In optimized conditions for each, which devices generally measures better - thd/gain/output impedance/etc
I was always led to believe that a triode was the optimum amplifying device. Hence the 4 day European Triode Festival over the last 24 years which gathers together the cream of valve designers.
The thd/gain/output impedance are system characteristics, not those of a device.
Triodes may be relatively linear, but have low gain, low output impedance and low power output compared to pentodes.
Does this make them superior?
Does this make them superior?
Each type of device has its advantages, and its disadvantages.
1. Solid State
BJTs, MOSFETs, (and do not forget JFETs) characteristics drift versus temperature.
BJTs ft, frequency cutoff varies versus current (bandwidth is not constant versus current).
MOSFETs and JFETs capacitance varies widely versus voltage.
2. Tubes
Most tubes and tube applications do not experience the problems listed in 1. above;
Or to a much lower degree.
3. The trick is to use each type of amplifying device where its advantages are maximized, and its disadvantages are reduced.
Pick the proper device for the job at hand.
Do not use a Ball Peen Hammer to install Phillips Screws in wood.
Engineering is an Art and a Science.
Just my $0.03
1. Solid State
BJTs, MOSFETs, (and do not forget JFETs) characteristics drift versus temperature.
BJTs ft, frequency cutoff varies versus current (bandwidth is not constant versus current).
MOSFETs and JFETs capacitance varies widely versus voltage.
2. Tubes
Most tubes and tube applications do not experience the problems listed in 1. above;
Or to a much lower degree.
3. The trick is to use each type of amplifying device where its advantages are maximized, and its disadvantages are reduced.
Pick the proper device for the job at hand.
Do not use a Ball Peen Hammer to install Phillips Screws in wood.
Engineering is an Art and a Science.
Just my $0.03
Since gain can always be traded for distortion reduction (by negative feedback) the right metric to compare would be the gain/distortion ratio. Transistors easily beat tubes in this regard.
Last edited:
But what if your criteria is the distortion of the actual device? You don't have to accept that negative feedback is an automatic choice.
Then we would have to exclude triodes since they acheive their low distortion through inherent internal feedback
Some threads on whether triodes have internal feedback
https://www.diyaudio.com/community/...resistance-internal-negative-feedback.178688/
http://ken-gilbert.com/images/pdf/Inherent_FB_inTriodes.pdf
https://www.diyaudio.com/community/...resistance-internal-negative-feedback.178688/
http://ken-gilbert.com/images/pdf/Inherent_FB_inTriodes.pdf
1. I believe the only non-arguable internal feedback that Pentodes and Beam Power tubes have is in the following operating mode:
Triode Wired Mode.
2. As long as feedback has been mentioned, here is a form of negative feedback for Class A push pull:
A push pull pair of tubes, either in Triode Wired mode, Or in Ultra Linear mode:
For those modes, the plate impedance, rp is low to medium (a resistive drive to the primary).
(Pentode and Beam Power tubes in native mode are current sources; so do not apply here).
Suppose one tube plate voltage goes down too far; that raises the other tube's plate voltage too much; when that plate voltage
goes up too far, that tube's rp pulls more current, partially cancelling the other tubes excess current.
Then, the operation is reversed for the opposite polarity of the signal.
"Push Pull intrinsic Plate to Plate negative feedback".
. . . Did you ever think of it that way?
Triode Wired Mode.
2. As long as feedback has been mentioned, here is a form of negative feedback for Class A push pull:
A push pull pair of tubes, either in Triode Wired mode, Or in Ultra Linear mode:
For those modes, the plate impedance, rp is low to medium (a resistive drive to the primary).
(Pentode and Beam Power tubes in native mode are current sources; so do not apply here).
Suppose one tube plate voltage goes down too far; that raises the other tube's plate voltage too much; when that plate voltage
goes up too far, that tube's rp pulls more current, partially cancelling the other tubes excess current.
Then, the operation is reversed for the opposite polarity of the signal.
"Push Pull intrinsic Plate to Plate negative feedback".
. . . Did you ever think of it that way?
Last edited:
Vacuum tube characteristics drift with time and age, more so than most solid state devices.
No single devices do all that great as single stage voltage amplifiers, but combinations like cascodes and Zlikai connections can act very linear, and having both P and N types can give a significant advantage.
No single devices do all that great as single stage voltage amplifiers, but combinations like cascodes and Zlikai connections can act very linear, and having both P and N types can give a significant advantage.
6a3Summer,
I'm not sure I'd include option #2, because in a push-pull amplifier, if the B+ remains at full voltage all of the time, then it is the out of phase grid inputs driving the (non-linear) outputs, rather than an element of one of the outputs influencing the input of the same device. You're making a feedback argument, but not an internal feedback argument. It's more just a quirk of the circuit overall being a bit asymmetric at its limits.
kind regards
Marek
I'm not sure I'd include option #2, because in a push-pull amplifier, if the B+ remains at full voltage all of the time, then it is the out of phase grid inputs driving the (non-linear) outputs, rather than an element of one of the outputs influencing the input of the same device. You're making a feedback argument, but not an internal feedback argument. It's more just a quirk of the circuit overall being a bit asymmetric at its limits.
kind regards
Marek
MarekH,
I know that I am 'stretching' the meaning of the word 'negative feedback'.
Can we agree that negative feedback is merely a method of using a corrective signal path, which reduces the linearity errors of part of, or of all of an amplifier?
I say the push tube plate is correcting one error of the pull tube plate, and vice versa.
In global negative feedback, the secondary of the output transformer is the source of the negative feedback signal.
Whereas, instead . . .
In the case of one plate feeding a correction signal to the other plate, the path is merely the mutual coupling of the two halves of the primary windings.
It is not internal to the tube, it is from tube to tube.
The non-linearity of each tube in push pull begins far before the tubes are near their dynamic limits.
Each tube's dominant non-linearity is 2nd Harmonic, but the signal of one plate feeds correction through the primary to the other plate; and in the other direction for the second tube to the first tube.
I know that I am 'stretching' the meaning of the word 'negative feedback'.
Can we agree that negative feedback is merely a method of using a corrective signal path, which reduces the linearity errors of part of, or of all of an amplifier?
I say the push tube plate is correcting one error of the pull tube plate, and vice versa.
In global negative feedback, the secondary of the output transformer is the source of the negative feedback signal.
Whereas, instead . . .
In the case of one plate feeding a correction signal to the other plate, the path is merely the mutual coupling of the two halves of the primary windings.
It is not internal to the tube, it is from tube to tube.
The non-linearity of each tube in push pull begins far before the tubes are near their dynamic limits.
Each tube's dominant non-linearity is 2nd Harmonic, but the signal of one plate feeds correction through the primary to the other plate; and in the other direction for the second tube to the first tube.
Last edited:
bwaslo,
Correct, tubes drift with time and age.
But . . . Solid state devices drift with temperature; and they can be destroyed by very brief overloads; and secondary breakdown too.
Choosing between these poisons is not easy.
P and N devices have some differences, complimentary devices are not always purely complimentary
PNP versus NPN:
Typically, a P device has a larger chip in order to have some of the same specs as an N device chip.
But that means the P device is slower than the N device, because of the larger capacitance of the larger P device chip; more area = more capacitance = slower device.
Or, to gain back the speed, allow for the P device to be lower in some specs, perhaps beta, power, or some other spec.
In terms of comparing solid state devices to vacuum tubes, this often applies:
"All generalizations have exceptions" In reference to Star Trek . . . Mr. Spock, try using your logic on that quote.
Correct, tubes drift with time and age.
But . . . Solid state devices drift with temperature; and they can be destroyed by very brief overloads; and secondary breakdown too.
Choosing between these poisons is not easy.
P and N devices have some differences, complimentary devices are not always purely complimentary
PNP versus NPN:
Typically, a P device has a larger chip in order to have some of the same specs as an N device chip.
But that means the P device is slower than the N device, because of the larger capacitance of the larger P device chip; more area = more capacitance = slower device.
Or, to gain back the speed, allow for the P device to be lower in some specs, perhaps beta, power, or some other spec.
In terms of comparing solid state devices to vacuum tubes, this often applies:
"All generalizations have exceptions" In reference to Star Trek . . . Mr. Spock, try using your logic on that quote.
Last edited:
True for semis. Otoh, a single DHT providing voltage amplification into a high impedance load can have astonishingly low distortion. Doesn't work so well in practice without a follower of course.
analog_sa,
Which brings us back to the circuit surrounding a tube or solid state device, that brings out the best of the device.
Which brings us back to the circuit surrounding a tube or solid state device, that brings out the best of the device.
I'd think you are on safe ground there and could even expand the scope to say that feedback is happening anytime the output is merely exerting some influence on the input, whether "good" or "bad".
What you are saying is that the grey lines are straighter and more parallel and more evenly spaced at the very high and the very low voltages than the red or green lines, when considered alone, suggest they ought to be.
https://www.diyaudio.com/community/attachments/composite-ul-pp-ll9-png.875398/
kind regards
Marek
I can't answer to the measured aspect, however in my particular experiment comparing BD139 and ECC88, measured distortion was similar but different. As similarly rated devices, I had a circuit where I could easily interchange between them. Driving each one, common base / grid, the tube sounded significantly better than the transistor. It was a bit of a watershed moment for me, as a lazy person, I stopped trying to perfect my transistor amps, and got back into valves. If the solitary transistor sound bad compared to the solitary tube doing the same job in the same circuit, doesn't bode well for making a good sounding circuit with bipolar transistor. It was a surprise, I thought they would sound the same.
You can make power beam pentodes quite linear using "Crazy Drive", pic1 below. Just takes two resistors. It's actually a form of feed-forward, driven at grid2. ( One resistor, approx. 6K from grid2 to grid1, and another resistor from grid1 to cathode, approx, 4K, values depend on tube type, curve tracer a plus for setting up ) It will typically still need some local or global N Fdbk to lower the output Z. Pic2 is standard grid1 drive.
And you can make a power beam pentode behave like a HV xmit triode using UnSet type mode. But still uses normal B+ and standard OTs. pic3 below using 6HJ5 tube. Requires two resistors, plate to grid1 and grid1 to ground reference and a P Chan follower to drive the cathode. It's an improved form of "Schade plate Fdbk". Pic4 below is the same 6HJ5 B pent. tube with conventional triode configuration.
You could also just start with a B pent. that's pretty linear inherently. Pic5 is a JJ KT77 in standard grid1 drive.
And some tubes will linearize fairly well with just some modest +V on grid3. Pic6 below of 4P1L with +15V on grid3. ( doesn't work well on most tubes )
Pic7 is the 4P1L without the fix.
And you can make a power beam pentode behave like a HV xmit triode using UnSet type mode. But still uses normal B+ and standard OTs. pic3 below using 6HJ5 tube. Requires two resistors, plate to grid1 and grid1 to ground reference and a P Chan follower to drive the cathode. It's an improved form of "Schade plate Fdbk". Pic4 below is the same 6HJ5 B pent. tube with conventional triode configuration.
You could also just start with a B pent. that's pretty linear inherently. Pic5 is a JJ KT77 in standard grid1 drive.
And some tubes will linearize fairly well with just some modest +V on grid3. Pic6 below of 4P1L with +15V on grid3. ( doesn't work well on most tubes )
Pic7 is the 4P1L without the fix.
Last edited: