And knowing what poison ivy looks like 😀 🙂 🙂
Crossover distortion (at low bias - 20mA) is hard to see in a
conventional way but it seems to affect THD in a big way (attached)
Also the spectral component of that distortion changes (H2 vs.
H3/H5).
I can't see where you could not detect this distortion (even audibly) as I can hear an underbiased amp from the next room.
OS
Attachments
janneman said:
Hi Jan,
Yeah, no, I often don't understand myself so we are in the same boat! 🙂
Well, i said I couldn't SEE it on a scope...I haven't actually listened to any of my amps when they weren't running a healthy idle current.
My ears are not what they used to be so I may not hear it anyway.
MJL, you have to look at the amount of 7th harmonic distortion in a Class B amp, and THEN you will understand. NO SCOPES, because you have to be below -100db at all levels to be OK.
scott wurcer said:
I'm not sure that I follow you here. All I was trying to point out is that the voltage developed across either emitter degeneration resistor by the load current WILL reverse bias the base-emitter of the Darlington not passing the load current. Also, 200uA at 20kHz (OS's sim) is off as far as I am concerned - cross conduction effects due to driver transistors with floating emitters do not make a "non-switching" OPS!
Cheers,
Glen
Hi Steve,
The reason I didn't read your thread originally was the reference to Krell amplifiers. Once I started to read about your work, I quickly realized how close your work was to mine. That includes the idea of a self contained output stage. It is capable of very good performance, but I think your transistors at the signal input might be starved for current. I used high beta devices than you show, and mine very definitely were current starved. I only use one stage and a diode now to avoid that problem.
Hi John,
-Chris
The reason I didn't read your thread originally was the reference to Krell amplifiers. Once I started to read about your work, I quickly realized how close your work was to mine. That includes the idea of a self contained output stage. It is capable of very good performance, but I think your transistors at the signal input might be starved for current. I used high beta devices than you show, and mine very definitely were current starved. I only use one stage and a diode now to avoid that problem.
Hi John,
Completely agree! This effect (class A) was very strong in some of my prototypes using a high power "diamond buffer" type output stage.
-Chris
anatech said:
Steve stated that he will no longer reply in this thread, only in the Krill thread.
I have seen the discussion move there.. steve is a valuable
contributer to DIY.
This is a discussion about preamps/general topology.. isn't it ??
GK says this topology can't be "non -switching" and my sims are
wrong. I will acknowledge that 2 ohm loads will render my sim
to fault ,but I've done N. pass's circuit (patent), tanaka's works,
and all the "super a's" from technics. As well I have been
VERY objective , using Andy C's models , more advanced
spice directives,different devices, etc.
My conclusions by all this are that 1. the total distortion of
the amp is less (especially using global NFB)
2.The type of distortion is different (much less H3 /5/ no 7 ! )
3. having both prototyped / ran active loads (speakers &
headphones) it sounds better , more like an "A".. 🙂
4. Since I don't feel right about "stealing" steves design ,
I won't build his , but am seriously considering using
this "alternative biasing" for headphone/preamp use.
5. NONE OF THE ABOVE SWITCH.
OS
Good post OS,
empirical data can't be beat by all the theories in the world - this is real diy, methinks & you represent the epitome of what this forum is about - discussing theory, analysing/simulating circuits & then BUILDING the things to measure & listen.
Excellent work
empirical data can't be beat by all the theories in the world - this is real diy, methinks & you represent the epitome of what this forum is about - discussing theory, analysing/simulating circuits & then BUILDING the things to measure & listen.
Excellent work
john curl said:
Oh, if I only had the tools...🙂
I tried to make a deal for a real distortion analyser but that fell through. I'll make do with RMAA.
joshua_G said:
What's wrong? Not getting enough attention anymore? Sound like a peed off school girl.
jkeny said:
Mistakes beat empirical data.
Not saying that anyone made any mistakes, but they do happen.
ostripper said:
Indeed, and it's a legitimate discussion.
However, Steve stated that he will no longer reply here, only on the Krill thread.
ostripper said:5. NONE OF THE ABOVE SWITCH.
OS
Horseraddish.
Your sim shows a piddly 200uA of crossconduction at 20kHz. You can do this with a D.Self double EF just by floating the emitters of the driver transistors as well.
You are aware that floating the driver emitters is actually BAD for overall linearity? (by floating I mean not adding an emitter-emitter resistor to set a driver quescient current).
Run it at 200Hz and put in Steves self-defeating 1uF cap across the driver bases - it switches. You can't get around the spanner thrown into the works by the voltage developed across the emiter degeneration resistors.
I could explain this until I'm blue in the face, around here it probably wouldn't make any difference.
GK , I am open to differing opinions. I HAVE "floated" EF's
"for the hell of it" to see cause and effect.
So what you say is that pass , tanaka , technic's are all
liars , pushing fraudulant designs ???
I did do them all , with the same results , much different
than any EF .
I have seen the shortcomings in steves design by the way
a modified it accordingly.. (cap)
OS
"for the hell of it" to see cause and effect.
So what you say is that pass , tanaka , technic's are all
liars , pushing fraudulant designs ???
I did do them all , with the same results , much different
than any EF .
I have seen the shortcomings in steves design by the way
a modified it accordingly.. (cap)
OS
I think that this 'Krill' design is very creative. I prefer simplicity to complexity, but it is an interesting study.
Although all the results are not in, it would appear that the large NPO ceramic caps measure about 10 times worse than polystyrene or Teflon, but 10 times better than Mylar.
A vs B
How do you know that? Have you ever build a class-B OPS with perfect crossover (which doesn't exists, btw)? If class-heat was still better, then the crossover was simply less then perfect.
Nevertheless, with today knowledge and techniques, it's possible to reduce all kind of distortions of a class-B amp below 1ppm. So no excuse any longer for building class-heat thingies.
john curl said:
How do you know that? Have you ever build a class-B OPS with perfect crossover (which doesn't exists, btw)? If class-heat was still better, then the crossover was simply less then perfect.
Nevertheless, with today knowledge and techniques, it's possible to reduce all kind of distortions of a class-B amp below 1ppm. So no excuse any longer for building class-heat thingies.
Law of physics, Edmond. All else being equal, class A is better. Of course, if you have a PERFECT output stage, it doesn't matter.
- Status
- Not open for further replies.