disaster electric: XVI: The Tower Vacuum-Tube Preamp and Demodulator.

REVISION NOTE:

This design has been revised. An updated schematic is available. These principles of operation still apply, however the revised design eliminates excessive high-frequency noise (the kind that sounds like nails on a chalkboard)

PDF on Github

Click on the image to embiggen, and then right-click and select 'open image in new tab' for full size.

XVI: The Tower is a trio of pissed-off pentodes that hate you and whatever you put through it. It will send your audio signal into the pits of an alien hellscape, and then give whatever is left of it back to you. If you plug CV into it, it will turn into a ring modulator. If you don't have CV plugged into it, it will modulate itself.

I am pretty proud of this monstrosity, I won't lie. It's insane. It's got it all.

Are you a boring, middle-aged white dude who just wants some 'WARM TOOB SATURATION ON MY MIX BUS, BRO, JUST TO TAKE THAT digital edge OFF?" This can do that, sure.

Are you a boring middle-aged white dude who USED to be hardcore back in the late 80's and early 90's, but you never let go of your inner rivethead?

Holy shit do I have just the thing. This is going to set a fire to your loins, let me tell you.

Do you like birds? Oddly enough, this is for you, too, I guess? Because it can make some killer bird sounds.

Look, this isn't for everyone. This is weird shit that uses weird tubes in weird ways. This isn't some regurgitated HM-2 "inspired" clone, or some Buchla or Moog "inspired" thing. The only comparison that I can give you is Metasonix. If you like Metasonix, you will love this. If you hate Metasonix, you will hate this.

I am not a sales-person, nor am I a business-person (if I was I wouldn't be just vomiting this shit all over the Internet, I would be trying to sell this god damn thing) so that's my pitch. It's a DIY project, anyway. Either that sounds like your jam or it doesn't.

I. Bill of Material.

Click on the image to embiggen, and then right-click and select 'open image in new tab' for full size.

II. Power Supplies.

I designed, tested, and intended for this circuit to work with a B+ voltage of 48VDC. This is low in comparison to most tube circuits, which often require a B+ voltage of at least 200VDC. You may get it to work at 12VDC or 120VDC but you will need to make adjustments that are beyond the scope of this writeup.

The power supply I used is a self-contained AC/DC supply from Meanwell. The part number is IRM-90-48ST. It takes 1.88A at 100-240VAC, 50 or 60 Hz directly and puts out +48VDC at 1.9A. It is fairly small and unobtrusive. As I built my Tower in an old test equipment enclosure (a Ballantine 300D VTVM if you're curious), I had plenty of room to spare.

To get the roughly 1.4A of 6V required for the filaments, I installed a 6.3V, 3A transformer in the enclosure, as seen below. (Click on the images to expand, right-click and select 'Open image in new tab' to see full size.)

As with most things, there are a thousand different ways you could get filament power and B+ voltage to the circuit. DC/DC buck or boost converters, switch mode supplies, linear regulators, etc. Most of the 'Nixie'-style SMPS boards available will work just fine without any added noise. I just used what I had in my parts bin.

Ultimately, I did what most designers seem to do - I left the power supply out. This serves multiple purposes. It is much simpler to draw; it leaves the question of power supply to the end builder to answer; and it tends to ward off anybody who wants to build this, but doesn't really have the knowledge or ability to build it correctly and troubleshoot the circuit when it inevitably goes pear-shaped.

Is it gate-keeping? Yes, it is, but it isn't frivolous, nor is it born of any ill-intent. This isn't a hobby project in the same vein as, say, scrapbooking. Or whatever the hell normal people do.

III. Block Diagram.

(Why did I make a block diagram? I don't know. They had one in all the service manuals for every old, battered piece of test equipment that I've restored...)

IV. Input Stage and First Stage Amplifier.

I will not lie - the old adage 'good artists copy, great artists steal' has been taken almost literally in regards to this first part of the circuit. This input stage is entirely lifted from the Metasonix TM-7 Scrotum Smasher (and subsequently the TM-5 preamp). It is a shit resistance-coupled amplifier stage, and it is designed to be shit on purpose. Guitar nerds and hi-fi snobs would probably have an apoplexy if they saw this. I mean, the plate current is only 222 fucking MICRO amps?! That can't POSSIBLY be ANYWHERE near the LINEAR REGION, RIGHT?!

Piss-taking aside - Component values have been changed based on my personal tastes and what I was trying to achieve. I used 1uF for cathode bypass as opposed to 47uF, for example, because a higher bypass cap muddied things up too much for what I was doing with it in this circuit.

More examples: C1 in the TM-7 is 47nF, while I made mine quite a bit larger at 220nF. VR1 and R2 are sized differently as well. I chose 500K for the input level pot to give finer control, as the circuit can and will burst into self-oscillation and the character of destruction changes dramatically at seemingly random points. R13 simply shifts the adjustment band of VR1 by 100K, so the control range is limited to 100K when turned up all the way. This ensures a minimum grid leak resistance of 100K going into V1. On the TM-7, these are 100K for VR1 and 470K for R13.

The tone control serves two purposes here. It acts as a filter and, depending on how one sets the rest of the controls, it will interact with VR3 (Gain) as a frequency tuner.

V. 'Demodulator' Stage.

In the interest of candor, I do not know exactly what is going on here, so I am going to make like an American prescription drug commercial and explain how I believe it to be functioning. At least the idea of what I want it to be doing.

Our now-amplified signal splits off to the CV jack and also to pin 9 (control grid 1) of the 6LE8. Going back to the datasheet, this is our 'reference signal.' Pro tip about CV: This thing LOVES square waves.

The CV input (at the top left) is a switched jack. Without anything inserted, the signal from the output of VR3 (Gain) is filtered and fed to pin 2 of the 6LE8. When CV (either audio or CV) is injected, that CV is filtered and then fed to pin 2. This is one 'phase' of our 'chroma information.'

The two plates are tied together. Our 47K load resistor gives our tube a mighty 1.02 milliamps to play with (I tried different plate loads all the way up to 1.5 megohms but settled on 47K). That's not much for a tube that has an average plate current of 17 mA specified in the datasheet, and is rated What this does is greatly impedes the tube's ability to 'demodulate' anything. It makes the demodulation effect erratic, at best. It's like asking someone with one arm to juggle. If you give the tube more juice, it is going to do its job better - which is exactly the opposite of what we want!

The output of pins 1 and 6 is fed through C10 and into another gain control (VR5) identical to the LEVEL control. After VR5, the signal is split. One leg goes into V3, and the other branches back as two feedback loops. This means that VR5 not only controls the gain going into V3, but also the amount of signal going into the feedback loops.

The 'Color Killer' control (VR6) is so named because it looks kind of similar to the color killer circuits used in old color televisions, and also because it sounds cool. In reality, it works with C11 as a high-pass filter. The filter's primary purpose isn't to filter, though - it's to introduce phase shifts. This filtered signal is sent to pin 7 and is the second 'phase' of our reference signal.

VR7 is our positive feedback control. I used the term 'Regeneration' because it sounds cool. The signal, having had its phase inverted twice in the first two tubes, is back in phase with the input signal. This control, along with VR6 and VR4 (Bias) wildly alters the effect this entire thing has on whatever poor signal you passed through it.

VR4 does exactly what it says on the tin - it shifts the bias of the tube, which changes frequency response and the behavior of the entire circuit. C10 increases the overall gain of the tube, especially at lower frequencies. I experimented with this value a great deal, but any lower than 100nF tended to make a lot of AM radio noise and any higher had no notable overall effect on frequency response.

The diodes (D1 and D2) are there along with R10 and R11 to bias both of the number-3 grids. I used 1N645 because I inexplicably have a large amount of them that I got on eBay in a lot of miscellaneous components. For our purposes, they are nearly identical to 1N4148, except with higher surge current and reverse breakdown voltage ratings, which are inconsequential for our use.

VI. Final Gain Stage and Output.

Since the previous two stages introduce a ludicrous amount of harmonics, oscillations, and erratic noise to the signal that go into painfully high frequencies and beyond, I added a simple low pass filter with a cutoff of around 15 kHz at the input of the third and final tube, right after the gain pot.

The rest of this portion is lifted directly from the TM-5/TM-7, with minor component value changes much like the first stage. The output is divided down via R8 and R18 to make sure the outgoing signal doesn't damage anything plugged into the output.

VII. Tube Types.

The tubes used for V1 and V3 are extremely negotiable. Several varieties of small sharp-cutoff 7-pin pentodes, of the 7CM, 7BD, or 7BK base type, may be used to varying effect. Some suggestions would be EF95, 6AK5, 6AU6, 6Ж1П(6J1P), 6J1, 6BZ6, or 6AW6. Mind your filament draw.

V2 is non-negotiable. There are other 'twin pentode' types, such as the 6BU8, 6GS8 or 6HS8, but they are a different pinout and will not work. If you have those tubes and want to experiment, you'll have to rewire the socket (which isn't hard). Both the 6GS8 and 6HS8 are meant to be sync separator-clipper/AGC tubes, and they use far less filament current than the 6LE8. There are also 'twin tetrodes' - such as the 8457, meant for use with battery-powered radio - which the enterprising experimenter may wish to employ, however they are rare and expensive.

VIII. Waveforms.

The following photos are my attempts to photograph my oscilloscope with my DSLR, so go easy on me. Click on the images to expand, right-click and select 'Open image in new tab' to see full size.

These first three photos are the output of the Tower when it is being fed a sine wave from a function generator, with the controls in various positions. The third and fourth images show some nice self-oscillations mixing with the input sine wave: