okay, thanks.
Your amp is a little bit of that and a little bit of this.
But although not much is revolutionary new in details
.... this is one unique amplifier >>>> The Mooly's Musical Blend
The first thing that comes into my mind,
if looking only at your input transistor and feedback arrangement,
is Rod Elliott Project 36 from 1999:
Death of Zen (DoZ) - A New Class-A Power Amp
http://sound.westhost.com/project36.htm
I would recommend any body interested in
How to Develop one Power Amplifier along with testing
to Read all this ESP Project article.
Last updated as late as 2005 .. 6 years after first version.
Your amp is a little bit of that and a little bit of this.
But although not much is revolutionary new in details
.... this is one unique amplifier >>>> The Mooly's Musical Blend
The first thing that comes into my mind,
if looking only at your input transistor and feedback arrangement,
is Rod Elliott Project 36 from 1999:
Death of Zen (DoZ) - A New Class-A Power Amp
http://sound.westhost.com/project36.htm
I would recommend any body interested in
How to Develop one Power Amplifier along with testing
to Read all this ESP Project article.
Last updated as late as 2005 .. 6 years after first version.
The special blend of our amplifier designs
right, moolie
i have said exactly same as you just did
in one post some year back
there is nothing really new in one
Lineup Audio LAB, John Curl, Nelson Pass or Mooly amp
it has all been tried / patented in some paper already .. and many years ago
what is different is the combination .. the special blend
of each designer
this is what our creativity & findings give us
a new blend
and hopefully a good working & good hi-fi sounding mixture
..... Lineup Audio special flavour amplifiers Designs .....
right, moolie
i have said exactly same as you just did
in one post some year back
there is nothing really new in one
Lineup Audio LAB, John Curl, Nelson Pass or Mooly amp
it has all been tried / patented in some paper already .. and many years ago
what is different is the combination .. the special blend
of each designer
this is what our creativity & findings give us
a new blend
and hopefully a good working & good hi-fi sounding mixture
..... Lineup Audio special flavour amplifiers Designs .....
I think we might have another satisfied customer
http://www.diyaudio.com/forums/showthread.php?postid=1649849#post1649849
http://www.diyaudio.com/forums/showthread.php?postid=1649849#post1649849
I'm starting to build one. Wondering what will be the difference if the output resistors (0R22) are put on the supply rail (or source pin of the mosfets).
Also, I think I haven't read about tweaking the small transistors' output/emitter resistors (R19/R20/220R) instead of tweaking the R8.
Also, it will be interesting to have the input to become pure symmetrical single ended with another transistor (the n-channel 2N5551).
Also, I think I haven't read about tweaking the small transistors' output/emitter resistors (R19/R20/220R) instead of tweaking the R8.
Also, it will be interesting to have the input to become pure symmetrical single ended with another transistor (the n-channel 2N5551).
Hi Mooly, yes of course I will build it as is. If not how can I compare any modification result? As usual I will build one channel or two channels with slight difference, for example amount of supply rail capacitance, cap type/brand, bridge type/brand, basically different implementation but still the same circuit. I will use the best channel for further comparisons.
I have almost a thousand of 2N5401/2N5551 I bought when I built the Stochino. Many of them have a very high hfe. I will put the highest hfe on the output drivers (Q6/Q7).
I have seen your post regarding splitting the front-end supply, and this driver is grouped with the output mosfet. I think the key to many mosfet amp is the driver itself. That's why I prefer the output mosfet is taking it's input from the emitter of the driver.
Long time ago, with the main objective to take the output from the drain of the mosfet (drain follower) based on the assumption that the capacitance will be minimized, I tried to find a configuration that also enable proper drive of the mosfet. For the drive I decided it has to be an emitter follower. This will affect the number of gain stage. The only usable topology (cannot be more complex) was the one similar to the simplified SKA posted few months later. That's why I didn't understand when it was said that the SKA was something new??? May be the detailed circuit.
Since then I temporarily "concluded" that if I want to build a drain follower from the lateral mosfet, I have to build something with the SKA topology.
But the way you appreciate the musicality of your amp and also the way you do the tuning made me want to build this amp. Even more when Hugh encouraged to do so. This is because I believe what djk has posted:
"I have found 90% of people cannot hear in the audiophile sense, and of the 10% that can, 90% of those pick the worst unmusical sounding gear. The 1% that could tell whether something was musical never seemed to have any money. I would describe the APT as uncolored, but lacking in dynamics, lifeless."
And also Carlos comment:
"Fourty years researching and having not too much idea that "what" makes a good sound!"
So when I have finished with this amp, don't expect me to give a comment regarding the quality of the sound
I have almost a thousand of 2N5401/2N5551 I bought when I built the Stochino. Many of them have a very high hfe. I will put the highest hfe on the output drivers (Q6/Q7).
I have seen your post regarding splitting the front-end supply, and this driver is grouped with the output mosfet. I think the key to many mosfet amp is the driver itself. That's why I prefer the output mosfet is taking it's input from the emitter of the driver.
Long time ago, with the main objective to take the output from the drain of the mosfet (drain follower) based on the assumption that the capacitance will be minimized, I tried to find a configuration that also enable proper drive of the mosfet. For the drive I decided it has to be an emitter follower. This will affect the number of gain stage. The only usable topology (cannot be more complex) was the one similar to the simplified SKA posted few months later. That's why I didn't understand when it was said that the SKA was something new??? May be the detailed circuit.
Since then I temporarily "concluded" that if I want to build a drain follower from the lateral mosfet, I have to build something with the SKA topology.
But the way you appreciate the musicality of your amp and also the way you do the tuning made me want to build this amp. Even more when Hugh encouraged to do so. This is because I believe what djk has posted:
"I have found 90% of people cannot hear in the audiophile sense, and of the 10% that can, 90% of those pick the worst unmusical sounding gear. The 1% that could tell whether something was musical never seemed to have any money. I would describe the APT as uncolored, but lacking in dynamics, lifeless."
And also Carlos comment:
"Fourty years researching and having not too much idea that "what" makes a good sound!"
So when I have finished with this amp, don't expect me to give a comment regarding the quality of the sound
There's really only one question to ask with audio equipment.
When playing the most emotive, engaging music you like, does it make your foot tap and a tear roll down your cheek?
If yes, then you have an amp which is not damaging the music, and permitting you to engage with the performers. In my experience, speakers and even CD players do not damage the musical engagement, only the extremes of frequency.
When playing the most emotive, engaging music you like, does it make your foot tap and a tear roll down your cheek?
If yes, then you have an amp which is not damaging the music, and permitting you to engage with the performers. In my experience, speakers and even CD players do not damage the musical engagement, only the extremes of frequency.
Ah-ha! I completely agree with the engagement/emotion characteristics. But I don't agree with the speaker as having less effect. But... it is probably because you can afford to have the best speakers in the world.
To me speaker is the most important factor and it should be seen as part of the amplifier. Yes you can easily enjoy the music with the involving sound of tube amps, but what if the amp cannot drive the speaker, or the speaker is the efficient type, built with no filter. Yes, you can enjoy the music but at the end of the day you will have red hot ears...
Also it is difficult to get the real impact of live music with small woofer, and tweeter that cannot be driven with full power (without fatiguing peaks)...
Add/Edit:
I think this is why Mooly have always asked the speakers used by his amp's builders when they report the sound of this amp. I found that the preamp is very important (and a necessary evil) for driving class-B mosfet amplifiers. It should be seen as part of the amplifier also.
Last edited:
Hi Jay, I remember the Stochino amp when it first appeared in Wireless World, a somewhat frightning beast, and all those IN4148's too. Didn't some have to be selected for low leakage at voltages above their ratings.
It's good you are having a go, and I don't think you will be disappointed with the sonics.
It's good you are having a go, and I don't think you will be disappointed with the sonics.
That's a good way of describing things, and certainly with the MOSFET amp it does allow the music to shine through.
In the UK we have DAB radio at appaling bit rates... radio 3 (classical) and "Classic FM" are the only two that use rates of 192kbs (barely adequate) and 160kbs respectively. Other popular stations use 128, and some down to 64kbs in mono.
And this is where the amp still shines... the music come through. Yes the transmission imperfections are quite audible, but the musical experience still comes across... something that absolutely eludes other amps I have used/built.
It's hard to put into words... you'll know when you hear it
Yes it must withstand 150V or so, but I didn't use 1N4148.
If you can live with it (for years!), why cannot I? I don't have difficult requirements from solid state amplifiers. I mean many good amplifiers in this site will do as long as the speaker is good.
But I don't know, I want a mosfet amplifier. Very difficult amplifying devices but I have a high hope on them. Tubes are too slow for me (except probably those I cannot afford). Lateral mosfets are faster than tubes but slower than the other transistors (and linear too). Slower devices tend to be musical and good on vocal and jazz. Faster devices (such as implemented in the Stochino) are better for orchestra, but this kind of music requires big multi-way loudspeakers, something that I don't have, and a music that I don't listen to.
Thanks Mooly. Now I'm confused whether I have to go with permanent "slow" building or one day point to point so I can hear it ASAP For permanent building I will need days to work on only the layout.
I have prepared the heatsink and the fiberglass veroboard. The holes are made for TO-5 because I want the final one to use K135/J50. But for testing I will use TO-220.
And I'm worrying about the pre-amp. I will guess that this amp will need a preamp, which I don't like to use, especially opamp based which seems to work well with amp like this. I'm thinking about building an integrated, with 5678 Aikido (or those 12AT7s that I should have not bought). But I will think about it later. I have a plan to build my own preamp.
About DC servo, I don't have TL071, but I have TL072s that I can use mono, but I prefer to use my other single opamps. Since I want the highest supply rail possible for the amp, I think I will have a dedicated off-board PSU for both on-board opamps (or may be to put the PSU board on the veroboard). I think I will use LF356H that I think is a fet-input one, or those from Analog Devices.
BTW, do you think I should build a speaker protector first? May be I should because sooner or later I will need it to protect my discontinued drivers. But I'm very reluctant even tho the components and PCBs I have are more than enough. Ah, stupid question
Mooly, can you describe your speaker? I think "mosfets" don't work at its best with small speakers.
For permanent building I will need days to work on only the layout. I have prepared the heatsink and the fiberglass veroboard. The holes are made for TO-5 because I want the final one to use K135/J50. But for testing I will use TO-220.
And I'm worrying about the pre-amp. I will guess that this amp will need a preamp, which I don't like to use, especially opamp based which seems to work well with amp like this. I'm thinking about building an integrated, with 5678 Aikido (or those 12AT7s that I should have not bought). But I will think about it later. I have a plan to build my own preamp.
About DC servo, I don't have TL071, but I have TL072s that I can use mono, but I prefer to use my other single opamps. Since I want the highest supply rail possible for the amp, I think I will have a dedicated off-board PSU for both on-board opamps (or may be to put the PSU board on the veroboard). I think I will use LF356H that I think is a fet-input one, or those from Analog Devices.
BTW, do you think I should build a speaker protector first? May be I should because sooner or later I will need it to protect my discontinued drivers. But I'm very reluctant even tho the components and PCBs I have are more than enough. Ah, stupid question
Mooly, can you describe your speaker? I think "mosfets" don't work at its best with small speakers.
If your tweeter is fatiguing in my experience there can be a couple of causes:
1) Inadequate crossover that allows too much low frequencies -> moves tweeter dome too much -> creates huge distortion. Quite a common one that. Decent pro-audio piezo can be very smooth funnily enough - its natural roll off usually prevents any chance of same happening.
2) A non linear amplifier (class B rather worse than class A) with high gain being held together with a big loop feedback loop. Distortion can only be moved - not reduced, so if you need to run more than about 10dB to make your amp linear then it will always sound fatiguing, partly from the non linearities and partly from the feedback.
Hi Jay,
Speakers I use are these, which have a 3 ohm minimum impedance.
View the 703 at Bowers & Wilkins - The World's leading Hi Fi Brand
http://www.bowers-wilkins.com/display.aspx?infid=1098&terid=1118
The amp will work just fine without a preamp if used with a CD player etc. Just use a 10 or 22k pot as a volume control. The feedback factor is low, and corresponding input sensitivity high. 22k and 470 ohm for the feedback equates to a voltage gain of 48, or 33db if you prefer.
The DC servo. First it must be a FET opamp that you use, not a bjt. The power supply for opamp only has to be sufficient to provide the bias for the input transistor, so -12 volts is perfect as the output of the opamp will be around -5 from memory. If you use an IC that can't work with the + input (pin 3) tied to the positive supply (pin 7) when running on a single supply then you will have to provide a split supply for it, in other words -/+12 volts. The TL071 and single zener seemed the most elegant engineering solution. The LF356 should also work correctly I suspect. Remember the opamp isn't in the audio path as such and it's AC characteristics are not important... apart from the high input currents a 741 would be just as good here. The opamp won't affect the sonics. A separate PSU for it is not required.
It's perhaps good practice to include DC offset protection, however lateral FET's are amazingly rugged, so if you have confidence in your construction abilities then I would say it's up to you. I did include offset detection, but that was me wanting to make a complete "job" of it all.
What is ESSENTIAL for a finished amp is the relay delay at power up as the servo takes a few seconds to settle. This can be a really simple design, little more than a power transistor with an RC network on the base to give the delay. As simple as that.
Speakers I use are these, which have a 3 ohm minimum impedance.
View the 703 at Bowers & Wilkins - The World's leading Hi Fi Brand
http://www.bowers-wilkins.com/display.aspx?infid=1098&terid=1118
The amp will work just fine without a preamp if used with a CD player etc. Just use a 10 or 22k pot as a volume control. The feedback factor is low, and corresponding input sensitivity high. 22k and 470 ohm for the feedback equates to a voltage gain of 48, or 33db if you prefer.
The DC servo. First it must be a FET opamp that you use, not a bjt. The power supply for opamp only has to be sufficient to provide the bias for the input transistor, so -12 volts is perfect as the output of the opamp will be around -5 from memory. If you use an IC that can't work with the + input (pin 3) tied to the positive supply (pin 7) when running on a single supply then you will have to provide a split supply for it, in other words -/+12 volts. The TL071 and single zener seemed the most elegant engineering solution. The LF356 should also work correctly I suspect. Remember the opamp isn't in the audio path as such and it's AC characteristics are not important... apart from the high input currents a 741 would be just as good here. The opamp won't affect the sonics. A separate PSU for it is not required.
It's perhaps good practice to include DC offset protection, however lateral FET's are amazingly rugged, so if you have confidence in your construction abilities then I would say it's up to you. I did include offset detection, but that was me wanting to make a complete "job" of it all.
What is ESSENTIAL for a finished amp is the relay delay at power up as the servo takes a few seconds to settle. This can be a really simple design, little more than a power transistor with an RC network on the base to give the delay. As simple as that.
Should mention the B&W's are what I have used for the last 5 years or so. When the amp was first designed I used Celestion SL100's and SL6's
Stereophile: Celestion 100 loudspeaker
Stereophile: Celestion 100 loudspeaker
It is possible to greatly improve rail efficiency (output swing) of this amp by making CFP an over-unity gain stage with a gain of about 1.15-1.25. Easily done by adding two 1.5k 1W resistors from emitters of Q6/Q7 to the ground. Such small gain will have negligible effect on both THD and phase margin.
It is particularly suitable for this amp since Q6/Q7 run very little current, thus power dissipation in Q6/Q7 emitter dividers (for over-unity CFP gain) is small.
It is particularly suitable for this amp since Q6/Q7 run very little current, thus power dissipation in Q6/Q7 emitter dividers (for over-unity CFP gain) is small.
Just FYI:
In my experience B&Ws can be improved greatly in the midrange by adding some decent stuffing behind the mid driver. Usually there are a few wisps of white poly foam, whereas pure wool would be the way to go here. Extra braces are also good.