G.Kleinschmidt said:
I think the real problem with laterals is not so much the source resistors to equalize the drain currents - it's the instantenous output into a short circuit. I got a paper testing a single set of hitachi's 2SK135/J50 into a short. Something over 100A in a short burst. I would use some DC overcurrent protection. Actually I will dig up a current limiting circuit for Laterals. No V limiting part however that's usually done with zener's anyway. Think it does have source resistors also...🙂
Matching P Chl's to N chl's seems to have not much merit according to some more experienced guys than myself. Lateral's are quite good compliments anyway.
Best Regards
Have a Great Easter!! even the sun is out in england......
Kevin
Fanuc said:
You have a great Easter too!
No shortage of sunshine where I am. The religiously inclined are praying for rain this year.
Cheers,
Glen
G’day all.
This should liven this thread up a bit. Have a read of this quasi-technical marketing guff; viz BJT’s vs FET’s.
http://www.borbelyaudio.com/products.asp
I can respect the general technical competence evident in the designs (except for a few PCB layouts) for sale via that website, but this generalised, unspecific guff on the alleged “graininess” and low “resolution” of BJT’s (which is clearly a marketing ploy) is just plain daft. And what is “resolution” supposed to refer to in a purely analogue system? Duhhh. Can someone please explain to me how to measure a BJT amplifier’s resolution? this, I'd truly love to know.
Allegedly, these traits hinder the sonic performance ultimately attainable from BJT amplifiers, despite the fact that using modern design techniques, a BJT amplifier producing truly inaudible levels of distortion can be made quite easily.
I’ve got several amplifiers without any FET’s in ‘em at all, and they sure don’t exhibit any miraculous degree of “graininess” not evident in those with a few FET’s inside.
What do ya’ll think?
This should liven this thread up a bit. Have a read of this quasi-technical marketing guff; viz BJT’s vs FET’s.
http://www.borbelyaudio.com/products.asp
I can respect the general technical competence evident in the designs (except for a few PCB layouts) for sale via that website, but this generalised, unspecific guff on the alleged “graininess” and low “resolution” of BJT’s (which is clearly a marketing ploy) is just plain daft. And what is “resolution” supposed to refer to in a purely analogue system? Duhhh. Can someone please explain to me how to measure a BJT amplifier’s resolution? this, I'd truly love to know.
Allegedly, these traits hinder the sonic performance ultimately attainable from BJT amplifiers, despite the fact that using modern design techniques, a BJT amplifier producing truly inaudible levels of distortion can be made quite easily.
I’ve got several amplifiers without any FET’s in ‘em at all, and they sure don’t exhibit any miraculous degree of “graininess” not evident in those with a few FET’s inside.
What do ya’ll think?
G'day Glen,
What do I think? I think that you don't realize/ know who Erno Borbely is. Measuring resolution? Easy, just use your ears,lots. And remember its all just entertainment.
Cheers,
Steve.
What do I think? I think that you don't realize/ know who Erno Borbely is. Measuring resolution? Easy, just use your ears,lots. And remember its all just entertainment.
Cheers,
Steve.
srh said:
Hmmm....I have read a lot of Erno Borbely's stuff over the years. I'm reasonably sure that I do realise who he is.
Well you see, that is the problem. I’ve designed and built many amp’s and listened to plenty more (with my ears too, as you suggest). Some had MOSFET output stages. Some had BJT output stages. There are JFET LTP’s and BJT LTP’s and JFET LTP’s cascoded with BJT’s and visa-versa. Not to mention BJT VA stages and MOSFET VA stages and BJT VA stages cascoded with MOSFET’s and vica-versa. There were a few BJT outputs with MOSFET drivers. A few BJT outputs with BJT drivers too!…… I’m sorry to report that there is no distinctly (or even mildly) obvious group that qualifies as having either more or less “resolution” (least of all “graininess” ) over any or all of the others what-so-ever 🙄
I agree. Allegations of “graininess” and low “resolution” against BJT’s are just silly. We shouldn’t take them seriously.
Cheers,
Glen
PS
Still holding my breath in desperation for a coherent technical argument in support of the alleged intrinsic audio “resolution” limitation of bipolar transistors.
srh said:
Steve,
then just use your ears, not your eyes.
Borbely doesn't serve shelf consumers, he can do without the bogus product brochure crape(-e).
Hi Glen,G.Kleinschmidt said:
I think the ‘graininess’ of BJT’s is easily explained if one considers that the active charge carriers consists of muons, while FET’s use the much lighter electrons. The problems with these evil muons is that they are not only 206.7 times heavier than electrons, hence a much lower ‘resolution’, but they also have a mean life time of a mere 2.2 microseconds. As a result, the propagation time delay back EMF induced distortion at a collapse of a muon, although not measurable, surprises the negative feedback loop in a graininess like fashion. Get me? else, ask…. (no names)
Cheers, Edmond.
BTW, What happened to your beautiful pair of eyes?
estuart said:
Hi Glen,
I think the ‘graininess’ of BJT’s is easily explained if one considers that the active charge carriers consists of muons, while FET’s use the much lighter electrons. The problems with these evil muons is that they are not only 206.7 times heavier than electrons, hence a much lower ‘resolution’, but they also have a mean life time of a mere 2.2 microseconds. As a result, the propagation time delay back EMF induced distortion at a collapse of a muon, although not measurable, surprises the negative feedback loop in a graininess like fashion. Get me? else, ask…. (no names)
LOL! Be careful. Next thing you know, that post will be quoted verbatim as a technical reference in AudioXpress.
estuart said:
I haven't been getting a lot of sleep lately, so they're a bit dark a baggy underneath. BTW, I did ask you not to speak to me that way in public, Edmond.
G.Kleinschmidt said:
Any fame is a good fame 😀
G.Kleinschmidt said:I haven't been getting a lot of sleep lately, so they're a bit dark a baggy underneath. BTW, I did ask you not to speak to me that way in public, Edmond.
I'm sorry. Of course, I mean her eyes.
Hi, Glen,
Is it right that "Graininess" from BJT amp, it is likely coming from clipping of BJT itself? Mosfets can come out of clipping differently than BJT (have no base Q to discharge). BJT amp can be made clean sounding, no graininess, just make sure every transistor in the signal path including the big output transistor, never reach VCE<=0V3, using baker clamp variations or the likes.
Is it right that "Graininess" from BJT amp, it is likely coming from clipping of BJT itself? Mosfets can come out of clipping differently than BJT (have no base Q to discharge). BJT amp can be made clean sounding, no graininess, just make sure every transistor in the signal path including the big output transistor, never reach VCE<=0V3, using baker clamp variations or the likes.
For output stage BJT it is quite simple actually, just put RC or other filter from the same voltage supply (that is used also by the output transistors) so that the front stage (differential and VAS)have lower voltage rail than the output stage. This can keep the output transistors (emitor follower assumed) not reaching low Vce. Big transistors (output transistors) are more difficult to discharge the base' Q from clipping condition, better prevent it before happening.
Or, more extreme using split power supply for the front-end and output transistors, but unlike the common approach, make the output transistors have higher voltage than the front stage like JLH-80w power amp.
Or, more extreme using split power supply for the front-end and output transistors, but unlike the common approach, make the output transistors have higher voltage than the front stage like JLH-80w power amp.
For VAS bjt, we can put something like this. The VAS transistor Q106 in worst case will have Vb=Vc. If needed more, change D101 to 2x1N4148, so Vbc of Q6 will be 0V6 worst.
"Graininess" can also come from bad impedance matching inside power amp stages. I don't understand much of impedance theoritically, but I feel that a good connection is low source-impedance connected to high receiver-impedance.
Like low impedance preamp output connected to high input impedance of power amplifier.
I read John Curl's post about emitor resistor
http://www.diyaudio.com/forums/showthread.php?postid=228105#post228105
He connected it with noise.
But I see it from impedance matching POV.
For example is the VAS connection.
Drawing1 is simple connection between differential resistor to VAS' base.
In drawing1, the differential's load = 1kohm, this is the "source" impedance, point A.
The "receiver" impedance is point B. Since there is no emitor degeneration here, I calculate the internal emitor resistor 25ohm/Ic. Eg : Ic=5mA, then the internal re=5ohm. Point B will be 5ohm x hfe, (eg : hfe=100), then point B = 500ohm.
This is not so good connection because A>B. Ideally we want A<<<<B (low source impedance compared with receiver impedance)
Then what happens if we put RE in VAS' emitor. Eg : we put 100ohm emitor resistor in VAS. Point A is still 1kohm, but point B (the receiving impedance) becomes 100ohm x hfe, B=10kohm.
This is already better than drawing1, because the source impedance is now lower than receiving impedance, A<B.
In drawing3, we put buffer T2 like DougSelf. (Eg : hfe of T1 and T2=100), and also we neglect the parrarel impedance effect. (both to internal re or outside resistor, we only calculate the impedances resultant of multiplication/division of hfe).
Point A=1kohm,
Point B becomes 2k2 x hfe; B=220kohm.
Point C will be 1kohm/hfe=1k/100=10ohm.
Point D will be internal re x hfe = 500ohm.
This drawing3 also good, because A<<B and C<<D
Drawing 4 is another buffer, but NPN type with 100kohm towards ground emitor load. In here
A=1kohm,
B=100kohm x hfe=10Mohm,
C=1kohm/hfe=10ohm,
D=100ohm x hfe = 10kohm.
This drawing 4 is better than drawing3 because A<<<<B and C<<<<D.
This drawing4 impedance matching is the one that DartZeel uses.
Like low impedance preamp output connected to high input impedance of power amplifier.
I read John Curl's post about emitor resistor
http://www.diyaudio.com/forums/showthread.php?postid=228105#post228105
He connected it with noise.
But I see it from impedance matching POV.
For example is the VAS connection.
Drawing1 is simple connection between differential resistor to VAS' base.
In drawing1, the differential's load = 1kohm, this is the "source" impedance, point A.
The "receiver" impedance is point B. Since there is no emitor degeneration here, I calculate the internal emitor resistor 25ohm/Ic. Eg : Ic=5mA, then the internal re=5ohm. Point B will be 5ohm x hfe, (eg : hfe=100), then point B = 500ohm.
This is not so good connection because A>B. Ideally we want A<<<<B (low source impedance compared with receiver impedance)
Then what happens if we put RE in VAS' emitor. Eg : we put 100ohm emitor resistor in VAS. Point A is still 1kohm, but point B (the receiving impedance) becomes 100ohm x hfe, B=10kohm.
This is already better than drawing1, because the source impedance is now lower than receiving impedance, A<B.
In drawing3, we put buffer T2 like DougSelf. (Eg : hfe of T1 and T2=100), and also we neglect the parrarel impedance effect. (both to internal re or outside resistor, we only calculate the impedances resultant of multiplication/division of hfe).
Point A=1kohm,
Point B becomes 2k2 x hfe; B=220kohm.
Point C will be 1kohm/hfe=1k/100=10ohm.
Point D will be internal re x hfe = 500ohm.
This drawing3 also good, because A<<B and C<<D
Drawing 4 is another buffer, but NPN type with 100kohm towards ground emitor load. In here
A=1kohm,
B=100kohm x hfe=10Mohm,
C=1kohm/hfe=10ohm,
D=100ohm x hfe = 10kohm.
This drawing 4 is better than drawing3 because A<<<<B and C<<<<D.
This drawing4 impedance matching is the one that DartZeel uses.
Hi all
Perhaps "graininess" means noise?
In a differential pair, graininess might mean resolution in terms of offset voltage, implying some sort of digital equivalent like resolution in terms of bits. But then we need to know how many bits...
Such possibilities still make no sense to say FETS are better. Unless exceptionally well matched, bipolars have lower offset voltage and "resolution" than FETS. Generally FETS may have an advantage on noise but then their gain is lower so additional stages may be required to match gains between designs.
Without any explanation of what on earth this means we cannot take it seriously.
cheers
John
Perhaps "graininess" means noise?
In a differential pair, graininess might mean resolution in terms of offset voltage, implying some sort of digital equivalent like resolution in terms of bits. But then we need to know how many bits...
Such possibilities still make no sense to say FETS are better. Unless exceptionally well matched, bipolars have lower offset voltage and "resolution" than FETS. Generally FETS may have an advantage on noise but then their gain is lower so additional stages may be required to match gains between designs.
Without any explanation of what on earth this means we cannot take it seriously.
cheers
John
G.Kleinschmidt said:
Hi Glen,
I'm surprized and disappointed hearing such a load of baloney from Borbely's site. Its all in the details, no matter what underlying technology one uses. And, yes, he is remiss in being quasi-technical on the one hand, while on the other hand not giving a hint as to what he means by "resolution".
We all have our preferences for certain devices in certain locations of amplifiers, but broad generalizations like those on that site are just marketing hype. Me, I like JFETs at the input, bipolars in the middle, and MOSFETs in the output; but those are just my choices in the context of the tradeoffs that I have chosen and am comfortable with. Others make different choices and that's OK too.
Cheers,
Bob
G.Kleinschmidt said:
Thank you Glen. And yes, it did liven up the thread 🙂.
estuart said:I think the ‘graininess’ of BJT’s is easily explained if one considers that the active charge carriers consists of muons, while FET’s use the much lighter electrons. The problems with these evil muons is that they are not only 206.7 times heavier than electrons, hence a much lower ‘resolution’, but they also have a mean life time of a mere 2.2 microseconds. As a result, the propagation time delay back EMF induced distortion at a collapse of a muon, although not measurable, surprises the negative feedback loop in a graininess like fashion. Get me? else, ask…. (no names)
ROFL! That was great 🙂. For a similar style of humor, check out the posts by "Swedish Chef" in this thread.
I think that 'graininess' mentioned by Erno, is the higher order harmonic distortion that bipolar devices normally make, compared to fets.
Secondly, why attack Erno? Hasn't he been very open about his schematics and such? What have you folks done lately? 😀 PS I think that tubes might be even less 'grainy' than fets.
Secondly, why attack Erno? Hasn't he been very open about his schematics and such? What have you folks done lately? 😀 PS I think that tubes might be even less 'grainy' than fets.