Hi,
I should propably post this in solidstate forum, but aikido is more known to tube related guys so....
I would like to built aikido headphone amplifier with FETs and with some buffer on the output to drive most variety of headphones.
But I do not have any clue about schematics, what type of FETs to use, what B+ voltage, what are the currents requierements.
Thank you.
I should propably post this in solidstate forum, but aikido is more known to tube related guys so....
I would like to built aikido headphone amplifier with FETs and with some buffer on the output to drive most variety of headphones.
But I do not have any clue about schematics, what type of FETs to use, what B+ voltage, what are the currents requierements.
Thank you.
First off: John Broskie, whose design the Aikido is, has almost certainly touched on this subject once before, and if he has my guess is that he concocted a (probably untested) schematic to go with the idea. I would check his blog.
Secondly, however, I would ask: what's the point? The Aikido exists because it is a flexible circuit that balances the imperfections of tubes against those of other tubes. You get good PSNR and a number of other benefits as bonuses; but if you're going to make a deal with the silicon devil for your headphone amplifier, I imagine that there are better topologies.
By no means let my comments stop you from experimenting. You may come up with an amplifier that does a darn good job. All that said, though, I don't view "taking a good tube circuit and replacing the tubes with transistors" as valid design strategy.
-k
Secondly, however, I would ask: what's the point? The Aikido exists because it is a flexible circuit that balances the imperfections of tubes against those of other tubes. You get good PSNR and a number of other benefits as bonuses; but if you're going to make a deal with the silicon devil for your headphone amplifier, I imagine that there are better topologies.
By no means let my comments stop you from experimenting. You may come up with an amplifier that does a darn good job. All that said, though, I don't view "taking a good tube circuit and replacing the tubes with transistors" as valid design strategy.
-k
Hi,
I want have FET version because of smaller voltages, cost, space. I already have tube aikido in my system. I mainly listen to speakers, but sometime headphones are handy, so I would like to built the headphone aikido topology in simpliest way.
Aikido topology should also work with FETs (or other amplifying device), not just tubes.
I will check tubecad.com for some more info.
I want have FET version because of smaller voltages, cost, space. I already have tube aikido in my system. I mainly listen to speakers, but sometime headphones are handy, so I would like to built the headphone aikido topology in simpliest way.
Aikido topology should also work with FETs (or other amplifying device), not just tubes.
I will check tubecad.com for some more info.
Koolatron is right: in various blogs Broskie already wrote about the inclusion of sand in the aikido, mostly as current output buffer in addition to the standard aikido, but I remember there was also a schematic using sand in the second stage. So the first thing would seem to be to go through the blogs again.
To keep it 'tube like' and simple I would start looking at depletion mode Mosfets.
Erik
To keep it 'tube like' and simple I would start looking at depletion mode Mosfets.
Erik
Hi,
I know this isn´t the Aikido and I will absolutely support the opinion of Erik de Best and the other tube-fellows.
Have you considered the Borbely FET headphone amplifier?. It might not be the simplest thing in the world to make, but Mr. Borbely has a very good reputation indeed.
Have a look at his circuit here:
http://www.borbelyaudio.com/pics/Web309line.pdf
At the Borbely-site you can find prices for his different kits as well.
If you really feel that your headphone-amp must be with FETs I am quite sure that you will get a lot of positive reactions and advice in the solid-state forum.
Good luck
Karsten
I know this isn´t the Aikido and I will absolutely support the opinion of Erik de Best and the other tube-fellows.
Have you considered the Borbely FET headphone amplifier?. It might not be the simplest thing in the world to make, but Mr. Borbely has a very good reputation indeed.
Have a look at his circuit here:
http://www.borbelyaudio.com/pics/Web309line.pdf
At the Borbely-site you can find prices for his different kits as well.
If you really feel that your headphone-amp must be with FETs I am quite sure that you will get a lot of positive reactions and advice in the solid-state forum.
Good luck
Karsten
To fake a Pentode, use depletion FETs. I do
not think you can make a straightforward
solid state Aikido in this way. The "Triode"
region of a starved FET is not sufficiently
like a real Triode for a proper Aikido either.
To make a FET misbehave Triode like, abuse
an enhancement FET. And bias the gate from
voltage divider strapped from drain to source.
This is the basic building block of the FETron.
Now you might be able to get somewhere.
A bit like O.H.Schade (RCA)'s Triode emulator
of 1938. Abusing the 6L6 Pentode with screen
referenced to B+ rather than Triode strapped.
Plate negative feedback interacting with G1
through an external network.
Honestly, I don't see the point. I believe other
circuit topologies than Triode emulation may be
more optimum for a FET. But there is certainly no
reason an Aikido could not be done if you were
merely trying to prove something.
not think you can make a straightforward
solid state Aikido in this way. The "Triode"
region of a starved FET is not sufficiently
like a real Triode for a proper Aikido either.
To make a FET misbehave Triode like, abuse
an enhancement FET. And bias the gate from
voltage divider strapped from drain to source.
This is the basic building block of the FETron.
Now you might be able to get somewhere.
A bit like O.H.Schade (RCA)'s Triode emulator
of 1938. Abusing the 6L6 Pentode with screen
referenced to B+ rather than Triode strapped.
Plate negative feedback interacting with G1
through an external network.
Honestly, I don't see the point. I believe other
circuit topologies than Triode emulation may be
more optimum for a FET. But there is certainly no
reason an Aikido could not be done if you were
merely trying to prove something.
kacernator said:
flg said:
In type of semi conductor N-JFET is similar to Triode.
Most any circuit using triodes could be built with N-JFET.
At suitable lower voltages of couse.
Should not be very diffincult to make one Aikido JFET Clone.
The performance would not be the same. But may not be at all bad.
In fact, it is very probable, it may be real good!
There are N-JFET with current ratings from 1 mA and as high as 100 mA.
Even there are a few Higher Power JFET
Lovoltech LU1014D power JFET
http://www.diyaudio.com/forums/showthread.php?threadid=71633
Lineup
FET JFET whatever. They essentially behave like Pentodes.
And the "Triode Region" is not the same thing as a Triode.
If you sub in depletion FETs (or JFETs), even if you get it
"working", it will not behave the same as an Aikido. Aikido
relies upon the low plate resistance of real Triodes for the
whole noise rejection thing to cancel out properly.
FETs won't do that without additional external feedback.
You have to fake the intrisic Mu feedback of a real Triode
or you can forget about Aikido as a viable FET topology.
I've already drafted a schematic just to prove to myself
that Aikido with FETrons is do-able. But I'm not going to
show it here without a real tube in it somewhere. I may
post to solid later. I am tired right now, its not a priority.
And the "Triode Region" is not the same thing as a Triode.
If you sub in depletion FETs (or JFETs), even if you get it
"working", it will not behave the same as an Aikido. Aikido
relies upon the low plate resistance of real Triodes for the
whole noise rejection thing to cancel out properly.
FETs won't do that without additional external feedback.
You have to fake the intrisic Mu feedback of a real Triode
or you can forget about Aikido as a viable FET topology.
I've already drafted a schematic just to prove to myself
that Aikido with FETrons is do-able. But I'm not going to
show it here without a real tube in it somewhere. I may
post to solid later. I am tired right now, its not a priority.
JFET Aikido with 4 N-JFET and great working!
I setup one circuit in my simulator
Using 4 common N-JFET and 24 VDC.
Most anybody with some knowledge of Spice can do this.
I has expected to see somebody here doing this before myself.
What is Aikido?
- first stage one input device with constant current collector load
- signal to output follower from the collector of input device
- output follower biased by one constant current source
Two N-JFET for first stage. Two N-JFET for second stage.
That's all.
At a voltage gain same as Aikido, +15 dB, and 1 Vrms output
my circuit shows just great performance!
In fact, simulates so great,
I have decided not to 'give away' this circuit here.
Will be one 'goodie' for my dear members at my own place.
Lineup
I setup one circuit in my simulator
Using 4 common N-JFET and 24 VDC.
Most anybody with some knowledge of Spice can do this.
I has expected to see somebody here doing this before myself.
What is Aikido?
- first stage one input device with constant current collector load
- signal to output follower from the collector of input device
- output follower biased by one constant current source
Two N-JFET for first stage. Two N-JFET for second stage.
That's all.
At a voltage gain same as Aikido, +15 dB, and 1 Vrms output
my circuit shows just great performance!
In fact, simulates so great,
I have decided not to 'give away' this circuit here.
Will be one 'goodie' for my dear members at my own place.
Lineup
Re: JFET Aikido with 4 N-JFET and great working!
Firstly: Aikido input stage loading is matched SRPP,
it is intentionally NOT a constant current. You have
already made it into something else, not an Aikido.
Secondly, output source loading is matched SRPP,
it is also intentionally not a constant current.
Third: Input noise is intentionally common to ground,
drive noise common halfway between ground and B+,
output noise cancels out common to ground again.
"Mu follower" of a high impedance device that has no
intrinsic Mu is not an Aikido. You don't even propose
to fake it.
lineup said:
Firstly: Aikido input stage loading is matched SRPP,
it is intentionally NOT a constant current. You have
already made it into something else, not an Aikido.
Secondly, output source loading is matched SRPP,
it is also intentionally not a constant current.
Third: Input noise is intentionally common to ground,
drive noise common halfway between ground and B+,
output noise cancels out common to ground again.
"Mu follower" of a high impedance device that has no
intrinsic Mu is not an Aikido. You don't even propose
to fake it.
"Aikido input stage loading is matched SRPP"
I'm pretty sure JB has made a strong statement that the input stage of the Aikido is NOT SRPP. The distinction does seem to be a fine point however. Only differences I can see is the lack of a cathode Rk bypass cap on the bottom tube and the output coming from the bottom tube plate instead of the top tube cathode. I think the intended Aikido mode of operation is that the top tube has the same impedance as the plate impedance of the bottom tube, so ends up producing roughly Mu/2 (ignoring the cath. Rk degeneration).
If one were to put a pentode on the bottom of the 1st stage with identical gm, looks like one would get roughly Mu gain. So a Mosfet could be used there in pentode mode if the top loading tube were still a strapped triode or just a diode. Although there would be some linearity mis-match due to the top being 3/2 power non-linear and the bottom being roughly 2 power non-linear (unless operated at enough current to get into the Mosfet 3/2 power region).
Seems likely Lineup is using the LU1014D Lovoltech part which could be in its triode like region, but probably has a 2 power gain characteristic still. So a current source would work for the loading device, but would not really be Aikido style and would not divide the PS noise in half either for canceling in the 2nd stage.
If the bottom device is just an ordinary Mosfet (pentode like) with a current source load, one would be seeing extremely high and non-linear gain.
The "Fetron" idea (resistive divider neg. feedback from drain to gate) could provide 2 power triode characteristics for the bottom gain device. (and could be used for the top loading device too for a totally "fake-ode" emulation as Ken said.)
Similarly, the output stage supposedly depends on the non-linear Rp of the bottom current source to help linearize the cathode follower above it (and for the PS noise cancellation scheme). Not sure how well that works though, since the top device is already 100% degenerated in cathode follower mode.
Just to throw in another option, Ken P, Michael K and myself have looked at "anti-triode" schemes that would take the output of the 1st stage at the mid point of the top Rk. But does require a triode like device on the bottom for the gain device. This ideally requires a high gm device on top then to get "accurate" complementary P-P with SE sound character.
Don
I'm pretty sure JB has made a strong statement that the input stage of the Aikido is NOT SRPP. The distinction does seem to be a fine point however. Only differences I can see is the lack of a cathode Rk bypass cap on the bottom tube and the output coming from the bottom tube plate instead of the top tube cathode. I think the intended Aikido mode of operation is that the top tube has the same impedance as the plate impedance of the bottom tube, so ends up producing roughly Mu/2 (ignoring the cath. Rk degeneration).
If one were to put a pentode on the bottom of the 1st stage with identical gm, looks like one would get roughly Mu gain. So a Mosfet could be used there in pentode mode if the top loading tube were still a strapped triode or just a diode. Although there would be some linearity mis-match due to the top being 3/2 power non-linear and the bottom being roughly 2 power non-linear (unless operated at enough current to get into the Mosfet 3/2 power region).
Seems likely Lineup is using the LU1014D Lovoltech part which could be in its triode like region, but probably has a 2 power gain characteristic still. So a current source would work for the loading device, but would not really be Aikido style and would not divide the PS noise in half either for canceling in the 2nd stage.
If the bottom device is just an ordinary Mosfet (pentode like) with a current source load, one would be seeing extremely high and non-linear gain.
The "Fetron" idea (resistive divider neg. feedback from drain to gate) could provide 2 power triode characteristics for the bottom gain device. (and could be used for the top loading device too for a totally "fake-ode" emulation as Ken said.)
Similarly, the output stage supposedly depends on the non-linear Rp of the bottom current source to help linearize the cathode follower above it (and for the PS noise cancellation scheme). Not sure how well that works though, since the top device is already 100% degenerated in cathode follower mode.
Just to throw in another option, Ken P, Michael K and myself have looked at "anti-triode" schemes that would take the output of the 1st stage at the mid point of the top Rk. But does require a triode like device on the bottom for the gain device. This ideally requires a high gm device on top then to get "accurate" complementary P-P with SE sound character.
Don
"Faking this one right, is actually quite tricky."
I was just thinking about the "anti-christ" I mean "anti-triode" Aikido scheme (input stage), and it occured to me that the bottom device could be a high gm part (Mosfet) and the top load device a low gm tube (or diode) and still get SE triode results (output off the center tapped top Rk). This is similar to the "flipped drive" scheme using a P-P xfmr with tube and Mosfet devices but driving the Mosfet instead of the tube. Finally some use for 6AL5s maybe.
Don
I was just thinking about the "anti-christ" I mean "anti-triode" Aikido scheme (input stage), and it occured to me that the bottom device could be a high gm part (Mosfet) and the top load device a low gm tube (or diode) and still get SE triode results (output off the center tapped top Rk). This is similar to the "flipped drive" scheme using a P-P xfmr with tube and Mosfet devices but driving the Mosfet instead of the tube. Finally some use for 6AL5s maybe.
Don
Maybe Lineup is using the Lovoltech parts for both top and bottom devices of the 1st stage and they are operating in their low voltage triode region. Would want matched parts then as SY indicated. Same for the output stage too. Top and bottom just have to match characteristics, even if 2 power law instead of 3/2 power devices.
Don
Don
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.