Mike, Ich sprache nichst Englander..I do not speak english...so, have to make some
adaptations of my words.
Metalic when metal instruments are played...hitting metal, sound is metalic...grumpf!....grrrrr.
Or, explaining better; Metalic spoken as the quality, founded in some good amplifiers, that make them reproduce metalic sounds correctly, with all their harmonics, this reproduction makes them reproduce as bell as a real bell playing, not a bell sounding alike a plate beeing hitted.
hummm...think about that Miky...have to make some interpretation...some decodification... a gift for ya!
Uncle Charlie.
adaptations of my words.
Metalic when metal instruments are played...hitting metal, sound is metalic...grumpf!....grrrrr.
Or, explaining better; Metalic spoken as the quality, founded in some good amplifiers, that make them reproduce metalic sounds correctly, with all their harmonics, this reproduction makes them reproduce as bell as a real bell playing, not a bell sounding alike a plate beeing hitted.
hummm...think about that Miky...have to make some interpretation...some decodification... a gift for ya!
Uncle Charlie.
Attachments
Hi Mike,
Good design, very like the 1972 Hitachi AN double diff amp.
Point of difference is the cascoded drone side of the second diff pair; the Vbe multiplier, and the charge suckout on the output stage.
I'm a little concerned that you have only between four and five mA on the VAS stage, which is only 2.5mA at most driving the double emitter follower. Is this enough? You have measured 133 watts from the amp and strong imaging performance, which means phase shift is minimal, but is this drive current sufficient for current hungry loads?
Aside from that, and your conventional use of CCS and mirrors, it's a competent, well refined circuit and I give you credit. BUT, it's pretty standard fare, which in view of your earlier emitter drive for the VAS really surprised me!
Cheers,
Hugh
Good design, very like the 1972 Hitachi AN double diff amp.
Point of difference is the cascoded drone side of the second diff pair; the Vbe multiplier, and the charge suckout on the output stage.
I'm a little concerned that you have only between four and five mA on the VAS stage, which is only 2.5mA at most driving the double emitter follower. Is this enough? You have measured 133 watts from the amp and strong imaging performance, which means phase shift is minimal, but is this drive current sufficient for current hungry loads?
Aside from that, and your conventional use of CCS and mirrors, it's a competent, well refined circuit and I give you credit. BUT, it's pretty standard fare, which in view of your earlier emitter drive for the VAS really surprised me!
Cheers,
Hugh
Hi Hugh !
Thanks for the compliments,
I use this standard ccs to be independent from supplyvoltage,
and practical testings showed me that this 2bjt-ccs is the most
stable for PSRR.
I was also surprised that a low current vas is able to drive a
doubledarlington, even with these rloads and 660pf to gnd.
I have big 4ohm vented speakers, beeing quite hungry and the
syamasym plays effortless at high levels with these speakers.
It's correct that the idlebias of the vas is 2.8ma, but because of
the symetric nature of the vas this current goes up to 5.6ma
when swinging. Because of some interesting ft-effects in 2nd
diffamp this current peaks to 15ma when reaching slewratelimit.
So i have the advantage of using fast and linear smallsignal-bjts
for the VAS.
Yes, the amp is pretty standard topology, but outperformed my
other constructions, looks like old prooven concepts work best ?
The really difficult part was getting this gainmonster stable without
any overshooting or ringing...
Finetuning the feedbackcompensation seems to be the key to
good sounding amp.
The audible difference between symasym4 and 5 is not subtile,
but both are stable.
The phasehift of this amp is not very low, but very constant,
enforced by the "big" caps to gnd in the vas.
This seems to hide dynamic phasehifts, which might be the reason
for some amps to have poor soundstage ?
Mike
Thanks for the compliments,
I use this standard ccs to be independent from supplyvoltage,
and practical testings showed me that this 2bjt-ccs is the most
stable for PSRR.
I was also surprised that a low current vas is able to drive a
doubledarlington, even with these rloads and 660pf to gnd.
I have big 4ohm vented speakers, beeing quite hungry and the
syamasym plays effortless at high levels with these speakers.
It's correct that the idlebias of the vas is 2.8ma, but because of
the symetric nature of the vas this current goes up to 5.6ma
when swinging. Because of some interesting ft-effects in 2nd
diffamp this current peaks to 15ma when reaching slewratelimit.
So i have the advantage of using fast and linear smallsignal-bjts
for the VAS.
Yes, the amp is pretty standard topology, but outperformed my
other constructions, looks like old prooven concepts work best ?
The really difficult part was getting this gainmonster stable without
any overshooting or ringing...
Finetuning the feedbackcompensation seems to be the key to
good sounding amp.
The audible difference between symasym4 and 5 is not subtile,
but both are stable.
The phasehift of this amp is not very low, but very constant,
enforced by the "big" caps to gnd in the vas.
This seems to hide dynamic phasehifts, which might be the reason
for some amps to have poor soundstage ?
Mike
Thank you Michael,
I will think hard on your comments. You have mixed in a lot of careful calculation and measurement; again, I give you full credit.
I agree that capacitor choice and dimensioning have the most profound effects in audio amplifier design. Of course, there are minor differences in designer style, but I recognise this as an excellent amplifier. You might now like to try voicing it for something really musical - an art, rather than a science - this is a wonderful approach that Carlos will help you with as well!
Thank you for sharing,
Hugh
I will think hard on your comments. You have mixed in a lot of careful calculation and measurement; again, I give you full credit.
I agree that capacitor choice and dimensioning have the most profound effects in audio amplifier design. Of course, there are minor differences in designer style, but I recognise this as an excellent amplifier. You might now like to try voicing it for something really musical - an art, rather than a science - this is a wonderful approach that Carlos will help you with as well!
Thank you for sharing,
Hugh
Hi thanh !
Yes, we have met long time ago...
Yes, i am a programmer, the last days i hate it...
My TIM calculation is not very precise, that's why i've put a "~".
I mixed a 4khz and a 7khz signal (50:50), both having 6v swing
at output. In fft i've checked the frequencies, the peaks at 3khz
and 11khz are the IMD, all other freqs were much lower.
the 2 peaks were 150uv. 150uv/6v -> 0.0025%.
This is for sure not an accurate method...
This amp is easy to build, the transistors are matched by pairs,
the 2 mpsa18, the 2 2n5401 in 2nd diffamp, and the 2 2n5551
in currentmirror. Matching between 2n5401 and 2n5551 is not
necessary. Layout is critical because of the high gain, inputdiffamp
must use exactly the same gnd-level for input and feedback,
or will get massive hum.
For first powering up, i set the pot for bias to middle, and then
adjust the quiscent current to 50ma.
Mike
Yes, we have met long time ago...
Yes, i am a programmer, the last days i hate it...
My TIM calculation is not very precise, that's why i've put a "~".
I mixed a 4khz and a 7khz signal (50:50), both having 6v swing
at output. In fft i've checked the frequencies, the peaks at 3khz
and 11khz are the IMD, all other freqs were much lower.
the 2 peaks were 150uv. 150uv/6v -> 0.0025%.
This is for sure not an accurate method...
This amp is easy to build, the transistors are matched by pairs,
the 2 mpsa18, the 2 2n5401 in 2nd diffamp, and the 2 2n5551
in currentmirror. Matching between 2n5401 and 2n5551 is not
necessary. Layout is critical because of the high gain, inputdiffamp
must use exactly the same gnd-level for input and feedback,
or will get massive hum.
For first powering up, i set the pot for bias to middle, and then
adjust the quiscent current to 50ma.
Mike
Mike, having some graphics, extracted from your Simulator, publish them please.
If you have not time enougth, send them to me, so i will publish for you.
Of course Michael, if you think that can be good to our forum friends to have those informs.
And if you have time...this is more difficult, in you actual situation, facing Microsoft Bureaucracy and deadlines.
regards,
Carlos
If you have not time enougth, send them to me, so i will publish for you.
Of course Michael, if you think that can be good to our forum friends to have those informs.
And if you have time...this is more difficult, in you actual situation, facing Microsoft Bureaucracy and deadlines.
regards,
Carlos
I am anxious to hear how this one sounds compared to the other. The folded cascode is faster. I don't see any compensation though. Have you ever considered using lateral mosfets as drivers. It elliminates a stage, and temperature compensates the outputs as well, if you choose the drain current carefully.
J.
J.
more details please Carlos
Carlos: can you elaborarte further on your previous comment. I read this and I am not certain what you mean. Thanks Moray James.
Quote from Carlos
"Also has a magnetic coupling between the output coil with the input..... always near the oscilation.....near the threshold."
Carlos: can you elaborarte further on your previous comment. I read this and I am not certain what you mean. Thanks Moray James.
Quote from Carlos
"Also has a magnetic coupling between the output coil with the input..... always near the oscilation.....near the threshold."