Vacuum tubes and electronic oscillators

[Jon.G]

I've been reading up about vacuum tubes and (more specifically) the Audion, and how they were used for instruments/amplifiers. This isn't anything I'm learning about on my degree, just things I'm reading up on myself so forgive me if I'm a little slow to grasp some parts.

I understand how the vacuum tube converts an AC signal to a DC signal, and how the Audion amplifies the signal but I've come across a few things which I'm finding a little bit confusing.

1. Does an Audion also convert the AC signal to DC or does it simply amplify it? I have found some sources which say different things :/

2. I read that the Audion was, among other things, used to create an electronic oscillator, but when I read up on the electronic oscillator circuits I am seeing that they convert DC from a power supply to an AC signal, not AC to DC as I thought the vacuum tube does.

3. How would these be applied to synthesizers? (This is what I'm reading about, the early development of synths) I've grasped how the Telharmonium and tone wheel organs (Hammond organ etc.) work, but I'm a little unsure with this. I think there would be an oscillator, which can have its frequency altered by playing. The original signal would not need to be too great as the Audion could amplify it. If this is right can you help me understand how the frequency would be altered.
My ideas from looking around are that playing would alter the capacitance of the system (capacitor & inductor), so that the AC current flips more/ less often, altering the frequency. But if this is right how exactly would the keys works? would there be multiple oscillators?

Sorry if this isn't very clear.
Any information on any of these would be greatly appreciated :)

 

[Svein]

I understand how the vacuum tube converts an AC signal to a DC signal

Well, it doesn't. The reason why it seems so, is that a vacuum tube introduces a very large DC offset which is usually removed by a capacitor.

2. I read that the Audion was, among other things, used to create an electronic oscillator, but when I read up on the electronic oscillator circuits I am seeing that they convert DC from a power supply to an AC signal, not AC to DC as I thought the vacuum tube does.

You can create an oscillator based on anything that has some gain. The rest of your statement I don't understand.

How would these be applied to synthesizers? (This is what I'm reading about, the early development of synths) I've grasped how the Telharmonium and tone wheel organs (Hammond organ etc.) work, but I'm a little unsure with this. I think there would be an oscillator, which can have its frequency altered by playing. The original signal would not need to be too great as the Audion could amplify it. If this is right can you help me understand how the frequency would be altered.

There were electronic instruments in the vacuum tube days, but apart from the Hammond organ, they were very primitive. See http://en.wikipedia.org/wiki/Electronic_musical_instrument.

 

[analogdesign]

3. How would these be applied to synthesizers? (This is what I'm reading about, the early development of synths) I've grasped how the Telharmonium and tone wheel organs (Hammond organ etc.) work, but I'm a little unsure with this. I think there would be an oscillator, which can have its frequency altered by playing. The original signal would not need to be too great as the Audion could amplify it. If this is right can you help me understand how the frequency would be altered.
My ideas from looking around are that playing would alter the capacitance of the system (capacitor & inductor), so that the AC current flips more/ less often, altering the frequency. But if this is right how exactly would the keys works? would there be multiple oscillators?

Sorry if this isn't very clear.
Any information on any of these would be greatly appreciated :)

From an abstract standpoint the function of vacuum tube and transistor circuits isn't that different (in their details they differ greatly).

For a vacuum tube synth, anything that changes the delay in the oscillator will change the frequency. You could use a control voltage (just like with a typical transistor VCO). You could alter the physical capacitor as you suggest but most likely it would be much simpler to alter the tube bias, which would change it's gm (and therefore frequency because it would be proportional to gm/C).

It turns out modern synths (starting with Moog/Buchla et al) were done after transistors were available so I haven't seen any vacuum tube synthesizers. There isn't any reason you couldn't do one, though. In music, you find tubes in output amplifiers typically, because some people claim the even-order distortion characteristic of a tube is preferable to the third-order-dominated distortion of a transistor.

 

[nsaspook]

You can do a lot with tubes.
One of the first true electronic music scores "Electronic Tonalities" was on 'Forbidden Planet' by http://en.wikipedia.org/wiki/Bebe_and_Louis_Barron#Forbidden_Planet

This was long before true transistor synthesizers.
http://www.janebrockman.org/BebeBarron/BebeBarron.html

 

[nsaspook]

Forgot to include the opening music to the film.


How that sound was created from samples of the audio circuits.

 

[meBigGuy]

The tube does not convert AC to DC unless you are building a rectifier. (like a rectifier with 5U4's)

An analogy would be saying that a single transistor common emitter amplifier converts AC to DC. As Svein said, the input signal is imposed on a DC offset.

The audion is a simple triode, where the grid voltage modulates the plate current. Think of a FET, where the gate voltage modulates the drain current. Same function.

Generally a circuit that can be made with a FET can be made with triodes. (some limitations apply, read the fine print).

 

[dlgoff]

How that sound was created from samples of the audio circuits.

I'm betting you've done this before. I have.

 

[nsaspook]

I'm betting you've done this before. I have.

Oh yes. The thing about tube pre-amp stages and analog mag-tapes in the feedback loop was they didn't hard limit so the sounds generated had a distinct analog feel. You can trick digital into doing the same thing but the mechanics of things actually moving, heating and changing electrical transfer characteristics while altering the sound is hard to mimic.

Most of these systems can be understood as a type of audio Attractor.

 

[FactChecker]

can you help me understand how the frequency would be altered.
My ideas from looking around are that playing would alter the capacitance of the system (capacitor & inductor), so that the AC current flips more/ less often, altering the frequency. But if this is right how exactly would the keys works? would there be multiple oscillators?

There are simple ways to physically alter capacitance or inductance.
You can vary them continuously: The plates of a capacitor can slide to have more or less close overlap. An inductor is usually a coil around a tube and metal sliding in and out of the tube would change the inductance. or
You can vary them in fixed, discrete amounts: Keys on a keyboard can complete connections to specific fixed capacitors.
You can feed all your frequencies through the same amplifier. It sounds like you might not be comfortable with integrated circuits, They are cheap and easy to use. These days, a 50 cent Texas Instrument chip can amplify signals with very little design effort on your part.

 

[jim hardy]

JonG:

get yourself a RCA Receiving Tube Manual

the 1960 edition was my high school textbook.

http://www.ebay.com/itm/RCA-RECEIVING-TUBE-MANUAL-TECHNICAL-SERIES-RC-20-VACUUM-TUBE-BOOK-1960-Rare-/251971322933?pt=LH_DefaultDomain_0&hash=item3aaaa94435

a few bucks at Ebay or Amazon

 

[FactChecker]

JonG:

get yourself a RCA Receiving Tube Manual

the 1960 edition was my high school textbook.

http://www.ebay.com/itm/RCA-RECEIVING-TUBE-MANUAL-TECHNICAL-SERIES-RC-20-VACUUM-TUBE-BOOK-1960-Rare-/251971322933?pt=LH_DefaultDomain_0&hash=item3aaaa94435

a few bucks at Ebay or Amazon

Ha! I haven't seen one of those in a long time! Nostalgia is hitting me hard.

 

[dlgoff]

Nostalgia is hitting me hard.

It can be like that around here.

 

[nsaspook]

The Navy tube bible.
http://www.phy.davidson.edu/instrumentation/Files/NEETS/Mod06 - Electronic Emission Tubes and Power Supplies.pdf

Space charge tubes:
http://www.rfcafe.com/references/ra...-tubes-january-1957-radio-television-news.htm

 

[jim hardy]

Space charge tubes:

now that's interesting...
my 1962 Chrysler radio had 12 volt B+ tubes and one output transistor , just as in that article... i always wondered how they did that.

 

[anorlunda]

Ah yes Jim,

I too had one of those RCA Receiving Tube Manuals, soon followed by a parts catalog from Digital Equipment. I used tubes to build analog computers. I applied for a job at Hammond Organs, but got turned down.

But I got seduced into the software side when I discovered that I could write machine language to make the computer play Old MacDonald on a transistor radio by the RF leakage. Never went back to analog after that.

 

[jim hardy]

I could write machine language to make the computer play Old MacDonald on a transistor radio by the RF leakage.

Now there's a practical use for the doggone frustrating things.

seduced, you say ? How apt.
Most computers today are used solely for entertainment...

do you think the industry would be better or worse off today had IBM settled on DEC or Motorola architecture rather than Intel ? I found those 'rolled nibbles' infuriating.

 

[Svein]

industry would be better or worse off today had IBM settled on DEC or Motorola architecture rather than Intel ?

To make it even worse, IBM (on the PowerPC) had its own way of numbering bits and bytes. To summarize:

• Intel: Bit 0 is the least significant bit, byte 0 is the least significant byte.
  
• Motorola: Bit 0 is the least significant bit, byte 0 is the most significant byte.
• IBM: Bit 0 is the most significant bit, byte 0 is the most significant byte.

Both the Intel and the IBM way is consistent, Motorola's is not. Therefore the Motorola way was chosen as the standard way to represent multibyte objects on the net.

 

[anorlunda]

do you think the industry would be better or worse off today had IBM settled on DEC or Motorola architecture rather than Intel ? I found those 'rolled nibbles' infuriating.

You're probably right. Many many sins were committed in attempt to spare a few bi ts of expensive memory. But the savings in reduced memory were squandered in more difficult software.

 

[jim hardy]

Many many sins were committed in attempt to spare a few bi ts of expensive memory. But the savings in reduced memory were squandered in more difficult software.

I've read that the decision resulted from a management edict " Thou Shalt Be Different from Apple".

Even the lowly TMS9900 had superior context switching (any memory location can be assigned as program counter , next n locations as stack)

well it is what it is and it's working (sorta) .

 

[Svein]

Even the lowly TMS9900 had superior context switching (any memory location can be assigned as program counter , next n locations as stack)

Well, not quite (I have programmed quite a large amount of code for the 9900 and that was a frustrating experience).

The TMS 9900 (I hope I remember correctly, check me at http://en.wikipedia.org/wiki/Texas_Instruments_TMS9900) had 2 physical registers (plus a "register containing status bits): The Program Counter and the Workspace Pointer. The Program Counter was more or less standard, but the Workspace Pointer was not. It pointed to an area in main memory containing 16 16-bit "registers". It had no stack pointer, which meant that you had to implement the equivalent of a stack in software.

The TMS 9900 was odd, but the development system (a TI 990 computer) was horrible. Thankfully it disappeared after a short time.

"In typical comparisons with the Intel 8086, the TMS9900 had smaller programs. The only disadvantage was the small address space and need for fast RAM."

 

[jim hardy]

mmmm okay , my memory isn't photographic
program counter and workspace pointers point to an area of memory used for registers as opposed to having them dedicated in hardware.
Simplifies context switching , per section 2.1 here

http://www.textfiles.com/bitsavers/pdf/ti/TMS9900/TMS9900_DataManual.pdf

i defer to your greater experience. I do know some programmers who loved it, though.

 

[Svein]

Simplifies context switching , per section 2.1 here

Yes, since exceptions carry their own workspace pointers. But you need to be careful, since the return address is saved in a "register". This is why normal subroutine calls are tricky, since there is no stack and the return address is saved in a "register". I had to invent the concept of "moving register windows" several years before it appeared in RISC processors.

 

[nsaspook]

To make it even worse, IBM (on the PowerPC) had its own way of numbering bits and bytes. To summarize:

• Intel: Bit 0 is the least significant bit, byte 0 is the least significant byte.
  
• Motorola: Bit 0 is the least significant bit, byte 0 is the most significant byte.
• IBM: Bit 0 is the most significant bit, byte 0 is the most significant byte.

Both the Intel and the IBM way is consistent, Motorola's is not. Therefore the Motorola way was chosen as the standard way to represent multibyte objects on the net.

It depends on the physical layout of the shift registers, you always want the higher analog order values first if the data is used to generate analog data on the fly. The analog circuits slew rate limits large changes so by the time the lower order bits have been clocked in the voltages from the higher order byte(s) have settled.

 

[anorlunda]

It depends on the physical layout of the shift registers, you always want the higher analog order values first if the data is used to generate analog data on the fly. The analog circuits slew rate limits large changes so by the time the lower order bits have been clocked in the voltages from the higher order byte(s) have settled.

Now that is interesting. I recall reading an entire special issue of IEEE Computer magazine about the Big Endian - Little Endian debate. In all those pages, nobody mentioned what you just said.

 

[Svein]

It depends on the physical layout of the shift registers, you always want the higher analog order values first if the data is used to generate analog data on the fly. The analog circuits slew rate limits large changes so by the time the lower order bits have been clocked in the voltages from the higher order byte(s) have settled.

Two points:

1. There is no rule that says what bit should arrive first out of a shift register (which has to be external to the CPU in any case).
2. Most DACs take parallel inputs anyway (I know, there are some using SPI and a small number using I²C)

 

[nsaspook]

Two points:

1. There is no rule that says what bit should arrive first out of a shift register (which has to be external to the CPU in any case).
2. Most DACs take parallel inputs anyway (I know, there are some using SPI and a small number using I²C)

Right, there is no rule but there is best practice where serial data is used to stream multi-byte clocked data where the analog signal is created from each bit in step (on the fly). Today almost everything (at the consumer level) but the fastest converters are serial.