Barkhausen tubes and other unicomponent oscillators

[nsaspook]

In the end it's tubes that get the last laugh. Every transistors daddy is a tube, it might be a big tube with lots of vacuum pumps and controllers but most of the machines that form the basic layers and do the doping have the basics of filaments, cathodes, control grids, suppression grids or plates and anodes in their guts.

 

[EinsteinKreuz]

@EinsteinKreuz.
A research news item constitutes a dream, not a movement. I still work with VTs in the real world and I do not have your rosy view. I put my effort into removing relays, monostables and VTs rather than looking for opportunities where I might bring them back.

Making free running oscillators is not the problem, we now need to generate and receive modulated data streams. Optical data links employ semiconductor transducers, so they really can not need VTs. VTs will never match a FET semiconductor on speed, or noise figure at room temperature. How might you cool a VT amplifier to 5°K.

There are situations where VTs are still the answer. There are also situations where a VT has no application. For example, there are “non-linear transmission lines” made from semiconductors that sharpen pulse edges and make data links, sampling mixers and oscilloscopes with quite ridiculous speed and resolution. That cannot possibly be done with VTs.

Well yes. These microscopic vacuum tubes are still in the R&D phase and haven't fully been put into use. And I do not dispute that there are electronics applications where VTs aren't useful and can only be done with semiconductors. And do not misunderstand me: I was not saying that semiconductors are useless!

 

[EinsteinKreuz]

There are semiconductor designs that extend the range into the thz range with quantum mechanical effects.

http://en.wikipedia.org/wiki/Resonant-tunneling_diode

True. But the thing is that electrons in semiconductors have certain allowed energy states(energy bands)whereas free electrons can have any energy. The higher the oscillation frequency of an electron or a photon the higher the energy it carries( E = h$\nu$)so if you want to go up to frequencies that are in the 100 Thz or Petahertz range and beyond it would be easier to accomplish this with free electron tubes. There is a team of scientists in Deutschland developing an opto-electronic petahertz oscillator using femtosecond laser pulses.

 

[EinsteinKreuz]

Since WW2 we have gradually migrated to spread spectrum and digital, both of which require wide bandwidth. But we do not use VTs since semiconductors are more reliable and use less power for signal processing.

In the case of audio electronics, digital will never truly replace analog because without conversion back into analog, you wouldn't be able to hear digitized recorded sound!
The problem with analog transistor amplifiers is that semiconductors are nonlinear in terms of the output current response to input voltages( collector current plotted against base-emitter voltage) and have very low impedance. This causes unwanted distortion of large signals(the clipping effect which is very unpleasant to the human ear). To minimize the distortion of large signals you need bypass capacitors and swamping resistors. Buuuuuuuuuuuuuuuuut...this has a filtering effect which can remove desirable frequencies from the signal.

VTs, when operated below their maximum voltage rating, have a linear response to changes in input voltages with the notable exception of the Tetrode. So while their power requirements are larger, you have much higher fidelity to the input signal unless you overdrive your tubes the way a guitar amp does(where distortion is desirable). The intentional distortion of triode guitar amps is pleasing to the human ear which is why tube amps are preferred to transistor amps. I've also noticed as an audiophile the superior sound of tube stereo amplifiers when playing records because there isn't that obnoxious humming sound coming through.

In the end it's tubes that get the last laugh. Every transistors daddy is a tube, it might be a big tube with lots of vacuum pumps and controllers but most of the machines that form the basic layers and do the doping have the basics of filaments, cathodes, control grids, suppression grids or plates and anodes in their guts.

I'm with you. And the advantage of tubes for high voltage AC, RF, and military electronics is that semiconductors are highly sensitive to electromagnetic interference. This makes them vulnerable to electromagnetic weapons and in the case of high-voltage semiconductors used in the power grid, geomagnetic storms. The reliance of solid state high powered switches makes the US power grid extremely vulnerable to EMP weapons and large Solar Flares.

 

[Svein]

For those who want to reminisce on the "good old tube days", this was a famous tube amplifier: http://www.tubebooks.org/file_downloads/stancor_ul_schematics.pdf

 

[Baluncore]

In the case of audio electronics, digital will never truly replace analog because without conversion back into analog, you wouldn't be able to hear digitized recorded sound!

Not true. That is where the semiconductor H-bridge, class D amplifier, works well.

The problem with analog transistor amplifiers is that semiconductors are nonlinear in terms of the output current response to input voltages( collector current plotted against base-emitter voltage) and have very low impedance. This causes unwanted distortion of large signals(the clipping effect which is very unpleasant to the human ear).

Not true. You are assuming class A, without feedback. In reality, analogue transistor amplifiers are way more linear than VT amplifiers. That is because the circuit can be designed using many more integrated transistors than could possibly be done with VTs. It it possible to build ultra-linear SC amplifiers that require only milliwatts of power. Saturation of any device will always cause harmonic distortion. VTs and SCs both have the same saturation problem.

To minimize the distortion of large signals you need bypass capacitors and swamping resistors.

Semiconductors do not require coupling capacitors or level-shifting because two polarities of device are available.
A VT is an N-channel FET with a pilot lamp. There is no equivalent to the P-channel FET in VTs.

It is a mistake to argue that VTs are better than SCs. There are places where each has it's place.
Emotions are not rational. People who think VTs are better will buy higher priced VT audio amplifiers.

 

[EinsteinKreuz]

Not true. That is where the semiconductor H-bridge, class D amplifier, works well.Not true. You are assuming class A, without feedback. In reality, analogue transistor amplifiers are way more linear than VT amplifiers. That is because the circuit can be designed using many more integrated transistors than could possibly be done with VTs. It it possible to build ultra-linear SC amplifiers that require only milliwatts of power. Saturation of any device will always cause harmonic distortion. VTs and SCs both have the same saturation problem.

Semiconductors do not require coupling capacitors or level-shifting because two polarities of device are available.
A VT is an N-channel FET with a pilot lamp. There is no equivalent to the P-channel FET in VTs.

It is a mistake to argue that VTs are better than SCs. There are places where each has it's place.
Emotions are not rational. People who think VTs are better will buy higher priced VT audio amplifiers.

Well when it comes to emotions and the sound quality of audio amplifiers, what matters in this context is which is more pleasing to the human ear. Conversion from analog to digital results in a loss of information because an ADC has a *finite* sampling rate(whereas if it had a countably infinite number of clock cycles per second it would be essentially continuous)and so values between those sampling points will not be discarded. Now with respect to your claims about Class-D amplifiers, I can't say that I'm entirely convinced about the sound quality and & I'm evidently not alone.

Now in the case of saturation and harmonic distortion, the kind that is produced in transistors is known to be very unpleasant to the human ear(clipping, crossover, and in the worst case avalanche noise)compared to that produced by thermionic valves.

 

[Baluncore]

Audio amplifier quality is all quite off topic for "barkhausen-tubes-and-other-unicomponent-oscillators".

Conversion from analog to digital results in a loss of information because an ADC has a *finite* sampling rate(whereas if it had a countably infinite number of clock cycles per second it would be essentially continuous)and so values between those sampling points will not be discarded.

That is a complete fallacy. The bandwidth of an audio amplifier limits the highest frequencies present. You clearly do not understand the Nyquist–Shannon sampling theorem, it says that if the data rate is more than twice the audio bandwidth, then there should be no difference between analogue and reconstructed digital signals. Your ears limit the bandwidth more than the technology. http://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem

Now in the case of saturation and harmonic distortion, the kind that is produced in transistors is known to be very unpleasant to the human ear(clipping, crossover, and in the worst case avalanche noise)compared to that produced by thermionic valves.

Saturation or clipping does not happen in a properly designed and operated amplifiers. Clipping should never occur, (except deliberately in some guitar amplifiers).

Harmonic distortion is evidence of non-linearity. It is easier to design and build an ultra-linear SC amplifier than a linear VT amplifier. Harmonic distortion is quite irrelevant to the VT versus SC argument.

Crossover noise is only a problem when an amplifier is driving a load with higher impedance than it was designed to drive. Again it is quite irrelevant to the VT versus SC argument.

Avalanche noise is produced by reverse biassed SC junctions in breakdown. That is why Zener diodes should not be used for voltage references or level shifting in audio amplifiers without sufficient low-pass filtering. Avalanche noise is not a problem.

If you name enough possible problems you will cause unfounded fears, uncertainty and doubt in the uneducated. When something fails because of poor design or out-of-spec operating conditions then that cannot be used as evidence in an argument between VTs and SCs.

 

[jim hardy]

One needs to compare analog to digital using same program material.

I have Fiedler;s "Pops Roundup" on both vinyl and CD.

That is a classic recording because the quality is so good.
Tony Salvatore was the recording engineer.

When RCA made the CD they restored the original Ampex vacuum tube tape recorder they'd used to make the original masters... digitized that analog signal
so the quality of both is about as good as either medium can get.

The CD version just plain outshines the vinyl even though both come from same master recording played on same equipment..
i think it's because the dynamic range that's available on a CD exceeds what you can do with a needle and groove
irrespective of whether the speaker cone is driven by class A, AB or D .

Doubtless cheap digital won't be as good as expensive analog.

Is my test flawed?

old jim

 

[EinsteinKreuz]

Audio amplifier quality is all quite off topic for "barkhausen-tubes-and-other-unicomponent-oscillators".

That is a complete fallacy. The bandwidth of an audio amplifier limits the highest frequencies present. You clearly do not understand the Nyquist–Shannon sampling theorem, it says that if the data rate is more than twice the audio bandwidth, then there should be no difference between analogue and reconstructed digital signals. Your ears limit the bandwidth more than the technology. http://en.wikipedia.org/wiki/Nyquist–Shannon_sampling_theorem

Yes, I am aware of the celebrated Nyquist theorem but there again, that is a mathematical theorem which has been proven mathematically but electrical engineering is not pure mathematics and the theory doesn't always correspond with real data.

Also, speaking of wikipedia, did you see the section on Aliasing in the article were analog and digital recording technologies are compared?
From the article:

Technical difficulty arises with digital sampling in that all high frequency signal content above the Nyquist frequency must be removed prior to sampling, which, if not done, will result in these ultrasonic frequencies "folding over" into frequencies which are in the audible range, producing a kind of distortion called aliasing. The difficulty is that designing a brick-wall anti-aliasing filter, a filter which would precisely remove all frequency content exactly above or below a certain cutoff frequency, is impractical.[8] Instead, a sample rate is usually chosen which is above the theoretical requirement. This solution is called oversampling, and allows a less aggressive and lower-cost anti-aliasing filter to be used.

Unlike digital audio systems, analog systems do not require filters for bandlimiting. These filters act to prevent aliasing distortions in digital equipment. Early digital systems may have suffered from a number of signal degradations related to the use of analog anti-aliasing filters, e.g., time dispersion, nonlinear distortion, temperature dependence of filters etc. (Hawksford 1991:8). Even with sophisticated anti-aliasing filters used in the recorder, it is still demanding for the player not to introduce more distortion.

 

[Svein]

I have built my share of both vacuum-tube and solid-state amplifiers, and I have a couple of comments to the "Hi-Fi debate" that is starting to show. here.

As to harmonic distortion: The link in my previous post refers to the "Williamson amplifier", widely known as the best tube amplifier ever. Look at the schematic. After the tube amplifier sits the output transformer. Big and heavy. And - as all iron-core transformers - introducing a fair amount of distortion due to hysteresis, leakage inductance and wiring capacitance. Yes, we tried to do feedback around the output transformer, but it was not very successful.

As to other kinds of distortion: Read through this paper: http://www.linearaudio.nl/linearaudio.nl/images/pdf/otala%20low%20tim%20amp.pdf
This is the famous paper by Matti Ottala that discusses Transient Intermodulation Distortion and shows an example of an amplifier with very little TIM.

As to digital vs. analog audio: Yes, digital audio is sampled - this introduces a digitizing noise floor of -84dB. But vinyl is not without its problems either - the signal is run through a complex high-pass filter before engraving, and must therefore be run through a matching low-pass filter (the well-known RIAA stage) before amplification. This circuit is not easy to design.

 

[Baluncore]

Yes, I am aware of the celebrated Nyquist theorem

It is actually “Shannon's sampling theorem”. It should not be referred to as the “Nyquist Theorem”. Others have attached Nyquist's name to parameters such as the “Nyquist frequency” and the “Nyquist rate”.

theory doesn't always correspond with real data.

That is a problem with competency to take measurements of real data, or an inability to understand and apply the theory correctly. This is a Physics Forum, so the real data and the theory do correspond. If they did not, then I would want to know why. Can you give an example where “theory doesn't always correspond with real data”.

Unlike digital audio systems, analog systems do not require filters for bandlimiting.

Wiki is correct. Analogue circuits are inherently low-pass filters so they do not need filtering. Band limiting immediately prior to digitisation is part of the analogue to digital conversion process. It takes place at such a high frequency that it is well outside the audio band.

 

[EinsteinKreuz]

I have built my share of both vacuum-tube and solid-state amplifiers, and I have a couple of comments to the "Hi-Fi debate" that is starting to show. here.

As to harmonic distortion: The link in my previous post refers to the "Williamson amplifier", widely known as the best tube amplifier ever. Look at the schematic. After the tube amplifier sits the output transformer. Big and heavy. And - as all iron-core transformers - introducing a fair amount of distortion due to hysteresis, leakage inductance and wiring capacitance. Yes, we tried to do feedback around the output transformer, but it was not very successful.

As to other kinds of distortion: Read through this paper: http://www.linearaudio.nl/linearaudio.nl/images/pdf/otala%20low%20tim%20amp.pdf
This is the famous paper by Matti Ottala that discusses Transient Intermodulation Distortion and shows an example of an amplifier with very little TIM.

As to digital vs. analog audio: Yes, digital audio is sampled - this introduces a digitizing noise floor of -84dB. But vinyl is not without its problems either - the signal is run through a complex high-pass filter before engraving, and must therefore be run through a matching low-pass filter (the well-known RIAA stage) before amplification. This circuit is not easy to design.

Regarding Vinyl, I have no doubt that these circuits are not easy to design. But Hi-Fi stands for High Fidelity(the accuracy to the original sound/recording).

That being said, I just got a brand new tube amp for my record player and WOW! I can definitely hear big improvement in sound quality to a transistor amplifier when it comes to vinyl. Playing a record through transistor amplifiers that I've heard results in an obnoxious, low frequency hum coming from the turntable whereas with a tube amp that noise is completely eliminated. I'm definitely sticking with VTs for audio amplifiers.

 

[Svein]

Playing a record through transistor amplifiers that I've heard results in an obnoxious, low frequency hum coming from the turntable whereas with a tube amp that noise is completely eliminated.

Well - either the hum comes from the turntable or it does not.

• If the hum comes from the turntable it should be reproduced by a good amplifier. If the hum isn't present through the tube amplifier I suspect a 60Hz high-pass filter somewhere, since tube amplifiers usually had problems with hum (the filaments used AC, and that AC had a tendency to insert itself in the audio chain).
• If the does not come from the turntable, it is harder to say what happens. It depends on the pickup, the cable and whether or not the RIAA stage is inside the amplifier or outside. A relevant test is to short-circuit the phono inputs on the amplifiers and listen for hum.

As an aside: Hi-fi pickups are usually very low-impedance things. Therefore, using shielded cables does not keep hum from entering the wires, you need good quality twisted pair cables. For the same reason tube preamps have a hard time handling raw pickup signals.

Observe: I do not say that transistor amplifiers are "better" than tube amplifiers. What I try to say is that a tube amplifier is a voltage amplifier, and needs an impedance transformer to match the low-impedance loudspeakers. A transistor amplifier is a current amplifier and has a very low impedance output, interfacing nicely with the loudspeakers.

I am somewhat surprised that no one has mentioned MOSFET-based amplifiers. A MOSFET has the same voltage-amplifier characteristics as a tube, can operate at high voltages and can interface directly with loudspeakers.