All tube digital delay

[R.G.]

This one came up when I was discussing the state of the music industry with my amp-tech friend.

Imagine this: an all-tube digital delay. Incredibly good advertising copy. I can see it now "the delay that takes the hard edge off digital! Our NEW all-tube digital delay restores the smooth tone of tubes to the delay abilities that only digital can do!"

Quantization noise is reduced about 6db per bit; for a reasonable sound fidelity, you need 14 to 16 bits. One dual triode can make one bit of storage flipflop, so you need sixteen dual triodes per delay location.

For reasonable digital delay you need maybe 4K locations, so you're talking 64 thousand dual triodes. I bet I could design a reasonable A/D for less than a thousand tubes, and a similar amount for a D/A, so we need 66K dual triodes.

Each triode uses 12x0.15= 1.8W of heater power, and maybe 1/2 W of operating power, so we get 2.3W per tube. Total DC power needed is then
66K*2.3 = 151.8kW.  The conversion efficiency of the power supply is probably around 75%, so we need 202.4kW of power to run the thing.

All the DC power in comes out as heat, so we need an additional 202.4kW of air conditioning for the beast. AC is about 4:1 efficient on a thermal basis, so we need another 25% to cool it, up to about 250kW for total power, about one modest sized substation.

I guess we need some tube sockets, cabinets, and so on. A dual triode needs about two square inches of surface area and about 3" of length for the tube and circuit parts. OK, lessee, 66,000 times 2 times 6 = 396K cubic inches, about 229 cubic feet. Not too bad, but let's put it into 19" relay racks. There's 17" of usable width, maybe 6" of usable depth, so lets call it 12 tubes wide by 4 deep, 48 tubes per rack. That's 1375 racks of about two rack units (1RU= 1.875"), or 5156 inches/429 feet of rack space front. That's 72 6' high racks.

You could get this thing in a semi trailer if you insisted on being able to walk between racks to replace tubes; getting the AC and power supplies in is a maybe - I'm an EE, not an ME.

Anybody want to go into business?  I think we have a winner.

[gez]

Anybody want to go into business?  I think we have a winner.

And if you DO want to go into business, do you also want to buy a used car? (one careful owner...a nun I believe)

[Transmogrifox]

Yeah, I'll go in business.  I have a better idea, though, we'll make a digital delay with semiconductors, then put a triode in the box with the heater connected so it glows, then put a transparent window on the front so the glowing tube is visible.  

Then we can market a second version that actually uses the triode for an input and output buffer to add the "warmth" of a pure vacuum tube digital delay. That's what I call an "all tube" digital delay with some kind of a digital IC to regulate the tubes' pure delay tone.

[puretube]

why not go back in time for about 50 years and use techniques that were
avaible then and known to those skilled in the art... :

1 bit adaptive PCM A/D ("sigma/delta") : ~3 x 12AX7 (or equiv. miniature);
4k locations: 8k sub-mini pencil-tubes (long life), no sockets (solder in),
filaments: 0.625V@10mA; (size: imagine a neon-indicator);
plate-voltage 15V @ <1mA*;
Simple integrating D/A ~ 2 x 12AX7 (or equiv.).

* : for "1"s, for"0"s : 0.0...mA;

Anybody calculate...?

[Peter Snowberg]

With 1 bit, quantizing much be done much faster (>1M samp/sec). Now you need many more stages. Why not use BBD techniques? Analog tube delay is just more at home to me.

I worked a bit on a delay that was built in 1957. It used a GIANT piezo crystal (almost a foot in diameter) as the delay medium. All the processing was tube too. It was the most top secret circuit in there at the time. The application? The Moving Target Indicator signal processor in a Nike Hercules missile system acquisition radar. The analog guidance computer used over 70 dual tube op-amps.

-Peter

[puretube]

...R.G. & his friend wanted it digital - completely;

of course they know everything about storage disks, tubes, and the oil-can stuff being comparable to BBD;

but that`s just chopped in the time-domaine;

they want it the hard way... at least 256 voltage-steps...

btw.: how about Tube-to-Midi or Midi-to-Tube?


(seriously: haven`t heard about that piezo-storage yet - interesting!
was that used kinda like a reverb-spring/plate, but at high frequency?).

[Peter Snowberg]

Ah.... all digital.... I see.

It's still a cheat, but what about using BBD and then a 256 step comparitor and resistor ladder at the end to posterize the signal? It would be "almost as good as digital" . LOL. It might have "space age digital vibe".  :?

Does it have to be ALL tube? Could you use core memory for storage? It's digital.

Midi to tube? a synth module?

Maybe a big wavetable made of diodes?


The piezo delay would subtract returns from the radar that were moving at speed of less than ~350 knots. The US was expecting the USSR to use very fast or supersonic bombers so they only watched for high speed targets. It was very similar to a spring reverb unit is some ways.

-Peter

[puretube]

Maybe a big wavetable made of diodes?

Yes, of course tube-master-clock, tube dividers, and tube-diodes (triodes wired as~) for the matrix.

Instead of resistors, for the ladder-thing I`d take Varistors ("Magnatone").



Hey, how about tube-digitizing (PCM again) and recording that on tape?
That would be the all-in-one-solution:

"DIGITAL-TUBE-TAPE-DELAY" (D-T-T-D may become a new reg. trademark...);

(including all-analog wow&flutter - saves an LFO/random-gen.)