Hi Everybody
I need your help on the following issue.
I have to construct a Class AB - push pull amplifier with MOSFET
Amplifier should be able to provide 21Watt output power (10.5
Watt RMS) at 8 ohm load (i.e speaker) .
The circuit suppose to be able to amplify input signal from 1 Vrms at
a range of 10 Hz to 40 KHz .
Requirements
1) Vdc should not exceed 50 Volts
2) Dc Current not to exceed 2 Amperes.
3) Construction of Amplifier should be with MOSFET , even if we can find
easier ways!
As a matter of fact is is a school assignment , and there is a plan given by teacher to work on it. Ofcourse we can propose our new schematic , if there is any , that will absolutely comply with the above rules.
Your opinion would be reall help for me .
Regards
George Vasiliou
GREECE
I need your help on the following issue.
I have to construct a Class AB - push pull amplifier with MOSFET
Amplifier should be able to provide 21Watt output power (10.5
Watt RMS) at 8 ohm load (i.e speaker) .
The circuit suppose to be able to amplify input signal from 1 Vrms at
a range of 10 Hz to 40 KHz .
Requirements
1) Vdc should not exceed 50 Volts
2) Dc Current not to exceed 2 Amperes.
3) Construction of Amplifier should be with MOSFET , even if we can find
easier ways!
As a matter of fact is is a school assignment , and there is a plan given by teacher to work on it. Ofcourse we can propose our new schematic , if there is any , that will absolutely comply with the above rules.
Your opinion would be reall help for me .
Regards
George Vasiliou
GREECE
In no order:
Q4 will blow up, needs to be rated at peak to peak voltage swing at a minimum. 2N3904 is a 40V part, try a MPSA06 (80V).
Get rid of the 20dB pad on the input. Make R1 1K5, R2 15K
D1N4003 needs to be a fast diode, 1N4934.
R18 needs to be more like 4R7~10R
C6 and C7 will need to be more like 8000µF for a 50hz line, and should be paralleled with about 10µF.
R15 should be more like 2K0, and R14 100R.
C3 should have an RC time constant one octave above C6 or C7 and the load Z, or about 5hz, or 330µF.
C5 will need to be increased, probably to between 22pF~100pF, use a 'scope.
Mount Q3 to track the heat of the outputs.
R3 looks like about 1K.
Q4 will blow up, needs to be rated at peak to peak voltage swing at a minimum. 2N3904 is a 40V part, try a MPSA06 (80V).
Get rid of the 20dB pad on the input. Make R1 1K5, R2 15K
D1N4003 needs to be a fast diode, 1N4934.
R18 needs to be more like 4R7~10R
C6 and C7 will need to be more like 8000µF for a 50hz line, and should be paralleled with about 10µF.
R15 should be more like 2K0, and R14 100R.
C3 should have an RC time constant one octave above C6 or C7 and the load Z, or about 5hz, or 330µF.
C5 will need to be increased, probably to between 22pF~100pF, use a 'scope.
Mount Q3 to track the heat of the outputs.
R3 looks like about 1K.
* Change R1 and R2 like prev post suggests but also change R15 to match R2 and scale R14 appropriately. This will help keep the base resistances of Q1/Q2 matched.
* C1 should have some kind of R with it to filter just frequencies above 100Khz or so (RF stuff). Othewise you'll shunt your entire input signal.
* C3 should probably be a non-polar cap or 2 bipolars back to back.
* What is C4 for? I don't understand bootstrapping so maybe it's a good thing to have. Seems like you can achieve your goals without needing anything like that.
* One idea you might consider, purely up to you, is swapping R9, R10, C4 for a current source. This will boost your open loop gain pretty significantly which is good especially since you seem to have a near unity or very small gain front end.
* There's a lot of oppinions about thermal runaway with MOSFETS. I don't understand a lot of it but if you're planning on your Vgs multiplier following your output devices then you might consider swapping the BJT for a small FET instead. It'll should work the same but will have more similar thermal tracking. On the otherhand you might not need to track your output devices at all.
Good luck, looks like a pretty straight forward uncluttered topology. Let us know how it sounds!!
--
Danny
* C1 should have some kind of R with it to filter just frequencies above 100Khz or so (RF stuff). Othewise you'll shunt your entire input signal.
* C3 should probably be a non-polar cap or 2 bipolars back to back.
* What is C4 for? I don't understand bootstrapping so maybe it's a good thing to have. Seems like you can achieve your goals without needing anything like that.
* One idea you might consider, purely up to you, is swapping R9, R10, C4 for a current source. This will boost your open loop gain pretty significantly which is good especially since you seem to have a near unity or very small gain front end.
* There's a lot of oppinions about thermal runaway with MOSFETS. I don't understand a lot of it but if you're planning on your Vgs multiplier following your output devices then you might consider swapping the BJT for a small FET instead. It'll should work the same but will have more similar thermal tracking. On the otherhand you might not need to track your output devices at all.
Good luck, looks like a pretty straight forward uncluttered topology. Let us know how it sounds!!
--
Danny
Azira, regard.Vbe multiplier
Azira,
I believe FETs usually have a positive temco and BJTs suffer from
a negative one.
Maybe it's OK in this app. it just struck my mind.....
Regards / Mattias
btw seems like a lot of fun in school nowadays.
Azira,
I believe FETs usually have a positive temco and BJTs suffer from
a negative one.
Maybe it's OK in this app. it just struck my mind.....
Regards / Mattias
btw seems like a lot of fun in school nowadays.
Re: Azira, regard.Vbe multiplier
I'm aware of that perception a lot of people have. In general many MOSFET amps have been developed with no temp comp involved but I suspect it's not that simple. There have been quite a few active threads on this subject in the past few months. My take on it is that MOSFETS seem to have a pos temp coeff after a certain current but that current rating can change between different types of FETS. In some cases it's 100mA and in other cases it's 2A. That's a huge disparity that could made a difference in this design.
Add to this, you can also deal with thermal runaway in other ways than just a temp tracker such as larger emitter resistors or sufficient heatsinking.
--
Danny
ekaerin said:
I'm aware of that perception a lot of people have. In general many MOSFET amps have been developed with no temp comp involved but I suspect it's not that simple. There have been quite a few active threads on this subject in the past few months. My take on it is that MOSFETS seem to have a pos temp coeff after a certain current but that current rating can change between different types of FETS. In some cases it's 100mA and in other cases it's 2A. That's a huge disparity that could made a difference in this design.
Add to this, you can also deal with thermal runaway in other ways than just a temp tracker such as larger emitter resistors or sufficient heatsinking.
--
Danny
Danny,
I was referring to, what I thought, was your idea of using a FET
for thermal tracking.
I have never seen that before. (did I get it all wrong (usually))
Regarding output stages I agree with you, some type of MOS FET devices change sign of temco at certain levels of current.
I don't think everybody is aware of that.
Personally I have a feeling that the DC operating point of the devices is critical weather to have positive temco or not.
/ Mattias
I was referring to, what I thought, was your idea of using a FET
for thermal tracking.
I have never seen that before. (did I get it all wrong (usually))
Regarding output stages I agree with you, some type of MOS FET devices change sign of temco at certain levels of current.
I don't think everybody is aware of that.
Personally I have a feeling that the DC operating point of the devices is critical weather to have positive temco or not.
/ Mattias
I misread your post Mattias. Turns out you were just talking about using the FET for a Vgs multiplier. I've never actually tried doing it in a real circuit but here are my thoughts on it.
In terms of creating a "rubber diode" or a device which holds a constant voltage accross it's terminals, a FET or BJT can be hooked up in a similar fashion to do so. The difference being that inherently the FET creates a Vgs drop and the BJT creates a Vbe drop. This I've done many times in simulations.
I assumed that since the small FET was more like the output devices it would track tempurature related effects more similarily to the output devices than a BJT would. If the temperature effects of the output FETs make it self regulating then the FET Vgs multiplier should self regulate and eventually settle at a value that won't drift too much.
On the other hand, if the output FETs turn out to suffer from thermal runaway... then a FET Vgs multiplier might perform better drift tracking to a FET output device.
I suppose that is more of a theory of mine than an actual fact, maybe I shouldn't have suggested it to him 🙂
--
Danny
In terms of creating a "rubber diode" or a device which holds a constant voltage accross it's terminals, a FET or BJT can be hooked up in a similar fashion to do so. The difference being that inherently the FET creates a Vgs drop and the BJT creates a Vbe drop. This I've done many times in simulations.
I assumed that since the small FET was more like the output devices it would track tempurature related effects more similarily to the output devices than a BJT would. If the temperature effects of the output FETs make it self regulating then the FET Vgs multiplier should self regulate and eventually settle at a value that won't drift too much.
On the other hand, if the output FETs turn out to suffer from thermal runaway... then a FET Vgs multiplier might perform better drift tracking to a FET output device.
I suppose that is more of a theory of mine than an actual fact, maybe I shouldn't have suggested it to him 🙂
--
Danny
"* Change R1 and R2 like prev post suggests but also change R15 to match R2 and scale R14 appropriately. This will help keep the base resistances of Q1/Q2 matched. "
That's all fine and good, but don't forget to change C3 if you change R14. With his DC trim pot the impedances don't have to match.
"* C1 should have some kind of R with it to filter just frequencies above 100Khz or so (RF stuff). Othewise you'll shunt your entire input signal."
What do you think R1 is for?
"* What is C4 for? I don't understand bootstrapping so maybe it's a good thing to have. Seems like you can achieve your goals without needing anything like that.
* One idea you might consider, purely up to you, is swapping R9, R10, C4 for a current source. This will boost your open loop gain pretty significantly which is good especially since you seem to have a near unity or very small gain front end."
A bootstrap is a constant current source, that's one of the main reasons its used. The other reason is to get higher drive voltage swing, very important in this kind of circuit with low power supply voltage and an output that needs very high bias voltage.
"* There's a lot of oppinions about thermal runaway with MOSFETS. I don't understand a lot of it but if you're planning on your Vgs multiplier following your output devices then you might consider swapping the BJT for a small FET instead. It'll should work the same but will have more similar thermal tracking. On the otherhand you might not need to track your output devices at all."
I always wonder why people make remarks like this. Pull the data sheet and look. Then there is less chance you will make an inappropriate remark. By accident you almost made a good remark, but it shouldn't be a small FET, it should be of the same type the outputs are.
I'll give you a hint, in addition to the tempco of the RDS, you also must consider the tempco of the Vgs.
Lateral MOS types are the only easy ones to use in a class AB amplifier, they are easy to spot because the Source is connected to the case. Vertical MOS are easiest to use in class A designs, the Drain is connected to the case.
Required Reading:
http://www.passdiy.com/pdf/citation.pdf
That's all fine and good, but don't forget to change C3 if you change R14. With his DC trim pot the impedances don't have to match.
"* C1 should have some kind of R with it to filter just frequencies above 100Khz or so (RF stuff). Othewise you'll shunt your entire input signal."
What do you think R1 is for?
"* What is C4 for? I don't understand bootstrapping so maybe it's a good thing to have. Seems like you can achieve your goals without needing anything like that.
* One idea you might consider, purely up to you, is swapping R9, R10, C4 for a current source. This will boost your open loop gain pretty significantly which is good especially since you seem to have a near unity or very small gain front end."
A bootstrap is a constant current source, that's one of the main reasons its used. The other reason is to get higher drive voltage swing, very important in this kind of circuit with low power supply voltage and an output that needs very high bias voltage.
"* There's a lot of oppinions about thermal runaway with MOSFETS. I don't understand a lot of it but if you're planning on your Vgs multiplier following your output devices then you might consider swapping the BJT for a small FET instead. It'll should work the same but will have more similar thermal tracking. On the otherhand you might not need to track your output devices at all."
I always wonder why people make remarks like this. Pull the data sheet and look. Then there is less chance you will make an inappropriate remark. By accident you almost made a good remark, but it shouldn't be a small FET, it should be of the same type the outputs are.
I'll give you a hint, in addition to the tempco of the RDS, you also must consider the tempco of the Vgs.
Lateral MOS types are the only easy ones to use in a class AB amplifier, they are easy to spot because the Source is connected to the case. Vertical MOS are easiest to use in class A designs, the Drain is connected to the case.
Required Reading:
http://www.passdiy.com/pdf/citation.pdf
"R6 anyone?"
Probably doesn't need to be there, but if you yank it be sure to increase C5.
Here is where the original schematic was ripped from:
http://www.irf.com/technical-info/appnotes/an-948.pdf
Probably doesn't need to be there, but if you yank it be sure to increase C5.
Here is where the original schematic was ripped from:
http://www.irf.com/technical-info/appnotes/an-948.pdf
djk said:
Hmm... after having posted the logic behind my suggesting a FET, what makes it such an "accident"? Not everyone on here agrees what "thermal runaway" is or if MOSFETs suffer from it...
What I am saying is straight forward:
IF you are of the classification that believes FETs suffer from thermal runaway then a FET Vgs multiplier might be better than a BJT one.
On the other hand, iF you are NOT of that classification, then it doesn't matter what you use...
Really, how many BJT amplifiers do you see use the same device for the output device AND Vbe multiplier?
Not the Citation 12, Not the Leach amp, Not P3A
You still don't get it do you?
"IF you are of the classification that believes FETs suffer from thermal runaway then a FET Vgs multiplier might be better than a BJT one.
On the other hand, iF you are NOT of that classification, then it doesn't matter what you use..."
Whether it needs it or not is not determined by anyones' opinion, but by the construction of the part.
The manufacturers data sheet will show the tempco of the RDS and the Vgs.
"Really, how many BJT amplifiers do you see use the same device for the output device AND Vbe multiplier?"
If you use the same part then the Vgs tempco will be the same, what does a TO220 FETcost? Maybe $0.50?
On CFP amplifiers I always use the same exact transistor I use as the driver, diode connected and in contact with the driver.
Look at a Mark Levinson ML3 schematic, that's where I got the idea from.
"IF you are of the classification that believes FETs suffer from thermal runaway then a FET Vgs multiplier might be better than a BJT one.
On the other hand, iF you are NOT of that classification, then it doesn't matter what you use..."
Whether it needs it or not is not determined by anyones' opinion, but by the construction of the part.
The manufacturers data sheet will show the tempco of the RDS and the Vgs.
"Really, how many BJT amplifiers do you see use the same device for the output device AND Vbe multiplier?"
If you use the same part then the Vgs tempco will be the same, what does a TO220 FETcost? Maybe $0.50?
On CFP amplifiers I always use the same exact transistor I use as the driver, diode connected and in contact with the driver.
Look at a Mark Levinson ML3 schematic, that's where I got the idea from.
djk said:
Yes, I do. You are one that doesn't believe MOSFETS can have thermal runaway. Or atleast what you interpret thermal runway
to be.
If that was the case, why bother putting a heatsink on it?
Here's a link that states that power MOSFETS can thermally runaway.
http://www.smpstech.com/tmos0000.htm
Here's a link to a thread I had a while ago in which the participants couldn't agree what thermal runaway WAS.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=25795&highlight=
At the risk of repeating myself:
"Lateral MOS types are the only easy ones to use in a class AB amplifier, they are easy to spot because the Source is connected to the case. Vertical MOS are easiest to use in class A designs, the Drain is connected to the case."
That should make it simple for those who can't find or read the data sheet.
"Lateral MOS types are the only easy ones to use in a class AB amplifier, they are easy to spot because the Source is connected to the case. Vertical MOS are easiest to use in class A designs, the Drain is connected to the case."
That should make it simple for those who can't find or read the data sheet.
MANY THANKS
Dear all
Your replys were a great help for me.
thank you all for you comments and for your help.
I will let you know for the results of my amp as soon as possible.
Regards
George Vasiliou
Dear all
Your replys were a great help for me.
thank you all for you comments and for your help.
I will let you know for the results of my amp as soon as possible.
Regards
George Vasiliou
Azira
Yo are right about this. Contrary to public believe FETs run away.
I have experienced this many times at work when we constantly monitor
the losses in different transistors during evaluation.
The mechanism, however, is different from BJTs but the bottom line are
always exploding transistors.
This almost deserve a new thread I think.
/ Regards / Mattias
Yo are right about this. Contrary to public believe FETs run away.
I have experienced this many times at work when we constantly monitor
the losses in different transistors during evaluation.
The mechanism, however, is different from BJTs but the bottom line are
always exploding transistors.
This almost deserve a new thread I think.
/ Regards / Mattias
- Status
- Not open for further replies.