There is the problem that 2.8Mz delta-sigma (SACD-format) amps need feedback correction because transistors are not fast enough to do this full-digital. Are HF tubes better?
And there is the point that volume control in digital amps is done by changing the rail voltage. Is this a problem for HF tubes? In what region can the anode voltage be varied?
And there is the point that volume control in digital amps is done by changing the rail voltage. Is this a problem for HF tubes? In what region can the anode voltage be varied?
A little pontificating:
1. Tubes are not good at on/off switching except at very high voltages. Loudspeakers don't want high voltages. A transformer could help but then we may as well stay analog throughout instead of class D .
2. Good square waves need edge speeds of 10 to 100 times the square wave repetition frequency, at 2.8 Mhz SACD rates this means 28 to 280 Mhz edges, transistors are usually superior at these moderate frequency, high current and moderate voltage requirements.
3. Symmetry in rise and fall times is important, complemetary solid state devices do this well. There is no such thing as a complementary tube.
While any of these problems can be reduced, it seems reasonably clear that transistors offer the best alternative as a solution to the digital amplifier concept.
ps. Anyone looked into driving an ESL panel and it's associated high voltage directly with class D tube amplification? ESL panels are on the whole capacitive in nature presenting a real charge/dischare rate problem for tubes, but maybe not insurmountable.
1. Tubes are not good at on/off switching except at very high voltages. Loudspeakers don't want high voltages. A transformer could help but then we may as well stay analog throughout instead of class D .
2. Good square waves need edge speeds of 10 to 100 times the square wave repetition frequency, at 2.8 Mhz SACD rates this means 28 to 280 Mhz edges, transistors are usually superior at these moderate frequency, high current and moderate voltage requirements.
3. Symmetry in rise and fall times is important, complemetary solid state devices do this well. There is no such thing as a complementary tube.
While any of these problems can be reduced, it seems reasonably clear that transistors offer the best alternative as a solution to the digital amplifier concept.
ps. Anyone looked into driving an ESL panel and it's associated high voltage directly with class D tube amplification? ESL panels are on the whole capacitive in nature presenting a real charge/dischare rate problem for tubes, but maybe not insurmountable.
This mostly true for many popular audio tubes that were originally designed for TV sets and made in very high volumes. There are any number of specialty tubes such as transmitter tubes that are useful at much higher frequencies and power levels.Eva said:
See for an extreme example a ceramic tube the 4CX1000 made for 110 MHz at a Kilowatt of plate dissapation. Used in class C radio transmitters or class D audio this tube could probably put out about 2-4 KW. There were also the tiny tubes such as the nuvistor that as a small signal amplifier could reach hundreds of megahertz.
Phillips actually built a class D tube amp for PA applications using TV line scan output tubes. It gave 100 volt line direct, and, since the comparator ramp was a simple RC (ie anything but linear) it's distortion figures were nothing to write home about.
However, lacking both mains and output transformers it was lightweight, (not cool, and definitely not high efficiency; dropping the heater volts with series resistors, practically universal with "p" series valves, is not an efficient way of working, if cheap) and the isolation on the input transformer had to be excellent (live chassis, remember? No, you probably don't remember, and much the better for that). They were probably built in the early sixties, considering the quantity of dust on the rackful I worked with at the end of the decade.
I can see no particular interest (apart from intellectual) in reviving the technology; it seems to collect together most of the disadvantages of a wide range of techniques, with none of the advantages (not even price, nowadays)
However, lacking both mains and output transformers it was lightweight, (not cool, and definitely not high efficiency; dropping the heater volts with series resistors, practically universal with "p" series valves, is not an efficient way of working, if cheap) and the isolation on the input transformer had to be excellent (live chassis, remember? No, you probably don't remember, and much the better for that). They were probably built in the early sixties, considering the quantity of dust on the rackful I worked with at the end of the decade.
I can see no particular interest (apart from intellectual) in reviving the technology; it seems to collect together most of the disadvantages of a wide range of techniques, with none of the advantages (not even price, nowadays)
Except maybe ESL direct drive, as already mentioned, since the Class A versions are extremely inefficient. What about adapting this circuit for Class D?
http://users.forthnet.gr/ath/sifakhs/images/amp838_main.GIF
Yes, my model works quite nicely, as you can see from the transfer curve. I'm planning a PP model which should do 160W from a pair of 31LQ6 and 6CL3, at about 80% plate efficiency.
Using pentodes or cascodes, low value plate resistors and high currents (10mA+), it's quite easy to get 200ns edges, corresponding to bandwidth of 5MHz or so. The 38HE7 "switches" in 100ns.
Tubes, in comparison to MOSFETs, are very similar. They are transconductance devices and they have a well-defined "on" resistance. I'm getting around 100 ohms "Rpk(on)" from the 38HE7. The 31LQ6 should be about half that. Fundamentally, "on" resistance depends on the cathode; higher grid and screen voltages can get you higher plate current, but always at a higher saturation voltage.
At suitable voltages (remembering that tubes always prefer higher voltages, while SS can direct drive a loudspeaker), tubes are only about 2-5 times lower in performance than MOSFETs: consider that the tube acts like a 1000V, 1A, 50 ohm "channel", whereas a comparable MOSFET might be 10 ohms or so.
Note that I still require an output transformer. This is not Dave Berning's patented chopper amp.
Tim
Using pentodes or cascodes, low value plate resistors and high currents (10mA+), it's quite easy to get 200ns edges, corresponding to bandwidth of 5MHz or so. The 38HE7 "switches" in 100ns.
Tubes, in comparison to MOSFETs, are very similar. They are transconductance devices and they have a well-defined "on" resistance. I'm getting around 100 ohms "Rpk(on)" from the 38HE7. The 31LQ6 should be about half that. Fundamentally, "on" resistance depends on the cathode; higher grid and screen voltages can get you higher plate current, but always at a higher saturation voltage.
At suitable voltages (remembering that tubes always prefer higher voltages, while SS can direct drive a loudspeaker), tubes are only about 2-5 times lower in performance than MOSFETs: consider that the tube acts like a 1000V, 1A, 50 ohm "channel", whereas a comparable MOSFET might be 10 ohms or so.
Note that I still require an output transformer. This is not Dave Berning's patented chopper amp.
Tim
INTERESTING
Looks like good old tube amp, but driven by ClassD, or full digital.
you can amaze your friend.
I simulated for a change.
Schematic shows very simplified 2A3 single, each wave are supplied PWM(blue) generated by sin wave(green), and output result(red).
It looks like 2A3 can eat PWM signal and generates audio output.
(Transformer reduces HF, but I added LPF just in case.)
Anyone please try, to supply raw class-D output PWM to your tube amp?
Looks like good old tube amp, but driven by ClassD, or full digital.
you can amaze your friend.
I simulated for a change.
Schematic shows very simplified 2A3 single, each wave are supplied PWM(blue) generated by sin wave(green), and output result(red).
It looks like 2A3 can eat PWM signal and generates audio output.
(Transformer reduces HF, but I added LPF just in case.)
Anyone please try, to supply raw class-D output PWM to your tube amp?
Attachments
