Designing an AA5 AM tube radio for use on 240V AC

5ive methods, just below is the "heater current in series with the plate current" method. Second method is the "higher B+ on the 50L6 audio output tube, AC cap in the heater string" method. A third is to use a capacitor in series with the entire radio circuit. And a 4th radio where the usual tubes are replaced by those whose heater currents are 90 (117L7) and 70ma (three 26x6). Yet another using 100ma heater tubes.

My objective here is more of an electrical engineering "what-if" than just being a way to convert an existing radio. Like a "How would the engineers at the XYZ Radio Corp design an "AA5" for sale and use in 240V countries". Given a large enough production volume, the cost of the 26V tubes used in radio four would come down to nearly that of the common AA5 tubes.

The diagram on the right shows the modifications done to an otherwise standard AA5 AM radio circuit. How a manufacturer might have designed an AA5 for 240V AC countries.

If it is desired to convert a normal "All American 5ive" 120V AC AM tube radio design to run off of 240V AC, the following is a relatively low energy consumption way to do it. Especially for the collector living in Europe or other parts of the 240V AC powerline world. Olympic back in '53 made an adaptor cord, which consisted of a 430 ohm power resistor around 40W in series with the entire set, for their AA5 sets. Curtain Burner? Cheap, but burns a lot of power (about 30W in the resistor plus the 30W for the AA5). My following mod saves about 20W. It does require the use of some "sand", though. Not totally out of place, as solid state silicon rectifier diodes did exist in the last decade of AA5 production. Or use a selenium rectifier for earlier sets, to be "authentic" to the technology available when your radio was manufactured. Also need a 500 to 600 ohm power resistor, of about 20 watt rating (about twice the power rating needed, but I like to run them cool). What we are going to do is place the heater string in series with the B+ current of the radio, and the diode and resistor (wired in series) is to provide heater current for the opposite direction of the flow of the radio B+ current. Thus, after warm-up, the heater string will see a reasonably balanced AC current. The diode and resistor also provides a critical start-up function to get the tube heaters warmed up enough so the radio B+ current starts pulling more current thru the heater string. Can't do that until the cathodes are reasonably hot first. It takes several minutes before everything gets warmed up and balanced out. Idea is to have the heater string act as the series dropping resistance to operate the radio's B+ circuits from, and also to use the radio's B+ circuits current as a series dropper for the heater string. Rather than burning more power by passing all the heater current AND B+ current thru a single big resistor.

More detail: first step is to make sure the radio works before doing any mods. Then, at the rectifier tube, the rectifier plate needs to be separated from the top of the heater string. The top of the heater string gets connected to one side of the 240V AC line (let's call it "A"), the rectifier tube's plate is connected to the other side of the 240V AC line (call this "R"). The heater string connection to the radio's B minus line stays in place as it was. This places the heater string in series with the radio's B+ supply circuit. Now take the silicon rectifier diode (1N4007 for example) and connect its cathode to the point where the rectifier tube's plate is connected, point "R". The anode of this diode connects to the power resistor, and then the other end of the power resistor connects to the radio's B minus line. Be sure to connect to the spot where the heater string itself connects to the B minus line. This avoids inducing hum into the radio circuits due to ground loop currents.

You may note that the AC current thru the heater string has a DC bias current (in other words, more current in one direction than the other). This is not a problem, heaters are just resistive elements. But it can make taking AC voltage measurements difficult to get meaningful information from. However, you can meaningfully measure the B+ from B minus voltage, I read 114V DC, which is reasonable. As for the heaters, use your experience and look at the color of the heaters and cathodes to see if they "look" the right color and brightness. Or use a 'scope:

Radio now consumes about 40 watts, vs. 30 watts for a standard AA5 set.

To design the above mod, I used a bit of calculus type math. I enjoyed calculus class in college so much, I took it twice! :-) But the following is fairly easy calculus. To determine power consumption over one cycle of the AC powerline, integrate, from 0 to 2p, (150(sine2(X))) dx. Sine squared because power is current squared times resistance of the heaters. This gives a reference number we need to match with the B+ and diode/resistor currents. Number was about 470.

We now figure the power dissipated in the heater string by passing the radio's B+ current thru it. Figure 60ma. This is effectively an RMS value, the actual waveform is anything but even. So we do an integral, from 0 to 2p, (70x) dx. This yielded a number of 200 or so, and the below diode/resistor needs to make it up to 470. A reality check would be:
heater power = 120V * 150ma = 18 watts. Radio B+ power would be 150V (don't forget losses in the rectifier tube) * 60mA = 9 watts. Half the heater power. We are going to put in "series" the heater power, and the B+ power plus an extra 9 watts or so of power in the power resistor.

Now we figure the power for a half cycle of the powerline, as the diode/resistor gives this. The diode makes the other half cycle zero power. So we do the integral, from 0 to p, (Y(sine2(x))) dx. And just keep picking a value for Y that makes things add up with the above 240 to get a total of 470. Once you nail down a value for Y, do: 120/Y= value of power resistor for use with the diode. I got about 600 ohms.
Reality check: figure (1/2)(diode cuts out half the cycle)*(120v2/600)=12 watts, ahhhh... a little high. Turns out I need to run a bit more than half the usual heater current thru the tubes on turn-on to get them hot enough to start functioning and conducting current thru their cathodes. After that starts, more heater current starts warming up the tubes more, eventually reaching a steady state condition.

Your calculus prof in college would probably shudder in horror at the crudeness of the above use of calculus, but it yielded an answer that made a working circuit when built the first time! And the above is roughly what real engineers in the real world actually do when designing stuff. No rigour, no proofs.

2nd method if radio uses octal tubes

This method rectifies the 220V AC line directly, to create higher B+ voltage for the audio output stage (50L6). The 50L6 is designed to operate at this plate voltage; however the 50C5 is not. The load on the output tube needs to be doubled for use with this higher B+. This can easily done by replacing the 4 ohm speaker with an 8 ohm speaker. The output transformer reflects this change to the output tube. There will be less bass response, but that is usually not an issue in an AA5. You should get more audio output power. Also the cathode resistor should be increased as per the diagram.

An alternative to changing the speaker is to change the output tube to a 50C6. That's right, fifty-cee-SIX. But be sure it can physically fit inside the radio, as it is bigger. Same basing as the 50L6. You'll need to change the cathode resistor to 220 ohms. The 50C6's load impedance is rated for 2600 ohms at 200V plate voltage, close enough to the 2000 ohm load for the 50L6 at 120V plate installed in the radio. The 50C6 is nearly the same as the 6Y6. BTW the 50Y6 is a rectifier.

A new B+ capacitor rated for 250VDC will be needed. And a 6K seven or so watts will need to replace the old 1K resistor to run the output tube screen grid and the rest of the set. Take the old section of the old capacitor you replaced just now and add it to the B+ running the screen grid and rest of set.

If you decide to use the 50C6, use a lower value power resistor (around 2.2K) in series with the old 1K resistor. The screen grid of the 50C6 goes to the node between the 2.2K and 1K resistors, along with the section of the filter cap that used to filter directly the rectifier output when the radio ran off of 120VAC. The screen grid tap should operate at about 140VDC, so the 2.2K resistor value may need adjusting. The 50C6 can have up to 135V on the screen, and 200V on the plate (referenced to the cathode). As the cathode in this set is at 12V above ground, this gives some margin. I used a 100 ohm 7W power resistor between the rectifier plate and first filter cap to drop the supply voltage down some, in this case with the 50C6 I got 195VDC. You'll still need this resistor (but size it to 220 ohms) if you keep using the 50L6, by the way.

What about the heater string? If there is no pilot light, we will insert an AC rated 2uF capacitor in the series string, between the 35Z5 and hot side of the line. Remove the tapped heater section connection from the rectifier plate, and connect the rectifier plate directly to the hot 220VAC line thru a 5 to 10 ohm resistor. The heater-cathode voltage rating will still be high enough (360V) to support this mod.

If there is a pilot light, we will have to move it (I've had problems with the pilot bulb popping if power is briefly interrupted to the set) and do the above, or arrange things so the set's plate current flows thru the heater tap section of the 35Z5 on the B minus supply line side. One possibility is to run the pilot light (a #47) from the powerline thru a 1. 5uF 250VAC capacitor. Maybe use a neon bulb run off the line or B+ (with series resistor) to light the dial as an alternative. If you don't mind modern "sand", a pair of 8V zener diodes (wired cathode-to-cathode in series, and this in turn in parallel with the pilot lightbulb, zener anodes connecting to both sides the bulb) may keep the bulb from popping. Don't forget, it's 6.3VAC RMS, 8.3V peak. The first 8V zener plus the 0.7V drop of the other zener should make for 8.7V, high enough to not normally conduct. Haven't tried it though.

We can't place the 35Z5 heater on the high side of the line, or else the heater-cathode voltage rating (360V) will be exceeded. When the AC line hits its negative peak, and the cathode is charged to B+ running voltage. So, we place the rectifier heater on the cold side of the line. {Do this if you do the "zener diodes across the pilot bulb" trick: "Cold" side of line to 6V section of the 35Z5 heater, also to the pilot with zeners in parallel. Other side of the pilot with zeners tied to the heater tap on the 35Z5. This becomes the radio's "ground". The other portion of the 35Z5 heater continues onto the rest of the tubes' heaters.} So the heater-cathode interface sees essentially just the B+ voltage, which is less than 360V. This means that the heater on the 12AV6 (or 12SQ7) will be on average 40VAC above ground instead of about on average 6VAC, but I didn't hear any extra hum because of this. We will insert an AC rated 2uF capacitor in the series string, between the 50L6 and hot side of the line. See the diagram and this should become clear. If you run off of a 50Hz powerline (mains), you may need to add 0.1 or so to get the voltage on the heaters right.

If the radio uses 6 tubes (RF, converter, IF, detector and first audio, 35L6, 35Z5) you could substitute a 50L6. The heater string will add up to higher than 120V, but we have 240V available. One radio I modified, my second GE 422 (a 6 tube radio), I used a 12SG7 for the RF, and the 50L6 in place of the original 35L6. Used a 19HR6 in place of the 12BA6 in the IF. Added up to 140V heater string. Used 2.15uF capacitance to drop the 240V to 140V to run this heater string. The pilot light was replaced (and rewired) with a neon bulb that has a bayonet base, and the 35Z5 stayed at the hot top end of the heater string, just "under" the capacitor. That is, the capacitor went between the 35Z5 and the "hot" 240V line. This keeps the heater-cathode voltage below the rated max of 360V.

Another separate mod is to use the higher B+ we now have available on the plates of the tubes in the RF, IF and converter stages. The screens need to be run on the lower voltage though. Note the 22K resistor on the screen grid of the IF tube. As the AVC reduces the current thru this tube, the screen voltage will go up. This actually decreases the AVC action of the IF tube itself. The AVC being a feedback system, this causes the RF and converter tubes to work harder to reduce gain on strong stations. The signal level at the IF stage's input is thus lower. This keeps the strong stations from getting too distorted by the remote cutoff curve of the IF tube, as the signal strength is quite high there. And higher and more prone to distortion if we didn't do this. And the IF tube acts partly as an AVC amplifier to boost the amount of detected AVC signal. This extends the radio's ability to handle strong signals. Another way to do this is to use a high impedance voltage divider on the AVC line to bias the IF tube grid circuit. High impedance to avoid loading the AVC detector circuit.

The radio with the 50C6 option consumes about 50 watts.

The radio with the elevated voltage on the 50L6 consumes about 40 watts.

The 12BA7 pentagrid converter is a separate modification from the 240V modifications (i.e., not needed for 240V operation). Click here for details. The 12BA7 yields about twice the signal gain of the 12SA7 or 12BE6.

Click here for a radio using a compactron 38HK7 power pentode/diode with 270VDC rectified directly from the 250V powerline (mains). This tube replaces the 50C5 and 35W4 in an otherwise ordinary AA5.

A third method using a 2. 5uF 250VAC rated capacitor in series with the entire AA5 set circuitry is possible. At powerup, the heater string is the only thing that will conduct. After the heaters have the tubes warmed up, the cathodes of all the tubes start to emit electrons. At first thought, one would think the AC line dropping capacitor would interfere with the half wave rectifier (35W4) supplying B+ in the AA5 radio. Because capacitors won't pass DC. As it turns out, the DC current circulates thru the heater string from ground to the point where the rectifier tube plate is connected. This current heads into the capacitor for half of the powerline cycle, and back out during the other half. Then it goes thru to the rectifier tube cathode, and after filtering, supplies the B+ to the set. Then the DC returns thru the circuits back to ground. (Remember current is in the direction opposite to electron flow). The AC capacitor supplies the drive to make this circulation happen. The B+ will be a bit low, somewhere around 90V instead of 110V, but the set will operate. If you (carefully!) use an oscilloscope to view the voltage waveform across the heater string, you'll find a crude sine wave with a negative DC bias on it. The positive peaks will be about +100V, the negative peaks around -180V. A tradeoff to consider, if the capacitor is increased to boost B+, the heater string will get too hot with excessive voltage. I did this to an FM only set that had a selenium rectifier. A 3uF cap along with a 35W4 added to the top of the heater string (for a total of about 150VAC) replaced the selenium rectifier, and its heater allowed higher B+ to be developed. However, one needs to be mindful of the max cathode-heater voltage (which for the 35W4 is 360V). A 2. 5uF capacitor on 250V 60Hz lines is about right, about 3uF value capacitor should be right for 50Hz mains for an ordinary AA5; the FM set needed a little more. The AA5 radio consumes about 23W, the FM set uses about 30W.

In an AA5, to get higher B+, use a power resistor of about 300 ohms between the rectifier tube heater and the rest of the string. And a 3. 7uF cap. I got 125V B+ with 350 ohms and 4uF cap. Using a scope to measure heater to cathode voltage in the 35W4, I'm slightly exceeding the -360V rating. Tubes are a little more forgiving on this sort of thing than solid state is. Looking at the voltage waveform across the heaters and resistor with a scope, I see a 157V RMS somewhat distorted sine wave with a DC bias of -75V DC. Doing some crude approx. math to calculate the power in the heaters and resistor (total 1150 ohms), let's say the positive portion of the waveform looks like a sine wave of 105VAC RMS, and is active for 40% of the time. I get an answer of 3.8W from P=0.4(Vsquared/R). And for the other 60% of the time (negative portion), the waveform looks like a somewhat distorted sine wave of 210V RMS. That gives me 23W. Adding them together gets me 26.8W, call it 27W. As the 350 ohm resistor is just slightly less than half the resistance of the heater string (which should consume 18W), it looks to be about the right amount of power is heating the heaters. The color of the cathodes and heaters looks about right. Even with the error range dictated by the crudeness of these measurements and calculations, we seem to be pretty close.

Ian Robertson in Australia writes to tell me of a Japanese made AA5 modified for 250VAC operation. It's somewhat similar to the above, except that the heater cap is in parallel with the rectifier circuit. This yields better results, as there is a higher reactive voltage across the cap to feed the rectifier tube 35W4. Thus yielding higher B+ voltage without needing the power resistor in the heater string. Below is a detail of the set's diagram he drew up and I added a few notes to. The heater string also provides peak charging current limiting to avoid exceeding the peak current rating of the rectifier tube. The 35W4 heater tap with or without pilot light does this in regular AA5s.

Yet another radio:

Well this time I used 4 tubes:
117L7 audio output/rectifier, 26A6, 26C6, 26D6 IF stage, audio driver/detector, pentagrid converter. That's 195V worth of heater string. An 800 ohm 3W resistor pads it to the 240V line. But the 117L7 draws 90 ma of heater current, and the 26x6 tubes draw 70ma. Well, what I did was pass the plate current thru the 117L7 heater so as to divert it from the 26x6 tubes. The radio's ground is a point between the 117L7 and the 26x6 string. The 117L7 rectifier plate is connected to one side of the 240V line, and filter cap goes to the radio ground. So all the B+ current to operate the circuits of the radio (RF, IF, audio) comes off the line, rectified and filtered, and then finds its way to the radio ground. And then it goes thru the 117L7 heater to reach the other side of the 240V line. Only thing the 26x6 current plus the plate current from the radio circuits is a little more than what the 117L7 wants, so there is a 4K3 3W resistor across this heater to take the excess. When the set is first warming up, the 26x6 tubes see a bit more than the usual heater current. This isn't a real problem as tube heaters are designed to withstand higher than rated current, and this condition goes away as soon as the 117L7 is warm enough to start conducting. Visual inspection shows that the tubes don't get too bright during warmup anyway. Some oscilloscope measurements and some calculus (below) show that the heaters are getting about the right amount of power.

A note on the heater-cathode voltage as seen by the 26A6. It appears to be exceeding the 90V max rating on the spec sheet, but I haven't had any problems with the tubes I'm using. These tubes were mil surplus made by RCA and packed in Dec 1962. The spec sheet is dated July 1946. My tubes may have improved heaters, as manufacturers would use heater stock used for many different types of tubes (some with higher H-K ratings) in production. Or they never characterized the 26A6 beyond 90V figuring nobody would ever need it higher. Or else the tube just hasn't blown up yet (though I have put about 50 hours on it so far)... If this design were ever to go into large scale production, another tube would be used (maybe 26CG6) unless the 26A6 were characterized to take the higher H-K voltage.

The 26C6 triode has a much lower mu (16) than a 12AV6 or 12SQ7 has (100), so the grid needed to be biased to about -3V. I tapped the pentagrid converter tube (26D6) oscillator section to get some negative bias. This bias is usually -5 to -8V. The orginal grid leak resistor of 10 megs is used along with another 10 meg resistor tapping the pentagrid tube to create a voltage divider to get the -3V for the triode grid. The plate resistor was reduced to 150K as well to have the triode operate in a reasonable region. To make up for the lost gain, I added a cathode bypass cap to the audio output pentode (117L7). The 26C6 triode is nearly the same as that of the 6SR7. There's also another tube, the 26BK6 (heater of 26. 5V and 70ma), that is nearly the same as a 12AV6. With a mu of 100, plate resistance of 80k, gm of 1250, and grid bias of -1v (same as the 12AV6). Soon as I get some, it should fit nicely in this radio without the 26C6 mods... Just got them in... The 26BK6 works well, just like a 12AV6, once the 26C6 mods are removed. Also removed the bypass cap on the 117L7 pentode cathode, as there is plenty of audio gain now. But the 26BK6 is as tall as a 35C5. The detector diodes don't arch around the plate, but are housed inside a shield which is tied to the cathode (thus to ground). This would cut down on IF harmonics from jamming reception of stations at 910KHz. The diodes are near the tube base, and the triode up top.

The radio using the 26BK6.

This radio had a pilot light, but I couldn't figure out a way to incorporate a pilot light in the new heater string. Without putting undue voltage stress on it upon power up. So I decided to use the pilot light as a cathode resistor on the audio output pentode tube (117L7). The pilot light is a 1819 which is rated for 28V at 40ma. I'm using it at 15V and it passes about 30ma at that voltage. The 117L7 usually develops -5V of cathode bias by itself. Here I rigged a voltage divider across the cathode resistor (the pilot light) with a tap at 2/3 from ground. This tap is what biases the control grid. Resistors 470K and 1.2meg. This makes the tube pull enough current thru the bulb to get the difference between the tap and the cathode to be 5V. Thus the 15V.

The output impedance of the 50L6 is rated at 2Kz. The 117L7 is rated at twice that, 4Kz, for similar B+. An easy way to accomodate that is to replace the 4 ohm speaker with an 8 ohm speaker so the output transformer will now look like 4Kz instead of 2Kz. However there will be less bass response in the audio. With audio transformers, when operating at impedances below those specified, the frequency spectrum is shifted downward, conversely, at higher impedances frequency spectrum will be shifted upward. To a first order approximation, in this radio, the bass response thru the transformer will be reduced to about twice the old bass frequency. However, the 117L7 will only produce about 3dB less audio power than the 50L6. These factors do not cancel out, and you will still lose that bass. The inductance of the output transformer's primary Lp doesn't change even though the load impedance (as seen at the primary via the impedance transformation of the transformer) Rload did. Also the winding resistance of the primary stayed the same, but here it's about a hundred ohms and can be ignored. So this looks like an RL high pass filter which is not affected by voltage or power level. But this radio had a larger than average transformer, so this loss of bass isn't quite as serious as with a set with a tiny transformer.

I later decided to "do it right" and change the output transformer to one that has impedances to go from 4K to 3.2 ohm speaker, and reinstall the old 3.2 ohm speaker. I hate rivits.... TV sets tended to use audio output transformers with higher impedances like this.

With these oddball heater voltage waveforms going on, and with the radio ground not tied to either side of the 240VAC line, one should check to see if the rectifier portion of the 117L7 is not being overvoltaged for peak inverse voltage (PIV) rating.

I had a couple power resistors below deck of the chassis, down below the local oscillator coil (which is also below deck). Seems this coil doesn't like getting warmed up, as this caused drift severe enough to detune completely a station at 1250KHz (from initial turn on to being on for 20 or so minutes). I repositioned the power resistors to a spot above the chassis, away from this coil. Drift is less severe now.

A sub for the 117L7 looks like the CV2556. This set lends itself for switchable or wire-able 117V or 240V service. There's two heater strings designed for 120V, and the rectifier circuit that wants 120V input. Change the connections from the 240V configuration and configure it for 117V:

Or use a slide switch of the sort found on PC power supplies, the ones marked "115V/230V" and require a small screwdriver or such to change:

And yet another:

Here we are using 100ma heater tubes.
The front end, IF, and detector/AVC/audio driver tubes are the 18V 100ma versions of the miniature AA5 tube line up. Their B+ is around 110V. But, as the 100ma AA5 tubes (32ET5 or 34GD5, and 36AM3) are not designed to take the higher voltages from a 250V AC line, we need better tubes here. There's the 60HL5, which can take 330V plate voltage, and the 38A3 (AKA
UY85) which can accept the 250V AC line. This set develops 260VDC for the output tube, and a resistor dropper provides the screen voltage and supply for the rest of the set, 110VDC. We could also use a UY82 rectifier as well. The heater string doesn't come up to 250V, but more like 152V (with the 38A3) or 170V (with the UY82). So we use a plain old fashioned power resistor of 850 ohms to drop the line down. Or a cap of 1.4uF 250VAC line rated. The cap avoids 8.5 watts of heat inside the radio.

The 60HL5 needed a "grid stopper" resistor of around 4.7K to avoid oscillations at supersonic and RF frequencies. This tube has a high gm. The resistor and the stray capacitences in the tube form a low pass filter that kills gain at frequencies above audio, thus stopping the oscillations. These oscillations can cause audio distortion and can also trash some AM band frequencies. I used a surface mount resistor to bridge across a cut I made in the circuit board trace feeding the 60HL5's control grid. Makes for a neat install.

This particular set is a clock radio, with the usual Telechron 120VAC 60Hz drive. I added a 0.33uF X2 rated cap (good to 1.4KVDC) in series to run it off my 250VAC 60Hz line. The capacitor will see around 340VAC, as the field coil has a fair amount of inductance. This saves about 4 watts of heat vs using a power resistor. Below is the modified circuit board:

18FX6, 18FW6, 18FY6, 60HL5 (look for 8HL5)

In the USA and other parts of the world using a pair of 120V powerlines (240V grounded center-tap), you don't have to wonder if the chassis is "hot", either way the powerline is connected, the chassis is always "hot". -->