From a friend of mines, in Italy. He is the same person that winds my transformers. I design transformers for him (not all of them) and I get mines for free....
That PCL86 amp started out as 2W amp with the triode strapped pentode and no fbk. It worked like that for several years. Then I modified it to pentode with cathode fbk. Several years back was also sold as a kit.
That PCL86 amp started out as 2W amp with the triode strapped pentode and no fbk. It worked like that for several years. Then I modified it to pentode with cathode fbk. Several years back was also sold as a kit.
Last edited:
In this example, signal voltage at grid is 3.4V, and at anode is 153V for a sum of 156.4V. This appears across a real resistor of 100K Ohms, and real current of 156.4V/100K equals 1.564mA must enter it, from both ends. At the grid end, this is 3.4V/1.564mA equals about 2K Ohms. Pure Ohm's law and just like Miller C, which isn't a physical capacitor, but still must be driven (charged and discharged) and appears as the driving valve's loadline. Miller C expands the ellipse and the Schade resistor rotates everything clockwise, each unto its nature.
All good fortune,
Chris
All good fortune,
Chris
BTW, that 2K Ohm load also effects the driving valve's gain. So, it has a gain of a couple or three, not dozens. Very easily measured.
Chris
Chris
I don't know. I find easier to think in the same terms I use for cathode feedback. The gain of the driver is surely affected by the load but it would be few percent of the mu with 2K load. It would not be reasonable with lower gain tubes like 6l6 etc in terms of overall input sensitivity and drive
In (series, negative) feedback, like feedback to the output valve's cathode, the driver pays a penalty in voltage. In (parallel, negative) feedback, like Schade, the driver pays a penalty in current. Both penalties are linearly proportional to feedback expressed linearly (as a factor, not log like dB).
All good fortune,
Chris
All good fortune,
Chris
The driver surely pays a penalty for both current and voltage with cathode fbk despite the fact the grid resistor in front of the power tube is multiplied by the feedback factor. In theory (with best quality tube) it would be more but in the real case of my PCL86 amp (that uses rather weak Polamp tubes) the feedback is just over 5 dB which means about 4% of the output voltage. Despite the 220K grid resistor in front of the power tube becomes 400K (because of the cbf), the total load for the driver is not much better (it's in parallel with 100K anode resistor) and it has to deliver about twice the voltage, hence more current too. But it can manage because the total amount is well within its possibilities.
Last edited:
Here you mean 40% of course, typo. But actually the grid resistor is outside the feedback loop, so isn't bootstrapped. Miller C is however in the feedback loop, so is decreased by the factor of the feedback, as is the grid-cathode capacitance. If a vacuum valve had any significant input resistance, it would be increased by the same proportion.
You obviously don't believe me about "Miller R", so you must test it for yourself. Measure the signal voltages at grid of driver, anode of driver (= grid of output valve) and anode of output valve. That will be a definitive test. Is the signal voltage at driver anode (= output grid) 3.4V at full output or some much larger number?
All good fortune,
Chris
You obviously don't believe me about "Miller R", so you must test it for yourself. Measure the signal voltages at grid of driver, anode of driver (= grid of output valve) and anode of output valve. That will be a definitive test. Is the signal voltage at driver anode (= output grid) 3.4V at full output or some much larger number?
All good fortune,
Chris
The load with cfb is 5K at the plate and 8R at the cathode, seen individually. The plate-to-secondary ratio is 25:1, the cathode-to-secondary ratio is obviously 1:1. So the total load distributed between anode and cathode is (25+1):1 and the feedback voltage is 1/26 of the total or 3.85% (about 5.6 Vrms for 4W output). The normal driving voltage with this actual tubes is some 6V+ rms because they are not really like the best PCL86 of the golden era. On the Hickok 539 they can barely reach 8.5-9 mA/V while the original Philips rockets at 11.5-12 mA/V. That results in less feedback at just over 5 dB.
Regarding the grid resistor. The gain of the output tube is now 12.4, without cfb is 23.6. That loss of gain boosts the input impedance. It's the same as a cathode follower, except in this case it's a partial cathode follower. However the load seen by the input triode is still anode resistor (100K) in parallel with input impedance of the power tube and so not much better in relation to the new driving voltage.
Regarding the grid resistor. The gain of the output tube is now 12.4, without cfb is 23.6. That loss of gain boosts the input impedance. It's the same as a cathode follower, except in this case it's a partial cathode follower. However the load seen by the input triode is still anode resistor (100K) in parallel with input impedance of the power tube and so not much better in relation to the new driving voltage.
Last edited:
Ah, I see. You meant 4% of output anode voltage, effectively summed primary voltage. Got it.
In the Crowhurst cathode follower case the grid resistor was bootstapped by signal from the cathode. In ordinary amplifiers the grid resistor goes to ground, so isn't bootstapped, just a dead loss. It's said to be "parasitic" because outside of the feedback loop.
All good fortune,
Chris
In the Crowhurst cathode follower case the grid resistor was bootstapped by signal from the cathode. In ordinary amplifiers the grid resistor goes to ground, so isn't bootstapped, just a dead loss. It's said to be "parasitic" because outside of the feedback loop.
All good fortune,
Chris
Then the load for the driver is the same and so it's even (slightly) worse than I thought!
Last edited:
The goal of taking feedback from the OPT primary is a worthy one. Most of the primary will be across the anode, even with as much cathode feedback as we can (easily) get. In the simplest two stage designs, output anode signal voltage is the correct polarity (for negative feedback) to two places: the driver's anode and the driver's cathode. Parallel feedback (to the anode) causes heavy loading. Series feedback (to the cathode) has a DC voltage issue.
The DC voltage issue could be approached in several ways, but the obvious answer, a big series capacitor, has its own issues. Nothing important is easy, but that's why we like 'em.
All good fortune,
Chris
The DC voltage issue could be approached in several ways, but the obvious answer, a big series capacitor, has its own issues. Nothing important is easy, but that's why we like 'em.
All good fortune,
Chris
This PCL86 was a cheap experiment and it's well worth trying. In general, I now only use pendotes as such with cfb (which is effectively an UL configuration as I do not refer g2 directly to the cathode by means of a capacitor) but I do not use the secondary to get cfb and only use fixed bias. The cfb is on a separate winding. It's a primary winding in other words. There is not DC voltage issue, trust me. The best compromise with most pentodes is between 8% and 20% of output voltage, IMHO. This way the efficiency is almost the same as a pure pentode, the reduction is distortion quite dramatic and the drive requirement is manageable with the usual signal tubes.
P.S.
A decisive factor is also the transformer design that must not be compromised by the presence of the cfb winding. Such winding must be integrated into it. Only this way the overall performance can be improved in all aspects like FR extending to around 100KHz, for example, and instability is more a theorical problem. I have not seen any instability so far without paying too much attention to the circuit time constants and the usual stuff critical for applying ground feedback.
P.S.
A decisive factor is also the transformer design that must not be compromised by the presence of the cfb winding. Such winding must be integrated into it. Only this way the overall performance can be improved in all aspects like FR extending to around 100KHz, for example, and instability is more a theorical problem. I have not seen any instability so far without paying too much attention to the circuit time constants and the usual stuff critical for applying ground feedback.
Last edited:
Please do not say that cathode negative feedback is a form of Ultra Linear Mode.
It is not.
Cathode negative feedback is when: The negative feedback voltage changes the Cathode to g1 control grid voltage.
That kind of negative feedback involves g1 voltage versus the cathode voltage (even though in this case it is the cathode voltage that is moving with negative feedback, it is not the grid voltage, g1 moving with negative feedback.
It is important to note, with cathode negative feedback, that only the plate voltage changes, the screen voltage is constant (just as it is in Pentode mode / Beam Power mode).
"Cathode versus g1."
Ultra Linear is defined as: The Screen Voltage changes whenever the Plate voltage changes. It is a function of the UL Tap percentage (or without an actual UL Tap, an extra tube and a voltage divider from the plate to ground that drives the extra tube grid, and that causes the extra tube's plate to drive the screen voltage at a percentage [%] of the output tube's plate voltage swing).
In either case of UL (UL Tap or extra tube and plate voltage divider, the screen voltage moves as a Percentage(%) of the output tube's plate voltage, at the designed % of the UL Tap or extra tube's voltage divider and circuitry.
"Screen versus Plate."
I hope that clears up the differences between cathode negative feedback versus UL negative feedback.
Then . . .
Some will say that cathode negative feedback also changes the cathode to screen voltage; that is true.
But, I bet g1 mu (u) is larger than g2 mu (u), which means the majority of the effect of cathode negative feedback voltage change, is predominantly due to g1, and not to g2.
Or . . shew me the g2 u curves, if the u of g2 is larger than the u of g1.
u as defined g1 to cathode voltage change versus plate voltage change.
u as defined g2 to cathode voltage change versus plate voltage change.
I do not mean the g1 to g2 u; that is totally a totally different u.
It is not.
Cathode negative feedback is when: The negative feedback voltage changes the Cathode to g1 control grid voltage.
That kind of negative feedback involves g1 voltage versus the cathode voltage (even though in this case it is the cathode voltage that is moving with negative feedback, it is not the grid voltage, g1 moving with negative feedback.
It is important to note, with cathode negative feedback, that only the plate voltage changes, the screen voltage is constant (just as it is in Pentode mode / Beam Power mode).
"Cathode versus g1."
Ultra Linear is defined as: The Screen Voltage changes whenever the Plate voltage changes. It is a function of the UL Tap percentage (or without an actual UL Tap, an extra tube and a voltage divider from the plate to ground that drives the extra tube grid, and that causes the extra tube's plate to drive the screen voltage at a percentage [%] of the output tube's plate voltage swing).
In either case of UL (UL Tap or extra tube and plate voltage divider, the screen voltage moves as a Percentage(%) of the output tube's plate voltage, at the designed % of the UL Tap or extra tube's voltage divider and circuitry.
"Screen versus Plate."
I hope that clears up the differences between cathode negative feedback versus UL negative feedback.
Then . . .
Some will say that cathode negative feedback also changes the cathode to screen voltage; that is true.
But, I bet g1 mu (u) is larger than g2 mu (u), which means the majority of the effect of cathode negative feedback voltage change, is predominantly due to g1, and not to g2.
Or . . shew me the g2 u curves, if the u of g2 is larger than the u of g1.
u as defined g1 to cathode voltage change versus plate voltage change.
u as defined g2 to cathode voltage change versus plate voltage change.
I do not mean the g1 to g2 u; that is totally a totally different u.
Last edited:
@6A3sUMMER , Crowhurst used to call it "modified ultralinear". Explanation down here in the specific paragraph and referred to fig.16.
In my experience, is a much better way to get performance. The classic ultralinear with the same amount of cathode fbk might result in lower Zout but is worse, a little or a lot, in all other aspects. Not a good deal for me, as I don't consider a Zout as low as possible a necessity at expense of the rest.
Note how with the "strict uoltralinear condition" for the modified ultralinear, the output power is the same as the classic UL with the same g2 percentage. However the drive is in most cases quite large..
In my experience, is a much better way to get performance. The classic ultralinear with the same amount of cathode fbk might result in lower Zout but is worse, a little or a lot, in all other aspects. Not a good deal for me, as I don't consider a Zout as low as possible a necessity at expense of the rest.
Note how with the "strict uoltralinear condition" for the modified ultralinear, the output power is the same as the classic UL with the same g2 percentage. However the drive is in most cases quite large..
Last edited:
45,
Thanks.
Modified UL
Key word: Modified
It is Not a Classical UL circuit, it is a combination circuit.
I hope Newbies do not confuse those two different things.
In the case of Unity Coupled plate and cathode, such as a MacIntosh amp, that is a design unto itself (and deserves its own unique name).
Looking at Unity Coupled as if it is Ultra Linear . . .
. . . Is like looking at a long tailed phase inverter pair with the second grid grounded,
and then calling it a cathode follower and grounded grid stage; It is Not.
. . . Since when does a cathode follower have a very large un-bypassed plate load resistor? Never, or it is not a cathode follower.
Call it a "Concertina" circuit? . . . (do not worry that the gain of the cathode and plate are not, equal in that circuit).
Calling an LTP phase inverter with the second grid grounded, anything other than an "LTP phase inverter", is not a good idea.
But if you want to use original fundamental circuit names, then it would be far more accurate to call it a combination concertina and grounded grid phase inverter. There is no cathode follower there.
Combination circuits are their own thing. They are unique, even though they use some elements of original fundamental circuit names; they quickly depart from that, due to the changes/modifications they posses.
I recommend calling them by their usual (new) names of the combination circuit; and not calling them by the unmodified fundamental circuit names.
Just my opinions.
Thanks.
Modified UL
Key word: Modified
It is Not a Classical UL circuit, it is a combination circuit.
I hope Newbies do not confuse those two different things.
In the case of Unity Coupled plate and cathode, such as a MacIntosh amp, that is a design unto itself (and deserves its own unique name).
Looking at Unity Coupled as if it is Ultra Linear . . .
. . . Is like looking at a long tailed phase inverter pair with the second grid grounded,
and then calling it a cathode follower and grounded grid stage; It is Not.
. . . Since when does a cathode follower have a very large un-bypassed plate load resistor? Never, or it is not a cathode follower.
Call it a "Concertina" circuit? . . . (do not worry that the gain of the cathode and plate are not, equal in that circuit).
Calling an LTP phase inverter with the second grid grounded, anything other than an "LTP phase inverter", is not a good idea.
But if you want to use original fundamental circuit names, then it would be far more accurate to call it a combination concertina and grounded grid phase inverter. There is no cathode follower there.
Combination circuits are their own thing. They are unique, even though they use some elements of original fundamental circuit names; they quickly depart from that, due to the changes/modifications they posses.
I recommend calling them by their usual (new) names of the combination circuit; and not calling them by the unmodified fundamental circuit names.
Just my opinions.
Last edited:
Correct, cathode negative feedback is neither Pentode mode, nor Beam Power mode.
I said:
"It is important to note, with cathode negative feedback, that only the plate voltage changes, the screen voltage is constant (just as it is in Pentode mode / Beam Power mode)."
I said the Screen voltage is constant (it is, it is often regulated, or at least should be from a low impedance voltage source).
The fact that the cathode voltage changes Versus the screen voltage, the screen is sill at a constant voltage (300V, for example).
Some Newbies should not get the idea that unity coupling changes the screen voltage, it does not.
I said:
"It is important to note, with cathode negative feedback, that only the plate voltage changes, the screen voltage is constant (just as it is in Pentode mode / Beam Power mode)."
I said the Screen voltage is constant (it is, it is often regulated, or at least should be from a low impedance voltage source).
The fact that the cathode voltage changes Versus the screen voltage, the screen is sill at a constant voltage (300V, for example).
Some Newbies should not get the idea that unity coupling changes the screen voltage, it does not.
I have thought about it but I still disagree on the "effective" value of the feedback resistor. The effective value of the feedback resistor must depend on the actual feedback factor. 2K is true only if you have 100% feedback.
In that circuit beta = R/R+Rf where R =ra//Rg. In numbers, ra =52K approx, Rg=200K and Rf=100K. So beta=0.28.7 or 28.7%. The feedback factor (A*beta +1), where A=45 is the gain of the EL84, is 13.86 and so the effective value of Rf is 7.2K. If one used a pentode, with typical 150K anode resistance, beta would be higher at 46%, the effective value of Rf would be some 4.6K and so not necessarily better. It all depends on the capability of the driver tube (anode voltage and anode current) on this I think we all agree.
Regarding the 52K for the anode resistance of the ECC81 and not 100K....
With 310V supply and 360R cathode resistor the anode current will likely set around 2 mA and the anode voltage around 110V anode voltage. This alone would already be a no no as the bias would be scarce and the ECC81 will soon run into grid current.....
Anyway, in this conditions mu=60 and gm is approx 2 ma/V so ra=30K with bypassed cathode. With un-bypassed cathode 360x60=21.6K and so ra=52K.
That Universal RH is suggested with lower gain pentodes like KT88 (11W into 2.5K) where the resultant feedback factor should be in the region of 5.3 and so the effective value of Rf would be around 19K. Better but overall still questionable with such wimpy driver.