hi lumanauw !
At least sims showed, that bjts like mjl1302 want to be used as EFs.
Seems they are optimized for this.
Using them in a cfp-style output gave large 3rd harmonics.
I should try in real world, but my prototype is too compact to make
that big changes...
You can't say bad sound with this style, but tripledarlingtonoutput
sounds better to me. My primary intention to build my own amp is the
fact that my "Yamaha DSP-A2" has poor bass-performance, and poor
trebles. As this circuit showed the same bass like my yamaha, i don't
like it. Maybe i try tuning a bit, at least checking if mids and trebles
are good. As it now has some conditionally unstability, i can't really
determine quality.
Has anybody an idea, why bassperformance is so different between
topologies ? I would understand it with mids and trebles, but low freqs ?
Mike
At least sims showed, that bjts like mjl1302 want to be used as EFs.
Seems they are optimized for this.
Using them in a cfp-style output gave large 3rd harmonics.
I should try in real world, but my prototype is too compact to make
that big changes...
You can't say bad sound with this style, but tripledarlingtonoutput
sounds better to me. My primary intention to build my own amp is the
fact that my "Yamaha DSP-A2" has poor bass-performance, and poor
trebles. As this circuit showed the same bass like my yamaha, i don't
like it. Maybe i try tuning a bit, at least checking if mids and trebles
are good. As it now has some conditionally unstability, i can't really
determine quality.
Has anybody an idea, why bassperformance is so different between
topologies ? I would understand it with mids and trebles, but low freqs ?
Mike
Hi MikeB,
The schematic is a magic to me.
I just can't make out how it is working.
First of all, the voltage across c75 is not constant, even it is designed to be. And c75 presents a heavy load to Q59 and Q60. And thus cause a pole at low frequency. Together with the dominant pole @ Q52 and Q53, It oscilates.
Secondly, the amp seems running in a strange DIRECTION. My reasoning is as below. When input goes from 0 to positive, the current flows through Q52 increases, ant thus Q55 and Q59. The the gate voltage of M6 lowers. This contradicts the expection that the output voltage should goes up.
I would think this may work connecting the gate of M6 to the emitter of Q59, and the gate of M5 to the emitter of Q60, with C75 sticking to both gates. The capacitors c67, c68, c71, and c72 and R81, R82 may better be removed. The bias voltage set aroud Q58 may have to be increased to bias the MOS @ class AB for high Vgs value.
Hope this helps. And please correct me iif there is anything wrong in the above.
The schematic is a magic to me.
I just can't make out how it is working. First of all, the voltage across c75 is not constant, even it is designed to be. And c75 presents a heavy load to Q59 and Q60. And thus cause a pole at low frequency. Together with the dominant pole @ Q52 and Q53, It oscilates.
Secondly, the amp seems running in a strange DIRECTION. My reasoning is as below. When input goes from 0 to positive, the current flows through Q52 increases, ant thus Q55 and Q59. The the gate voltage of M6 lowers. This contradicts the expection that the output voltage should goes up.
I would think this may work connecting the gate of M6 to the emitter of Q59, and the gate of M5 to the emitter of Q60, with C75 sticking to both gates. The capacitors c67, c68, c71, and c72 and R81, R82 may better be removed. The bias voltage set aroud Q58 may have to be increased to bias the MOS @ class AB for high Vgs value.
Hope this helps. And please correct me iif there is anything wrong in the above.
robert_chien said:Hi MikeB,
The schematic is a magic to me.
Which part is magic to you ? The outputstage ? You might have not
seen that the mosfets are connected with their source to rails, and
with drain to output. This means that if the driver opens, the mosfet
also opens, as Vgs raises. Vgs is across the 150ohm.
Mike
hi robert !
If i remember correct, openloopgain is ~1:10000, pole freq don't
remember, phaseshift ~400ns.
I'm at work, so have no acess to simuresults...
Many voltageamplifications ? I count three, the only one with high
gain is the vas itself.
I think, as long properly compensated with lead & lag-caps, you can
use any number of amplifications ?
BTW, the amp is not finished, these caps are not adjusted / missing.
Mike
If i remember correct, openloopgain is ~1:10000, pole freq don't
remember, phaseshift ~400ns.
I'm at work, so have no acess to simuresults...
Many voltageamplifications ? I count three, the only one with high
gain is the vas itself.
I think, as long properly compensated with lead & lag-caps, you can
use any number of amplifications ?
BTW, the amp is not finished, these caps are not adjusted / missing.
Mike
Hi MikeB,
I think Q52/Q53 contributes the first pole, and Q59/Q60 a second pole. I am curious about the pole location of the M5/M6 stage. The other buffer stages have poles much farther away. There got 3 apparent poles so I believe it less easy to compensate this amp.
And do you have any idea about why adding C75 will cause oscillation? It just looks fine.
I think Q52/Q53 contributes the first pole, and Q59/Q60 a second pole. I am curious about the pole location of the M5/M6 stage. The other buffer stages have poles much farther away. There got 3 apparent poles so I believe it less easy to compensate this amp.
And do you have any idea about why adding C75 will cause oscillation? It just looks fine.
robert_chien said:
Yes, i observed 3 poles in the AC-sweep, the pole for outputstage was
at ~100mhz. All 3 poles were far below unitygain.
I think c75 only triggers an already existing instability. I wouldn't finish
this amp without c75, it's too dangerous without compensation for
crossconduction. In sims crossconduction was at 40A for 200khz.
With c75 = 47nF it was completely gone. (I had no 47nF, thus 100nF)
Mike
Hi, Robert Chien,
I just want to add, the strange C you are reffering is sugested by Sajti, for eliminating/minimizing crossconduction if you built CFP. I also read this directly from Sajti here, you cannot find that in books. Sajti suggested value of 10nf-22nf, but Mike found that value.
Hi, Mike,
Yes, I also experienced many bass sounds differs ALOT with different cct. I also cannot explain why, but sometimes it is frustating, that when I built an amp that is expected to be perfect, the result is an amp with only mids and highs (no overall tonal balance). For now this overall tonal balance is more important to me than detailed trebles, or super bass. Perfect tonal balance makes an amp enjoyable, even if the trebles are not so detailed as other design.
AKSA said once, that distorted low frequency gives more pleasant bass, like his design. This is also said by many of his amp builders. I dont know, maybe an "Accurate" amp gives bad bass?
Bass usually determined by the behavior of output stage. Like non feedback amp, they usually critized for not-controlled bass.
What is your opinion on Error Correction? Will it make the bass better, no matter what topology you use? Or this Error correction is more heard in trebles than bass? What do you think?
PS: Mike, how old are you? And what job are you with right now? You likes audio amps alot. Is it have to do with your current job?
I just want to add, the strange C you are reffering is sugested by Sajti, for eliminating/minimizing crossconduction if you built CFP. I also read this directly from Sajti here, you cannot find that in books. Sajti suggested value of 10nf-22nf, but Mike found that value.
Hi, Mike,
Yes, I also experienced many bass sounds differs ALOT with different cct. I also cannot explain why, but sometimes it is frustating, that when I built an amp that is expected to be perfect, the result is an amp with only mids and highs (no overall tonal balance). For now this overall tonal balance is more important to me than detailed trebles, or super bass. Perfect tonal balance makes an amp enjoyable, even if the trebles are not so detailed as other design.
AKSA said once, that distorted low frequency gives more pleasant bass, like his design. This is also said by many of his amp builders. I dont know, maybe an "Accurate" amp gives bad bass?
Bass usually determined by the behavior of output stage. Like non feedback amp, they usually critized for not-controlled bass.
What is your opinion on Error Correction? Will it make the bass better, no matter what topology you use? Or this Error correction is more heard in trebles than bass? What do you think?
PS: Mike, how old are you? And what job are you with right now? You likes audio amps alot. Is it have to do with your current job?
hi lumanauw !
If you look for an amp with good balance, my symamp might be
interesting for you, it has very good bass, clean trebles, and dynamics.
It does not prefer certain tones. But i'm still finetuning...
I can't imagine that the bass from my symamp is distorted. It's too deep
and precisely controlled to be distorted. (Deep means ouch...)
The interesting thing is, my speakers have the lowest impedance
at ~50hz, close to the freq for deep bass from music.
I will try an asymamp these days, just to compare topologys and find
the best for me. I gave up trying to reinvent the wheel.
My job ? I am programming computergames (Xbox and PC). I have
some experience from programming soundengines, but audioamps
always fascinated me, don't know why... And my age is 35.
Mike
If you look for an amp with good balance, my symamp might be
interesting for you, it has very good bass, clean trebles, and dynamics.
It does not prefer certain tones. But i'm still finetuning...
I can't imagine that the bass from my symamp is distorted. It's too deep
and precisely controlled to be distorted. (Deep means ouch...)
The interesting thing is, my speakers have the lowest impedance
at ~50hz, close to the freq for deep bass from music.
I will try an asymamp these days, just to compare topologys and find
the best for me. I gave up trying to reinvent the wheel.
My job ? I am programming computergames (Xbox and PC). I have
some experience from programming soundengines, but audioamps
always fascinated me, don't know why... And my age is 35.
Mike
Hi, Mike,
Waiting forward your experiment result with Asymetrical amp. I've tried both, the result,......I like the Asym. I experimented with more "not hi-tech" things, like R only for CCS. Cannot give statement yet. Too many "gurus" do not use this R loading for their commercial products, got to find out why they prefer CCS. In a glance this R gives better sonics.
The next experiment will be EC. Steven's design is the one I'm going to built. It is full complementary, non-feedback, have opamp servo, and EC output stage.
Output stage is very important, when you built a big rated amp, such as Pro-Amps. If you read the specs, like MC2 pro amp, the damping factor is very big, like 1000. I dont know what makes this, when I see inside the MC2 amp, it only employ small amount of output devices. You can get big damping with parrareling so many output stages, or with Error Correction (EC). I suspect this MC2 uses EC. The sound with MC2 is fine, bass is very well controlled. Maybe this is due to its damping factor.
I really wanted to try your design. But I cannot buy the complementary Jfet you use here. I can only buy K30 here. That means I lack of matched Pfets. Small good/complementary N/Pfets are hard to buy here.
I found that the best frequency for bass is 45hz. You boost at this point, it is obvious that this frequency is better than other bass frequencies. Your speaker is close to tis frequency. Maybe even it is small it will give good balance.
Waiting forward your experiment result with Asymetrical amp. I've tried both, the result,......I like the Asym. I experimented with more "not hi-tech" things, like R only for CCS. Cannot give statement yet. Too many "gurus" do not use this R loading for their commercial products, got to find out why they prefer CCS. In a glance this R gives better sonics.
The next experiment will be EC. Steven's design is the one I'm going to built. It is full complementary, non-feedback, have opamp servo, and EC output stage.
Output stage is very important, when you built a big rated amp, such as Pro-Amps. If you read the specs, like MC2 pro amp, the damping factor is very big, like 1000. I dont know what makes this, when I see inside the MC2 amp, it only employ small amount of output devices. You can get big damping with parrareling so many output stages, or with Error Correction (EC). I suspect this MC2 uses EC. The sound with MC2 is fine, bass is very well controlled. Maybe this is due to its damping factor.
I really wanted to try your design. But I cannot buy the complementary Jfet you use here. I can only buy K30 here. That means I lack of matched Pfets. Small good/complementary N/Pfets are hard to buy here.
I found that the best frequency for bass is 45hz. You boost at this point, it is obvious that this frequency is better than other bass frequencies. Your speaker is close to tis frequency. Maybe even it is small it will give good balance.
hi lumanauw !
"Waiting forward your experiment result with Asymetrical amp"
Me too !
I had an asymamp before, my experience was, that the symamp
sounds cleaner and more balanced. But it had some flaws i need
to correct and test.
"Maybe this is due to its damping factor."
Yes, i believe also that high dampingfactor gives this controlled deep
bass. BTW, what's the exact formula for calculating dampingfactor ?
I only checked the voltagedrop between load and noload. I can also
calculate the outputimpedance from this. This gave 0.3mohms.
But the ratio to 8ohms would be 26666. This can't be dampingfactor,
do i need to divide by 100 ?
"That means I lack of matched Pfets."
Hmm, that's a pity !
"Maybe even it is small it will give good balance"
Small ? The speakers have 40liters ! Or did you mean the outputpower
from my amp ?
Mike
"Waiting forward your experiment result with Asymetrical amp"
Me too !
I had an asymamp before, my experience was, that the symamp
sounds cleaner and more balanced. But it had some flaws i need
to correct and test.
"Maybe this is due to its damping factor."
Yes, i believe also that high dampingfactor gives this controlled deep
bass. BTW, what's the exact formula for calculating dampingfactor ?
I only checked the voltagedrop between load and noload. I can also
calculate the outputimpedance from this. This gave 0.3mohms.
But the ratio to 8ohms would be 26666. This can't be dampingfactor,
do i need to divide by 100 ?
"That means I lack of matched Pfets."
Hmm, that's a pity !
"Maybe even it is small it will give good balance"
Small ? The speakers have 40liters ! Or did you mean the outputpower
from my amp ?
Mike
lumanauw said:
Hi lumanauw !
I checked the dampingfactor precisely now. (in sim) This results into
a dampingfactor of 4300 into 4 ohms. I think this is far enough for any
kind of weird load...
Need to do some real world checks, but i am lacking equipment.
hi thanh !
I did some more checkings on the asymamp i posted here, it's kind
of useless. When clipping to + , it completely breaks down, seems
that currentmirror is not too easy. I could remove the buffer, but
this is bad for distortions. BTW, it's nearly identical to the blameless,
not so blameless at all... I would never build an amp showing these
symptoms !
So i will continue finetuning my symamp first.
Mike
Hi thanh,
yes i did, and i replied 2 times to your email, ??? Is this yahoo-adress
wrong ?
Okay, what i wrote, the posted circuit was an older one, i added a
buffer before vas since then, giving following results: (20khz)
old amp:
Inputvoltage: 0.3v
Outputvoltage: 13.5v into 4 ohms
2nd: 1.75mV
3rd: 424uV
Amp with buffer: (symamp2.gif)
Inputvoltage: 0.3v
Outputvoltage: 13.5v into 4 ohms
2nd: 128uV
3rd: 121uV
Sorry, had to zip the gif, was too big...
BTW, this amount of distortion is not really bad, the difference
in thd was not really audible, only trebles got slightly better.
There are very good sounding amps with much higher distortions...
Have you replaced the mjl3281/1302 models in spice ? This is a must,
the original mods from onsemi are bad.
Look at the outputstagedistortionthread, there you can get corrected
models.
Mike
yes i did, and i replied 2 times to your email, ??? Is this yahoo-adress
wrong ?
Okay, what i wrote, the posted circuit was an older one, i added a
buffer before vas since then, giving following results: (20khz)
old amp:
Inputvoltage: 0.3v
Outputvoltage: 13.5v into 4 ohms
2nd: 1.75mV
3rd: 424uV
Amp with buffer: (symamp2.gif)
Inputvoltage: 0.3v
Outputvoltage: 13.5v into 4 ohms
2nd: 128uV
3rd: 121uV
Sorry, had to zip the gif, was too big...
BTW, this amount of distortion is not really bad, the difference
in thd was not really audible, only trebles got slightly better.
There are very good sounding amps with much higher distortions...
Have you replaced the mjl3281/1302 models in spice ? This is a must,
the original mods from onsemi are bad.
Look at the outputstagedistortionthread, there you can get corrected
models.
Mike
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.