In any case this improvement is according to this schematic.Super pair is still alive......up to now.
Is it time killing them?
Opinions please.
Attachments
Last edited:
Try it , please. Exactly as I posted.
You will get saturation instead of sticking.
You still might be getting a minute amount of 19mhz oscillation (super-pair)
even as it is not prominent (easily visible). I actually used the tian for
this one. The slewmaster OPS's actually need a little VAS capacitance at the
pd+/Nd- to remain stable.
This would affect the audible quality of the amp.
Some built the original sansui (no super) and said it had a special character.
If you do (and it does) I will re - layout the Kypton v for release with
the eyesee and EFA.
The new ones WILL NOT oscillate at 19 or any other Mhz ! 😡
OS
H bridge is truly excellent !
It clips like the CFA kypton. (self clamping ??)
Thimios will have a good showing with this one. It will even
run WITHOUT compensation.
truly a superior "leach" type VFA. BUT , it has 640V/us slew !
No wonder guru Curl likes this topology.
Toner for this - tomorrow. 🙂
OS
It clips like the CFA kypton. (self clamping ??)
Thimios will have a good showing with this one. It will even
run WITHOUT compensation.
truly a superior "leach" type VFA. BUT , it has 640V/us slew !
No wonder guru Curl likes this topology.
Toner for this - tomorrow. 🙂
OS
Attachments
How hard is the Eyesee/Alexander opamp loaded.?? I mean just as an experiment try to download to 100 ohm and listen to different op-amps, most sound awful when that is done, but here I am a little in doubt just how low the load really is as it's also part of the FB network and thus is the node driven from two sides, so most of the needed current may be from the output
Last edited:
Eyesee is eeezzzz , isn't it ?
112R is the resistive load. But being at the current feedback node with
just a small voltage there ..... Basically , just 100's of uA per device(below).
(servo barely needs 10's of uA to correct offset)
We are not listening to the output - but to the collectors of the IC's
internal output stage.
OS
112R is the resistive load. But being at the current feedback node with
just a small voltage there ..... Basically , just 100's of uA per device(below).
(servo barely needs 10's of uA to correct offset)
We are not listening to the output - but to the collectors of the IC's
internal output stage.
OS
Attachments
😎
I forgot to mention - R9-11 might be bigger (1W) - this is actually
a true CFA . 😀
OS
Last edited:
kypton-v
No super pair.
First clipping picture C6=5P6 ,C15=3p9
Sec. C6=12pf,C15=3p9
Third C6= 12pf, C15 unpopulated
No super pair.
First clipping picture C6=5P6 ,C15=3p9
Sec. C6=12pf,C15=3p9
Third C6= 12pf, C15 unpopulated
Attachments
Last edited:
Now , the sticking might be discouraging. 😱 But listen to it ... there
will be none of the 19mhz (super-pair) garbage.
Try a square wave test to see if there is any parasitic oscillations at the
rising/falling edges .
To deal with the "sticking" ,I will make a baker clamp for the VAS.
PS - I know what is happening ... it is sticking after coming out of saturation
LT predicted this. BUT - no X mhz BS, better SW , and WAY better THD/bandwidth.
Edit - there is more to an amp than clipping behavior. That can be easily dealt with.
The "tradeoff" will be in audible quality - actually no tradeoff ,after it's "clamped".
OS
Last edited:
Hi OS from first square wave test i can't see any parasitic oscillation.
I will try a full square wave test, may be this afternoon, and i will post all results again.
Thanks.
Try to connect two antiparalell connected 1N4148 diodes s between Q1/Q2 colectors, the same for Q5/Q6 and Q7/Q8.
