Will this work as audio amplifier?

[fab]

Originally posted by MikeB
...
But maybe there is a reason, why nearly noone uses CFP in the
diffamp ? (Except steven and some cyrus-amp)

Mike

hi Mike

At least Halcro, Lavardin and me use CFP in the input diff amp. I have also seen other amp circuits from what I remember. If it is not widely used it does not mean that it is not good. If it was so obvious to know what is the best topology in an audio amp then everybody would use the same basis.

Fab

[lumanauw]

I read this somewhere here. Why is that for very low rail voltage (like +/-15V), a differential is better fed with CCS? While for higher rail (like +/-40V), it is ok to use CCS or plain resistor?

FAB,
How big is the CFP resistor can be (how low the current minimal in the first transistor)? Smaller than 50uA is OK?

[MikeB]

Hi lumanauw, the reason is very simple: the two emitters of
the diffamp have a voltageswing of the middle of both inputs,
that means nearly exact the inputvoltage (plus some harmonics)
for a voltageswing of 0.7v at voltagerail 15v, this means that
the current fed into the diffamp differs about 4.6%.
The higher the voltagerail, the less this effect. At some voltage
the plain resistor can give a better constant current then a ccs.

I finished my amp this night, but this late in the night i was
not able to make extensive listeningtests. But i can say, that
i'm very impressed by the quality. The sound is crystalclear and
clean, and the instrumentseparation is a dream. Also it's like
you can see/feel the singer. (I tested with "brother in arms")
(hmm, i hate monolistening)
And no sign of any hard sound.

I used 2K for the CFP-Resistors, the CFP-BJTs are bc556b, the
input transistor are sk170. I skipped the currentmirror, to get
maximum bandwidth (-3db at ~1mhz )
Outputdevices are MJL1302A & co.
I already hear some noise, but only when pressing the ear directly
to the tweeter. It seems that i already have some thermal
problems in the diffampstage, the DC drifts ~60mv with a
frequency of ~30secs.
I still have to investigate this phenomen, maybe the diffamp is
not balanced enough. (2.2mA to 2.8 mA)
In spice i tested with upto 10k for the CFP-resistors, but the
difference in THD was minimal to the 2k.
Maybe i will downgrade to 1k, if the thermal drift keeps...
Or i reduce the current to the diffamp.
I have to recheck the datasheet for the sk170...

Mike

[lumanauw]

Hi, Mike,
Thanks for the explenation.
Now back to the sch in post #1. The main idea of this sch is to advoid 2nd harmonic cancelation, so I dont use differential for input stage. But the VAS is forming differential. Will the harmonics be canceled here or not?

[lumanauw]

the current fed into the diffamp differs about 4.6%.

The current fed to differential will varied if the rail voltage is low (like only +/-15V and using only plain resistor, not CCS). But is this good or bad (audibly and measurement)

This is kind of scary but we must know. Finding good sounding is not the same as finding good measurement.

True story -- There's a guy named Andrew Hefley that is a very good solid state designer. Do a web search and you'll see he has a long and distinguished resume. You can even see some of his circuits in an app note he wrote for Motorola (now available at On Semiconductor).

About six or eight years ago, he heard his first single-ended triode. And do you know what he did? He quit the business. Now he makes instrumentation for light aircraft.

He told me that after 30 years of designing solid state audio amps that he figured he knew it pretty well. But after hearing the SET, it showed him that pretty much everything he knew and believed was wrong. Instead of going back to the drawing board, he literally quit the field.

There are other guys here, like Nelson, who didn't quit. Instead, they started exploring a completely different approach than "perfect steady-state performance on a test bench into a resistive load".

Thinking is good. Listening is better. Listening and thinking about what you are hearing is best.

[fab]

Originally posted by lumanauw
FAB,
How big is the CFP resistor can be (how low the current minimal in the first transistor)? Smaller than 50uA is OK?

hi lumanauw,

Unfortunately, I have not preformed a study on the optimal value for that resistor. I use 2k2 value. I think the resistor value depends also on the ratio of first input transistor collector current versus the second transistor collector current. Gerrard Perrot (Lavardin Technology) in his patent 5,635,874 suggests to use 1/4 of second bjt collector current in the first bjt collector current as an example and I believe that he may have performed a thourough analysis and testing to determine this ratio of collector current. I also think that if you use a too small current you may have an increase of noise depending on the input transistor noise characteristic.

Fab

[thanh]

The odd harmonics are not phaseshifted

Ok! This is a reason which odd harmonics is increase by a factor of 2

[thanh]

Finding good sounding is not the same as finding good measurement.

Ok! I re-say some word from a member of diyaudio
"Some one like music and some one like distortion"
I think "some one like HIFI and some one like distortion"
I hate odd harmonics too!

[MikeB]

Hi lumanauw !

I did some checkings on VAS with Spice, it seems that the
VAS mainly creates 2nd harmonics, but i don't know if this
is still so if the transistors are matched. Cascoding the VAS
had a very big increase in qualitity.

I will check with the ccs to the diffamp this evening, but from
my imagination, using a resistor with low voltagerails should
create 2nd harmonics. (but maybe only with asymetrical topology)

BTW, i did some extensive listeningtests with my CFP-testamp,
and i must say: Wow, CFP rulez ! is it really patented ?
Of course it's still preferable to find a topology for ClassAB that
cancels out odd harmonics.

Mike

[MikeB]

Originally posted by thanh

Ok! I re-say some word from a member of diyaudio
"Some one like music and some one like distortion"
I think "some one like HIFI and some one like distortion"
I hate odd harmonics too!

Yes, odd harmonics are agony... and the human ear is extremely
sensitive to them. I experimented a little, add 3% 3rd harmonic
to music (done with software), and you instantly puke, add 3%
2nd harmonic, and you wonder if there really is a difference !
(Okay, played only through pcspeakers) But this means that
simple THD-measurements are useless to describe the soundquality
of an amp. (Not a new discovery at all...)
I still don't know what i prefer, HIFI (ultraprecise reproduction),
or some nice harmonics... I am still on the quest to a ultraprecise
ClassAB. (Maybe found it last night) I think i "simply" need to
construct a single ended ClassA, not showing any odd harmonics.
(And then compare them)
Mike