[patrick-turner]

On Friday, 9 November 2012 19:07:56 UTC+11, Alex Pogossov wrote:
"John Byrns" wrote in message ... RCA published a paper describing the 6L6 beam power tube and its development in the Proceedings of the IRE and also the RCA Review back in the mid 1930s. Besides describing the development of the 6L6, they showed the op amp like connection you are speaking of and what the characteristic curves of the 6L6 look like when this connection is used, basically triode like IIRC. I don't think the connection was very common although I believe I have seen a couple of amps that used it. Much more common is a variant on the connection where the resistor from the output tube plate is feedback to the driver tube cathode, I have seen this connection quite a bit over the years.. Interesting to calculate will such feedback be deeper tham the op-amp connection? Possibly, if Rfb is relatively small (and here you start loosing AF power) and driver plate load resistor is large. As J.L.Stewart mentioned, a pentode in the driver will be an advantage. The problem with either of these schemes is that it destroys the pentodes inherently good power supply rejection. That means that we either need to provide greater power supply filtering to reduce the noise on the power supply line, or go to parallel feed to improve the power supply noise rejection, either solution adds to the cost. Yes, but all 300B / 2A3 lovers are bound to face the same issue. If I were doing this and was free to spec the output transformer I wanted I would go with the cathode feedback scheme in the output stage. Taking the feedback from the plate doesn't improve the low frequency stability issues, it only helps with high frequency stability, so what I would do if I couldn't spec the output transformer I wanted, would be to take the negative feedback from the secondary at low frequencies and from the plate/primary at high frequencies. I did the two branch feedback exactly in the same manner. One solution is just to throw a small capacitor (10pF) from the plate to grid. It will work as a compensation cap in the op-amp, creating a dominant HF pole. In other words it makes 6V6 work as integrator at HF. Later I realised that far better is to throw a larger cap from 6V6 plate to unbypassed cathode of the previous driver stage. (To the same cathode a second resistive (regular) NFB is fed from the OPT secondary.) Regards, Alex

I've tried caps between OP anodes and g1 or to driver cathodes, but usually it just causes more open loop phase shift of 90 degrees at a lower F than do the existing amounts of C shunting Ra in the input-driver amp stages. So with the ultimate OLG phase shift of 180d occuring at a lower F the amp becomes harder to stabilise with other networks with a pure C load. The C you put there adds a high Miller effect where it was smaller to begin with. But perhaps its possible to use an R&C zobel to shelve the HF that way to avoid the 180d shift at such a low F. The R&C zobel is best as a passive network across V1 anode to 0V, ie, it shunts Ra, and this works better IMHO to reduce ringing on C loads and keep away from FB C having to load down any driver stages.

I'll do it your way after I have seen your many schematics which show/prove your method to be superior. When I test many amps it is revealing to set them up with a 0.22uF cap load and with 5kHz square wave at the -12dB level. One will often see some ringing at Vo which is to be expected. But if you examine the anodes of OP tubes, the ringing is often a lot worse than at Vo.. Just what is done to reduce the ringing is debateable, but I recall having tried every possible way without reducing the widest full power BW with pure R load but while making the amp unconditionally stable with an possible load or none at all. Whay appears in schematics at my site is a distilation of ideas that worked, and priciples that can be applied to nearly all amps without wasting too much of my precious time.

Patrick Turner.