A first question (before thinking further about it):
Why are you using two input signals? Connect the earth at another point (won't make a difference, at least with the Spice simulation) and you need only one source.
This does not yet prove that it works. What is the gain you expect with this? And the noise (as Nelson suspects) ?
I built (and am still enhancing) a MOSFET Circlotron amplifier some years ago. It works fine but uses some more devices... Only later I learned from a tube journal, that they call it Circlotron.
Cheers,
Gerhard Wolf
Why are you using two input signals? Connect the earth at another point (won't make a difference, at least with the Spice simulation) and you need only one source.
This does not yet prove that it works. What is the gain you expect with this? And the noise (as Nelson suspects) ?
I built (and am still enhancing) a MOSFET Circlotron amplifier some years ago. It works fine but uses some more devices... Only later I learned from a tube journal, that they call it Circlotron.
Cheers,
Gerhard Wolf
After having a closer look:
This is definitively a buffer, having gain below 1:
The left side of the load follows the right input, reduced by Vgs2 and V3 noise, the right side follows the left input, reduced by Vgs1 and V5 noise.
So you will also hear the power supply ripple.
Two hands clapping is not enough. You need also your feet (driver transistors) for making enough noise
Gerhard
This is definitively a buffer, having gain below 1:
The left side of the load follows the right input, reduced by Vgs2 and V3 noise, the right side follows the left input, reduced by Vgs1 and V5 noise.
So you will also hear the power supply ripple.
Two hands clapping is not enough. You need also your feet (driver transistors) for making enough noise
Gerhard
Mr. Pass, Gerhard:
Thanks very much for your replies. I was expecting both voltage and current gain from the common source configuration, instead I get unity voltage gain as if it were a common drain stage. I was driving the circuit balanced, but it also works with the negative input grounded.
I'm interested in the circuit because (if I understand correctly) it can operate in class AB beyond its quiescent bias, without more complex biasing circuits, which gives it a sort of Zen appeal. But, without voltage gain, it isn't a very practical project.
I'm not giving up tho. Do I need to offset the gates somehow?
Thanks again.
Mike
Thanks very much for your replies. I was expecting both voltage and current gain from the common source configuration, instead I get unity voltage gain as if it were a common drain stage. I was driving the circuit balanced, but it also works with the negative input grounded.
I'm interested in the circuit because (if I understand correctly) it can operate in class AB beyond its quiescent bias, without more complex biasing circuits, which gives it a sort of Zen appeal. But, without voltage gain, it isn't a very practical project.
I'm not giving up tho. Do I need to offset the gates somehow?
Thanks again.
Mike
mrothacher said:
Whether it's practical depends, I think, on your power requirements and on the voltage output available from your preamp. For example, you can drive a circlotron (or zirclotron, but I'm not sure there's any advantage) output stage from a balanced Zen line stage for up to 100W or so output. Surely this is at least a "Zen compatible" approach. Also, the circuit will work class AB, but I find that it sounds best, using IRF output devices at least, at 1A or more idle current.
Hey Joe!
How have you been? I like your idea. I'm going to finish my balanced Zen linestage and try the zirclotron like a power follower. I'd forgotten that the BOSOZ can swing a lot of voltage. In fact, my first attempt at a two-stage design was really just a zen linestage with a circlotron hung on the outside.
The idea of a 100 Watt "Zen-like" power buffer that idles at 1 A or so is rather appealing after all!
Thanks
Mike
How have you been? I like your idea. I'm going to finish my balanced Zen linestage and try the zirclotron like a power follower. I'd forgotten that the BOSOZ can swing a lot of voltage. In fact, my first attempt at a two-stage design was really just a zen linestage with a circlotron hung on the outside.
The idea of a 100 Watt "Zen-like" power buffer that idles at 1 A or so is rather appealing after all!
Thanks
Mike
mrothacher said:
Hi Mike,
I looked at your circuit, but I'm missing some kind of ground reference for the amp. I mean, the whole circuit sort of floats between the signal sources, which are referred to gnd, but the amp isn't. That may be the reason there is no volt gain. If this works at all it may be because spice often assumes implicit gnd refs in device models.
Any expert views on this?
Jan Didden
Hello Jan!
Wow, I'm really getting some big-time help with this circuit. Thank you all. I'm in the process of wiring one together right now. I've tried this before, but not with the load between the drains and the supplies flipped. Even without voltage gain, the inverting action might prove useful.
I'm not sure about other spice simulations, mine shows only current gain. I think I could throw in some resistors to ground at the zero V points in practice.
I'll let you know what happens!
Mike
Wow, I'm really getting some big-time help with this circuit. Thank you all. I'm in the process of wiring one together right now. I've tried this before, but not with the load between the drains and the supplies flipped. Even without voltage gain, the inverting action might prove useful.
I'm not sure about other spice simulations, mine shows only current gain. I think I could throw in some resistors to ground at the zero V points in practice.
I'll let you know what happens!
Mike
Jan,
there is nothing strange with a floating circuit, having no ground reference for its own. Only Spice needs a ground, for knowing where voltage 0 is.
But there could be a problem if noise is injected through the power supply (inter-winding capacitance) and finds its path to ground only through the input leads. What will happen, depends on your circuit topology.
Cheers,
Gerhard
there is nothing strange with a floating circuit, having no ground reference for its own. Only Spice needs a ground, for knowing where voltage 0 is.
But there could be a problem if noise is injected through the power supply (inter-winding capacitance) and finds its path to ground only through the input leads. What will happen, depends on your circuit topology.
Cheers,
Gerhard
I don't think it will work the way you hope because the feedback resistorsR8 and R9 are connected to AC ground, i.e. they connect to their own fet's source except lifted up by 34 volts.mrothacher said:
You can make into a nice sounding power follower by changing it to two source followers, and running a 100 ohm resistor from each end of the load to ground. I can't see an obvious way to give the cct voltage gain.
Yeah, 1 amp quiescent current would be nice. N-channels and circlotrons were made for each other - don't need any of those girlie-man P-channels!
The value of these resistors to ground is flexible. From a DC perspective, 10K ohms here works as well as 100, and you can then get away with using a 1/4 watt part. However, depending on construction, you may then see a lot of induced power supply noise on the outputs. You can attenuate this noise by reducing the resistance to ground, and this may account for Graham's recommended 100 ohm value.Circlotron said:
Personally, I don't like to waste more than 1% of the available output current in these resistors. So, assuming a nominal 6 ohm load, I might use about 300 ohms to ground on each side, or a total of 600 ohms in shunt with the load. If the noise on the outputs is still higher than I like to see, I look for other ways to reduce it.
Circlotron: On second thought, I think you're right. I don't think the latter circuits accomplish anything special.
Circlotron, Joe: What do the resistors to ground do? Would you bias the source follower version with the usual voltage divider bias?
Is the attached circuit what you have in mind?
Thanks
Mike
Circlotron, Joe: What do the resistors to ground do? Would you bias the source follower version with the usual voltage divider bias?
Is the attached circuit what you have in mind?
Thanks
Mike
powercirc.bmp
R8 and R9 provide a ground reference for each follower on the assumption that each phase of the input signal is also referenced to ground (or to a very clean DC source which is itself ground referenced). The lower the value, the less supply noise you are likely to see on each output to ground.
If you have the parts available, you can raise the values of R1-R4 by an order of magnitude and reduce C1 and C2 by the same factor. This may let you use a better quality cap, and by raising the impedance of the supply-noise source represented by R1 & R3, allows the 470 ohm output impedance of your drive source to better swamp out this noise.
R8 and R9 provide a ground reference for each follower on the assumption that each phase of the input signal is also referenced to ground (or to a very clean DC source which is itself ground referenced). The lower the value, the less supply noise you are likely to see on each output to ground.
If you have the parts available, you can raise the values of R1-R4 by an order of magnitude and reduce C1 and C2 by the same factor. This may let you use a better quality cap, and by raising the impedance of the supply-noise source represented by R1 & R3, allows the 470 ohm output impedance of your drive source to better swamp out this noise.
powercirc.bmp (cont'd)
I just noticed that Nelson Pass included some source resistance for each output MOSFET in the Zen V5 circuit. As your bias scheme also isn't temperature compensated, you may want to do the same to help stabilize the bias level. Even then, unless your output stage power supplies are regulated, you may see more bias variation than you'd like.
Approaching this from the Zen perspective, I hesitate to suggest anything fancier until the need becomes apparent. But if you want to do so later, you could replace the divider networks with a Vgs multiplier type bias source. I've done this several different ways, and all work well in my experience.
In passing, I should note that my earlier statement that R1 and R3 represent the power supply noise source impedance is but a crude approximation, meant to suffice for present purposes only.
I just noticed that Nelson Pass included some source resistance for each output MOSFET in the Zen V5 circuit. As your bias scheme also isn't temperature compensated, you may want to do the same to help stabilize the bias level. Even then, unless your output stage power supplies are regulated, you may see more bias variation than you'd like.
Approaching this from the Zen perspective, I hesitate to suggest anything fancier until the need becomes apparent. But if you want to do so later, you could replace the divider networks with a Vgs multiplier type bias source. I've done this several different ways, and all work well in my experience.
In passing, I should note that my earlier statement that R1 and R3 represent the power supply noise source impedance is but a crude approximation, meant to suffice for present purposes only.
Joe:
Thanks. Your explanations are always clear and helpful. I've wired one together with source resistors, gate stoppers, and all. I think I damaged a mosfet though as only on side biases up correctly. The other side pegs the current meter when I bring up the voltage. I'll give it another try tomorrow.
I think it will work!
Thanks. Your explanations are always clear and helpful. I've wired one together with source resistors, gate stoppers, and all. I think I damaged a mosfet though as only on side biases up correctly. The other side pegs the current meter when I bring up the voltage. I'll give it another try tomorrow.
I think it will work!
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.