48 Low mu=2.5 Tetrode

 Fellow Radiophiles:

The 48 Tetrode from 1932 by RCA was designed with an unusually low mu=2.5 between G1 and G2 or in triode connection. This low mu enables a high plate current flow with a relatively low voltage at G2. The standard curve traces in the RCA data sheet were plotted with G2=96V, and show a 1.5kΩ load line for audio power output of 2W with the plate also powered from 96V. The 30V/400mA heater was designed for series string operation in 117VAC/DC radios.

The unique plate structure of this tube uses six ridges inside the plate to trap secondary emission. This nearly completely eliminates the tetrode saddle and widens the plate voltage swing to lower voltages than would be possible with a conventional power pentode.

In a recent email exchange, tube historian Ludwell Sibley contributed this note about the origins of the 48 Tetrode and Edward Herold's invention of it's anode structure in US patent 2018362 granted to RCA:

---

 

Haw!  I have an RCA 48 whose design of the top mica makes it hard to look inside, but under the right lighting the eight ridges are indeed visible.  The mount design differs in some details from that in the patent but that’s not fatal. 

I also have a couple of Philco-brand 48s (most likely of Sylvania or National Union origin) that are easier to observe.  These have six ridges. 

Herold got the patent, but ascribes design of the tube to another.  His historical article in “AWA Review” Vol. 7, says, “I was asked to determine whether the 1929 tetrode screen-grid tube, type 24, could be recommended as a negative-resistance called the dynatron.  The dynatron made use of secondary emission to produce an anode current that decreased as the voltage increased.  After testing hundreds of tubes, made at different times, often with different anode materials, it was clear that no such recommendation could be made.  However, I did learn a lot about secondary emission and about negative resistance.  The secondary-emission knowledge helped me make an early invention [cites your patent], an anode with a series of slats to suppress secondaries.  Such an anode was used by Sam Dodge in designing the type 48 power tetrode.  The negative-resistance aspect stimulated a side-line interest in that subject and led to a Proceedings of the IRE publication [cites Vol. 23, 1935, pp. 1201-1221] and to a new use for the types 57 and 6D6 to obtain negative transconductance and negative resistance [cites RCA App Note 45].”

  Lud

---

Our own Jacob Roschy had this to contribute in the same email exchange:

---

Hi Joe,

This morning I took a few shots of the 48, I hope you enjoy ! [ see uploaded top views in 48 page]

I wonder why this principle came to oblivion after the 48 ? Probably the producing costs were higher as of the usual beam power tetrode ?

Some 30 years later, these plate ridges became reinvented, as someone must have thought, if one combines these plate ridges with the beam focus frame of the BPTs, a much better screen-grid output tube must result.
This was achieved at least in Nov.1960, as the TV- sweep tube PL500 was introduced.
Valvo, the German tube division of Philips, called this special plate construction as Kammeranode (chamber anode), see more under 40 Jahre Valvo - Röhren post 3 : Die Fernsehröhren seit 1953

The PL300 has also this construction, as well the PL509 and the PL519.  [...]
Best Regards, Jacob

---

Lud had a further contribution in the Tube Collector Association Yahoo newsgroup on the 48:

---

You might compare the 48 with three beam-power tubes specifically intended
for 28-volt B+: the RCA 26A7 (dual), the Tung-Sol 26E6WG, and the Sylvania
28D7. The latter was pretty specifically promoted as a replacement for the
48. Also, the Western Electric 375A beam-power tube was optimized for
48-volt supply and may be a contender.

I once reprinted an article on use of tubes by the Western Union Telegraph
co. ("TC," Feb. 2005). In it, the author fairly swoons over the 48: ". . .
let me say that in the early days of the application of electronics to
telegraphy, the good old '48' found many uses. This tube literally has not
been beaten yet. Many hundreds have emitted 70 milliamperes continuously
for many years and are still going strong; and in multichannel carrier it
made a more linear push-pull class A amplifier than we have been able to
build with any more modern tube."

Lud

---

I obtained a NOS 48 by National Union from Jim Cross at Vacuumtubesinc.com. Jim is also the current Chairman of the board of directors of the Tube Collectors Association. Fortunately, this National Union 48 has a similar structure to the RCA. (Raytheon also marketed a 48, but it was a Pentode of comparable characteristics, but a characteristically softer Plate knee region)

In an email exchange with RM member Paul Reid, he pointed out the use of this tube in 32VDC farm radios. See the Silvertone 4612A from Sears, for example. This application takes advantage of the low mu=2.5, in operation with only 32V at the screen and plate, as might be found in an American farm radio that got it's power from wind mill charged batteries. While the plate draws a nominal 52mA in class A operation from 96V, it still draws a usable 15mA from 32VDC between cathode and G2+Anode. This much plate conduction at 30V would be difficult to obtain even with a space charge grid, especially if the low plate knee of the 48 is taken into consideration.

I plot here the curve traces for my NU 48 operating with the screen G2 at 96V, 32V and 26V.

 

Note how dramatically reduced the available power output is, when the G2-cathode voltage drop is cut 32V. The 48 was usually configured in class AB push-pull to deliver about 3x more power, up to 400mW. Some designs included a bias resistor at the cathode that raised the cathode about 6V, thus reducing the cathode to G2 drop to 26V, and the single ended class A power was thus dropped to 80mW and the class AB push-pull power was down to 250mW.

A reasonably efficient speaker can deliver plenty of volume to fill a room with 250mW to 400mW of driven power.

 

Paul Reid reminded me that the Langmuir-Child 3/2 power law for plate current as a function of Screen grid Vg2 and control Grid Vg1 voltages predicts the DC bias point quite well. If there is a set of Screen and control grid voltages and plate current, then a new plate current can be calculated for a new set of voltages without having to know the perveance of the tube. When reducing the Screen voltage, the control grid bias should be reduced in a similar proportion, so that it remains approximately half-way to cutoff.

The 48 curves show 15mA plate current for 32V and 10mA for 26V at the screen G2. This agrees well with the calculation result of 17mA and 10.6mA.

 

A 50C5 is compared here in the last calculation because it draws similar Plate current levels from similar screen grid voltages, yet the much higher mu=6 reduces the current disproportionately at 32V. Note that the grid swing of the 50C5 will be also limited by the start of conduction around -1V and complete cutoff will be experienced at 32V/6=-5V with a 32V supply.

Regards,

-Joe

To thank the Author because you find the post helpful or well done.

Radio Age editor Ed Lyon had this to say in a private email about the Pentode variant and the application of the 48 in American farm radios:

I had thought that there were more brands that copied Raytheon’s design, but maybe they were all made by Raytheon and were rebranded.  I notice that RCA and Sylvania differed slightly, one (RCA) having 8 ribs in the anode, the other having 6.

Incidentally, the use of the 48 in farm radios was for a specific power supply on farms – the Delco Light Plant.  This was a bank of 16 2-volt lead-acid cells in glass containers, charged by a one-cylinder forced-air-cooled GM engine with a Delco d-c generator direct coupled to it, all in one casting.  The generator was used as a starter if the batteries had enough charge left, but there was a crank on the flywheel just in case...  My father got one of these outfits from another farmer who had got lucky and had power installed by the utility company just before the copper shortage of 1941.  This was my best ever radio-constructing power system – 32 volts for B+ and 2, 4, or 6-volts for A supply, obtained by tapping from the negative end of the battery bank.  We collected rain-water regularly in an enameled pail to re-fill the cells as the electrolyte evaporated.   Prior to getting the Delco Plant, my A-battery was “borrowed” from the tractor, and I made B cells with vinegar, copper and zinc strips, and cardboard separators in ½-pint jelly jars.  A bank of 40 of these made about 45 volts, good for a few milliamps.

Ed

That is some pretty amazing resourcefulness on Ed's part, and a very understanding mother to give up so many jelly jars.

The house wiring for the 32V lighting system must have been quite heavy.

Regards,

-Joe

To thank the Author because you find the post helpful or well done.

> G2 drop to 26V, and the single ended class A power was thus dropped to 80mW and the class AB push-pull power was down to 250mW.

The Silvertone 4612A schematic linked here has a Power Output spec:

Undistorted: 0.15 watts
Maximum: 0.32 watts

On your graphs, I figure 20V 20mA peak swing. In class B (or AB) this is 0.4W peak (or squarewave), 0.2W Sine. With transformer and other stray loss, and conservative interpretation of "undistorted", this 0.2W matches the 0.15 watts claim. The 0.32 watts number is likely gross distortion, essentially square-wave.

> Langmuir-Child 3/2 power law for plate current

It may be more useful to think "cathode current", not "plate current", when applying the space-charge equation. With just one positive electrode, as in a diode, and in a negative-grid triode, no difference. When you start throwing multiple electrodes positive, you must consider their relative leverages (Mu) and voltages to derive an equivalent voltage, then their relative capture-areas (not commonly known) to estimate the division of current between electrodes.

Yes, the engine-generator lighting system AND the wind-generator system were common on wealthy US farms 1920s-1940s. The engine only needs gasoline and runs any time. The windmill works for free if you have wind. In the Midwest, wind was intermittent and fuel was readily available. In Kansas it is a long way to the store but the wind never stops.

Either source wants storage. Inefficient to idle a 600W engine for one 25W lamp. Even in Kansas the wind may stop. Storage comes in 2V units and at the time a separate glass jar for each 2V unit. OTOH copper costs more than insulation so distribution favors higher voltage.

So the systems-engineering problem is to balance number of jars versus fat wires. In a homestead, the loads are mostly small and nearby. 32V (16 jars) seems to be the accepted voltage for small units. (There were also 110V DC units for larger systems.)

> The house wiring for the 32V lighting system must have been quite heavy.

Well, serious, but probably no fatter than urban houses. The "competition" was candles and kerosene lanterns. Only the biggest kerosene lanterns had the output of a "normal" incandescent. I have 240 Watts (equivalent incandescent) of lights on right now; a 1920s farmer might be thrilled with two or three 25 Watt bulbs. Also the NEC suggested 2% voltage drop is for negligible flickering in systems with large intermittent loads; a farm system may have no large loads and be quite understanding of flicker when loads are switched on/off. A 30-foot 2-wire circuit in #12 (same as good 120V wiring) is about 0.1 ohms. For 10% drop at 32V that is 29 Amperes or 900+ Watts. Turning 25W-50W lamps off and on would have nearly no effect on other lamps. I see Delco sold a clothes-iron... taking this at 600 Watts it would cause a perceptible dip when turned on (around 6%), but this may be preferable to firing-up the wood stove to heat a sad-iron. I could see a #12 wire run from cellar to kitchen, and #14 for general lighting. No more wire than a city-house, perhaps less because the large cost of fuel and batteries discourages excess use.

To thank the Author because you find the post helpful or well done.