scott wurcer said:
In a first approximation, in a field controlled device, constant beta means a constant Z/L x mobility product. Mobility goes down, Z/L has to go up, hence area has to go up, hence the capacitances have to go up. Its a little more complicated in JFETs because the substrate concentration comes into play as well, but essentially that's the drill.
Slowly but surely, as you read your old class notes or go on line to find the answers, all the blanks will be filled in.
Gee, turn your back and things get interesting.
Complimentary parts usually aren't. That's one reason why I don't really like complimentary differential pairs. Besides, matching all four together is a pain. That's the only way I've seen them sound good with low DC offset (no DC servo or offset control).
I do really like the sound of a J-Fet differential pair for amplification.
-Chris
Complimentary parts usually aren't. That's one reason why I don't really like complimentary differential pairs. Besides, matching all four together is a pain. That's the only way I've seen them sound good with low DC offset (no DC servo or offset control).
I do really like the sound of a J-Fet differential pair for amplification.
-Chris
OK, everyone, stop making your JC-2 line amps and JC-3 power amps. Everyone here knows (except me) that they don't work very well. 
All right, more usefully: I have found that a complementary differential input stage is 5 times lower in overall distortion than a single differential input stage with a single second stage drive transistor and a current source load;
I have also found that a single differential input stage to have approximately 2 times the distortion with a second stage of a differential amplifier and a current mirror, compared to a complementary differential input stage.
These were MY findings 35 years ago, your measurements may vary.
For the record, I have made hundreds of examples of all the above.
If someone here wanted to try this yourself, just make or simulate a JC-2 line amp circuit and get it running well. THEN, just add one large electrolytic cap (good quality) across the base emitter junction on 1 of the line output transistors. The DC will not change, but the distortion will rise. Check it out.
I have also found that a single differential input stage to have approximately 2 times the distortion with a second stage of a differential amplifier and a current mirror, compared to a complementary differential input stage.
These were MY findings 35 years ago, your measurements may vary.
For the record, I have made hundreds of examples of all the above.
If someone here wanted to try this yourself, just make or simulate a JC-2 line amp circuit and get it running well. THEN, just add one large electrolytic cap (good quality) across the base emitter junction on 1 of the line output transistors. The DC will not change, but the distortion will rise. Check it out.
Are you talking open loop or closed loop? Low frequency or high frequency?
I tried the matching chore for the complementary differential and decided it was not practical for the production processes I had available. (I have no patience for sorting or matching and no one to dump it on.) I used the single differential driving a second stage differential+current mirror with the mosfet output followers, using 2 P ch to match 1 N ch. I lifted it mostly from the LH0024. I still use essentially the same circuit with enhancements. Despite toying with many other circuits I keep returning to what works. I think that is true of most successful designers. I also think experience and refinement are more important than a magic circuit.
I tried the matching chore for the complementary differential and decided it was not practical for the production processes I had available. (I have no patience for sorting or matching and no one to dump it on.) I used the single differential driving a second stage differential+current mirror with the mosfet output followers, using 2 P ch to match 1 N ch. I lifted it mostly from the LH0024. I still use essentially the same circuit with enhancements. Despite toying with many other circuits I keep returning to what works. I think that is true of most successful designers. I also think experience and refinement are more important than a magic circuit.
Good questions Demain. I factored the 5 in the previous comparison, because the complementary differential gives 2 times more gain and therefore 2 times more feedback, but that leaves about 2.5 times distortion improvement.
As you know, I have made both designs with single fet pairs and complementary fet pairs. It has been my personal experience that ALL ELSE BEING EQUAL that the complementary differential input stage can give a factor of 2 improvement in distortion with about the same forward gain, but that could vary with other designs.
The design that I made for ST, years ago, was my lowest distortion design, and it only used a single fet pair at the input. However it used a lot of negative feedback to achieve it, and twice as many parts as my typical complementary differential design would use.
As you know, I have made both designs with single fet pairs and complementary fet pairs. It has been my personal experience that ALL ELSE BEING EQUAL that the complementary differential input stage can give a factor of 2 improvement in distortion with about the same forward gain, but that could vary with other designs.
The design that I made for ST, years ago, was my lowest distortion design, and it only used a single fet pair at the input. However it used a lot of negative feedback to achieve it, and twice as many parts as my typical complementary differential design would use.
Since this was discussed here before:
Cold fusion seems to come of age. Who would have thought that...
http://www.planetanalog.com/showArticle.jhtml?articleID=216200397&cid=NL_planet
PS The Editor of the magazine where this is published has stated upfront that there is NO 1-April joke in the magazine...
Jan Didden
Cold fusion seems to come of age. Who would have thought that...
http://www.planetanalog.com/showArticle.jhtml?articleID=216200397&cid=NL_planet
PS The Editor of the magazine where this is published has stated upfront that there is NO 1-April joke in the magazine...
Jan Didden
MIX
I assume you're also a fan of Donald Knuth. His work is fascinating and entertaining as well.
Recently, I found this on the web:
“Beware of bugs in the above code; I have only proved it correct, not tried it. (@JC: this is a joke)
His first three volumes still take a prominent place in my bookcase.
Nelson Pass said:
I assume you're also a fan of Donald Knuth. His work is fascinating and entertaining as well.
Recently, I found this on the web:
“Beware of bugs in the above code; I have only proved it correct, not tried it. (@JC: this is a joke)
His first three volumes still take a prominent place in my bookcase.
matching P and N FETs
(hint: look at one of your own amps: they still need a servo to get rid of the off-set)
The solution is pointless. 😀john curl said:
(hint: look at one of your own amps: they still need a servo to get rid of the off-set)
My amps work very well without servos, they just make it easier to have almost zero offset.
The JC-3 did not have a servo, for example, and neither did the JC-2.
The JC-3 did not have a servo, for example, and neither did the JC-2.
janneman said:
Jan, did you read the caption below the picture?
"Silvered Dewar calorimeter used by Navy researchers to detect neutron emissions from a cold fusion process" (italics by me)
Are you sure it's no April fools' joke?
john curl said:
Ok Teach, I have a question. On the one hand you seem to be complimenting the complementary i/p LTPs on the basis that THD is halved. On the other hand, your lowest THD has been achieved with a single i/p LTP. But you continue to use the complementary i/p.
Isn't that inconsistent with you goal?
How do you justify that?
janneman said:
The headline writer seems to have ignored the key word, "claimed." The paper's author is not exactly a giant in the field- I think before breathlessly touting "confirmation of cold fusion," perhaps it's worth seeing if anyone else can duplicate the results, though that doesn't sell magazines (or get clickthroughs). Remember that there were people back in '88-'89 who claimed to have confirmed Pons's work...
traderbam said:
A complementary differential design with no GNF may sound better than single FET pair with a lot of GNF, though the later will have lower THD.
Edmond Stuart said:
Jan, did you read the caption below the picture?
"Silvered Dewar calorimeter used by Navy researchers to detect neutron emissions from a cold fusion process" (italics by me)
Are you sure it's no April fools' joke?
My first reaction was: April fool!
But then I read this:
EDITOR'S NOTE
First, there is no April Fool's Day joke in this newsletter, for two reasons: we're not in the mood; also, most of these so-called "jokes" fall flat and are funny only to the originator, not the recipients. Enough said, we move on.
I don't know enough of physics to judge the claims. Over to you guys 😉
PS While I'm at it, check out this:
http://www.spectrum.ieee.org/mar09/7901
Jan Didden
A tube gain stage will use even fewer parts and show lower open loop distortion, but that's apparently beyond the pale.
>A tube gain stage will use even fewer parts and show lower open loop distortion, but that's apparently beyond the pale.<
Not here it's not. My top preamp uses cascoded differential TUBE gain stages - without any GLF. And THD is very hard to measure on my (admittedly) 16 bit audio specA.
Regards, Allen (Vacuum State)
Not here it's not. My top preamp uses cascoded differential TUBE gain stages - without any GLF. And THD is very hard to measure on my (admittedly) 16 bit audio specA.
Regards, Allen (Vacuum State)
- Status
- Not open for further replies.