Warmed up. You'll find that they drift a fair amount when the amp is just turned on, but after twenty or thirty minutes or so, you'll get pretty stable values.
The absolute value isn't really all that important unless you intend to run the amp near full power. The relative offset is the one to watch out for. Parts matching is your first line of defense. Then the circuit embellishments to fine tune things.
Around this point, someone frequently pops up asking about using a servo. Yes, it can be done. No, it's not necessary to go to those lengths.
Grey
The absolute value isn't really all that important unless you intend to run the amp near full power. The relative offset is the one to watch out for. Parts matching is your first line of defense. Then the circuit embellishments to fine tune things.
Around this point, someone frequently pops up asking about using a servo. Yes, it can be done. No, it's not necessary to go to those lengths.
Grey
GRollins
"The relative offset is the one to watch out for. Parts matching is your first line of defense. Then the circuit embellishments to fine tune things"
thats what ive said, if you aim for high precision match! and use
high precision components, high quality is the number 1 factor
here
if you cant attain these matches in components just follow
the 5 ways mr Pass has talked about
Mr pass thank you for the links
"The relative offset is the one to watch out for. Parts matching is your first line of defense. Then the circuit embellishments to fine tune things"
thats what ive said, if you aim for high precision match! and use
high precision components, high quality is the number 1 factor
here
if you cant attain these matches in components just follow
the 5 ways mr Pass has talked about
Mr pass thank you for the links
Petter
I would be very much interested to have a look at your last design of the x100 since Nelson doesn't mind about it.
Another thing though, using 12 pairs of fets for 100w per channel is indeed an overkill unless you plan to go up to 600w with the same circuit?
I would be very much interested to have a look at your last design of the x100 since Nelson doesn't mind about it.
Another thing though, using 12 pairs of fets for 100w per channel is indeed an overkill unless you plan to go up to 600w with the same circuit?
Nelson wrote: "...so on the X350 we used a current mirror for a level shifter."
Could anyone explain how to do it on X100?
Could anyone explain how to do it on X100?
B.L. said:
Well, you know what a current mirror is - there are plenty of
examples on this site. You source a certain amount of current
into it and you get a certain proportional amount out. The
ratio between input and output are arbitrary, and when the
ratio is unity, it's a level shifter.
Uhmm, could be something like this?
The frontend in this thread is single ended, so I drew it this way. Could also be complementary I guess.
Not sure if it would work.
But it's more a conceptual schematic anyway. All this resistor values guessing...
Ciao, Tino
The frontend in this thread is single ended, so I drew it this way. Could also be complementary I guess.
Not sure if it would work.
But it's more a conceptual schematic anyway. All this resistor values guessing...
Ciao, Tino
Attachments
How does the level shifter help achieving rail-to-rail output? Doesn't the voltage at the gate of the output mosfets have to be 4-5 volts higher than the rail?
I just wanted to draw peoples attention to this interesting push-pull X design. The author has built it and says good things about it.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=64612&perpage=10&pagenumber=1
Leve
http://www.diyaudio.com/forums/showthread.php?s=&threadid=64612&perpage=10&pagenumber=1
Leve
It's a good design and easily implemented. It really runs best
as a Class A design though, the distortion is a bit high with
AB bias.
😎
as a Class A design though, the distortion is a bit high with
AB bias.
😎
I did it; called it the Double Cross in one iteration, something else in another version. I ran mine class A.
Grey
Grey
I'll break the ice.
Wow, almost a year since a post here. Is this project THAT dead.? Is there a reason this went virtually nowhere? We've had a hot summer here so far, and the full dissipation of the class a stuff has made it even hotter for me.
(I don't complain in the winter though)
Does anyone have experiences with any of the designs detailed or mentioned in this thread? I am interested in trying one, but unfortunately don't have time to try them all.
Your input is appreciated,
Marc
Wow, almost a year since a post here. Is this project THAT dead.? Is there a reason this went virtually nowhere? We've had a hot summer here so far, and the full dissipation of the class a stuff has made it even hotter for me.
(I don't complain in the winter though)Does anyone have experiences with any of the designs detailed or mentioned in this thread? I am interested in trying one, but unfortunately don't have time to try them all.
Your input is appreciated,
Marc
Hi Marc,
I can just say for myself, that my design (LBHajdu mentioned few posts before) is working very well and currently I'm working on new 4 chanel version for active speakers. There are no radical improvements, just added cascode on differential stage, removed current limiters and removed temperature compensation. Bias is 3A, previously it was 1.5A.
You can find measurement on this page
Thats all
Regards, Robo
I can just say for myself, that my design (LBHajdu mentioned few posts before) is working very well and currently I'm working on new 4 chanel version for active speakers. There are no radical improvements, just added cascode on differential stage, removed current limiters and removed temperature compensation. Bias is 3A, previously it was 1.5A.
You can find measurement on this page
Thats all
Regards, Robo
a little more?...If you'd be so kind..
Well I hope you wouldn't mind me asking a question or two.
1) Do you have an updated schematic or more detail as to the mods you have done since the original schematic?
2) Did you find it necessary to double your bias in the interest of sound quality? Was it perhaps due to a difficult load?
My interest is the somewhat lower bias for some watts class A but of course much more AB. In other wirds, I already have a couple of ~3A amps around here...don't need another
3)What percentage (A to AB)would you say you are at at 3A?
Thanks in advance, I don't think this has got the attention it deserves. It's got mine😉(unless I'm really missing something) and I really would like to make it my next project if it's do-able.
Marc
Well I hope you wouldn't mind me asking a question or two.
1) Do you have an updated schematic or more detail as to the mods you have done since the original schematic?
2) Did you find it necessary to double your bias in the interest of sound quality? Was it perhaps due to a difficult load?
My interest is the somewhat lower bias for some watts class A but of course much more AB. In other wirds, I already have a couple of ~3A amps around here...don't need another
3)What percentage (A to AB)would you say you are at at 3A?
Thanks in advance, I don't think this has got the attention it deserves. It's got mine😉(unless I'm really missing something) and I really would like to make it my next project if it's do-able.
Marc
Percentage A A/B
Hi there,
I think the power of the X unit is that it allows you to run it without high bias, unlike the Aleph, Aleph-X and Zen variants.
I am not currently using the amplifier (not using any amplifier actually, pretty sad), but plan to get operational again this winter.
My plan is to have low supply voltage (enough for 100W output power into 8 Ohms) so that when I do bias high, I will not burn so much heat. The plan is to have about 1/4 of that in Class A. As I understand from NP, you only need enough bias to keep your FET output devices linear, and that appears to be the way that the commercial units are set up.
One note of warning though. When I built, I underestimated the heatsinking (not the power, that was calculated) needed for all FET's on the input stage.
I am also considering a Lovoltech design using switched mode power as I have gotten hold of some seemingly incredible units.
The schematic is pretty accurate. I would use a Vbe multiplier or equivalent on each side to reduce risk of thermal runaway though. There is also the small concern that you might end up with your input at 1 Vgs drop above ground unless you are careful, and that could cost you some output voltage swing since the unit is no longer symmetrical around zero (symmetrical around 1 Vgs drop above zero. As far as your speakers, that should not matter since most speakers are floating anyway and you are driving phase to phase rather than phase to ground.
The other thing to note is that it can be quite hard to figure out if your unit is actually working since you need the balance .... Next time around I will use slow servos to set up the current sources, but for now, it is convenient for me to set up partial circuits and tune them independently (a pain in the you know what)
Good luck!
Hi there,
I think the power of the X unit is that it allows you to run it without high bias, unlike the Aleph, Aleph-X and Zen variants.
I am not currently using the amplifier (not using any amplifier actually, pretty sad), but plan to get operational again this winter.
My plan is to have low supply voltage (enough for 100W output power into 8 Ohms) so that when I do bias high, I will not burn so much heat. The plan is to have about 1/4 of that in Class A. As I understand from NP, you only need enough bias to keep your FET output devices linear, and that appears to be the way that the commercial units are set up.
One note of warning though. When I built, I underestimated the heatsinking (not the power, that was calculated) needed for all FET's on the input stage.
I am also considering a Lovoltech design using switched mode power as I have gotten hold of some seemingly incredible units.
The schematic is pretty accurate. I would use a Vbe multiplier or equivalent on each side to reduce risk of thermal runaway though. There is also the small concern that you might end up with your input at 1 Vgs drop above ground unless you are careful, and that could cost you some output voltage swing since the unit is no longer symmetrical around zero (symmetrical around 1 Vgs drop above zero. As far as your speakers, that should not matter since most speakers are floating anyway and you are driving phase to phase rather than phase to ground.
The other thing to note is that it can be quite hard to figure out if your unit is actually working since you need the balance .... Next time around I will use slow servos to set up the current sources, but for now, it is convenient for me to set up partial circuits and tune them independently (a pain in the you know what)
Good luck!
I have to say that I could not hear difference between 1.5A and 3A bias on my current gear. Maybe with higher grade components, especialy loudspeakers, there's very little difference.
Updated schematics of single chanel version is availabe, just email me toudy@chello.sk
Regards, Robo
Updated schematics of single chanel version is availabe, just email me toudy@chello.sk
Regards, Robo
DarkOne said:
Setting aside the issue of what you can hear, there is a
real measurable difference. There isn't a lot of feedback on
these designs, and the performance depends quite a but on
the linearity (=bias) of the output stage. My rule of thumb for
Class AB X amps is that the idle dissipation should approach the
rated output at 8 ohms.