Member
Joined 2009
Paid Member
It's a subjective thing, people like different amps. But again, I would suggest reading about the SEWA amplifier as it has the kind of output you are interested in.
http://www.diyaudio.com/forums/pass-labs/66822-sewa-seven-watt-amplifier.html
go down to post 163:
About an op-amp as an input stage. If the feedback is taken from the output of the amp, won't then the op-amp try to compensate for some of the distortion then?
I'm hoping for an oscilloscope for christmas so I can finally start looking at the output of my amps and getting some THD values 😀 If I make an amp like this on a breadboard, will measuring distortion give an accurate value? Or do I have to solder it in order to get valid measurements?
I'm hoping for an oscilloscope for christmas so I can finally start looking at the output of my amps and getting some THD values 😀 If I make an amp like this on a breadboard, will measuring distortion give an accurate value? Or do I have to solder it in order to get valid measurements?
Hello, Jay
I think I should add some comments. In this discussion, the problem is that main terms are not well defined. For me,
- "tuby sound" is simply a sound without "harshness", it is very frequently very far from good sound, because of other sound properties. It can be obtained with SS or tube designs.
- "macro sound picture" is the domain where THD could play some role, with very high THD timbres could start to deviate from proper ones, but that must be really high THD, more than 1%. With many simple designs one can easily stay below 1%.
- "micro sound picture" is the domain where all designs differs most of all, and this sound property has almost no relation with THD. It seems to be more related with phase intermodulation, a kind of "jitter" caused by variation of gate-drain, gate-source, base-collector, base-emitter capacitances with voltages. Tube schematics have some natural benefits in this respect. Apart from "jitter", "micro sound picture" suffers from many other reasons, like PS hi-frequency spikes achieving main schematics, effects of capacitors, resistors, contacts, PCB design and electro-magnetic coupling, etc. Many factor tend to destroy "micro sound picture", but there is no THD among them.
Speaking about simple single stage designs, it is quite obvious that they have almost no reason to produce "harsh" sound, so it will be tuby in that or another way.
However, since they have no measures included against phase intermodulation, I find them all loosy at the "micro sound domain", very far worse than good quality tube amps.
I did attempts to suppress somehow the "jitter", by using special HF LDMOS transistors at power follower, with definite succsess. With proper output "cap", it is surely competitive with best tube amps. However, one is forced to use preamp in addition to follower, and any preamp I tried to use with such an advanced follower, made the combination preamp-advanced follower worse in "micro picture" than good integrated tube amp. I speak about listenings of high quality system, that allows to distinguish tiny differences.
For those who are looking for really top sound quality, integrated amp solution, but with external power supply, is preferable.
So, one stage amp is not a real amp, it is a "semi-amp" and can not determine final sound quality.
Many people put on the top three reqirements, and believe they go in correct way
1) output power
2) low THD
3) no output cap
From my point of view, for audiophile home listening, these requirements are terribly misleading and add enormous difficulties in achiving best sound. Good sound coud be obtained even with these postulates, but with terrible difficulties, bringing costs of a good sounding amp to 20000...50000 USD range.
Last edited:
As per my observation, "combined" output cap in NoGNFB or low GNFB designs adds less problems to sound (I do not say distortions, since they usually assosiated with THD) than caps at low-signal positions. Sound effects of "combined output cap" is comparable to effects of a relay contact. Due to this, I stay for a long time with SE class A schematics. Within this approach, I obtain some simplification of PS, no need in servo, no speaker protection, no crossover distortions, no deep GNFB, no sound "harshness". However, good "micro sound picture" is not an inherent benefit of simple class A, it must be achieved by careful selection of low Ciss jFETs, very high Ft output transistors, current type and not deep GNFB.
Last edited:
On the top of all said above, I should mention also a "bass signature" of the kind of output transistors used. All the good properties of lateral FETs and LDMOS (very high "effective" Ft) do not make them free from some specific signature at bass domain. There is no chance to remove this signature by deep GNFB and very low output impedans (high damping factor). Maybe this will not be the case with specific high efficiency low-power-tube dedicated speakers, but with typical 3-band speakers one could easy recognize bass features, depending on what kind of transistors is used at the output.
After many comparisons and weighing all pro and contra, now I prefer BJTs for HF applications, with Ft = 300...700 MHz.
SIT transistors are a bit another story, they are also very attractive.
After many comparisons and weighing all pro and contra, now I prefer BJTs for HF applications, with Ft = 300...700 MHz.
SIT transistors are a bit another story, they are also very attractive.
Thank you for the replys, I have a better understanding now 😀
I haven't learned about fourier analysis and calculating THD values yet so I don't know the math behind it, but are you saying that "tuby sound" and "micro sound picture" can't be expressed mathematically? I found it really attractive to think of a distortion meter that could directly and mathematically compare one amp to another 🙁
A little off topic here. Could any of you link a class AB design that is allmost as simple as the class A design I linked in the opening post? An op-amp input stage would be preferable and I really like mosfets. As much as I found class A design fancy, the power dissipation and extra demand on the transformer isn't that appealing 🙁 I was thinking of looking into class AB designs.
I haven't learned about fourier analysis and calculating THD values yet so I don't know the math behind it, but are you saying that "tuby sound" and "micro sound picture" can't be expressed mathematically? I found it really attractive to think of a distortion meter that could directly and mathematically compare one amp to another 🙁
A little off topic here. Could any of you link a class AB design that is allmost as simple as the class A design I linked in the opening post? An op-amp input stage would be preferable and I really like mosfets. As much as I found class A design fancy, the power dissipation and extra demand on the transformer isn't that appealing 🙁 I was thinking of looking into class AB designs.
I would say, math reflects general features, macro picture, but not the complete truth. Math is based on the model, and the model is definitely simplified compared to all details of the phenomenon (sound reproduction).
Additional advances in math is possible, but, again, up to definite point only, it will not explain all the details again.
I'm really curious about having an op-amp as an input stage with it's feedback taken form the output of the amp. I was under the impression that op-amps usually have great specs (PSRR, CMRR etc) compared to many descrete amplifier designs. Also with the feedback taken from the output of the amp I can't understand why this won't greatly reduce distortion.
Another question. In my next design I wan't to use a voltage regulator to reduce humming. If I then connect the gate bias resistor directly to the output of the regulator, won't this greatly increase the gain aswell? I remember doing this the first time I made the amp and the gain was insane compaired to after I connected the resistors to drain instead. One problam was that the operating point would shift dramaticly if the voltage or the bias current would shift, but with a regulator the voltage shouldnt change and neither should the current.
Another question. In my next design I wan't to use a voltage regulator to reduce humming. If I then connect the gate bias resistor directly to the output of the regulator, won't this greatly increase the gain aswell? I remember doing this the first time I made the amp and the gain was insane compaired to after I connected the resistors to drain instead. One problam was that the operating point would shift dramaticly if the voltage or the bias current would shift, but with a regulator the voltage shouldnt change and neither should the current.
I guess its among the best of its kind....didn't he also design the ONO, to start with
who was it ? norwegian guy, mats, mads, or something ?
just read the whole post, simpler better in class A, in my limited experience, yes
after more than a year hard tweaking my Aleph J clone, listening to many friends F5, F4 etc. and after 2 weeks enjoying CCSed 2SK82 SIT amp,
just as Vladimir says somewhere here about our perceptions on output power, THD, no output cap, my old perceptions is surely on a shaky ground now 🙂
after more than a year hard tweaking my Aleph J clone, listening to many friends F5, F4 etc. and after 2 weeks enjoying CCSed 2SK82 SIT amp,
just as Vladimir says somewhere here about our perceptions on output power, THD, no output cap, my old perceptions is surely on a shaky ground now 🙂
Member
Joined 2009
Paid Member
Does LM317 have enough bandwidth?
You said don't mind component value so I wont, but in my experience those switching MOSFETs have horrible non-linear gate capacitance, making simple circuit difficult to sound good.
You said don't mind component value so I wont, but in my experience those switching MOSFETs have horrible non-linear gate capacitance, making simple circuit difficult to sound good.
Not sure about the LM317's bandwidth. It doesn't have to be a switching mosfet. The circuit drawing just showed the design not saying anything about what kind of components that are used, the LM317 represents an adjustable regulator. I later got that a regulator might not be the ideal current source so I then wanted to assume an ideal current source instead just so I could understand what was "flawed" about the actual one-stage mosfet output. But I understand now that it is the unlinearity of the mosfet that's the bottle neck, but that's why I'm wondering how a op-amp input stage with the feedback taken from the output of the amp won't fix a lot of this.
The SEWA is basically a POWER FOLLOWER 99 by Andrea Ciuffoli, with normal/ordinary power supply (Ciuffoli used virtual ground). Key element is the use of IRFP150. A little improvement can be made with current source a-la VladimirK.
check out the pictures by post #1 and #2 about
http://www.diyaudio.com/forums/pass...itional-op-amp-ultimate-sounding-phl1230.html
the based schematic is here:
http://cygnus.ipal.org/mirror/www.passlabs.com/images/zenf2.pdf
the air coil on the images are 10 mH (choke) and the caps behind this choke 2x 10.000uF (Kendeil)
Supply voltage behind the 10 mH inductor is 25V and the idle current is 3,5A (3500mA)/each channel.
The outdoor power supply for one channel (no pictures) consist of two light transformers (each 12V 500VA), rectifier, 1x 10.000uF Siemens Sikorel and a small air coil inductor. The distance for wire lead between power supply and 10 mH choke on the actually ZEN amp from the mentioned images is 5 meters (~16 feet).
Additional very strong increase of the sonic quality is reached by introducing of additional choke and additional caps (6,8mH/75mF = additional 12db lowpass) so as additional caps of 200mF in parallel to the exist ones behind the 10 mH choke.
The unwanted capacities arround the Gate, Drain and Source from the gain MOSFET on ZEN (Grounded Emitter) seems to have had no negative unwanted and audible effects cause a unity gain preamp in front of this power amp with 1A idle current (Vdd: only 9V) - inside there is Andrea Ciuffoli's Power Follower (redesign for 9V supply voltage).
the Power Folllower
The answer to the headline is from my view clearly yes, despite any unwanted inherent properties like very bad power supply rejection ratio (PSRR) and large non linear input-reverse capacities of typical power MOSFETs from the good known IRF series.
Last edited:
Member
Joined 2009
Paid Member
I would guess that simple power followers go back a long way before Andrea but there is something about the SEWA that captured my imagination more than the others. I have a short list of amplifiers I want to build and this is one that's still on the list. Mads developed versions with and without current sources - somewhere there's a thread he started called OTA (one transistor amplifier) which might have been the inspiration for Mark's ZCA (zero component amplifier).
Take a peek over on the chip amp forum, lots of stuff over there along these lines.
- Status
- Not open for further replies.