe on ebay.
The resistance is in series with the choke impedance, which is pretty sure to be greater than speaker impedance over the intended frequency band. Relative to 1/Gm, the total is "high" and not too important.
Of course some speakers are happy with a touch of un-damping. A switch could select one point or the other.
Very likely. I would exhaust this venue by also experimenting with a simple resistor/zener replacement for the regulator and also various types of batteries.
Looks like ideally we would want a custom wound choke with a correct DCR. And a second grid type choke which ties the gate to ground through a low dcr and solves the gate leak issues.
I like the "as simple as possible" approach. However, if this was my amp, I'd try to get out as much as possible. I couldn't sleep well with using those rare SITs without being sure the the gate bias and gate current are controlled well (note that it's not about the pre-amp, since there is a DC blocking cap in front of the gate!). The first step would be to measure the gate current and input capacity to see what might be going on. If necessary, you can add a buffer to bias and drive the gate (à-la Zen Mod).
Not necessarily. It is a question of overall concept and integration. If you are using tubes for voltage gain an interstage transformer's secondary can be a good way to apply bias. With an opamp front end a preset offset can be used for bias.
Interestingly, no one so far seems to have built a balanced version of the choke loaded follower thus disposing with the output cap. The voltage drop across the chokes is so low that offset should not be a problem even without a servo.
If the SIT starts pulling gate current (DC), this current can't come from the preamp, since the DC blocking cap sits in between. That's why I wrote that the preamp won't make a difference. With the current simple configuratation, the DC gate current will have to flow through the gate bias resistor, offsetting the bias voltage.
I was not referring to the output cap. But since we're at it, I did build a balanced choke loaded SIT amp that does away with the output cap. It's still in breadboarding phase, but the chassis parts are sitting on the bench, waiting for a proper build. See here: https://www.diyaudio.com/community/threads/susy-t-puckfo.379901/
I gave two examples of preamps that require no coupling caps at the gates. If it is not obvious what i meant i could also draw little circuits
Pretty much what i was thinking. But with a PP : PP interstage instead of caps.
Do you find a centre-tapped choke advantageous compared to two separate chokes? Better symmetry?
It's not centre-tapped. It's two separate windings connected with their relative phase opposite from a centre-tapped winding. It's all about SuSy. Take a look at the SuSy T. PuckFo thread and the links in the first post for more on this. Please post any follow-up questions over there in order to not pollute this thread here.
^^I appreciate that, thanks!
These devices are very scarce. I want to use them fully and so if I have four of them, I'd rather make two amps of super high quality. I also think they deserve to showcase their special character without a lot of circuit tricks. They also don't need anything around them (think of a 300B or 2A3) AND they do show you what is around them.
This amp is a very easy way to get your feet wet with VFETs. This is nothing new. See the L'Amp thread all the way back from 2011:
https://www.diyaudio.com/community/threads/lamp-a-simple-sit-amp.201655/
I have built this amp and listened to it for a while. There is no problem with the bias and stability. I sleep very well at night. I did have problems with one device where it ran away at gate resistor of >10k. But it is fully stable at 10k.
I also tried myriad combinations of gate bias. At a minimum, you need:
1. Two supplies (bipolar).
2. Both supplies must be regulated to provide immunity from mains changes.
3. Low impedance drive.
Yes, these are not insurmountable, but they add complexity. In this particular design, I am CHOOSING to not add complexity (also I am nowhere as smart as ZM and don't want to spend another year exploring ideas just to do the same thing as he did differently... I want to hear it in action now!!!). And versions with a buffer already exist, so this is something different.
The preamp comment I made, to which you (mbrennwa) responded, is in relation to the fact that the gate of the VFET has significant capacitance and it wants a low impedance source. As long as you can drive it well (current) and provide low impedance it is happy. It doesn't matter if that comes from the buffer or the preamp. What you are talking about is doing the drive and gate bias in one go, which is a noble goal.
With the simple design, the amp works, bias is stable, it is simple to build, and the sound is almost completely of the VFET and you get to hear its character. Those are the goals here.
These devices are very scarce. I want to use them fully and so if I have four of them, I'd rather make two amps of super high quality. I also think they deserve to showcase their special character without a lot of circuit tricks. They also don't need anything around them (think of a 300B or 2A3) AND they do show you what is around them.
This amp is a very easy way to get your feet wet with VFETs. This is nothing new. See the L'Amp thread all the way back from 2011:
https://www.diyaudio.com/community/threads/lamp-a-simple-sit-amp.201655/
I have built this amp and listened to it for a while. There is no problem with the bias and stability. I sleep very well at night. I did have problems with one device where it ran away at gate resistor of >10k. But it is fully stable at 10k.
I also tried myriad combinations of gate bias. At a minimum, you need:
1. Two supplies (bipolar).
2. Both supplies must be regulated to provide immunity from mains changes.
3. Low impedance drive.
Yes, these are not insurmountable, but they add complexity. In this particular design, I am CHOOSING to not add complexity (also I am nowhere as smart as ZM and don't want to spend another year exploring ideas just to do the same thing as he did differently... I want to hear it in action now!!!). And versions with a buffer already exist, so this is something different.
The preamp comment I made, to which you (mbrennwa) responded, is in relation to the fact that the gate of the VFET has significant capacitance and it wants a low impedance source. As long as you can drive it well (current) and provide low impedance it is happy. It doesn't matter if that comes from the buffer or the preamp. What you are talking about is doing the drive and gate bias in one go, which is a noble goal.
With the simple design, the amp works, bias is stable, it is simple to build, and the sound is almost completely of the VFET and you get to hear its character. Those are the goals here.
