Hi all,
Me again.. Presenting my newest SE speaker amp project..
Its based on my old (2006) and newly modified SE headphone amplifier project;
http://www.diyaudio.com/forums/showthread.php?s=&threadid=130360
However a bit more powered...
In last year I worked some on this project but it was not succesfull. This is the working one..
I use NE5534 as input buffer and VAS stage because of its supply voltage range.. With +/-18v supply I can get 14v RMS swing on NE's output. So I can get (14^2) / 8 ~25W power into 8R load. Under that conditions 14/8 = 1.75A will be sourced into load.
I use a 150W MOSFET and a 5V6 zener as CCS. It sinks 2.25A constant current.
The output device is TIP142... Which has better sound characteristics than MOSFETs while driving with an opamp. If you consider it has not less than 1000x Hfe then in worst case the opamp will source under 3mA.. I think thats enough to keep it under Class A conditions.
And for the frequency compensation, I am using NE's COMP pins.. In the simulation I see ~90Khz (@-3dB) upper band limit..
The final sensivity will be ~650mV for full output and the power dissipation 80W on idle.. I think 0.5°C/W heatsink will be enough per channel.
I will use that amplifier with a 90dB/W fullrange speaker.. I hope it will be enough for my little room.
I am still working on details, but could run a prototype one.. Its too early to say exact things but just its working..
So.. Do you have any advices or critics on the circuit?
Thx in advance..
Me again.. Presenting my newest SE speaker amp project..
Its based on my old (2006) and newly modified SE headphone amplifier project;
http://www.diyaudio.com/forums/showthread.php?s=&threadid=130360
However a bit more powered...
An externally hosted image should be here but it was not working when we last tested it.
In last year I worked some on this project but it was not succesfull. This is the working one..
I use NE5534 as input buffer and VAS stage because of its supply voltage range.. With +/-18v supply I can get 14v RMS swing on NE's output. So I can get (14^2) / 8 ~25W power into 8R load. Under that conditions 14/8 = 1.75A will be sourced into load.
I use a 150W MOSFET and a 5V6 zener as CCS. It sinks 2.25A constant current.
The output device is TIP142... Which has better sound characteristics than MOSFETs while driving with an opamp. If you consider it has not less than 1000x Hfe then in worst case the opamp will source under 3mA.. I think thats enough to keep it under Class A conditions.
And for the frequency compensation, I am using NE's COMP pins.. In the simulation I see ~90Khz (@-3dB) upper band limit..
The final sensivity will be ~650mV for full output and the power dissipation 80W on idle.. I think 0.5°C/W heatsink will be enough per channel.
I will use that amplifier with a 90dB/W fullrange speaker.. I hope it will be enough for my little room.
I am still working on details, but could run a prototype one.. Its too early to say exact things but just its working..
So.. Do you have any advices or critics on the circuit?
Thx in advance..
I havent measured the actual value of bias current yet. But in the simulation it sinks 2.25A really.
C2 and C5 are the current values.. Not calculated exactly, whats your suggestion?
Concerning the RF attenuation capacitor. In real world I didnt hear any Russian Radio station with this configuration
. And in the simulation I cannot see any difference with 47pF on pin 3.. The frequency response was still 90Khz @-3dB.. And I dont want to put there a 120pF.
And about the output power. I can get Vcc - 2.5v to Vee - 2v from NE5534, this makes 31.5v peak. This is what NE5534 datasheet says.
PS: Why Vgs should be 2.6v, while a 5V6 zener is there?
C2 and C5 are the current values.. Not calculated exactly, whats your suggestion?
Concerning the RF attenuation capacitor. In real world I didnt hear any Russian Radio station with this configuration
. And in the simulation I cannot see any difference with 47pF on pin 3.. The frequency response was still 90Khz @-3dB.. And I dont want to put there a 120pF.And about the output power. I can get Vcc - 2.5v to Vee - 2v from NE5534, this makes 31.5v peak. This is what NE5534 datasheet says.
PS: Why Vgs should be 2.6v, while a 5V6 zener is there?
chee whiz, where do I start?
What is the voltage across the 0r75 emitter resistor when passing 2.25A? What is left to drive the FET gate?
What about adding a gate resistor?
RF attenuation, try 330pF to 1n5F on pin3. Adjust to taste.
C5 limits to ~-1dB @ 7Hz.
C2 should allow <1dB @ 4Hz.
I would change C5 to 4u7F, R4 to 20k, C2 to 150uF & Crf to 680pF. This sets the F-3dB passive filter passband to 1.7Hz to 234kHz and F-1dB ~ 3.3Hz to 120kHz.
If you have dual supply rails of 18Vdc and have confirmed that the 5534 can give 14Vpk from these supplies, then it follows that the maximum voltage to the load is 14Vpk not 31.5Vpk.
Vpk*Vpk/Rload/2 = 14 *14 / 8 / 2 = maximum power = 12.3W.
This needs a minimum of 1.75A of bias current as you have already stated.
The bias could be set in the range 1.8A to 2A to achieve this single ended ClassA output into 8r0. It will do no harm to set the bias higher but a sink of 0.5C/W is a bit small and even 1.8A will give rather high Tc values reducing the available SOA to cope with your speaker loading. If you go to 2.25A on +-18Vdc I suggest you use <=0.3C/W sink per channel.
What is the voltage across the 0r75 emitter resistor when passing 2.25A? What is left to drive the FET gate?
What about adding a gate resistor?
RF attenuation, try 330pF to 1n5F on pin3. Adjust to taste.
C5 limits to ~-1dB @ 7Hz.
C2 should allow <1dB @ 4Hz.
I would change C5 to 4u7F, R4 to 20k, C2 to 150uF & Crf to 680pF. This sets the F-3dB passive filter passband to 1.7Hz to 234kHz and F-1dB ~ 3.3Hz to 120kHz.
If you have dual supply rails of 18Vdc and have confirmed that the 5534 can give 14Vpk from these supplies, then it follows that the maximum voltage to the load is 14Vpk not 31.5Vpk.
Vpk*Vpk/Rload/2 = 14 *14 / 8 / 2 = maximum power = 12.3W.
This needs a minimum of 1.75A of bias current as you have already stated.
The bias could be set in the range 1.8A to 2A to achieve this single ended ClassA output into 8r0. It will do no harm to set the bias higher but a sink of 0.5C/W is a bit small and even 1.8A will give rather high Tc values reducing the available SOA to cope with your speaker loading. If you go to 2.25A on +-18Vdc I suggest you use <=0.3C/W sink per channel.
Hmm...
This is bad news for me! I was supposing to have at least 20W amplifier! This is not enough. However I am confused about the RMS output value of opamp..
- If I have 18v rails,
- If the opamp can give at least -2,5v from each rail,
- Then it makes 36-5= 31v peak to peak..
- The RMS value would be 31 / SQRT(2) ~22v under that conditions...
Where am I mistaken?
By the way,
Nico,
Hello my friend.... How are you? A long time passed since last mailing.. I heard that some of my students are mailing with you...
Take good care of you..
And thx again..
This is bad news for me! I was supposing to have at least 20W amplifier! This is not enough. However I am confused about the RMS output value of opamp..
- If I have 18v rails,
- If the opamp can give at least -2,5v from each rail,
- Then it makes 36-5= 31v peak to peak..
- The RMS value would be 31 / SQRT(2) ~22v under that conditions...
Where am I mistaken?
By the way,
Nico,
Hello my friend.... How are you? A long time passed since last mailing.. I heard that some of my students are mailing with you...
Take good care of you..
And thx again..
Dxvideo said:I think I am confused Peak with Peak to Peak...
This project is dead! RIP
Hi Ozgur, I am well thank you.
Peak is half of peak to peak, and rms is peak x 0.707
Andrew correct me if I am wrong.....
If you want 20 watt (rms) into 8 ohm class A, then you would need a supply voltage of 38V plus the Vds drop which can be guessed at 4V. Thus the supply should be +-21V.
The peak current necessary to flow through the 8 ohm speaker is 2.24 amp. (sorry I misplaced the decimal point in the last e-mail) and the resistor value would be 0.39 ohms (as Andrew indicates) assuming that you use the 5V6 zener.
The op-amp will control the bias of the top output transistor in an attempt to zero the output and about 2.24 amps will also flow through the top MOSFET.
Now assuming that your transformer is stable and maintains the +-21V supply across each output transistor, each would have to dissipate about 47watt.
190 watt (stereo) of heating is an awful lot of power especially if you consider your summer ambient temperatures to rise to 40 deg centigrade, you will need substantial heat sinks in order for you not to seriously burn yourself accidentally.
Adding fans is not an option as they are noisy and makes listening at moderate levels of music unpleasant.
Very few people listen to music at 20 watt levels and my suggestion is to have a switched bias, one for low power, say 5 watt and one for high power which would turn on some fans as well when you want to irritate the neighbors.
If you consider the suggestion of 1.5 to 1.7A bias and a supply voltage of around +-15V, this become a lot more manageable but you would end up with about 18 watt peak and more than adequate for most humans.
A final note: the op-amps is biased in class B (most op-amps are) and running only the output stage in class A is not going to give you the sound benefit of class A as the driving signal already contains cross-over distortion.
In order to get the benefit of class A you need to force the op-amp into class A. but this will mean that your amp bias arrangement has to change a bit.
Kind regards
Nico
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.