Need help about amplifier conception
Hi
I need some help,
I want to build the famous szekeres headphone amp, but with another mosfet: the hitachi 2ks216
the datasheet is here: http://www.hitachisemiconductor.com/sic/resource/japan/eng/pdf/transist/e2081241_2sk213.pdf
The simplified diagram is this one:
You can ignore Rg, since no current pass through it, it causes no voltage drop to the gate potential
Forget about DC offsets and things like this, this is a simplified diagram, off course it will be input and output capacitors
Now, I don't know where to start.
That is sure, is that Vcc=12V
After that, I don't know where to start
must I first define a bias current, and according to it, VDS and VGS
or define Vds, and calculate Ibias and Vgs from it
Or begin with setting Vg?
Second question: to be used as an audio amplifier, does a mosfet need to be used in saturated mode, or in non saturated mode?
Help me please
Hi
I need some help,
I want to build the famous szekeres headphone amp, but with another mosfet: the hitachi 2ks216
the datasheet is here: http://www.hitachisemiconductor.com/sic/resource/japan/eng/pdf/transist/e2081241_2sk213.pdf
The simplified diagram is this one:
An externally hosted image should be here but it was not working when we last tested it.
You can ignore Rg, since no current pass through it, it causes no voltage drop to the gate potential
Forget about DC offsets and things like this, this is a simplified diagram, off course it will be input and output capacitors
Now, I don't know where to start.
That is sure, is that Vcc=12V
After that, I don't know where to start
must I first define a bias current, and according to it, VDS and VGS
or define Vds, and calculate Ibias and Vgs from it
Or begin with setting Vg?
Second question: to be used as an audio amplifier, does a mosfet need to be used in saturated mode, or in non saturated mode?
Help me please
Just use a pot for R2 and R3. R4 will determine bias current. You probably want to set the source of the fet at about 6V. So the bias current will be 6/R4 amps. 30 ohms might be a good value for R4. It would need to be a power resistor (2W). You will need a cap for the output. You should download some form of spice and try it out. I have no experience with this circuit.
The mosfet is not saturated. Changing the fet should only require resistor value changes if anything. That is why pots would be good.
Darrell Harmon
The mosfet is not saturated. Changing the fet should only require resistor value changes if anything. That is why pots would be good.
Darrell Harmon
Rg should be 200 ohms or more.
R3 should be about 1.5 times the value of R2 if you don't want
to use a pot.
Good values might be 15K or R2 and 22K for R3.
This assumes a quiet 12 V, otherwise decouple the V+
to R2 with a 1K resistor and a 220 uF or so to ground.
You also need an input cap.
And don't forget that output cap.....
R3 should be about 1.5 times the value of R2 if you don't want
to use a pot.
Good values might be 15K or R2 and 22K for R3.
This assumes a quiet 12 V, otherwise decouple the V+
to R2 with a 1K resistor and a 220 uF or so to ground.
You also need an input cap.
And don't forget that output cap.....
Bricolo all you need to do is build the original and use a Vdd of 12 volts (the term Vcc is used for Bipolars not Fets). You are making this way more complicated than it should be. The basic design is already set. If you want to get complicated add a current source or something but don't reinvent the wheel. If you are using this to learn how to bias Fets and understand it you must get a text book and a calculator. I don't mean to sound harsh and I hope that I don't come off that way but your trying to design something that's already been built many times. Good luck with your headphone driver. Drop me a line if you need help.
G
G
I must applaud your efforts! I am impressed that you have asked so many questions, then worked to come up with better questions. It makes me happy that someone is interested in how thinks actually work. while it is true that you are at the point where a book is a good next step, I encourage you to keep going. Your last few questions have very complicated answers. For your data sheet look at the graph titled "Typical Transfer Characteristics" your bias point should be somewhere on the curve where the line is straight (a bent region will give you distortion). Your bias point should be high enough that when the negative peak of the input signal is added the resulting point is still on the straight region. My preamp's max output is +/- 1V, I would choose a bias point of 2.25V. My portable CD player has much less than this, I would guess about half so if I was going to use it as a source I would choose 1.75V. But, this is not a graph of Vds=12V, so this would be a very good starting point, but the key here is tinkering.
The best value for Rg
Nelson Pass listed these values and comment about a output cap;
Rg should be 200 ohms or more.
R3 should be about 1.5 times the value of R2 if you don't want
to use a pot.
Good values might be 15K or R2 and 22K for R3.
This assumes a quiet 12 V, otherwise decouple the V+
to R2 with a 1K resistor and a 220 uF or so to ground.
You also need an input cap.
And don't forget that output cap.....
End
I disagree with the value specified for Rg by Nelson Pass and the reasoning that this value can be made even higher than 200 ohms.
I acknowledge the fact that the value of Rg in a common collector configuration can be made somewhat higher than in a common source configuration, but that is not the point. The point is that all devices like to be driven from the lowest impedance source possible. This not only reduces distortion in the upper frequency range, but also reduces phase shift caused by the combination of Rg and the various capacitances within the MOSFET to a minimum.
The value of Rg should be slightly greater than the minimum value required to prevent high frequency oscillation. The lower the value the better. In actual circuit configurations and MOSFET characteristics will determine the best value of Rg. There there is no set best value for Rg.
Higher values of Rg are usually safer but may not allow for best performance due to increased upper frequency distortion and phase shift. High frequency roll off is related to phase shift, and thus there will be loss of high frequency response with high vales resistors for Rg. How much of Rg induced degradation can be tolerated will depend on how critical you are.
There is actually considerable amount of signal current flown though Rg in a typical wide band power amplifier. This current ranges from a few micro amps at 1 Khz and then rises rather sharply to the lower milliamp range for some of the larger MOSFETS at 100 Khz. The actual current through Rg will vary with type of devices driven, frequency and circuit configuration.
Your application is a rather simple one, but the principals for keeping the gate resistors as low a value as possible still apply. Making the gate resistor large will have the effect of increasing the input impedance of the circuit at higher frequencies at the expense of other parameters.
John Fassotte
Alaskan Audio
Nelson Pass listed these values and comment about a output cap;
Rg should be 200 ohms or more.
R3 should be about 1.5 times the value of R2 if you don't want
to use a pot.
Good values might be 15K or R2 and 22K for R3.
This assumes a quiet 12 V, otherwise decouple the V+
to R2 with a 1K resistor and a 220 uF or so to ground.
You also need an input cap.
And don't forget that output cap.....
End
I disagree with the value specified for Rg by Nelson Pass and the reasoning that this value can be made even higher than 200 ohms.
I acknowledge the fact that the value of Rg in a common collector configuration can be made somewhat higher than in a common source configuration, but that is not the point. The point is that all devices like to be driven from the lowest impedance source possible. This not only reduces distortion in the upper frequency range, but also reduces phase shift caused by the combination of Rg and the various capacitances within the MOSFET to a minimum.
The value of Rg should be slightly greater than the minimum value required to prevent high frequency oscillation. The lower the value the better. In actual circuit configurations and MOSFET characteristics will determine the best value of Rg. There there is no set best value for Rg.
Higher values of Rg are usually safer but may not allow for best performance due to increased upper frequency distortion and phase shift. High frequency roll off is related to phase shift, and thus there will be loss of high frequency response with high vales resistors for Rg. How much of Rg induced degradation can be tolerated will depend on how critical you are.
There is actually considerable amount of signal current flown though Rg in a typical wide band power amplifier. This current ranges from a few micro amps at 1 Khz and then rises rather sharply to the lower milliamp range for some of the larger MOSFETS at 100 Khz. The actual current through Rg will vary with type of devices driven, frequency and circuit configuration.
Your application is a rather simple one, but the principals for keeping the gate resistors as low a value as possible still apply. Making the gate resistor large will have the effect of increasing the input impedance of the circuit at higher frequencies at the expense of other parameters.
John Fassotte
Alaskan Audio
Thank you all for your answears (whao, Mr Pass himself has replyed )
I finally bought this famous 2SK214 (the only one that my resealer got. 213,214;215 and 216 are the same, except for their Vds breakdown voltage) and some resistors/pots/heatsinks and a lm317.
I started with the diagram I posted, with Rg=220R, R3=220K, R4=22R, R2 is a 1M pot (so Vg can be 2.16-12V)
Vdd is +-12.25V (coming from my computer PS. Yes, I know... but it's only for testing)
I made several measures of Vg (not Vgs, Vg means V(Gate to ground)) and V4 (the voltage across R4)
And came up with this:
This is beatifully linear
You can see that y=0.76x+b
This is annoying me: the voltage gain is 3/4, far from the 0.95 Mr Szekeres wrote :/
I rode some books at my university's library, about mosfets/amplifiers...
After having found complex formulas using physical constants, and atom-related calculations, i found a nice book that explains more usual things
This book told that, to make a mono-mosfet amplifier in optimum cunditions, you should operate this way:
-Use the datasheet's Id/Vds graph
-bound it on the right side with a maximum Vds voltage (let's use Vdd, since the 2SK's max is more than 100V, we are far enough from it)
-bound it on the bottom by a Vgs=x curve, which straight part is still parallel to the other's
-on the top by a Vgs=x straigh at our Vds values
-on the left by the gate saturation (vertical line)
now, we've got a "square" where we want our values to be
the distortion will be small, everywhere in this area
now, we have to draw the line on which our Vds and Id will always be.
Since we use a resistor R4, I will be linearly dependant of Vds
and when I=0, Vds is at it's maximum: Vdd
when Vds=0, V4=Vdd so I=Vdd/R4
in my case, Vdd=12.25, R4=22R so Imax=557mA (note that it is over the mosfet limit, but we don't care, it's out of our square)
we draw a line from (0/557) to (12.25,0)
this is my yellow line (hand made with paint, don't laugh)
Our Vds and Id will vary on this curve
the curve's equation is: a*x+b
a=1/R4
b=Vdd/R4
we'll set a "theorical optimum" (blue cross) bias point on the middle of this line (I mean, the middle of the line that is in the red square)
but we can put it everywhere we want, as long as the input voltage's variations doesn't make Id and Vds vary outside the square.
all measures and graphs can be found here:
http://roulz.free.fr/Divers/mosfet.xls
So far, I think I've understood the basics to finish my 2SK based szekeres
I've got a few IMMPORTANT remarks:
-My Id/Vds/Vgs (values don't match the datasheet. In example, My Id/Vds values match the Vds=3 curve when my Vgs=3.5V)
-R4 gets hotter than a drunk girlfriend (I'd rather say in our case)
Use at least 10W for R4. Mine was 22R 8W, only used 4W max. I don't even want to know how hot it is at 8W
-The mosfet don't get very warm. (I have a heatsink on it)
That's all
If you see an error, PLEASE TELL US
If you don't understand something in my french english explanations, ask me
)I finally bought this famous 2SK214 (the only one that my resealer got. 213,214;215 and 216 are the same, except for their Vds breakdown voltage) and some resistors/pots/heatsinks and a lm317.
I started with the diagram I posted, with Rg=220R, R3=220K, R4=22R, R2 is a 1M pot (so Vg can be 2.16-12V)
Vdd is +-12.25V (coming from my computer PS. Yes, I know... but it's only for testing
)I made several measures of Vg (not Vgs, Vg means V(Gate to ground)) and V4 (the voltage across R4)
And came up with this:
An externally hosted image should be here but it was not working when we last tested it.
This is beatifully linear
You can see that y=0.76x+b
This is annoying me: the voltage gain is 3/4, far from the 0.95 Mr Szekeres wrote :/
I rode some books at my university's library, about mosfets/amplifiers...
After having found complex formulas using physical constants, and atom-related calculations, i found a nice book that explains more usual things
This book told that, to make a mono-mosfet amplifier in optimum cunditions, you should operate this way:
-Use the datasheet's Id/Vds graph
-bound it on the right side with a maximum Vds voltage (let's use Vdd, since the 2SK's max is more than 100V, we are far enough from it)
-bound it on the bottom by a Vgs=x curve, which straight part is still parallel to the other's
-on the top by a Vgs=x straigh at our Vds values
-on the left by the gate saturation (vertical line)
now, we've got a "square" where we want our values to be
the distortion will be small, everywhere in this area
now, we have to draw the line on which our Vds and Id will always be.
Since we use a resistor R4, I will be linearly dependant of Vds
and when I=0, Vds is at it's maximum: Vdd
when Vds=0, V4=Vdd so I=Vdd/R4
in my case, Vdd=12.25, R4=22R so Imax=557mA (note that it is over the mosfet limit, but we don't care, it's out of our square)
we draw a line from (0/557) to (12.25,0)
this is my yellow line (hand made with paint, don't laugh
)Our Vds and Id will vary on this curve
the curve's equation is: a*x+b
a=1/R4
b=Vdd/R4
we'll set a "theorical optimum" (blue cross) bias point on the middle of this line (I mean, the middle of the line that is in the red square)
but we can put it everywhere we want, as long as the input voltage's variations doesn't make Id and Vds vary outside the square.
An externally hosted image should be here but it was not working when we last tested it.
all measures and graphs can be found here:
http://roulz.free.fr/Divers/mosfet.xls
So far, I think I've understood the basics to finish my 2SK based szekeres
I've got a few IMMPORTANT remarks:
-My Id/Vds/Vgs (values don't match the datasheet. In example, My Id/Vds values match the Vds=3 curve when my Vgs=3.5V)
-R4 gets hotter than a drunk girlfriend
(I'd rather say in our case)Use at least 10W for R4. Mine was 22R 8W, only used 4W max. I don't even want to know how hot it is at 8W
-The mosfet don't get very warm. (I have a heatsink on it)
That's all
If you see an error, PLEASE TELL US
If you don't understand something in my french english explanations, ask me
Here is a related question.
It should be should be possible to calculate the time constant of the circuit made by Rg and the gate capacitance. You would need about four times constants for the voltage to go back to the bias point. When doing this calculation would you use the peak to peak voltage? Or would you use the fact that you want a -3dB at some frequency, say 25kHz, and design it like a filter?
Nice work Bricolo
It should be should be possible to calculate the time constant of the circuit made by Rg and the gate capacitance. You would need about four times constants for the voltage to go back to the bias point. When doing this calculation would you use the peak to peak voltage? Or would you use the fact that you want a -3dB at some frequency, say 25kHz, and design it like a filter?
Nice work Bricolo
I can't tell if you figured this part out: The output impedance
of the Source pin is the inverse of the transconductance
of the Mosfet, and from your curve we see the transconductance
to be about .2 amps per volt, for an output impedance of
about 5 ohms. Your bias resistor plus headphone will load
this down, and it would appear that they represent something
like 15 ohms, forming a voltage divider that gives a .75 gain.
of the Source pin is the inverse of the transconductance
of the Mosfet, and from your curve we see the transconductance
to be about .2 amps per volt, for an output impedance of
about 5 ohms. Your bias resistor plus headphone will load
this down, and it would appear that they represent something
like 15 ohms, forming a voltage divider that gives a .75 gain.
I didn't understant everything you told, Nelson
I'll try to find more infos about this transconductance
I didn't take care of the mosfet own resistance.
The yellow line is only calculated by 1/R4
I've just done some mesures with the CCS modified circuit.
I've got some weired results
My V4=f(Vg) graph is a line going up slowly, still Vg=6V
from Vg=6 to 12, it is a line, going up faster
I think the 1st part is when the LM317 hasn't reached 250mA (I set it to 250mA)
The lm317 is not a perfect current source
it's more like a current limiter
that explains why we got 2 part for the V4=f(Vg), a first one when the 317 hasn't reached 250mA, and another one when it limits the current to 250mA
I can't mesure the current to prove this, but I'm quite sure it's that
I'll try to find more infos about this transconductance
I didn't take care of the mosfet own resistance.
The yellow line is only calculated by 1/R4
I've just done some mesures with the CCS modified circuit.
I've got some weired results
My V4=f(Vg) graph is a line going up slowly, still Vg=6V
from Vg=6 to 12, it is a line, going up faster
I think the 1st part is when the LM317 hasn't reached 250mA (I set it to 250mA)
The lm317 is not a perfect current source
it's more like a current limiter
that explains why we got 2 part for the V4=f(Vg), a first one when the 317 hasn't reached 250mA, and another one when it limits the current to 250mA
I can't mesure the current to prove this, but I'm quite sure it's that
general definitions
Reciprocals:
Resistance
Conductance
Reactance Susceptance
Impedance Admittance
Ratios:
V/V voltage gain, no units
V/I resistance or impedance, ohms
I/V conductance or admittance, mhos or siemens
I/I current gain or beta, no units
A trans[ratio] amplifier converts ratio denominator (input) to the ratio numerator (output).
Eg: transconductance of a FET is I(out)/V(in) or Id/Vgs
Reciprocals:
Resistance
ConductanceReactance
SusceptanceImpedance
AdmittanceRatios:
V/V voltage gain, no units
V/I resistance or impedance, ohms
I/V conductance or admittance, mhos or siemens
I/I current gain or beta, no units
A trans[ratio] amplifier converts ratio denominator (input) to the ratio numerator (output).
Eg: transconductance of a FET is I(out)/V(in) or Id/Vgs
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.