RCA Radiola III and III-A  

By Doug Criner

An interesting radio is the battery-powered, regenerative, RCA Radiola III, c. 1924 – along with its companion “balanced-audio” amplifier (RCA’s terminology for the 2-tube push-pull amplifier designed to boost the power of the Model III).  The Model III-A includes both the radio and the audio amplifier integrated into a single cabinet.  (A Radiola III without the audio amplifier is usually limited to earphone listening, although it can drive an efficient loudspeaker with a strong, local station; a III-A can drive a loudspeaker comfortably.)  For tuning and regeneration adjustment, these sets used two variometers, each a pair of coils with adjustable mutual inductances that are very interesting devices themselves.

Underneath the III-A chassis.  The two variometers are inside the large, brown tube.

These radios are quite plentiful and popular with collectors.  Their performance on the AM broadcast band is good; they are more compact than most radios of this vintage; and their appearance, with exposed tubes and retro-looking controls, is interesting and attractive.  They are a favorite of visitors seeing an assortment of antique radios. These sets could be used as props in a mad-scientist movie – possibly for a high-voltage controller of zombies!

Like most radios of this vintage, an external speaker (or earphones) and an external antenna are required.

Preliminary Repairs

These sets are fairly simple in construction, and repairs should be straightforward.  Things to inspect include:  checking audio transformers for open windings; measuring the cartridge-style, fixed tuning capacitors for leakage; cleaning the tube socket contacts; and testing the grid-leak resistor for increased value.  (The grid-leak resistor is inside one of the cylindrical capacitors.)  Replacement audio transformers are available from RadioDaze or Antique Electronic Supply.

If there is a problem getting the set to work, it’s simple enough to trace all the wiring with an ohmmeter to verify that connections are complete.  Be especially suspicious of the wiring to the variometers, which flexes as the set is tuned or the regeneration adjusted.  Schematics by John F. Rider are available at:  http://www.nostalgiaair.org/Resources/  Rider’s schematics and the schematics included in the original RCA instruction manuals are difficult to visualize, particularly in the RF tuning section.  The book Fixing Up Nice Old Radios, by Ed Romney has a redrawn schematic that is easier to follow.

Replacement parts:  audio transformer, rubber strips for tube sockets, WW2 surplus 864 Tubes

The tube sockets are rubber mounted for shock and vibration absorption, protecting the tubes and helping to prevent microphonics.  The original rubber strips, which are usually rotted, can be replaced with narrow strips, about ¹/₄" wide and ¹/₁₆" thick, cut from sheet rubber.  Possibly an old bicycle inner tube can be used, but I think two layers would need to be attached with adhesive to obtain sufficient thickness and rigidity.  Bill Turner sells replacement strips precut to the correct size:  http://www.dialcover.com/  Bill’s strips, which seem to be made of vinyl rather than rubber, have just the right rigidity so that the tube sockets do not deflect excessively when removing or inserting tubes.

 WD-11s Are Virtually Extinct

These radios are quite plentiful on eBay and at antique radio swap meets, but they are almost always sold without vacuum tubes.  These sets were originally equipped with WD-11 triode tubes, now virtually extinct. When WD-11s can be found, used ones sell for over $100 each; for a Model III-A requiring four tubes, a set of tubes costs more than the value of the radio itself.  Even dud WD-11s get snapped up by collectors that use them to make their radios more displayable.  Maybe it’s all just as well since WD-11 tubes reportedly have fragile filaments, rather weak performance, and tend to be microphonic.  (I say “reportedly” because I’ve never been able to afford to get anywhere close to a WD-11.)

In 1925, early in the Radiola III’s history, RCA issued a bulletin^* suggesting that these sets could be modified to substitute a better performing and more rugged UX199 detector and UX120s for the audio tubes, which themselves are now becoming expensive.  (Alternatively, UX199s can be used for all the tubes.)  The UX199 and UX120 tubes also had lower filament power requirements than WD11s, which extended dry-cell battery life – not an advantage today if using a battery eliminator for power.  Lately, collectors have identified other, more modern tubes to substitute for the original WD-11s.

When beginning work on a Radiola III or III-A, first determine if the filament circuit has been electrically rewired for UX-based tubes.  If it has been modified, I would prefer to leave the modifications intact; since such modifications were offered as a recommended option by RCA, they are historically correct (see the asterisked footnote below).

 WD-11 Sockets are Unique

A problem is that WD-11 sockets are unique and do not lend themselves to a plug-and-play tube substitution.  WD-11s have four pins, one of which is large diameter (about ⅛ inch) and the other three pins smaller (about ³/₃₂ inch).  No tubes have been manufactured with such basing since WD-11s were discontinued, eight decades ago.

Two options present themselves:  replacing the Radiola tube sockets with a more modern version or using an adapter.  Replacing the set’s original tube sockets ruins the originality of the radio, so that option can’t be recommended.

When RCA suggested the substitution of UX-based tubes, adapters were available that allowed the use of common 4-pin-based tubes (known as type-4D basing).  “Na-Ald” brand adapters were manufactured by Alden Manufacturing Co.; currently, such vintage adapters are almost as rare as WD-11s themselves.  Modern replica adapters were recently made, but that source dried up due to health problems of the Indiana man building them.  More recently, adapters have again become available from George H. Fathauer,  http://www.users.qwest.net/~tubes/  Adapters don’t just change the pin configuration – rewiring has to be incorporated within the adapter since the pin-out sequence for WD-11s does not match the more common 4-pin basing.  Review the basing of a WD-11 and the tube you intend to substitute at: http://tdsl.duncanamps.com/index.php 

For a Radiola III-A, you will need four such adapters.  For a Radiola III, without its companion audio amplifier, two adapters are needed.  The adapters will cause the tubes to stand about ³/₄ in. taller in their sockets.  The tips of tubes in a Radiola III or III-A are exposed on top of the cabinet, so with the adapters the tubes will stick out an additional ³/₄ in.; this may be objectionable to some collectors. 

It’s possible to build your own adapters, but it isn’t very simple.  See:  http://www.vcomp.co.uk/tech_tips/wd11/wd11.htm 

Substitute Tube Options 

What tubes are available to plug into the adapters?  Of course, you can use the RCA-suggested tubes:  a UX199 for the detector and UX120s for the audio tubes, or even all UX199s.  They should work fine, but the set itself might need to be modified per: http://www.antiquewireless.org/otb/radiola.htm  Such a modification was recommended by RCA to accommodate the different filament electrical requirements.^*  Another drawback to the UX199/UX120 substitution is that those tubes, while certainly more available than WD-11s, are still expensive. 

Another approach is to use 864 tubes, which can plug into the same adapters and thence into the radio without modifying the set’s filament circuit.  The 864’s glass bulb is about ³/₁₆ in. larger in diameter than the 1-inch diameter UX-type tubes or a WD-11.  This larger size does not cause any physical problem with fitting into the radio, but the difference is noticeable and potentially objectionable to purists.

Tubes (L-R):  864, UX120 and UX199

864 tubes, which were supposedly used in World War II aircraft equipment, are still relatively available; their electrical characteristics are quite close to the WD-11, but with a little higher amplification factor.  The 864s are much less prone to microphonics than the old UX-based tubes – which makes sense if the 864s were designed for use in WWII bombers.

There are several reputable tube dealers who regularly stock UX199, UX120, 864, and even WD-11 tubes.  If you are patient, you may find bargains on eBay or at radio swap meets.  Recently, I was lucky to win several NOS 864s on eBay for about $12 each; they were sealed in their original 1943 U.S. Army Signal Corps boxes.

With either UX120/199 or 864 tubes, you will need the same adapters to plug into the original sockets.  The most significant difference is the heater power requirements (3.3V, 60mA for the UX tubes, versus 1.1V, 250mA for the 864s).  If your set has already been modified for the UX-based filament requirements, then I would opt for UX120/UX199 combination of tubes.  If the set is unmodified, then I would go with 864s, which have the same filament requirements as WD-11s.  (I wouldn’t select all UX199s unless I already had them or I found a very inexpensive source.) 

(Regarding used tubes versus NOS – new-old-stock – tubes:  I prefer the lower-cost, used tubes, provided they have been tested and represented as good.  Most antique radios, especially of  1924-vintage, will not be played a lot, and used tubes should last indefinitely.) 

There are several other substitute tube possibilities, some more difficult to implement.  See, for example: http://www.vcomp.co.uk/tech_tips/wd11/wd11.htm 

Bill Turner builds WD-11 replacements with a subminiature tube concealed inside the glass.  These plug right into the original WD-11 sockets.  This is an elegant, cost-effective solution. 

I have tried only two line-ups for a Model III-A:  1) four 864s, and 2) three UX120s and one UX199.  Both of these line-ups produce ample volume with a loudspeaker and good overall performance.  Ignoring the physical size issue, the 864s have a slight edge because of their lower microphonics and a higher threshold for oscillation. 

Suggested Tube Voltages 

The following table summarizes my suggested voltages for alternative Radiola III-A tube line-ups.  (These are the voltages to apply to the set’s battery leads; the actual A-voltages will be less due to voltage drop and the setting of the filament rheostats.) 

The above WD-11 voltages are those recommended originally by RCA's Radiola III instruction manual.  WD-11s are dual rated for 90V and 135V B+.  Presumably, RCA recommended the lower B+ voltage to reduce the number of dry-cell batteries required.  With the current high cost of WD-11 tubes, few collectors today would risk using them except for brief periods and at the lower voltages originally recommended by RCA.  (If WD-11s were plentiful, I’d recommend jacking up their B+ and C- voltages for improved performance.) 

When using all UX199s, the recommended B+ for the audio tubes is 90V rather than the 135V for UX120s.  For some reason, the UX199 data sheet lists 90V as the maximum plate voltage, whereas UX120s are dual rated for 90V and 135V. 

For any of the tube line-ups, if excessive oscillation is encountered that cannot be effectively controlled with the set’s regeneration adjustment, a lower B+ voltage can be tried. 

Speakers 

A Radiola III, with just a detector and a single audio stage, wouldn’t be expected to drive a loudspeaker without its companion balanced-audio amplifier; yet, surprisingly, for strong stations, it will drive an efficient, high-impedance loudspeaker.  The Model III-A has plenty of volume to drive a high-impedance speaker, such as the vintage RCA 100-A.  A modern low-impedance speaker may be used with a suitable matching audio transformer. 

In the late-1920s, third-party radio dealers sometimes sold aftermarket wooden cabinets to house a Radiola III-A, and these are often encountered today.  The cabinets included a built-in horn speaker, essentially an earphone-style reproducer, and a wooden horn directing the sound through a front grille in the cabinet.  These sets are attractive, but do require a decision about whether to retain the original reproducer.

Aftermarket cabinet (with speaker) for Radiola III-A.  "Acre" brand.

Another aftermarket cabinet with speaker. No brand name identified.

Sometimes, the reproducer has an open winding, so it is shot.  But if it still works, after a fashion, the permanent magnet is likely weak and the reproducer may give marginal tone and volume.  My preference is to fit into the cabinet a modern, low-impedance 2-in. speaker and a small transistor-radio audio matching transformer.  The original reproducer can be stored somewhere inside the cabinet for historical purposes and to allow any future owner to restore the speaker to its original condition. 

There is a slight problem with the wood cabinet pictured above with its top lid open.  With WD-11 tubes, there was adequate clearance for the top to be closed.  However, with UX120 tubes and the extra ³/₄ in. tube height caused by socket adapters, the lid hits the tops of the tubes.  But with the 864 tubes, which are a little shorter than the UX120s, the lid closes with sufficient clearance.  The cabinet itself is quite nice – it appears to be made of solid mahogany. 

The other cabinet pictured is labeled “Acre, Chicago.”  It also is made of solid wood, probably mahogany.  Both of the cabinets have built-in speakers, and the original reproducers work. 

Operation of the Set 

Here is a helpful link that explains the operation of the Radiola III:  http://antiqueradio.org/radiola3.htm  Suffice it to say that tuning is a little arcane.  For example, the III-A has two filament rheostats for volume, a regeneration adjustment, a tuning lever, and several different antenna hook-ups to receive particular parts of the AM broadcast band.  Bill Turner sells reprints of the original Model III and III-A instruction manuals, which are helpful. 

The Radiola III has four antenna binding posts, numbered 1-4, with a grounded link that may be connected to either Post #3 or Post #4, or left disconnected.  RCA’s instructions listed six different antenna hook-up combinations, numbered 1-6, which involve optional binding posts to attach the external antenna and three different positions of the link.  The most selective antenna hook-ups are Combinations #5 and #6; based on my experience, you can forget the other combinations unless you live in an area with only very weak, distant stations.

Antenna hook-up Combination #5 (antenna on Binding Post #1, link on Binding Post #4).  For the upper half of the broadcast band. 

With antenna hook-up Combination #5, the external antenna wire is connected to Binding Post #1, and the link is connected to Post #4.  This should be good for tuning from about 800 kHz to 1540 kHz. 

With hook-up Combination #6, the external antenna wire is connected to Post #1 (the same as for Combination #5).  The link, however, is connected to Post #3.  This provides tuning from about 470 kHz to 970 kHz, so there is some frequency overlap between the two hook-up combinations.  The tuning scale is arbitrarily labeled from 1 to 10; turning toward a higher number reduces the tuned frequency (and increases the wavelength).

Antenna hook-up Combination #6 (antenna on Binding Post #1, link on Binding Post #3).  For the lower half of the broadcast band.

Adjustment of the regeneration (labeled “Amplification” by RCA) is similar to other regenerative sets. 

Turning the rheostats clockwise increases the filament voltages which, in turn, increases volume at the expense of shorter tube life.  There is not a separate volume control in the usual sense.

The Radiola III-A has two audio output tubes in push-pull.  If one of the two is removed from its socket, the radio's volume may not change perceptively.  When one tube is removed, the "A" filament current will decline -- which, in turn, will reduce the voltage drop through the associated filament rheostat.  This increases the voltage supplied to the remaining audio tube which tends to increase the volume.

Issues With the Fixed Tuning Capacitors 

Unlike modern radios, tuning is achieved by a variable inductance rather than a variable capacitance.  The tuning capacitors are fixed values mounted inside cardboard cylinders that are readily noticeable on the underside of the chassis.  These capacitors are connected into the tuning circuit based on the antenna hook-up that is selected.

The five cylindrical fixed tuning capacitors.  The grid-leak resistor is inside the lower cylinder.

Assuming that only Combinations #5 and #6 will be used, the critical values are for the two capacitors that are mounted under Binding Posts #3 and #4.  The value of these two capacitors will determine the frequency tuning ranges for antenna Combinations #5 and #6. 

Rider’s schematic for the Radiola III has the values of these capacitors penciled in by an unknown person, presumably long ago.  There have been discussions among some modern collectors about whether those penciled-in values are correct; no official RCA information is available (many manufacturers’ schematics from that era did not include component values).  However, I believe that the penciled-in values are correct, at least for the two critical capacitors that are mounted under Binding Posts #3 and #4:  250pF and 10pF, respectively.  The lower value capacitor allows tuning at the higher end of the broadcast frequency band, and vice versa. 

If the tuning ranges associated with antenna Combination #5 or #6 fall outside the frequencies specified by RCA, then the tuning can be brought into alignment by changing out the associated fixed capacitor.  A lower value capacitance will raise the frequency band, etc.  Silvered-mica capacitors may be used for replacements since they are available in small values and are very stable.  Alternatively, ceramic disc capacitors could be used, preferably NPO rated.  In any case, close tolerances are called for, certainly within 5%. 

Opening one of these cylindrical capacitors to re-stuff a replacement capacitor is a challenge.  There are metal end-caps that are firmly crimped to the cardboard body.  The end caps can’t be twisted off without damaging the cardboard body.  The only approach I’ve found is to cut the cylindrical cardboard body in two with a band saw or coping saw.  Then, after re-stuffing, it can be taped back together and wrapped with brown paper to conceal the joint. 

I tend to think that RCA’s selection of the frequencies covered by the lower band, the one enabled by antenna Combination #5, is not optimal, at least for the modern U.S. broadcast spectrum.  The lower limit of 470 kHz includes frequencies that are silent today.  Similarly, the higher band, the one enabled by antenna Combination #6, which ends around 1540 kHz, misses some of the modern broadcast spectrum's top end.  One could tweak the values of the capacitors beneath Binding Posts #3 and #4 to broaden the set’s overall effective coverage, but possibly reducing the frequency overlap in the middle. 

Grid-Leak Resistor 

The detector grid-leak resistor is inside one of the five cylindrical capacitors – the one that is stacked on top of another with its lead going to the detector tube’s grid.  Most likely the resistor will be drifted up in value or even open.  I do not know RCA’s original value for the grid-leak, but tube data sheets recommend 3-5M for the WD-11 and 2-9M for a UX199.  The value isn't critical. 

******************* 

^*Note: The modifications recommended by the RCA bulletin are due to different filament supplies required for a WD-11 (1.1V, 250mA) versus a UX120 or UX199 (3.3V, 60mA) .  The original A-voltage supply used with the Radial III or III-A was a 1.5-V dry-cell battery.  To supply UX120/UX199 filaments, 1.5V was too low, as was the next available voltage, 3.0V (two dry cells in series).  The next available voltage, 4.5V  (three dry-cells in series), combined with the lower filament amperage, resulted in ineffective volume control by the original 2.5-ohm filament rheostats. 

The RCA-recommended modifications involve either: 1) adding an additional, higher-value rheostat that would fit outside the cabinet or, 2) in the case of the Radial III-A, which has two 2.5-ohm rheostats, to put both rheostats in series. I suspect that Option 2 was used for most III-A sets, since it required no additional parts. (A disadvantage of wiring the two rheostats in series is that the detector and audio filament voltages can’t be adjusted independently.)  These RCA-sanctioned modifications are described here:   http://www.antiquewireless.org/otb/radiola.htm 

Nowadays, the filament supply is likely to be from a battery eliminator, not dry cells.  Many modern battery eliminators, such as the ARBE-III, have an adjustable, regulated A-voltage supply.  If, for example, the A-supply voltage is set at 3.5V, then when the rheostat is cranked down in resistance, the tubes will be provided near their rated voltage and the set will play at full volume.  When the 2.5-ohm rheostat is adjusted to full resistance, the voltage drop through it will be: 2 tubes x 0.06A x 2.5Ω = 0.3V. Will this drop be sufficient to effectively control the volume?  Possibly, in combination with adjustment of the regeneration control and trimming the A-voltage supply from the battery eliminator, but try it – if it works, then no modification to the set’s filament wiring would be necessary.  Otherwise the set will need to be modified along the lines suggested by RCA. 

There is an alternate modification that may be simpler than those suggested by RCA.  If each UX199/UX120 tube’s filament is shunted with a 22-ohm, 1-W resistor, the total current through the rheostats will be the same as for WD11 tubes.  This will restore the original rheostats’ effectiveness for volume control. The resistors can be soldered on the tube socket terminals under the radio’s chassis.  Or, a 12-ohm, 2-W resistor can be shunted across the terminals of each rheostat. 

RCA would not have suggested using shunt resistors since it would have increased the current and reduced A-battery life.  Besides, in those days, the required shunt resistors would have been bulky, wire-wound affairs.  With a modern battery eliminator, the filament current is not a concern. 

© Doug Criner 2006