apexaudio
aparatusonitus
Thank You very much. I do not have green leds in stock, I have only 4,7V and 5.1V zeners, but I will try 1n4148 too. I will try the modyfications and report back tomorrow.
regards Peter
Use 5x1N4148 in series instead LEDs or zeners,
Regards
That was probably the Sansui marketing department who called it so, the name appears here in 1979 :
Ralisation d'un amplificateur classe A de 20W
Historically interesting, in the same article, there is the schematics of one of the first CFA of the word, the Hiraga's class A power amp, which input circuit configuration is now known as "diamond".
Amp which input circuit configuration use buffer with common collector complementary pair (diamond buffer) named: 'diamond amp'
Amp with emiter connected full complementary differential circuit in second stage named: 'diamond drive'.
Sansui use non symmetrical j fet input in 'Sansui Diamond Drive' and I use symetrical full complementary dual differential input circuit named 'Apex Diamond Drive' 🙂
Been following this as well as this http://www.diyaudio.com/forums/solid-state/243842-gla-good-little-amp-survivor-8.html. Good progress 🙂
apexaudio
I have checked:
x50 - front stage
x20 - second stage
- VAS - single transistor
Above configuration was stable, no oscilations but clipping above 4k with sine (with square was ok) was very bad.
The enhanced cascode has had better performance on measurements so I am going to test it on monday. Bellow corrected test board an schematic (gains will be checked at real amp to get best performance).
Been following this as well as this http://www.diyaudio.com/forums/solid-state/243842-gla-good-little-amp-survivor-8.html. Good progress 🙂
I use similar topology (pure push-pull vas) in AX11, it can be upgrade with j-fet input stage like in Sansui AU317
Attachments
Quad Complementary Pair Transistor MPQ6700 can be used in input stage if available.
Attachments
Last edited:
'ostripper' suggestions:
"I don't say it won't work , you just don't need the miller comp. at cascoded
VAS.
I built a luxman cascoded VAS with double green diodes referenced
from rail to rail. It "sang" very nicely , but with more difficult loads - an EF2
won't do.
Using a EF2 with lateral FET's or an EF3 would allow a VAS to "sing"
freely with little loading.
On the subject of intentional loading (R23-24 on the GLA) ,this preloads
the VAS, making an external load (such as the output stage) less
of a factor over a wider range. It is also a convenient way to "trim" the
loop gain of the entire amplifier. (lower H3/5 as well.)
OS"
_____
"I don't say it won't work , you just don't need the miller comp. at cascoded
VAS.
I built a luxman cascoded VAS with double green diodes referenced
from rail to rail. It "sang" very nicely , but with more difficult loads - an EF2
won't do.
Using a EF2 with lateral FET's or an EF3 would allow a VAS to "sing"
freely with little loading.
On the subject of intentional loading (R23-24 on the GLA) ,this preloads
the VAS, making an external load (such as the output stage) less
of a factor over a wider range. It is also a convenient way to "trim" the
loop gain of the entire amplifier. (lower H3/5 as well.)
OS"
_____
If compare GLA (Sansui Z3900) and this HD50 in progress amps are similar only HD50 is symmetrical, amps use 2 bjt differential gain stages with local NFB. Sansui Z3900 use gain 39X in 1 stage and 8,2X in 2 stage, this give good stability with almost any usual transistors in DIY.Been following this as well as this http://www.diyaudio.com/forums/solid-state/243842-gla-good-little-amp-survivor-8.html. It would be nice to have someone compare these amps
Attachments
Last edited:
Another day of testing. I must say that it is not easy to stabilise fully the amp.
Bellow:
-1st stage x22
-2nd stage x4.7
-vas - standard cascode with zener 5.1 (bypassed with 4,7uF)
-feedback cap 33pF
What I like:
-sound of it
-the sound comming out of it
-sound
-and sound
-apex dd drive
What I do not like so far:
-clipping with siuns signal over 4kHz (square looks fine)
-I can not use the biggest advantege of it - ability to get very high gain at first stages (If multiplied gain of 2 first stages goes close to 200 it is hard to stabilise - for me).
Bellow:
-1st stage x22
-2nd stage x4.7
-vas - standard cascode with zener 5.1 (bypassed with 4,7uF)
-feedback cap 33pF
What I like:
-sound of it
-the sound comming out of it
-sound
-and sound
-apex dd drive
What I do not like so far:
-clipping with siuns signal over 4kHz (square looks fine)
-I can not use the biggest advantege of it - ability to get very high gain at first stages (If multiplied gain of 2 first stages goes close to 200 it is hard to stabilise - for me).
Attachments
-
y2qulb10u808ch4ke7fq_1384242802.jpg121.3 KB · Views: 98
-
oer5h7c5lg3d7tqjgf7b_1384242800.jpg124 KB · Views: 90
-
e9ubtakzfi14biomkgd4_1384242804.jpg125.5 KB · Views: 91
-
cwpxpamvzcj8x9lxdrly_1384242798.jpg126.8 KB · Views: 122
-
9v851sopjs2vl1e383mb_1384242811.jpg82.1 KB · Views: 104
-
mm29ssgztdb21icz4z7x_1384242809.jpg71.4 KB · Views: 107
-
iyhj7wwh20c7t5futy0d_1384242808.jpg69.7 KB · Views: 479
-
f33ibxsfuywyxgzgga6m_1384242806.jpg69.7 KB · Views: 520
-
ju0m9dfxvjpnl82wixi3_1384242796.jpg148.1 KB · Views: 569
Another day of testing. I must say that it is not easy to stabilise fully the amp.
Bellow:
-1st stage x22
-2nd stage x4.7
-vas - standard cascode with zener 5.1 (bypassed with 4,7uF)
-feedback cap 33pF
What I like:
-sound of it
-the sound comming out of it
-sound
-and sound
-apex dd drive
What I do not like so far:
-clipping with siuns signal over 4kHz (square looks fine)
-I can not use the biggest advantege of it - ability to get very high gain at first stages (If multiplied gain of 2 first stages goes close to 200 it is hard to stabilise - for me).
Thanks for your work on development this amp.
apex
I like chalanges so must work hard to get best out of it. THX
Bellow
gain1 at 47
gain2 at 4.7
compensation cap 33p
I think my sound card is not able to measure distortions anymore.
Perfection need time 🙂
Regards
If compare GLA (Sansui Z3900) and this HD50 in progress amps are similar only HD50 is symmetrical, amps use 2 bjt differential gain stages with local NFB. Sansui Z3900 use gain 39X in 1 stage and 8,2X in 2 stage, this give good stability with almost any usual transistors in DIY.
With symmetrical 2 stage circuit diamond drive VAS directly without add pure push-pull circuit (Q5 on Sansui Z3900 circuit)
borys,
did you try TEF output stage from post #71 ?
Yes ofcourse I have tryed it. It performed not as good as cascoded one, higher thd measures, harder to get it to stabilise but it sounded kind of better (maybe due to different thd spectrum). Another think is that cascode version is keepeng low thd up to very high output amplitude (tef version thd is rising with output amplitude). It looks like the voltage gain there is important.
I had to keep 100pF in feedback with TEF version.
Yes ofcourse I have tryed it. It performed not as good as cascoded one, higher thd measures, harder to get it to stabilise but it sounded kind of better (maybe due to different thd spectrum). Another think is that cascode version is keepeng low thd up to very high output amplitude (tef version thd is rising with output amplitude). It looks like the voltage gain there is important.
I had to keep 100pF in feedback with TEF version.
Did you try cascode VAS and TEF output stage?
Attachments
Last edited:
Mile
My mystake, I did not tryed TEF stage, I have tryed only the basic setup (same as bellow but without Q18 and Q19).
I tryed enhanced cascode and standard one --> I can not see any difference on the measurements. I still have the 1st protototype so will try the TEF this week.
The CFP can be tested too, but for now I do not have much time.
Another small question - can bias the cascode transistors with one resistor ? (R39 R40 --> put one instead and disconnect earth connection )
My mystake, I did not tryed TEF stage, I have tryed only the basic setup (same as bellow but without Q18 and Q19).
I tryed enhanced cascode and standard one --> I can not see any difference on the measurements. I still have the 1st protototype so will try the TEF this week.
The CFP can be tested too, but for now I do not have much time.
Another small question - can bias the cascode transistors with one resistor ? (R39 R40 --> put one instead and disconnect earth connection )
Attachments
Last edited:
Another small question - can bias the cascode transistors with one resistor ? (R39 R40 --> put one instead and disconnect earth connection )
Yes, this don't change anythig.
missing 1000uF cap
where is the 1000uF feedback cap, at your schematic it is missing...?
As I know we can omit it when we using DC-servo 🙂
Regards
Hi Borys,Mile
My mystake, I did not tryed TEF stage, I have tryed only the basic setup (same as bellow but without Q18 and Q19).
I tryed enhanced cascode and standard one --> I can not see any difference on the measurements. I still have the 1st protototype so will try the TEF this week.
The CFP can be tested too, but for now I do not have much time.
Another small question - can bias the cascode transistors with one resistor ? (R39 R40 --> put one instead and disconnect earth connection )
where is the 1000uF feedback cap, at your schematic it is missing...?
As I know we can omit it when we using DC-servo 🙂
Regards
- Home
- Amplifiers
- Solid State
- Diamond Differential Amplifier