Miscellaneous Projects

Some fun projects you might want to try:

4069 Theremin

A very simplistic Theremin, using only one 4069 IC. The reference oscillator uses a 455KHz ceramic resonator, and the varying oscillator uses a tunable 455KHz IF can. The beat frequency is demodulated an amplified before being fed to the speaker.

Neon Color Organ



Click here to see video

Neon bulbs halves modulate alternately with ambient sound. Can use any neon, but better with ones with electrodes of which either can be anode or cathode.

Phototube Amplifier

Makes the HV voltage needed to run an old phototube. These tubes use only a few tens of microamps, so the 2N5401 is plenty good for the job. Use the op-amp type shown. If you’re going to sub it, then use a ST TSH22. An LM358 is way too slow. The TSH22 and On-Semi MC34072 are both “single-supply” op-amps, and have high slew rates needed to operate this circuit.

Click Here to see a video of it in operation. The pulses are from the remote control, its detecting.

Driving a lot of LEDS

Need to run a  few new super bright LEDs, from a low voltage source. This will do it using the old reliable MC34063 switching supply chip. Modern GaN superbright LEDs (White, Blue, and true Green), run at higher voltages than the red and yellow LEDs. Often as high as 4.5V. That gives you little head room, if you want to run them off a 5V supply. Using this circuit you can tie a bunch of them in series, so they’ll all get the same current. Tie as many together, as long as the total voltage drop is less than 40V, the MC34063 limit. The circuit is rigged to regulate current, not voltage. That’s what makes LEDs happy. As shown, that current is 18mA. A safe value. The circuit will regulate to 18mA, and the voltage at the top LED will be whatever it needs to be to get 18mA thru.

Portable High Voltage Supply:

Get +12V and +200V from a 6V “gel-cell” (sealed lead acid) battery. You MUST use an MC34074 quad op-amp, for this to work Parts like the LM324, LF347 or TL084, will NOT work. The LM324 is too slow.  The MC34074 can swing all the way to ground, like the LM324, but its as fast as the LF347 and TL084. The LF347 and TL084 can not swing all the way to ground.

There’s this thing “specmanship” when it comes to datasheets. Op-amps have this thing called “gain-bandwidth-product”. That’s the frequency, where the op-amp’s gain is one (unity). Its a useless number. All the verbage in the first two paragraphs of most datasheets is useless, if not deceptive, marketing garbage. The real, useful data is after that, and needs someone with training and experience to properly make use of it. A LM324 may have a gain-bandwidth-product of 1MHz, but as a full-swing (rail-to-rail) oscillator, it can’t go much faster than 5KHz.

Toy Phasor:

Here is a little noise maker, with blinky lites, that reminds you of the old Star Trek Series:

Click Here to listen to a short audio clip.

Telephone Ringer Light

Here’s a little circuit, that makes a 120VAC light flash, when the telephone rings. Requires an old fashion land line to work:


Note: U1, a 4N35 opto-coupler, isolates the telephone circuit, from the 120VAC line circuit. J1 is an RJ11 modular telephone jack. The center-most two pins are to be tapped on this RJ11 jack. If wires come out of it, they are colored red and green, by convention.

Marshall JCM2000 Bias Board Settings:


Back to TOOBs !

I picked up a little … er … actually quite big 866 mercury rectifier. This tube was intended to be used in supplies that could produce a a couple of KV of DC at 250mA, continuous ! That’s quite a bit of oomf (>500W) ! Here is it powered up from a big transformer, I have lying around:


But I need that transformer a new bench supply, so I decided to make a little switcher, that converts 12VDC to the 2.5V (5amp) filament power needed to heat up this tube. Here is the circuit (click on image, to enlarge) :


Here’s is the link to the video, to see it operation. It emits a bit of UV, so take care, when viewing a real tube, lit.

On the other hand, the actual volatges are quite low. Nothing over 30V, or so. Ironic for a tube capable of rectifying 2500Vrms !

The circuit uses a TL494 chip, rigged so the 2 outputs switch ON, complimentary to each other (one ON, while the other is OFF). The chip drives two power FETs that route current into a transformer, in the classic push-pull arrangement. The transformer is a ferrite toroid with a center tapped primary (20 turns + 20 turns, 40 turns total), and a secondary of only 4 turns. On the input, only a ~1.3 amp is drawn, while the secondary delivers 5 amps, at ~2.5Vrms. The switching frequency is ~30KHz. A side benefit, is that the primary has a little “flyback” transient, that can be tapped for the ~30V plate voltage for the tube. With the parts shown, the plate current is ~60mA, so common 1N914 diodes can be used, instead of the MBR160 Schottky rectifiers. Note: both of these diodes (1N914 & Schottky) can switch at nanosecond speeds. Your crappy 1N400X cannot, and should never be used in a switching supply !

Cellphone Tube Amp Dock:

Here’s a little tube amp I made to play tunes off of my cellphone:



And the schematic for it:


Even though that output transformer is tiny, the bottom doesn’t roll off until just under 90Hz or so. Its feeding two 4 ohm, 2″ speakers, so don’t expect great bass response. It does sound a lot better than the phone alone. And it sounds better than most docks I’ve heard in stores. Plus how many docks have you seen that are made of wood !?

More Phototube Stuff:

Here’s a demo circuit that uses a phototube and a small thyratron, to switch a relay. This simple circuit was included as a sample app, for small RCA thyratrons like the 5727 (premium 2D21) and the 5696. Its drawn a little different here, buts its almost exactly the same circuit as shown in the datasheets, for the two tube types, above. Here is the circuit using a Cetron B-25 (Red, gas) phototube, and a 5663 thyratron. The 5663 specs similar to the RCA 5696, and even closer to the GE GL-546. Even has the same pinout as the GE tube:


The only difference between the drawing, above, and the datasheet circuits, are that the values are given here, and that the filament is shown being powered thru a 3.3uf film, current limiting cap. Also the load is an Omron G5LE-14 24V relay. The coil voltage is really unimportant. What matters, is that the current draw, to engage the relay, is less than the 20mA max average, that the 5663 can handle. In the circuit, I built, it was drawing ~15mA.

When Operating this circuit, apply the 120VAC, with S1 “OFF”. Keep it OFF, for at least 30 seconds. The filament is not switched, so this allows the cathode to heat up, and form the electron cloud around it. Once warmed up, then S1 can be closed (turned ON), and present plate voltage. Note that there are no rectifiers, and that this circuit works on mostly AC. The plate circuit, does rectify the current going in the relay, and the 47uf cap, filters it into steady DC, going thru the relay coil. Don’t do the “capacitor trick”, for current limiting, other than in the heater circuit. Current limiting using caps, works by shifting the phase between voltage and current. In thyratron circuits, that can wreak havoc. S2 just selects which contact is connected to the lamp load (NC or NO).

Here’s a video of the circuit.

Here’s data on the GE 5663.

Below is the original circuit from the RCA datasheet:


The resistors R1, R2, & R3 correspond between the two drawings.  This snippet, is out of both the datasheet for the 5727 and 5696. Since, I’m using a gas phototube, instead of a vacuum type, I tap in the middle of “R2”, so the voltage rating of that tube is not exceeded.

No data found on the Cetron B25. From the coating on the cathode, its response is towards the red side (~800nm). If I hit it with an IR source, in the dark, and see it slightly glowing, then its a gas phototube. Care must be taken with these. They are more sensitive, than vacuum phototubes, but should never be exposed to more than 90V peak !

 Homemade Plasma Globe / Generator

A quick way to see if your cold cathode tubes still have gas in them, is to place them next to a running plasma globe. Here is a circuit that will provide a high voltage AC source (click on 1st photo to view video):



Lighting up a Hollow Cathode (spectral reference) tube:


Here is the schematic (Click on drawing to enlarge):


The circuit is a modified MC34063 nixie power supply. The inductor has been replaced with a transformer, with a 20:1 turns ratio. The circuit generates 300Vpp of pulsed DC at the primary. Only the time varying signal gets transfers over to the secondary, and multiplied by 20, as ~6KVpp. The frequency is roughly 30KHz, but its anything but a sinewave, so the a lot of higher frequency harmonics.

Here’s a link to a vendor, that sells E-Cores, and bobbins, for the transformer:

Amidon Corp.

Test Amplifier

A quick solid state audio amp, when you need amplify a drive a speaker. It uses a National Semiconductor (now TI) LM1877 stereo 1W amplifier chip, hooked up in a bi-phase arrangement, so the output voltage swing can be near twice the supply voltage.


Remarkably, very good audio quality, for a single chip amplifier, that delivers less than 10W. No hum, noise, or noticeable distortion.

Joule Thief using Common Mode Inductor

Joule thieves are very common circuits found on the web. Most are based around the blocking oscillator, which goes back to the tube days. The problem is that it requires a transformer, or at least a coil with a centertap. This often means winding your own around a ferrite or powdered iron toroid. But there still exist common two coil inductors used in modern circuits. They are common mode inductors, and they are intended to be used in RFI/EMI filters, to keep the “hash” from modern switching supplies from polluting the AC line. Below is a circuit that uses such a coil:


Like all joule thieves, it boost the voltage from a single 1.5V dry cell battery and boosts it high enough to light new ultrabright GaN, blue, green, or white LEDs. But instead of requiring a custom coil, it uses an off the shelf standard Kemet SU9H-07010 (1mH) common mode choke. Some added features, in addition to the coil choice: (1) a PNP transistor. With a silicon transistor, this circuit can run on a battery as low as 0.7V. Germanium transistors can even operate lower. As low as 0.25V. When germanium transistors were common, they more often than not came as PNP. NPN germanium transistors were harder to make. (2) a voltage doubler rectifier, and filter circuit has been added. Usually the LED is just placed across the pulsed output. This wastes half the signal. This doubler circuit, uses the whole AC pulse generated. To waste even less, BAT43 schottky diodes are used. Even though common 1N914/1N4148 diodes will work, the BAT43, will squeeze out another half volt.

Voltage Regulator Tube Tester

Jig for testing old voltage regulator tubes. These are the tube version of zener diodes. and have numbers like 0A2, and 0B3. The leading character is a zero, not an “o”, but in the days before personal computers, “o” and zero, where regularly interchanged, in informal conversation.


A couple of warning notes. Only one  socket is to be occupied at any one time. Use an electrically isolated meter, since this circuit ties directly to the AC line. A battery powered portable is advised. There are two small sacrificial resistors (R1 & R2, 47ohm 1/4W) inserted in each AC leg. In case a mistake is made and a grounded bench meter is used, then one of these resistors will vaporize

CRT Deflection Amplifier:


With parts shown, tested to have bandwidth of ~200KHz. Can be used as part of solid state analog scope (using old CRT tubes), along with this HV kit.

Here is the example of that CRT power supply, using the HV Kit:


As shown, you’re stuck to the AC line, since the filament is powered by a 120V to 6.3V filament transformer. Why not just connect the filament to the 12V, with a limiting resistor, or 6V regulator, you say ? Can’t, its not that simple. That filament must not stray too far from the cathode potential. Even though the cathode (K of the CRT) and its filament (H-H) are isolated from each other, the voltage between them must not differ much more than ~100V. If it does, current will arc across the small gap between them. That cathode is more than a 1000V negative relative to ground (circuit common), and the low 12V input.

A low isolated 6V can be made using a Royer oscillator. It will convert 12V to a high frequency pulse train, that can pass thru a small toroid transformer, with very few windings, and is very compact. Here is the circuit:


This circuit’s function is to power a 6.3V filament, yet be electrically isolated, from primary to secondary, in excess of 1000V. The two windings are electrically separated from each other, and power is transferred thru the shared magnetic field. The transformer core is a small half inch diameter ferrite toroid. To enhance electrical isolation, the core is to be wrapped with several layers of Kapton tape (5mm wide), so that voltage does not arc thru the core, from primary to secondary.

Royer circuits, like this have been used commonly, for induction heaters, so they can transfer a lot of power. This circuit has been tested to deliver over one amp, with little voltage drop, and the transistors staying cool. The components that do warm up, are the transformer, and C1. Use a film cap with very low DF. Possibly even a foil cap. 250V rated or higher. Oscillator frequency is in the ballpark of 50KHz.

Extending Range on Chinese 3-Wire Meters

In recent years, inexpensive 3 digit LED, 3-wire panel meters have become available, out of China. These are descendants of LCD digital panel meters that have been around for sometime. The problem with those old meters, is that their power source needed to be electrically isolated from what was being measured. An unnecessary complication, with extra circuitry. Recently, from China, a simpler design came out, using a common reference, or ground, shared by both the power source, and the measured signal. They can be found on eBay, searching under 3-wire (or 3-line) digital panel meter. They come in various ranges, from 10V max, to 200V max. All DC. All need to be hooked up in the proper polarity. Their input impedance is rather low. Here is a circuit, that extends the range of a 0-10V unit. The measured signal is input to the blue wire, and the shared black common. Power (+4V to +30V) is applied to the red wire. The input impedance was measured at 28.4K:


A 100:1 voltage divider is added. The values of the resistors were chosen to yield an input impedance a tad greater than 1M ohm. Note R1 thru R8 make up the upper leg of the divider. This makes a 992K 2W resistor. several resistors are used instead of one 1M, to distribute both the power, and voltage. Up to 1000V may be applied to this divider. At 1000V, each resistor will dissipate ~1/8W and have ~125V across each part. Note 1/4W metal film resistors have a maximum voltage of 250V, in addition to their power limit of 250mW. Rs1, and Rs2 are shunted in parallel with the meter. The combine resistance is to be 1/99 of that of the sum of R1 thru R8. This works out to ~10.02K. Rs2 allows an adjustment range from 8.9K to 11K. A multiturn trimpot is desired. Trimpots with 10 or 25 turns are available. Instead of using an ohmmeter to set it up to 10.02K, apply a voltage >100V, to the unit working, and connect a good multimeter, in parallel, measuring the same voltage. Adjust Rs2, until the two readings match.

Note: The panel meter is reading 0-10V, while 0-1000V is being applied. That means the decimal point is off. So ignore the decimal point. It is invalid.