Wunderlich operational details

Fellow Radiophiles:

There is a misconception in the usual understanding of plate ripple performance with push-pull detection in the Wunderlich. With perfectly matched drives and grid gains, with the plate biased for linear audio amplification, there is no 2x carrier ripple at all at the plate, even without any capacitor filtering at the plate.

As conditions deviate from the ideal, the difference in gains between the two grid paths will produce a small ripple at the carrier frequency. With larger signals, say, when  enough of the 3/2 gm power law is traversed for 10% distortion, a 2x carrier signal will emerge at the plate as each positive peak is amplified with a different gain from the negative peak between the two grids.

Input loading of the peaks will also produce this slight distortion which results in 2x carrier content at the plate. All this is true for an unmodulated input carrier with RC biased grid leak detection. The capacitor in the RC grid leak is essential because it keeps the grids DC-biased at mid carrier amplitude. When modulation is applied, then there will be a small carrier frequency ripple that has the amplitude of the audio increment from cycle to cycle. The unfiltered plate ripple will be greatest with 100% audio modulation at fastest modulation frequency of 4kHz.
The center tap of the driving transformer, where the AVC voltage is picked up, has the usual detection ripple across the RC grid leak, but at 2x the carrier frequency. For these measurements, I chose a very large grid leak cap of 0.1uF to eliminate all ripple at the grid leak RC, because I wanted to look for ripple at the plate that was only caused by grid imbalance or non-linearity. The Carrier component in the second photo is due to the slightly different grid gains, while the smaller 2x carrier component is due to slight non-linearity caused by the 24Vp-p swing at each grid. With the audio gain from the center-tap running approximately at 10, the 24Vp-p carrier only survives as a 10Vp-p signal at the plate, this is a 1/24=27dB reduction in carrier amplitude at the plate. A manual trim of the relative input grid amplitudes could have reduce the output carrier, to the point that only the 2x carrier component survives at the plate. 

The low mu=12 for the two grids driven together, as they are driven by the audio signal developed at the grid leak, and the plate bias via a large resistor to a high Bplus and effective AGC, should eliminate the cutoff type of distortion that is common with triode grid leak detectors driven with large signals with a low fixed bias at the plate, as was the case with battery sets of the 1920's. It would take 250V/12=-21V at both grids to cutoff the Wunderlich plate in the schematic shown above.

If the capacitor in the RC grid leak were removed, and the resistor were of sufficiently low impedance for a fast time constant, including parasitic capacitance, to follow the carrier cycle by cycle, one of the two grids would always be forward biased and we would have a very good carrier frequency doubler, as the plate current would appear full-wave rectified.

This photo shows this principle with 60Vp-p at each grid but with an input frequency of 160Hz to insure accurate fullwave tracking at the grids.

This insight came up from my design effort with the class AB p-p design of the Gammatron output stage in Sputnik.

Curve traces

This curve tracing super-imposes the two grid families in two colors. Note that the mu=6 for each grid is about half that for both grids, which is 12 as seen in the plot on the left. The grid match is best at lower grid bias levels.

 

The following plots show that a negative bias on one grid causes the curve family for the other grid to be shifted to the right. The second plot has a -20V bias on one grid, which caused with curve family for the other grid to move it's zero bias trace to the -18V position. This shows that the fields of the two grids add fairly linearly to control plate current in this 0-20V range.

 

 

 

This plot shows Gammatron style operation of the Wunderlich, where the plate is tied to one grid for the output voltage sweep, while the other grid is stepped at -2V/step. Note that the horizontal scale is reduced to 2V/div. Nearly all current in this plot flows through the grid that is tied to the plate. The plate carries little current.

This illustrates the potential application of the Wunderlich as a more efficient alternative to a space charge tetrode in low plate voltage applications. Space charge tetrodes employ the first grid as a space charge grid that draws a wasted current that is much larger than the plate current.

 

The following three plots sweep one grid while stepping the other grid with negative steps, while the plate is held fixed at 0V, +50V and +200V. Note how the effect of one grid over conduction by the other grid is neutralized by positive Anode voltages. These plots are useful to predict detection behaviour at the grid.

 

 

It seems that the salient feature of the Wunderlich, as compared to a conventional grid leak detector, is that there is much less carrier signal current at the plate. A filter cap can be added to the plate, but this already functions as the second pole in the envelope filter.  The very effective rejection of the carrier means that a faster grid-leak can be used with higher ripple because there is a second opportunity to filter without danger of diagonal distortion at the plate. The nearly constant and low impedance of the plate eliminates all danger of diagonal distortion there, thus making it a preferable spot for filtering. The plate impedance will be constant on a steady state basis, but will increase for stronger signals, as the grid leak circuit generates a negative grid bias.

After seeing the Wunderlich in action, it is clear to me that it can be functionally duplicated with a 12AE6 or 6BF6, which are triode-duplex diode tubes with comparable mu and gm to the Wunderlich. The connection would be slightly different with the two diodes driven by the ends of the coil and the grid tied to the centertap of the coil. This kind of hookup should make the operation of these two tubes nearly indistinguishable from the Wunderlich. A triode with three built-in diodes could emulate the Wunderlich B. The extra diode in the Wunderlich B serves as positive clamp to eliminate the chance of driving the AGC voltage positive in an amplified AGC scheme, which would induce signal path attenuation, thus inverting the AGC polarity, with the potential for latchup or AGC instability.

References

Die Wunderlich Röhren by Wolfgang Holtman (I read this German language post with the Google translator)

Questions about the Wunderlich tube by Wolfgang Holtman

"The Wunderlich Detector" by Ludwell Sibley, Tube Colector December 2006 Vol-8, no6, p33. Lud's article includes an extensive list of 12 references, many of which are contemporary with the Wunderlich tube.

"Wunderlich's B tube" by Ed Lyon, Radio Age Jan 2000 pp. 1, 3, 8-9.

---

Ed Lyon was very kind to lend me his Wunderlich detector for these measurements. Thank you, Ed.

Best regards,

-Joe

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

At the moment we show about 40 models with Wunderlich tubes and 14 different types of tubes. In many cases it is the same tube with other designation but as a reference work we have to show all.

This type of a Wunderlich tube is the most known - with nearly 40 models. You find the models by clicking the tube and then scrolling down the tube page to click the presented models, sorted by model year.

But we find also a car radio from Sparks-Withingon, Sparton 34 with the type 70. This was confused with the type 70 current regulator - until a guest, Ron Wing, Wichita, KS, USA, told me. He added: "In the book "The Collector's Vacuum Tube Handbook" by Robert T. Millard, The Sparton Model 34 is given as the only known model that used a 70."
He used the contact form, an easy way to bring in such corrections.

By the way: The Wunderlich A with 5 pin and a top contact must have used also in radios I believe - but we have not yet listed a model for it.

Lud Sibley published an article "The Wunderlich Detector" in the TCA Tube collector bulletin Volume 8, Number 6 - December 2006 about "A discussion of circuits, variants, and radios that used them". His list covers:
Allied A6 and S-7 Auto (6 V).
Audiola 13S8, 23S8Q, and 32S8Q (2.5 V);
Audiola S-6, S-7, 23S7 (which alternatively used the 85), 33A6, and 33S6 auto sets (6 V).
Hoodwin 6-33 AVC (2.5 V).
Knight Audiola Auto (6 V).
Lincoln R-9, Deluxe Dual, and SW-33 De Luxe (2.5 V);
Lincoln SW-34 Battery usees nine 2.0-V tubes and a 2.5-V Wunderlich!
Mission Bell 5AC and 17A (6 V); 19 and 19A (2.5 V).
Pacific Radio Exchange Spero Super (2.5 V).
Patterson Radio 107AW which can use alternatively the 55 (2.5 V).
J & L Sara Co., Inc. S-39B, S-41L, SJ-4, and SJ-4K (2.5 V); S-42B (6 V).
Scott 1933 Deluxe AVC Super (All Wave 12) and All-Wave 14 (alternatively the 55, 2.5 V).
Sheldon Radio Co. A6 (6 V); Super (2.5 V).
Silver-Marshall K (6 V).
Sparton 34, auto uses a 70 as detector-AVC tube only, sp. 1st audio stage. (6 V).
Stewart-Warner R104A (2.5 V).
Voco V80 and V100 (2.5 V).

Grillecloth.com shows the
Allied Knight Audiola Auto with 6 tubes: 36, 39/44, 39/44, 89, BR, Wunderlich. (Sylvania = S)
Allied Knight A6 Auto with 6 tubes: 6C6, 6D6, 6D6, 89, BR, Wunderlich. (S)
Allied Knight Audiola Auto with 7 tubes: 36, 37, 39/44, 39/44, 79, BR, Wunderlich. (S)
Allied Knight S7 Auto with 7 tubes: 6C6, 6D6, 6D6, 37, 79, BR, Wunderlich. (S)
Audiola 33A6 with 6 tubes - .... 6D6, Wunderlich, 37, 89 (Yaxley = Y).
Audiola 33S6: 39, 39, 36, Wunderlich, 89, BR. (Y)
Audiola 32S-8-Q, 13-S-8 with 24A, 27, 27, 35/51, 35/51, 47, 80, wunderlich. (S)
Audiola 33-A-6: 6C6, 6D6, 6D6, 37, 89, Wunderlich. (S)
Audiola 33-S-6: 36, 39/44, 39/44, 89, BR, Wunderlich. (S)
Audiola S-6 Auto with 6 tubes: 6C6, 6D6, 6D6, 89, BR, Wunderlich. (S)
Audiola S-7 Auto: 6C6, 6D6, 6D6, 37, 79, Wunderlich, BR. (S)
Audiola 23S8Q: 35, 35, 35, 27, 27, 24, Wunderlich, 47, 80. (Y)
Auto-Vox 75: 6A7, 41, 78, 78, 79, 84/6Z4, Wunderlich. (S)
Hoodwin Aero 6-33 AVC: 58, 58, 58, Wunderlich, 47, 80. (Y)
Lincoln Radio Corp Lincoln DeLuxe Dual 2B6, 2B6, 2B6, 56, 56, 56, 58, 58, 58, 58, Wunderlich, 5Z3.
Lincoln R-9: 45, 45, 56, 56, 56, 58, 58, 58, 58, 80, Wunderlich. (S)
Linclon SW-33 Deluxe, SW-33 Improved DeLuxe, SW-34 Battery (10 tubes).
Mission Bell 5 AC: 36, 39/44 39/44, 47, 80, Wunderlich. (S).

From Mallory-Yaxley (3rd and 5th edition): 11 models:
Audiola 23S8Q, 33S6, 33A6
Hoodwin 6-33 AVC
Mission Bell 19, 19A
Pacific Spero Super
Patterson 107AW
Scott Radio 1933 DeLuxe AVC Super (12 tubes)
Stewart-Warner R104A, 11 and E (6 tubes), Voco V100 and V80.

Sylvania shows additional:
Mission Bell 17A wit: 1V, 36, 39/44, 39/44, 41, Wunderlich (IF 262).
Sara Co, J&L 5-42B with: 6A7, 6D6, 6D6, 18, 25Z5, 25Z5, Wunderlich (IF = 125).
Sara Co. J&L S-41L: 2A5, 2A7, 58, 58, 80, Wunderlich (IF 125).

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