After reading the title of the article, you may have thought: “But why ?!”, “The efficiency of tube amplifiers is very low!” or “What is the point of connecting a Bluetooth module to a tube amplifier?!” etc. There is no point. The device is assembled for the sake of the device, for the sake of entertainment. I love tube radios very much, I like to study their device, admire their appearance and even just contemplate at dusk the magical lights behind the back wall, radiating warmth, comfort, romance and a kind soulful voice from the loudspeaker … But unfortunately, in our time it is possible to receive mostly foreign stations and only at night. So sometimes I amused myself by listening to music by connecting to the amplifier of the radio. So the idea arose to make a portable Bluetooth speaker in retro style with a tube amplifier without any pretensions to High End.
Selection and test of radio tubes, the first layout
Initially, I wanted to keep the design of the speaker in the style of amplifiers and radios of the early 30s. Therefore, I was looking for lamps with spherical flasks “Globe”. The choice was made on pentodes, as they turned out to be much cheaper, more offers in online stores, higher output power than triodes at the same anode voltage, and also more economical heating. In addition, a lower negative bias voltage is required, and the output stage can be easily driven without an interstage transformer. To clarify, I am writing exclusively about “battery” lamps, it is much more difficult and expensive to find such a triode in a preliminary cascade with a high gain, and I didn’t want to increase the number of cascades, as well as use an interstage transformer, since this is a heavy, complex and expensive part. Of course, the pentode has disadvantages: higher non-linear distortion, a higher level of intrinsic noise and a sharp dependence of the gain on the load resistance. But we are not constructing a “High End”, we will survive somehow.
As a result, RCA-233 battery pentodes were purchased in a “Globe” flask for the output stage. Heating voltage 2 V, heating current 260 mA. Output power at an anode voltage of 135 V – 0.7 W. As you can see in the photo above, in terms of dimensions they are no different from network lamps. In fact, these lamps are designed for stationary battery radios, which were very popular with millions of farmers living far from the city and the mains. But unfortunately, it was not destined to use these lamps in the layout and the finished device. Both lamps turned out to be defective, the grid closed to the filament. Why this happened and in which bulb the radio tubes are better, we will consider below.
I also ordered more modern radio tubes of the same 33rd Tung-Sol 33 series in a figured flask “ST” (shoulde type). They were only going to be used for experimentation and in an amplifier layout so as not to damage the more vintage and expensive light bulbs. But it was they who became the heart of the wireless speaker. In addition, the curly flask also looks very beautiful.
In radio tubes with a “ST” bulb, the tapered upper part is used for additional fixation of the electrodes. This is achieved by inserting a mica disc which is clamped at the top and holds the electrodes. At the same time, in radio tubes with a “Globe” bulb, for example, in triodes, mica in the upper part may be completely absent. There is some speculation that the absence of mica in tubes improves their sound, as any additional materials in the bulb can release gases during their lifetime. Perhaps there are other reasons affecting the sound. For example, the difference in materials that were used in the early days of radio tubes. Let’s leave these assumptions on the conscience of “high-end players” and “audiophiles”. On the other hand, the lack of additional support for the electrodes makes the lamps in the “Globe” bulb more prone to microphonic effect.
The question of whether to use radio tubes in the ST bulb or look for options with the Globe is simply answered by the fact that the latter are becoming rarer and more expensive. Due to their “fragile” design, they are much more demanding in terms of storage and transportation. More prone to mic effect, which can be a problem when used as a driver or preamp tube, as more samples will need to be sampled. Also, all this will make it difficult to find the perfect pair.
To test the radio tubes, I assembled a simple layout. In the output stage, the Tung-Sol 33 pentode, in the preliminary tetrode JAN-CKR-32 (VT-44). I did not specifically choose manufacturers, I bought what was cheaper. The heating voltage is 2 V, the heating current of the pentode is 260 mA, the tetrode is 60 mA. The output power of the amplifier at an anode voltage of 135 V is 0.7 W. You can get 1.4 W if you increase the anode voltage to a maximum of 180 V. As you can see, the lamps are economical and at the same time give out good power. I assembled the chassis on a piece of cardboard from the box, there were no panels for five pins, I made it myself from “crocodiles”. The amplifier was powered from a homemade power supply, the voltage was applied to the filament through a DC-DC converter, the bias voltage was negative from lithium batteries. In the future, I plan to power everything from the converter (may the high-end people with their kenotrons forgive me). The loudspeaker is from a German electrophone, the transformer is also, but it is not suitable for this output lamp.
In the video below, I am testing a fully pulsed power supply with the exception of a -18 V bias voltage from two Krona batteries. I raised the anode voltage to 180 V, for this I used the DC-DC module from Nixie clock on the MAX1771 chip, there are a lot of circuits on the network, but I bought it ready. The output power of the amplifier has become a little more than a watt. I also tested a transformer specially designed for the Tung-Sol 33 lamp, it is very important to match the pentode with the load. The sound has become better, but with a simple speaker without design, it is difficult to fully appreciate it.
Dampers from a quadrocopter in the fight against the microphone effect
The first thing that did not suit me when testing was the microphone effect of radio tubes.
If you touch the chassis with your hand or a wire, you can hear the sound in the speaker. And since the amplifier and loudspeaker will be in the same housing, the vibrations from the speaker will be transmitted to the radio tube. There may even be acoustic feedback. In radio receivers, the lamp is often damped only in the first stage, or the entire chassis as a whole. I decided to dampen both tubes in my amplifier. Usually there is no problem with the output lamp. But since I use very old and not intended for portable equipment tubes, dampers will help protect them from accidental impacts on the speaker case during transportation and operation. And perhaps even when the device falls to the floor. (I am writing from the future. Of course, I did not drop the column, but I tried to put it sharply and somersault on the table, damping panels protect the lamps well.)
I had factory sockets with protection against microphone effect, but only for tubes with four pins:
Long elastic metal contacts swallowed slight vibrations of the radio receiver case when it was touched, turned the tuning knobs or used the switches on the control panel.
For the radio tubes in my amplifier, I decided to make protection from the microphone effect myself, since suitable sockets were rarely found on sale.
I picked up dampers for softness, since Ali has a large selection in size and rigidity (these dampers are designed to protect the video camera from vibrations of quadrocopter engines). Designed in such a way that nothing touches the dampers and does not interfere with their work.
The design is assembled from textolite plates, the base is made in two layers (so as not to break when installing the lamp) and riveted with turrets, which serve as contacts. They are connected to the socket pins with a soft wire, which is used to connect the voice coil to the loudspeaker terminals. This will prevent vibrations from passing through the wires. Already to the turrets, you can solder any wire and even radio components.
Despite the fact that the screws do not pass through the dampers, we will not tear out the panel when removing the lamp, since it will rest against the chassis, and when installing the lamp, it will not be possible to transfer these “gentle” dampers, since this will be prevented by the pressure washer of the panel, it 6 mm high, and the working height of the dampers is 9 mm, in total – the compression will be only 3 mm (see video below).
Yes, the design turned out to be not simple, the photo below shows the number of parts for one design (I forgot to put a flexible wire), but it works well, now even if you “star” the chassis with a screwdriver, there will be no ringing in the dynamics.
In the workshop, a chassis was made of 1.2 mm sheet metal on a sheet bender, holes were cut out with a laser.
I ordered a TR-57-S0346 level indicator with LED backlight from China.
Improved it a bit. To make the backlight more uniform, I installed LEDs with a wide emission angle instead of the usual ones. Also removed the built-in diode bridge with resistor and replaced them with a level meter circuit from Radio #8 1986 (Page 46). This greatly improved the performance of the dial indicator and increased the input resistance.
The only thing I changed in this circuit is the capacitance of the capacitor C3, reduced it to 50 uF. Thanks to the high-capacity capacitor, the pointer slowly returned to the beginning of the scale and facilitated the work of the operator, increasing the convenience of reading the readings. But I installed the indicator purely for decorative purposes and the sluggish operation of the arrow did not suit me, I wanted it to dance more actively to the music.
I have such a dislike for the process of etching printed circuit boards that I can’t eat. But I’m ready to rivet at least all day, so I did the installation on textolite with tubular rivets.
It’s time to paint the chassis black, install the sockets, lamps, transformer, indicator and do the surface mounting.
The amplifier was changed again, I abandoned the fixed bias. And replaced the lamp in the pre-cascade. Instead of a JAN-CKR-32 (VT-44) tetrode, I installed an Adzam 15 pentode. They are similar in characteristics, but the second one has indirect heating. This made it possible to simplify the filament circuit, connect the lamps in parallel to one power source and reduce the effect of interference from the DC-DC converter on the preliminary stage. But the current consumption has increased, the Adzam 15 glow consumes almost 4 times more – 220 mA, but this light is on and this light will warm my soul. The fact is that economical battery lamps do not glow, at most they barely smolder in complete darkness. Another advantage of this light bulb is that it is even less prone to the microphone effect, and also has a smaller bulb, which made it possible to slightly reduce the speaker housing in height.
Since I was constantly testing different tubes and bias types, the circuit kept changing. For the article, I drew the “final” from memory. The pre-stage is set to high sensitivity so that you can connect vintage piezo-needle EDAs, receiver detectors, and other experiments. But nothing waves to switch the lamp to the triode mode and improve the sound quality, you can also transfer to the triode and the second lamp, but the output power will drop significantly. Or enter a negative feedback.
The panel has LED indicators for power and Bluetooth. In the center is a dial indicator of the battery level with a power button. Two large knobs – volume and channel selector combined with power on. Made it so that in analog mode, completely turn off the Bluetooth circuits. I also love the big control sticks, and by combining two switches into one, I made room for them.
At the bottom, the RCA input jacks are a bit illogical. The main one in the middle, two along the edges, is a stereo-to-mono adder with resistors so that any stereo cable can be used.
Near the Bluetooth pairing button and the charging connector.
The panel itself is assembled from three printed circuit boards with a black mask ordered in China. Signatures and frames are made of copper with lead-free tinning. To fasten the racks in the first layer, I made holes for the M3 screw, on average with a larger diameter, so that a flat screw head and a disc washer that prevents unscrewing fit inside, I also applied a red thread lock to the thread, it turned out reliably.
On the reverse side of the panel on the racks there is a Bluetooth module and a voltmeter board with an extended scale.
The voltmeter board with an extended scale was assembled on the K140UD601 “opamp” according to the scheme from the magazine Modelist Constructor No. 12 1990:
Power supply and case
None of the previously tested DC-DC converters completely satisfied me. I decided to test various modules with galvanic isolation and assemble a power supply on them. For the anode voltage, I used WRH12200S-8, it produces 200 V, with a maximum current of 40 mA. To be honest, I didn’t have to choose, this is the only high-voltage module that I found on sale. But he completely arranged me, though I had to glue a radiator on it.
I will not draw a power supply circuit, since I connected all the modules according to the datasheet, additionally added EMC filters and fuses. I tried to power the filament circuit from a small module above in the photo, it whistled and made a lot of noise, so I replaced it with an AM3T-1203S-RZ, it produces 3.3 V, at a maximum current of 0.9 A. The glow is not powered directly from DC-DC, but through a linear LD1084V stabilizer, it is with low voltage drop, LM317 simply won’t start here, it requires almost 2 volts of difference. Of course, the result is not ideal, you can hear the faint noise of the incandescent DC-DC if you bring your ear close to the speaker. The third converter, AM1A-1212SZ, outputs 12V at a maximum current of 83mA. Needed to power the input signal level meter and the Bluetooth module.
The result was a power supply with an input voltage of 9-18 V, at the output: anode 200 V, incandescent 2 V, and for powering the rest 12 V.
The 3S3P battery was assembled from lithium “cans” 26650 of 5000 mAh each and protection boards with balancing. I decided to fix lithium cells in holders, so that if necessary, it would be possible to install not all 9 pieces, but only three or six, and also to easily replace any if necessary.
The case was made from 12 mm plywood and covered with wax, the electronics compartments were pasted over with copper foil.
I really wanted to make a protective grille out of wood and fabric, like on radios of the 30s and 40s. But a massive wooden grille is not very acoustically transparent than a rare metal or simply stretched fabric. Therefore, I made the grille openable, and so that the hinges did not rattle when playing music, I chose friction ones. As a bonus, you don’t need magnets or a latch, the hinges securely hold the grille in any position (see video below).
Another video with the first version of the column, a full demonstration of the work.
Amplifier pentode, automatic bias.
Output power: 1.2W,.
Power: Li-ion battery assembly 3S3P, 12.6V, 15000mAh,
Working time 19.5 hours,
Loudspeaker: 6.5″ Dayton Audio PS180-8,
Bluetooth aptX HD,
Body made of 12 mm plywood, waxed.
Dimensions: 680x210x210 mm, excluding protruding parts.
Weight: 11 kg.
Today we have clearly seen that in our time you can have good fun with radio tubes, if not just store them in a pantry in a three-liter jar, but try to collect something.