So, R5 (1k) should enable the use of symetrical currentmirrors ?
And avoid the instable/undefined biasing ?
But is this resistor not killing the benefits of using a currentmirror ?
But i personally don't like currentmirrors, they can have strange
effects for higher freqs by having a sligthly delayed response.
I never really understood the purpose of Q8 in leachamp, what is
it good for ?
Mike
And avoid the instable/undefined biasing ?
But is this resistor not killing the benefits of using a currentmirror ?
But i personally don't like currentmirrors, they can have strange
effects for higher freqs by having a sligthly delayed response.
I never really understood the purpose of Q8 in leachamp, what is
it good for ?
Mike
Mike : Q12 has a voltage base -rail. the current floving in R5 is then ~ (vbias-vbe) / R5 = x A wich is the bias current in this stage. The current modulated is then the current flowing from (and to if you like that expression) the mirrored differensial current from input stage. thus Delta I flowing in Q5 will double when mirrored compared to the original one, as the orginal will only sorce and not sink for the same differential input.
This is basic stuff man 😎 as there is no gain, there is no noise gain either. 😱
Q8 in the orginal schem is for current limmiting purp.
This is basic stuff man 😎 as there is no gain, there is no noise gain either. 😱
Q8 in the orginal schem is for current limmiting purp.
Mike,
have you tried to get rid of C3, split R6 into 2 devices of half value and feed back your signal from this point to the input stage?
Also this will give much higher thd levels there are many people who prefer global feedbackless designs. The main advantage is that your are not feeding back your signal from the slow output devices but from the much faster driver stage.
Only my $0.02.
have you tried to get rid of C3, split R6 into 2 devices of half value and feed back your signal from this point to the input stage?
Also this will give much higher thd levels there are many people who prefer global feedbackless designs. The main advantage is that your are not feeding back your signal from the slow output devices but from the much faster driver stage.
Only my $0.02.
Hi bocka, i will try !
But before, i need to satisfy my curiosity and will try a topology with
OLgain of ~1:1.000.000, want to know what really happens to the
sound with this amount of feedback. Need to know if the "clean" sound
has something to do with thd. I will adopt my existing amp, to ensure
all other parts are the same. This topology has harmonics at -139db.
I am planning to write a harmonizer, a small program applying 2nd & 3rd
harmonics realtime. Maybe i can determine this way an audible threshold
for thd-levels ?
Mike
But before, i need to satisfy my curiosity and will try a topology with
OLgain of ~1:1.000.000, want to know what really happens to the
sound with this amount of feedback. Need to know if the "clean" sound
has something to do with thd. I will adopt my existing amp, to ensure
all other parts are the same. This topology has harmonics at -139db.
I am planning to write a harmonizer, a small program applying 2nd & 3rd
harmonics realtime. Maybe i can determine this way an audible threshold
for thd-levels ?
Mike
hi all! mikeb and lumanuaw! I'm using a darlington output stage .Can the sound change if I 'm using a tripledarlington output? I want to listen yours expirient before I try to do it🙂
I am not too experienced with "normal" darlington outputstage,
i once had made one, and the sound was flat and powerless,
adding the predriver solved this. But i had lowcurrent-vas.
If you have highcurrent-vas, or a vas not needing highimpedanceload,
tripledarlington should be worse.
My recommendation: Try it !
In sims tripledarlington was always better.
Mike
i once had made one, and the sound was flat and powerless,
adding the predriver solved this. But i had lowcurrent-vas.
If you have highcurrent-vas, or a vas not needing highimpedanceload,
tripledarlington should be worse.
My recommendation: Try it !
In sims tripledarlington was always better.
Mike
Have you checked with low impedance loads ?
If sims say, no difference, you maybe shouldn't use it.
The extra devices add more delay and distortion.
Check with high freqs and heavy loads, let's say 20khz into 3ohm ?
If it's not too difficult, add predriver and listen...
Mike
If sims say, no difference, you maybe shouldn't use it.
The extra devices add more delay and distortion.
Check with high freqs and heavy loads, let's say 20khz into 3ohm ?
If it's not too difficult, add predriver and listen...
Mike
Hi, Thanh,
You should try both and tell us the difference 😀
If we consider minimum part, double darlington is better. But like MikeB said, you have to had sufficient current from VAS to final stage.
You should try both and tell us the difference 😀
If we consider minimum part, double darlington is better. But like MikeB said, you have to had sufficient current from VAS to final stage.
pershap FC use 2 diff stage so noise is very low. Evil! Have you finished making this schematic?
???
Not a sinewave ? What else ? With triple or double-darlington ?
I always check if my amps reproduce a proper sinewave with 200khz...
Mike
Not a sinewave ? What else ? With triple or double-darlington ?
I always check if my amps reproduce a proper sinewave with 200khz...
Mike
Hi thanh,
my sims and my real measurement show that a properly designed triple EF is able to reproduce sine waves up to 1MHz at least as well as square waves at 100kHz. Maybe there is something strange in your sim?
my sims and my real measurement show that a properly designed triple EF is able to reproduce sine waves up to 1MHz at least as well as square waves at 100kHz. Maybe there is something strange in your sim?
I observed the same in sims, that tripledarlington-EF has much less
troubles feeding low impedance with high freqs. With doubledarlington
20khz into 2 ohm it was very difficult to produce nice sinewave.
troubles feeding low impedance with high freqs. With doubledarlington
20khz into 2 ohm it was very difficult to produce nice sinewave.
with normal darlington output stage . I remark that with 1.5k load which you often use offset output voltage is higher than 3k3 but by using triple darlington offset output voltage will increase 🙂 I will simulate with 1k5 load and 20khz input sinal 😀
this is mu FC amp.It bad (distortion)
this is mu FC amp.It bad (distortion)
Hi thanh !
Don't you think that your outputstage is a "bit" underbiased ?
It looks like you try to do some EC...
I already tried it this way, but never got good biasing.
The output-waveform shows that your circuit is unstable, every
time it oscillates, the outputswing is reduced.
1st, you have 1.96pA through the biasing circuit, that is about NULL.
i think this should be at least 1mA, that's 500.000 times more.
2nd, for driverstage i prefer at least 40ma, but that could be taste...
3rd, ~100uA in the final stage is about 1000times less than it should be !
(But according to the output you have nice ClassB)
But i think the real problem is the feedback into the biasing beeing
unstable.
And the bjts q3/q4 could have a bit higher hfe, ~100 is not much...
the bc546b/556b i use have ~300 (measured for vce ~8v).
maybe try sa970 ? It's highvoltage and has hfe of ~500.
The biasing for my triple-EF is 2ma,40ma,100ma, fed from a vas
with 2ma.
Maybe try 10nF paralell to the biasing circuit ?
Mike
Don't you think that your outputstage is a "bit" underbiased ?
It looks like you try to do some EC...
I already tried it this way, but never got good biasing.
The output-waveform shows that your circuit is unstable, every
time it oscillates, the outputswing is reduced.
1st, you have 1.96pA through the biasing circuit, that is about NULL.
i think this should be at least 1mA, that's 500.000 times more.
2nd, for driverstage i prefer at least 40ma, but that could be taste...
3rd, ~100uA in the final stage is about 1000times less than it should be !
(But according to the output you have nice ClassB)
But i think the real problem is the feedback into the biasing beeing
unstable.
And the bjts q3/q4 could have a bit higher hfe, ~100 is not much...
the bc546b/556b i use have ~300 (measured for vce ~8v).
maybe try sa970 ? It's highvoltage and has hfe of ~500.
The biasing for my triple-EF is 2ma,40ma,100ma, fed from a vas
with 2ma.
Maybe try 10nF paralell to the biasing circuit ?
Mike
Hi thanh, i refered to your previous amp, not the FC.
I can post you later my FC, with OLgain of 27000,OLBW -3db@27khz,
i have harmonics below 100uv for 12volts swing. (1khz...20khz)
thats ~thd of 0,0007%
Mike
I can post you later my FC, with OLgain of 27000,OLBW -3db@27khz,
i have harmonics below 100uv for 12volts swing. (1khz...20khz)
thats ~thd of 0,0007%
Mike
- Status
- Not open for further replies.