For what it's worth, I have gone back to single-supply capacitor-coupled amplifier design with singleton input stage. I find that provided the capacitors are of "good" quality, nobody hears a difference. I am not worried about THD, I focus more on IMD. I also suggest the use of a ladder R-C power supply filter, rather than the humonguous capacitor banks (of course a choke would be superior, but very expensive for a solid-state amplifier, unless one is running a high-voltage ( >200V) design).
As long as your not going to use full voltage supply for max output swing, then quality of cap matters less. Using full voltage supply for max output swing will lower amplifiers ouput quality to bad no matter how good the signal amp can reject it, because the amp have used up its voltage supply and can only go into clipping with increased voltage supply ripple. Lets say you use 2/3rd of your supply voltage for max output swing with a 8 ohm load, then the last 1/3rd of the voltage not used from voltage supply will be equal to 4 ohm resistance between output and voltage supply. This will make any avarage to high quality capasitor fast enough to deliver current to the output (as cap ESR is way below 4ohm), which is the supply caps primary purpose in audio. Of course its there to remove ripple from 50hz AC or smps, but its still needed to deliver fast currents when needed which no other component can do. This is a reason why a high power amp can deliver better at same output power level as a low power amp, both amps equal in design, high power amp have only higher voltage supply. When useing 1/2 of your supply voltage for max output swing, you can say your cap bank is virtually twice of what you have, because it has twice the voltage supply needed for max output swing and can borrow from the unused voltage if needed.
Last edited:
ps caps vs output caps
to quote bigun "
It's interesting, but I now realize that I'm using the words 'signal path' in more than one way - specifically I am using it differently than you are, and possibly differently than the 'norm' which is likely a bad thing on a Forum.
Anyhow, I'm thinking in terms of the current loop that flows through the speaker as the 'signal path' as far as the load is concerned - and this is not the same thing, as you point out, as the 'voltage' signal that travels from left to right across a schematic."
The point of this thread to me, is that caps put between the output transistor and the speaker keep the speaker from being burnt out by an overheated output transistor, without an expensive microprocessor and current transformer to measure safe operating area violation on the transistor and open expensive precious metal contact relays. Caps in the power supply, no matter how high quality, cannot perform this function. The number of burnt up speakers and output transistors I see at the music resale shop and goodwill, have indicated to me that the entire industry has adopted the split supply amp in pursuit of tenths of a percent of imd or thd, while secretly being glad that the system burns up and has to be replaced every couple of years of hard use. It's as if you bought a car with seat belt and 6 airbags, but with a paper shell so that they made sure the other car killed you, not theirs. (Actually, micro cars are pretty close to that, I think). I'm glad to see another person, Balaboo, that thinks single supply amps with simple output capacitors are worthy of use if you are a hard headed DIyourselfer. Output caps sound funny on split supply amps, but not on single supply amps, as my experiment proved. Spending as I do 6 to 10 times as much money on the speaker that I do on the amp, I find split supply amps without SOA protection, output crowbars and speaker relays, to be a version of snake oil that almost the entire world thinks is "wonderful". I'm saying the emperor has no clothes.
to quote bigun "
It's interesting, but I now realize that I'm using the words 'signal path' in more than one way - specifically I am using it differently than you are, and possibly differently than the 'norm' which is likely a bad thing on a Forum.
Anyhow, I'm thinking in terms of the current loop that flows through the speaker as the 'signal path' as far as the load is concerned - and this is not the same thing, as you point out, as the 'voltage' signal that travels from left to right across a schematic."
The point of this thread to me, is that caps put between the output transistor and the speaker keep the speaker from being burnt out by an overheated output transistor, without an expensive microprocessor and current transformer to measure safe operating area violation on the transistor and open expensive precious metal contact relays. Caps in the power supply, no matter how high quality, cannot perform this function. The number of burnt up speakers and output transistors I see at the music resale shop and goodwill, have indicated to me that the entire industry has adopted the split supply amp in pursuit of tenths of a percent of imd or thd, while secretly being glad that the system burns up and has to be replaced every couple of years of hard use. It's as if you bought a car with seat belt and 6 airbags, but with a paper shell so that they made sure the other car killed you, not theirs. (Actually, micro cars are pretty close to that, I think). I'm glad to see another person, Balaboo, that thinks single supply amps with simple output capacitors are worthy of use if you are a hard headed DIyourselfer. Output caps sound funny on split supply amps, but not on single supply amps, as my experiment proved. Spending as I do 6 to 10 times as much money on the speaker that I do on the amp, I find split supply amps without SOA protection, output crowbars and speaker relays, to be a version of snake oil that almost the entire world thinks is "wonderful". I'm saying the emperor has no clothes.
Last edited:
Impressive results of both damping factor and harmonic distorsion from an amp using a singleton input transistor and an output coupling capacitor :
MJR-7 Mosfet Power Amplifier.
MJR8-Mk2
I built the mjr7-mk3 using author's boards.
My intellect and my ears enjoy the result.
MJR-7 Mosfet Power Amplifier.
MJR8-Mk2
I built the mjr7-mk3 using author's boards.
My intellect and my ears enjoy the result.
Member
Joined 2009
Paid Member
The MJR uses split feedback, with some of it coming before and some of it from AFTER the output cap. I think this makes a difference because the feedback error correction will reduce distortion from the cap.
In a dc-coupled speaker-amplifier system using split rail power supply the main capacitors through which the speaker current returns to the power supply are the rail capacitors. And if such amplifiers have high PSRR then these caps have negligible influence by definition of having high PSRR (unless PSRR is poor at some frequencies so be careful!). If instead, the amplifier in question has lower PSRR then these caps may have a real influence on the sound depending on how low the PSRR is because the speaker return currents flowing through the imperfect capacitors will generate some distortion on the power rails that can affect the amplifier performance by definition of having low PSRR (how low is another question).
Most amplifiers with global negative feedback have good PSRR, the feedback directly increases the PSRR and 'takes the rail caps' out of the equation if the simpler power supply design is desired.
Perhaps this means that if you have a v.low PSRR amplifier design (probably has no gnf), you might as well feel good about using a single rail supply with a good quality output capacitor.
In a dc-coupled speaker-amplifier system using split rail power supply the main capacitors through which the speaker current returns to the power supply are the rail capacitors. And if such amplifiers have high PSRR then these caps have negligible influence by definition of having high PSRR (unless PSRR is poor at some frequencies so be careful!). If instead, the amplifier in question has lower PSRR then these caps may have a real influence on the sound depending on how low the PSRR is because the speaker return currents flowing through the imperfect capacitors will generate some distortion on the power rails that can affect the amplifier performance by definition of having low PSRR (how low is another question).
Most amplifiers with global negative feedback have good PSRR, the feedback directly increases the PSRR and 'takes the rail caps' out of the equation if the simpler power supply design is desired.
Perhaps this means that if you have a v.low PSRR amplifier design (probably has no gnf), you might as well feel good about using a single rail supply with a good quality output capacitor.
Last edited:
check out this thread:
http://www.diyaudio.com/forums/soli...eaker-output-not-dc-wanted-3.html#post4235951
http://www.diyaudio.com/forums/soli...eaker-output-not-dc-wanted-3.html#post4235951
are there commercial amp devices available in the meantime, where the MjR topology is in use?
Thule Audio's amp design operates without global NFB - go to
go to last attachment in post 8 under
Ultimate ECC for Class B from Thule Audio? Topology also used by Teac
This means, the audible difference by listening test between low cost main capacitors and such with extrem low ESR and screw terminals (absolutely no very small cv product) should actually be very large.
That is indeed the case also in real live - that could be verified perfectly.
OTOH - if there is an amp with a very good PSRR character like that one in the attachment from post 1 under
Cyrus III power amp schematic here, but schematics of the later models wanted
then this audible difference between the same different classes of electrolytic caps must actually be much more smaller.
In real live, however, this was not the case - the perceived difference by listening test was of the same order of magnitude.
The same situation is to observe in the small signal area by comparing of various operational amplifiers - a high value of PSRR does not allow to realize only a cost-effective approach for the power supply.
I hear again and again that it is due to a bad GND management or an unfavorable PCB wire layout. OTOH nobody was able until know,
to convince myself that the very good audible results can be achieved even with a cheap power supply approach by the use of parts like the LME's from TI with very high PSRR.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.