Achim Dassow has simulated the K2-R and it's successor, the K2-SR, which as designed by the late Bruce Seddon.
Achim points out that the gain is very similar at 2.5kHz, but much higher at DC, so he concludes that 60Hz rejection is much better for the K2-SR, but the step response should be similar between the K2-R an the K2-SR.
As simulation results show, the K2-R, and the K2-SR have comparable gain-bandwidth product, but the elegant conductance subtraction achieved at the output with positive feedback, increases the DC gain of 64dB for the K2-R to 90dB, for the K2-SR. The higher gain of the K2-R was achieved with the exclusive use of hi mu=100 12AX7 triodes, which have one order of magnitude less intrinsic gain than the 6AN8 (mu=1400) pentodes in the lower gain K2-R.
Bruce Seddon designed the last and most enduring line of tube power supplies produced by Philbrick. These were the R-xxx series. He also improved the earlier SR-xxx power supplies by redesigning the K2-R into the K2-SR.
Bruce systematically replaced pentodes of existing designs with triodes in his improved or new designs. Bruce realized that the pentode could not realize it's high intrinsic gains in excess of 1000 because current source loads were needed. So pentodes were usually used with modest gain, as seen in the K2-R because of the necessarily low impedance resistive loads, to moderately high 300V supplies. But pentodes, running with fixed screen voltages always showed the full 1.5 power law (I =K*V^1.5) non-linearity of the thermionic diode equation, which made the low gain situation worse. The equivalent of the pentode screen in a triode is the plate and, if the plate is allowed to swing under a high impedance load, very little of the power-law of the triode transconductance is realized, and the voltage transfer function from control grid to plate becomes very linear. With triodes, like the 12AX7 and 12AU7, which have a very constant intrinsic gain mu of 100 and 17, respectively, it is possible to boost the tube voltage gain with positive feedback, which cancels out 90% of the output tube conductance, boosting the intrinsic triode gain of the 12AX7 from 100 to 1000, but with a real resistive load aided localized positive feedback.
Output (plate) conductance cancellation schemes were used in much of the Philbrick tube OPAMP line, and are the norm in modern IC opamp design, because DC open loop gain (AVol) gets increased without a loss of Gain Bandwidth product (GBW) or high frequency stability. Reduced AC stability usually results in designs that employ additional stages for DC gain.
-Joe Sousa (webmaster)
A few excerpts from email exchanges about the K2-S and K2-SR with Bob Pease and Achim Dassow:
In the meantime i've done some investigations with the K2-SR design, which offers a lot more gain when doing simulation. But at 1000Hz the gain is hardly in excess of the K2-R gain (only 2500).
So hum regulation should be excellent with the K2-SR, but not necessarily the step response or transient response.
*** I think you are right about that. The 5910 was a lot better, faster. It would hold nearly a gain of 1 million out to 100 Hz.
Date: Wed, 27 Aug 2008 21:03:59 +0200
If you didn't recognize yet, i'll have to point to a mistake i've made:
In the K2-SR Simulation i explained that the Input Value of the Stimulating Signal Source represents only Vpk but not Vpk-pk. So the calculation of the resulting gain factor has to be corrected. Gain should be 50000 instead of 100000. I'll send you a new screen plot where the values are correct.
Unfortunately the SwCadIII makes only a Vpk Voltage of the entered Value although i think, one needs more the Vpp Value to be used. If you are using the cursor measurement at SwCadIII you also would like more to make Vpp instead of Vp measurements.
Achim Dassow contributed the following simulation results, on which my discussion above is based (8-26-8).