12HL7 should make a great driver. It is sort of the 2nd runner up to the more popular and higher gm 12HG7/12GN7 which has been used for a driver at times.
The 6HB6 should work even better for the RH... stuff. They were $1 for quite a while too. So were the 29GK6's which are near equiv. to EL84 (just heater V diff.).
Clearly, Schade is the way to go for outputs. All the fluff around expensive triodes, and then $1 TV beam pentodes in Schade mode blow them away. And then Schade mode even gives the exact same benefits as CFB OT's, without having to wind a custom OT.
Putting even more transconductance into the local Schade loop, by taking the resistive feedbacks back to the preceding driver cathodes should make for obscenely linear outputs. The RCA 50 Watt amplifier schematic in their tube handbook does that, but comments are that it sounds too clinical like a SS amp. I think they effectively designed a low cost Mac substitute.
The internal Mu for pentodes like the 12HL7 refers to the grid1 to grid2 ratio of transconductances, which is what determines the Mu when triode configured. The 12HL7 is around Mu 25 in triode mode. You can make it any Mu you want, up to that "1155", figure in Schade mode, by choosing the Fdbk resistor values.
The 6HB6 should work even better for the RH... stuff. They were $1 for quite a while too. So were the 29GK6's which are near equiv. to EL84 (just heater V diff.).
Clearly, Schade is the way to go for outputs. All the fluff around expensive triodes, and then $1 TV beam pentodes in Schade mode blow them away. And then Schade mode even gives the exact same benefits as CFB OT's, without having to wind a custom OT.
Putting even more transconductance into the local Schade loop, by taking the resistive feedbacks back to the preceding driver cathodes should make for obscenely linear outputs. The RCA 50 Watt amplifier schematic in their tube handbook does that, but comments are that it sounds too clinical like a SS amp. I think they effectively designed a low cost Mac substitute.
The internal Mu for pentodes like the 12HL7 refers to the grid1 to grid2 ratio of transconductances, which is what determines the Mu when triode configured. The 12HL7 is around Mu 25 in triode mode. You can make it any Mu you want, up to that "1155", figure in Schade mode, by choosing the Fdbk resistor values.
Last edited:
Question, you said that internal mu of these pentodes are the grid1 to grid2 ratio of transconductance..........how does this affect output impedance? For example in triode mode is the plate resistance now 25/.021=1190?
I see that in the RCA schematic there are 2 resistors from each 7027 plate, one goes to the plate of the 6CB6, the other resistor to the cathode of the 6CB6. I am having a hard time seeing what is fed back where. I guess my question is if they are taking feedback from the plate of the 7027 to both plate and cathode of the same driver wouldn't one be negative feedback and the other be positive feedback? Sorry for the noob question my brain is having issues today "seeing" the bigger picture.
I see that in the RCA schematic there are 2 resistors from each 7027 plate, one goes to the plate of the 6CB6, the other resistor to the cathode of the 6CB6. I am having a hard time seeing what is fed back where. I guess my question is if they are taking feedback from the plate of the 7027 to both plate and cathode of the same driver wouldn't one be negative feedback and the other be positive feedback? Sorry for the noob question my brain is having issues today "seeing" the bigger picture.
Yes. Mu = gm x Rp and so for triode configuration Rp = 25/.021 = 1190 Ohm
Of course these are sensitive to current, since gm1 varies as approx. the square root of current and gm2 as approx. the cube root. (all in the non saturation region, saturation makes the exponents drop) Variation of gm2 and gm1 differently with current, also makes the Mu change with current slowly.
A resistive path from the 6CB6 plate to it's cathode would constitute positive feedback, but the mid point being the output tube plate (with inverted signal no less) with the load across it should minimise that.
It is puzzling as to just what the RCA design is trying to accomplish primarily. The Schade feedback to the 6CB6 plate kills its gain, which spoils the N Fdbk effect going to its cathode. I can only guess that maybe the N Fdbk to the cathode is being used like a bootstrap to linearize the 6CB6 V to I conversion for the top (plate) Schade process. In which case, this design would just be a tweaked conventional Schade design.
Really high correction would occur if the top Schade connection were removed, and a Gyrator pull-up placed on the 6CB6 plate instead. That would make the cathode Schade N Fdbk path extremely effective, by boosting the 6CB6 gain enormously. Distortion might drop by another 100X if the "local" loop stays stable. (two stages in a loop is not strictly "local" Fdbk, but the OT is not included in the path, which is another criteria for "local")
Some simulation of this circuit with an end to end gain plot versus signal voltage is called for to determine just what is being optimised with the double feedbacks to the 6CB6.
The Citation II also does an odd double feedback to the same driver input point, which doesn't make much sense conventionally (one cannot satisfy both error corrections simultaneously), except as one path is being used as a small distortion tweak to the primary Fdbk path. (Maybe the distortion spectrum was observed, while the feedbacks were adjusted with Pots, until they liked what they saw. Same for the RCA design too.)
Of course these are sensitive to current, since gm1 varies as approx. the square root of current and gm2 as approx. the cube root. (all in the non saturation region, saturation makes the exponents drop) Variation of gm2 and gm1 differently with current, also makes the Mu change with current slowly.
A resistive path from the 6CB6 plate to it's cathode would constitute positive feedback, but the mid point being the output tube plate (with inverted signal no less) with the load across it should minimise that.
It is puzzling as to just what the RCA design is trying to accomplish primarily. The Schade feedback to the 6CB6 plate kills its gain, which spoils the N Fdbk effect going to its cathode. I can only guess that maybe the N Fdbk to the cathode is being used like a bootstrap to linearize the 6CB6 V to I conversion for the top (plate) Schade process. In which case, this design would just be a tweaked conventional Schade design.
Really high correction would occur if the top Schade connection were removed, and a Gyrator pull-up placed on the 6CB6 plate instead. That would make the cathode Schade N Fdbk path extremely effective, by boosting the 6CB6 gain enormously. Distortion might drop by another 100X if the "local" loop stays stable. (two stages in a loop is not strictly "local" Fdbk, but the OT is not included in the path, which is another criteria for "local")
Some simulation of this circuit with an end to end gain plot versus signal voltage is called for to determine just what is being optimised with the double feedbacks to the 6CB6.
The Citation II also does an odd double feedback to the same driver input point, which doesn't make much sense conventionally (one cannot satisfy both error corrections simultaneously), except as one path is being used as a small distortion tweak to the primary Fdbk path. (Maybe the distortion spectrum was observed, while the feedbacks were adjusted with Pots, until they liked what they saw. Same for the RCA design too.)
Last edited:
Agreed. I don't have a 7027 model.........I think it's probably okay to go ahead and just use a 6L6 just for this exercise.
Last edited:
I would think that would be close enough too.
Seeing as this RCA design was published for public use, I am guessing they did not want a design that would be highly sensitive to layout or freq. compensation. So my take is that it was intended as just a simple conventional Schade design (plate to plate N Fdbk), with the extra path to the 6CB6 cathode added as a small (weak) touch-up to the distortion spectra. The instructions say it does 1% distortion, probably low order 3rd Harmonic.
Adjusting the design to be high loop gain through the 6CB6 cathode Fdbk exclusively (without the top plate to plate Schade path), would make it very sensitive to layout capacitance and likely would need a HF freq. rolloff network somewhere in the loop for stability (like maybe around 250 KHz). A differential (CCS tailed) driver stage would make it much more effective at removing class AB crossover distortion (by observing and correcting for the net sum of both outputs). Then it would give the Mac a run for the (now much less) money.
Of course we would be free to put in different (like frame grid) tubes for drivers and the front stage. And high current Sweep tubes with a low primary Z OT for cheap but high performing output duty. One could even put a touch of current feedback(s) from the output tube cathodes (current sense low value R's) back to the driver cathodes to null out the OT resistance effects, giving this a high damping factor just like SS. There is also room for some global Fdbk around the OT as well.
Then listen to all the complaints about it sounding like a SS amp.
The final design version might then include a (mis-) balance pot in the splitter for some 2nd Harmonic, and an adjustable tail resistor somewhere to set the 3rd Harmonic level. The current feedbacks from the output tube cathodes then might allow for inversion and level adjustment to set a finite output Z or damping factor. In other words: "Comfort controls" or more cleverly: speaker distortion nulling controls, prominently noted on the advertising sheet.
Seeing as this RCA design was published for public use, I am guessing they did not want a design that would be highly sensitive to layout or freq. compensation. So my take is that it was intended as just a simple conventional Schade design (plate to plate N Fdbk), with the extra path to the 6CB6 cathode added as a small (weak) touch-up to the distortion spectra. The instructions say it does 1% distortion, probably low order 3rd Harmonic.
Adjusting the design to be high loop gain through the 6CB6 cathode Fdbk exclusively (without the top plate to plate Schade path), would make it very sensitive to layout capacitance and likely would need a HF freq. rolloff network somewhere in the loop for stability (like maybe around 250 KHz). A differential (CCS tailed) driver stage would make it much more effective at removing class AB crossover distortion (by observing and correcting for the net sum of both outputs). Then it would give the Mac a run for the (now much less) money.
Of course we would be free to put in different (like frame grid) tubes for drivers and the front stage. And high current Sweep tubes with a low primary Z OT for cheap but high performing output duty. One could even put a touch of current feedback(s) from the output tube cathodes (current sense low value R's) back to the driver cathodes to null out the OT resistance effects, giving this a high damping factor just like SS. There is also room for some global Fdbk around the OT as well.
Then listen to all the complaints about it sounding like a SS amp.
The final design version might then include a (mis-) balance pot in the splitter for some 2nd Harmonic, and an adjustable tail resistor somewhere to set the 3rd Harmonic level. The current feedbacks from the output tube cathodes then might allow for inversion and level adjustment to set a finite output Z or damping factor. In other words: "Comfort controls" or more cleverly: speaker distortion nulling controls, prominently noted on the advertising sheet.
Last edited:
OH, I guess I didn't correctly catch the detail in your earlier comment:
"I guess my question is if they are taking feedback from the plate of the 7027 to both plate and cathode of the same driver wouldn't one be negative feedback and the other be positive feedback? "
The output feedback to the plate of the 6CB6 is N Fdbk for the grid of the output tube. The 6CB6, being a pentode has high output Z there and won't internally notice any feedback effects there (but it does greatly reduce the driver stage gain due to the low Z, from the Miller like feedback through the Fdbk R).
The output feedback to the 6CB6 cathode is negative feedback for the 6CB6 (it requires more input signal to overcome).
What is interesting is this N feedback to the 6CB6 cathode also acts like a bootstrap for the cathode resistor there (a shunt current bootstrap, instead of a series voltage bootstrap), making it look like a higher value, so making the V to I conversion of the 6CB6 more linear. This is a benefit for the linearity of the Schade Fdbk up top at the 6CB6 plate (from the output).
So the cathode N Fdbk path is not real-real effective in itself, due to the 6CB6's now having low gain (from driving the low Z of the upper Schade, but probably still greater gain than unity, so it still does have some useful N Fdbk effect), but it linearizes the 6CB6 V to I function for the benefit of the top Schade network.
Getting a linear current driver seems to be the major Achilles heel for the usual local Schade scheme. So this RCA linearizing scheme is not to be taken lightly. It well deserves to be noted by anyone planning on Schade experiments.
On the other hand, eliminating the top Schade network, and just using the driver in high gain mode with the lower cathode feedback, not only puts more loop gain in the lower "Schade" (or cathode) loop, but eliminates the need for a linear V to I converter altogether. So I would expect this to be much better at eliminating output stage distortion, maybe 50X better. But one now has a lot more touchy circuit to deal with, with all that loop gain.
"I guess my question is if they are taking feedback from the plate of the 7027 to both plate and cathode of the same driver wouldn't one be negative feedback and the other be positive feedback? "
The output feedback to the plate of the 6CB6 is N Fdbk for the grid of the output tube. The 6CB6, being a pentode has high output Z there and won't internally notice any feedback effects there (but it does greatly reduce the driver stage gain due to the low Z, from the Miller like feedback through the Fdbk R).
The output feedback to the 6CB6 cathode is negative feedback for the 6CB6 (it requires more input signal to overcome).
What is interesting is this N feedback to the 6CB6 cathode also acts like a bootstrap for the cathode resistor there (a shunt current bootstrap, instead of a series voltage bootstrap), making it look like a higher value, so making the V to I conversion of the 6CB6 more linear. This is a benefit for the linearity of the Schade Fdbk up top at the 6CB6 plate (from the output).
So the cathode N Fdbk path is not real-real effective in itself, due to the 6CB6's now having low gain (from driving the low Z of the upper Schade, but probably still greater gain than unity, so it still does have some useful N Fdbk effect), but it linearizes the 6CB6 V to I function for the benefit of the top Schade network.
Getting a linear current driver seems to be the major Achilles heel for the usual local Schade scheme. So this RCA linearizing scheme is not to be taken lightly. It well deserves to be noted by anyone planning on Schade experiments.
On the other hand, eliminating the top Schade network, and just using the driver in high gain mode with the lower cathode feedback, not only puts more loop gain in the lower "Schade" (or cathode) loop, but eliminates the need for a linear V to I converter altogether. So I would expect this to be much better at eliminating output stage distortion, maybe 50X better. But one now has a lot more touchy circuit to deal with, with all that loop gain.
Last edited:
My last comment about eliminating the need for a linear V to I conversion if the top Schade loop were removed is a bit over the top.
Any pentode driver stage still needs to drive its [ load R || next stage grid R ] with linearity, and so still needs to be a linear V to I. The low Z Schade load up top however really stresses this requirement by requiring high current swing or high voltage swing (to drive a follower and then series R), making for a challenging driver design.
The interesting question remains, however, whether the RCA driver stage suffers big time from the significant loss of it's pentode driver gain (the low Z Schade load), or whether the DUAL action at the driver cathode (shunt bootstrapped cathode R, and full loop N Fdbk) is unusually effective and makes up for that.
The RCA linear V to I driver setup could also be used for a Schaded SEP by just splitting it in half. (no P-P crossover Fdbk paths to complicate matters) I haven't seen that done.
Any pentode driver stage still needs to drive its [ load R || next stage grid R ] with linearity, and so still needs to be a linear V to I. The low Z Schade load up top however really stresses this requirement by requiring high current swing or high voltage swing (to drive a follower and then series R), making for a challenging driver design.
The interesting question remains, however, whether the RCA driver stage suffers big time from the significant loss of it's pentode driver gain (the low Z Schade load), or whether the DUAL action at the driver cathode (shunt bootstrapped cathode R, and full loop N Fdbk) is unusually effective and makes up for that.
The RCA linear V to I driver setup could also be used for a Schaded SEP by just splitting it in half. (no P-P crossover Fdbk paths to complicate matters) I haven't seen that done.
Last edited:
I think that is caused by the big cap I use in series with the Schade feedback resistor, drifting in voltage slightly as the plate curve peaks change during the grid V stepping. It is charged to the average V difference between plate and grid on the curve tracer setup, and the plate peaks are changing during the scan, causing some zero bias drift across the grid feed resistor from the tracer stepper. The big 20 uF cap keeps the drift small, but not zero.
So it's just a tracer setup artifact.
Normally, for Schade amplifier designs, the cap is put into the grid circuit, since the R feedback goes from plate to plate with similar voltages. And the grid nominally does not draw current (well, except for a tiny amount through the grid bias R), so the average cap voltage stays constant.
Since I don't have a plate voltage like stepper on the tracer, I can't do it that way.
So it's just a tracer setup artifact.
Normally, for Schade amplifier designs, the cap is put into the grid circuit, since the R feedback goes from plate to plate with similar voltages. And the grid nominally does not draw current (well, except for a tiny amount through the grid bias R), so the average cap voltage stays constant.
Since I don't have a plate voltage like stepper on the tracer, I can't do it that way.
Last edited:
I was kind of thinking that might be the problem.
With regard to cap placement in amplifiers, that's one of the reasons I really like using the p-channel fet to drive the feedback network. Then the feedback can all be direct coupled and capacitive coupling occurs at the gate of the fet and there is no cap charging even under overload conditions since there is never any gate current on the fet. Overload recovery is instantaneous. Of course, the drawback is you have to have another negative supply voltage. And of course, if overload is bad enough you may blow the fet as it is hard to get a p-channel fet with a bunch of voltage rating headroom. But so far, I have been successful doing 20% plate-to-grid feedback ratios with power tubes and 500V p-channel fets and haven't lost a single fet. So it doesn't seem to be a problem in my experience.
Any more feedback (>20% of plate signal) becomes very difficult to do with the direct coupled pfet approach because a huge negative supply is needed to accommodate the swing. You would have to cascode devices to get around the relatively low voltage ratings of the available parts and it gets pretty complicated pretty fast.
With regard to cap placement in amplifiers, that's one of the reasons I really like using the p-channel fet to drive the feedback network. Then the feedback can all be direct coupled and capacitive coupling occurs at the gate of the fet and there is no cap charging even under overload conditions since there is never any gate current on the fet. Overload recovery is instantaneous. Of course, the drawback is you have to have another negative supply voltage. And of course, if overload is bad enough you may blow the fet as it is hard to get a p-channel fet with a bunch of voltage rating headroom. But so far, I have been successful doing 20% plate-to-grid feedback ratios with power tubes and 500V p-channel fets and haven't lost a single fet. So it doesn't seem to be a problem in my experience.
Any more feedback (>20% of plate signal) becomes very difficult to do with the direct coupled pfet approach because a huge negative supply is needed to accommodate the swing. You would have to cascode devices to get around the relatively low voltage ratings of the available parts and it gets pretty complicated pretty fast.
Well, I tried repeating a Schade curve plot and I couldn't get any baseline tilt at first. Then I found that the vertical sensitivity has to be down at like 10 mA/div to really see it. Doubling up the capacitor had no effect on it then. I tried triode mode then, and that was totally flat on the bottom.
Then it dawned on me what was going on.
I powered down the tube heater, and the curves collapsed down to just the tilted baseline. Its from the 60 Hz AC plate voltage variation causing a small current through the Schade R and C feedback network back to the grounded step generator. Apparently the cap is big enough that it's 1/wC is not relevant at 60 Hz compared to the 100K feedback R. So the current is just the 60 Hz half sinewave plate voltage, divided by 100K Ohms.
I guess in a P-P output stage one would see something similar, since the tube (or even the other tube) would be making the plate voltage swing (even when this tube is off) and the Schade divider (feedback and driver plate load resistor) goes to B+ (AC ground). Well, its just a tiny bit of extra load on the output.
Then it dawned on me what was going on.
I powered down the tube heater, and the curves collapsed down to just the tilted baseline. Its from the 60 Hz AC plate voltage variation causing a small current through the Schade R and C feedback network back to the grounded step generator. Apparently the cap is big enough that it's 1/wC is not relevant at 60 Hz compared to the 100K feedback R. So the current is just the 60 Hz half sinewave plate voltage, divided by 100K Ohms.
I guess in a P-P output stage one would see something similar, since the tube (or even the other tube) would be making the plate voltage swing (even when this tube is off) and the Schade divider (feedback and driver plate load resistor) goes to B+ (AC ground). Well, its just a tiny bit of extra load on the output.
Last edited:
Ah yes, that makes sense. The feedback is a load and the characteristics we are measuring are for the overall circuit, not just the tube. When the tube is cut off, the overall circuit is not and indeed cannot be. If you were to double the resistance values of your feedback resistors, we should see that area under the curves get smaller, at the expense of bandwidth.
So in my push-pull amp that I built, when the power tubes are "cut off", there is still some current flowing through the output transformer on that side that goes through the feedback network. We can cut off current flow through the tube but not through the overall circuit.
I knew this of course when I designed it but it is interesting to see it visually depicted in the circuit characteristics. I'm a bit embarrassed that I didn't pick up on what was going on in the plots sooner.
So in my push-pull amp that I built, when the power tubes are "cut off", there is still some current flowing through the output transformer on that side that goes through the feedback network. We can cut off current flow through the tube but not through the overall circuit.
I knew this of course when I designed it but it is interesting to see it visually depicted in the circuit characteristics. I'm a bit embarrassed that I didn't pick up on what was going on in the plots sooner.
I just calculated the approx. V gain from grid1 of the 6CB6 drivers in the RCA design, and they are like 90!! (that would be for a grounded cathode case, not including the degeneration from the cathode N Fdbk and cathode R, which then cuts the g1 gain down to around 11, I think.) I had hastily assumed earlier that the Schade network Zin at the driver plates was really squashing the 6CB6 gain.
The N feedback network for the 6CB6 cathodes however attenuates the final output signal by 187.5. With about 5.5 gain in the output tube, that would give a loop gain of 2.64 for the cathode path (90 x 5.5 /187.5). (Hmmm, I may not be calculating that correctly though, since the g1 requires 9 times more voltage swing now than a grounded cathode case. So maybe 9 x 2.64 = 24 loop gain at cathode?) So I'm not sure how much correction is going on through the cathodes other than bootstrapping the cathode resistor. I think that makes the AC cathode resistance look like about 2330 Ohms.
The N feedback network for the 6CB6 cathodes however attenuates the final output signal by 187.5. With about 5.5 gain in the output tube, that would give a loop gain of 2.64 for the cathode path (90 x 5.5 /187.5). (Hmmm, I may not be calculating that correctly though, since the g1 requires 9 times more voltage swing now than a grounded cathode case. So maybe 9 x 2.64 = 24 loop gain at cathode?) So I'm not sure how much correction is going on through the cathodes other than bootstrapping the cathode resistor. I think that makes the AC cathode resistance look like about 2330 Ohms.
Last edited:
Just trying to play catch-up, I still can not wrap my head around the fact that the two connections are exactly equivalent... so I tried to plot out their characteristics using SPICE, since CFB only works for AC input, its curves have to swing around a selected bias, therefore I might not be making an apple-to-apple comparison. In any case, I just want to show that the two characteristics are a bit different, below is a comparison showing the 6L6 with 10% feedback for both connections. The blue curves are Schade and the red curves are CFB, but please bear in mind that the amplitude for the CFB is 2 times the Schade's, if using the same amplitude, its curves would end up on the lower right corner of the chart. I could very well messed up the sims, if so, please correct me...
Schade: -100V to 0V, 10V/step; CFB: +100V to -100V, bias = -50V
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
Hi Jazbo,
Could you show the schematics for each case?
In showing the equivalence earlier, I used a divider between B+ and the plate (R divider, or tap on the primary) to get the % CFB equivalent. (virtual ground put there) Although I don't see any reason why just adjusting the feedback R from the plate alone would give anything different. I'm puzzled why you are seeing a 2 to 1 difference in gain if they are supposed to be equivalent feedback. Also, don't forget, the Schade drive has to be a controlled current. (Putting a resistor from a controlled V is an approximation to current drive, since the grid voltage has to swing some, affecting the V to I linearity some. Although it seems to have worked quite well for the curve tracer plots.)
In any case, looks like it would just take a small offset shift and scale factor change between your curve sets to get them to overlay better.
Could you show the schematics for each case?
In showing the equivalence earlier, I used a divider between B+ and the plate (R divider, or tap on the primary) to get the % CFB equivalent. (virtual ground put there) Although I don't see any reason why just adjusting the feedback R from the plate alone would give anything different. I'm puzzled why you are seeing a 2 to 1 difference in gain if they are supposed to be equivalent feedback. Also, don't forget, the Schade drive has to be a controlled current. (Putting a resistor from a controlled V is an approximation to current drive, since the grid voltage has to swing some, affecting the V to I linearity some. Although it seems to have worked quite well for the curve tracer plots.)
In any case, looks like it would just take a small offset shift and scale factor change between your curve sets to get them to overlay better.
Last edited:
That's what I'm trying to figure out, but first please check the schematics below for errors, I may have messed something up...
An externally hosted image should be here but it was not working when we last tested it.
Here is the comparison with equal range used for the CFB, i.e., V1 sweeps from 50V to -50V:
An externally hosted image should be here but it was not working when we last tested it.
Last edited:
I'm a little puzzled by the plate voltage readout on the graphs. I assume V4 and V8 are being swept from 0 to 250 V. The ARB1 function seems to always use a static 250 V in its calculation, shouldn't it be just using Vout (or V4)?
And then, since ARB1 adds another 10% to the total voltage, I would think the total scan of V4 would need reducing by 10% to 225V.
I'm not sure how ARB1 should be calculated. Normally it would be 10% of the voltage drop of a total load R.
Maybe one needs to include the plate current in the ARB1 calculation. Or put in a load R. ??
That may be the easiest route, just put 0.9 Rload up top, and 0.1 Rload on the bottom with a fixed V4.
---------
Then for the Schade case, I assume V8 is sweeping 0 to 250V, otherwise the feedback through R2 would just be a near constant current. I'm not sure how to relate an Rload in the CFB case then to the Schade case.
And then, since ARB1 adds another 10% to the total voltage, I would think the total scan of V4 would need reducing by 10% to 225V.
I'm not sure how ARB1 should be calculated. Normally it would be 10% of the voltage drop of a total load R.
Maybe one needs to include the plate current in the ARB1 calculation. Or put in a load R. ??
That may be the easiest route, just put 0.9 Rload up top, and 0.1 Rload on the bottom with a fixed V4.
---------
Then for the Schade case, I assume V8 is sweeping 0 to 250V, otherwise the feedback through R2 would just be a near constant current. I'm not sure how to relate an Rload in the CFB case then to the Schade case.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.