They are on the wayback machine web archive, but in Japanese only? https://web.archive.org/web/20130502175118/http://ayumi.cava.jp/audio/ULKF.pdf
As we are examining this matter of where/how NFB is applied in an amplifier, I perhaps need to remind of another matter - said before but I cannot locate the thread now.
In comparing UL feedback with Schade feedback, one must not lose sight that it is in a sense oranges vs. apples. The pentode in the Schade circuit is not the same as the 'pentode' in the UL circuit.
The latter 'pentode' is no longer a classic pentode since the 'non-linear' screen feedback changes the tube (internal) characteriatics before we go anywhere else. As we know UL makes it a more 'triodish' tube than a straight pentode. Thus with Schade feedback one is linearly decreasing the distortion etc. of a pentode - the lowered/improved characteristics are still in essence a better pentode. The amount of feedback may be the same, but a UL stage with feedback might still be preferable to a straight pentode stage with an equal amount of NFB.
In comparing UL feedback with Schade feedback, one must not lose sight that it is in a sense oranges vs. apples. The pentode in the Schade circuit is not the same as the 'pentode' in the UL circuit.
The latter 'pentode' is no longer a classic pentode since the 'non-linear' screen feedback changes the tube (internal) characteriatics before we go anywhere else. As we know UL makes it a more 'triodish' tube than a straight pentode. Thus with Schade feedback one is linearly decreasing the distortion etc. of a pentode - the lowered/improved characteristics are still in essence a better pentode. The amount of feedback may be the same, but a UL stage with feedback might still be preferable to a straight pentode stage with an equal amount of NFB.
Last edited:
50AE,
M.... not necessarily.
Just to recap: Firstly, to begin with, one must realise that CFB is UL. That portion of the OPT primary between screen and B+ in UL, transferred to the cathode side in the equivalent diagram, still retains the same UL topology (cathode current flowing in the winding). Taking one thing a a time, then with CFB the 'internal' pentode sees the same conditions as in UL. The characteristics are no longer that of a classic pentode, but the well-known 'somewhere-between-pentode-and-triode' type of mode (although I hate that simplified description. The characteristics are not kind of midway between pentode and triode. They are a new set of characteristics.)
Next we come to the fact that the CFB arrangement also gives feedback just as an unbypassed cathode resistor would have done. The third point then regards that, for classic UL with say 43% taps, this leads to a rather high signal of low distortion having to be available to the final stages.
To overcome this the cathode feedback is reduced, but to still want UL as per 43% taps, one then needs to have a screen tap for the remainder of the UL requirement. Say one allows 20% cathode windings, another 23% remains. That is then catered for by 43 - 20 = 23% screen taps on the remainder of the OPT primary.
It then follows that the next consideration is, what signal amplitude can be supplied to he power stage with the necessary low distortion. Optimality would require a lot of mathematical spadework and testing, and then one still does not know the spread in tubes on the market.
In my own case (4 x p.p.p.6L6GC) I settled for 25% cathode windings, with no provision for further screen taps. The task of having to provide some 380Vpp signal from either driver stage anode was taxing. I used an E180CC twin triode power triode as driver.
Whether I made the right choices . . . "The proof of the pudding ..." - and the eating has yet to be done!
M.... not necessarily.
Just to recap: Firstly, to begin with, one must realise that CFB is UL. That portion of the OPT primary between screen and B+ in UL, transferred to the cathode side in the equivalent diagram, still retains the same UL topology (cathode current flowing in the winding). Taking one thing a a time, then with CFB the 'internal' pentode sees the same conditions as in UL. The characteristics are no longer that of a classic pentode, but the well-known 'somewhere-between-pentode-and-triode' type of mode (although I hate that simplified description. The characteristics are not kind of midway between pentode and triode. They are a new set of characteristics.)
Next we come to the fact that the CFB arrangement also gives feedback just as an unbypassed cathode resistor would have done. The third point then regards that, for classic UL with say 43% taps, this leads to a rather high signal of low distortion having to be available to the final stages.
To overcome this the cathode feedback is reduced, but to still want UL as per 43% taps, one then needs to have a screen tap for the remainder of the UL requirement. Say one allows 20% cathode windings, another 23% remains. That is then catered for by 43 - 20 = 23% screen taps on the remainder of the OPT primary.
It then follows that the next consideration is, what signal amplitude can be supplied to he power stage with the necessary low distortion. Optimality would require a lot of mathematical spadework and testing, and then one still does not know the spread in tubes on the market.
In my own case (4 x p.p.p.6L6GC) I settled for 25% cathode windings, with no provision for further screen taps. The task of having to provide some 380Vpp signal from either driver stage anode was taxing. I used an E180CC twin triode power triode as driver.
Whether I made the right choices . . . "The proof of the pudding ..." - and the eating has yet to be done!
Last edited:
Not sure that I follow the last two posts viz-a-viz UL .... ?
When the screen is connected in such a way that the cathode and screen are always at the same a.c. potential, yes, that will give normal pentode action, irrespective of what goes on in the outside world. In other words, one can have cathode feedback with a pentode, with a triode, or with UL topology connection. For UL operation the screen always needs be at an a.c. point somewhere between anode and cathode, irrespective of what goes on in the 'outside'world'.
When the screen is connected in such a way that the cathode and screen are always at the same a.c. potential, yes, that will give normal pentode action, irrespective of what goes on in the outside world. In other words, one can have cathode feedback with a pentode, with a triode, or with UL topology connection. For UL operation the screen always needs be at an a.c. point somewhere between anode and cathode, irrespective of what goes on in the 'outside'world'.
I do have technician who can rewire my trafo. I have now 50-50UL/triode setup, all time stay with UL. Very lean sound with triode conection. So I gonna ask him to do 35+8+7 secondaries, to have many options like 65-35UL, 57-42UL, 50-50UL, 42UL+8CFB, 43UL+7CFB, 35UL+15CFB. After all I will post my impressions here. Now I am looking on how LTP works.
Last edited:
So if CF is essentially a form of UL operation I would expect that if one wanted something akin to the various SE UL+Schade designs but had either an OPT without UL tap or a pentode with low screen voltage rating one possible approach would be to use the OPT secondary as CFB and combine with Schade. Obviously there would be less UL effect than a tapped tranny would provide but might it be enough to be useful?
Right. To preserve it's transconductance and linearity, then apply cathode feedback to make it more linear and decrease output resistance. Otto Schade in his famous article demonstrated it well. However, somebody called only resistive feedback from anode to control grid "Schade", while it is simplification and profanation. Schade meant just a local feedback to the control grid, and the example in his article was about application of the feedback from anode to the grid through an interstage transformer's secondary. Cathode feedback is just another way to skin the cat.
Exactly.
So combining plate to grid with CF would not be an unreasonable approach?
A related question regarding the use of the OPT secondary for CF. In so doing some of the isolation of the speaker from the output stage is lost. Is there a concievable failure mode that might put the speaker at risk with such an arrangement?
A related question regarding the use of the OPT secondary for CF. In so doing some of the isolation of the speaker from the output stage is lost. Is there a concievable failure mode that might put the speaker at risk with such an arrangement?
Not necessary, you can find a reason. Like you can play such a way with current and voltage drive requirements from the driver. Also, cathode feedback would include the transformer in the loop.
Do you mean isolation by DC? You can calculate voltage drop on the secondary caused by cathode current, it would be absolutely insignificant.