Feedback fun with K5881 - Re-visited!

[John Caswell]

Hi David,
You are correct in saying the 2K2 resistor has a dual function, but you are still under the impression that the AC signal is returned to the optx when in fact it is the other way around, as I have said before, so your idea is not a one good, as you will then have a long path from the first stage to star earth via the output transformer, along with any currents that are built up in the secondary being impressed upon the feedback and on the cathode currents.
It is far, far better to take the front end earth return as short and as direct to the star earth and leave the feedback return through a screened lead earthed only at the send ie transformer end. This keeps input and output currents separate and avoids "common impedance coupling" (see) This is a bugbear in valve amps and a major bugbear in SS amps where the input current may be µamps but the output current in amps and both may be run along a common cable. The large current then modulates the small one and the net effect is unwanted positive feedback with often dire results in SS amps (smoke fire etc).

John

[Richard]

Hi David,

Originally Posted by barnesdavid20

Hi Richard, Going back to the feedback mod on the K5881 mk2, I have just noticed that it looks like the modulated signal through the 1st valve now goes via the 2.2k resistor in the feedback wire and then through the secondary of the output TX and then to star earth, where as before it went directly to star earth.

No it's shorted to ground via the cathode bypass cap same as before

As this is the most sensitive part of the signal,would this additional routing introduce a higher noise level.

same as before

Also if you connect the screen on the feedback wire at both ends ie to negative on the output TX secondary and to the negative on the cathode bypass cap on the first valve, could you just have the one wire from the output TX secondary going to star earth and do away with the separate earth return wire from the first stage.

Why, you would be daisy-chaining the first stage ground onto the TX secondary ground and speaker lead ground along with any outside noise. I would definitely keep them separate with their own direct paths to ground which is the essense of star grounding so one isn't modulating or introducing noise to another.

David

[Richard]

Ah, morning John!

[barnesdavid20]

THank you both for explaining it to me, you must think that I am stupid.
I did not realise that the modulated part of the DC passing through the valve got back to star earth via the cathode bypass cap, I thought it went via the 2.2k resistor, hence my concern about noise pick up. it also now makes sense about leaving the Ist stage earth return as is.
Thank you both for your replies.
David

[barnesdavid20]

Something that I have done in that area, which reduced the mv across the output TX significantly, was to connect the screened feedback wire at both ends and do away with the star earth wire from the negative of the OP TX secondary.
From a safety point of view, the secondary is still connected to earth via a substantial screen and the earth return from the first stage.This change definitely reduced the noise level in my amp. Hence 0.6mv on each channel across the LS. Hope it is ok to do this.
David

[barnesdavid20]

Hi Richard, after some time I have just noticed having got hold of the latest circuit diagram for the KT88, that the feedback circuit seems to have the same configuration as the K5881. In your opening thread you said that the KT88 had the new feedback circuit incorporated, as the KT88 is the latest circuit design from WD, does it mean that the original feedback circuit as used on the K5881 is preferred.
Cheers
David

[Richard]

Interesting observation David. Yes KIT88 had the AG feedback arrangement and WD88VA has the more usual arrangement seen on Mullard 5-20.

Perhaps, rather than looking at a collection of design features, (for instance one is pentode the other UL etc) it may be more important to consider the complete design. Bottom line for me would be if you prefer the AG method then stick with it.

The feedback method differs iirc by rolling off the bass more quickly from the chosen low frequency point. I wonder whether it could have been used to advantage in WD88VA to prevent what has been reported as a LF hump from 7Hz downwards? From his various articles AG has always seemed conscious of not saturating the OPTXs and really a flat response down to 1Hz is not needed or desirable.

It would be interesting to try the AG fb on WD88VA to roll off the bass quickly below 20Hz and see if it sounds and measures better. Ref;

http://www.world-designs.co.uk/forum...ead.php?t=6706

http://www.world-designs.co.uk/forum...ead.php?t=6647

[barnesdavid20]

Hi Richard, thanks for reply, maybe someone will try it, I can only say that I would never go back to the original feedback circuit, my K5881 sounds superb.
David