 |
The 71A, also known by a myriad of equivalents such as:
|
It is a forgotten gem from the golden age of vacuum tube audio. This directly heated tube was originally designed to power portable radios, giving them a voice as early as 1926. What we could call the “mother” of the more famous 45, the 71A tube was the forerunner of some of the most beloved tubes in audio history, including the 2A3. With a filament operating at 5 volts and only 250 milliamps, the 71A was designed to work even on battery power, bringing music wherever portability was needed. Despite its modest power—just 0.8 watt output into a 4800 ohm load—the 71A still has surprises in store for audio enthusiasts. The true charm of these tubes lies in their sound. They are perfect for those who enjoy low-volume listening and can drive very high-efficiency loudspeakers, delivering a warm, enveloping and detailed sound. Every note and nuance is reproduced with a purity and naturalness that only tubes can offer. I was commissioned to develop a custom output transformer for the 71A tube, and I’m proud to share the result of months of research, design and testing. In the following lines, I will guide you through the process that led to the creation of this exceptional audio component.
Given the low internal resistance of the 71A tube, I made a deliberate design choice for the output transformer. I opted for a core of grain-oriented laminations for a precise reason: these cores provide exceptional performance in the mid-high range, where sound clarity is crucial. They ensure extremely pure and detailed sound reproduction, without any coloration compared to other theoretically more prestigious cores that may offer higher primary inductance but also bring some drawbacks.
For the secondary winding I chose silk-covered Litz wire to maximize signal transmission. Litz wire minimizes losses, further preserving audio signal integrity. This results in a sound rich in detail and free of unwanted interference, ensuring top-level reproduction for the most demanding audiophiles. In the photo you can see the two spools with the wire for the primary and the secondary (Litz).
For internal insulation of the transformer I used a thoughtful combination of materials, including vintage transformer paper inside the layers and polymer insulators mixed with paper to separate primary and secondary parts. This decision was guided by several considerations.
The vintage transformer paper was used for its ability to effectively dampen vibrations of the copper wire, helping to minimize any noise caused by these vibrations. However, even though the tube’s anode voltages are quite low, it was still essential to maintain reliable insulation between primary and secondary. For this reason I integrated polymer insulation, which ensures effective separation of the transformer’s two sections. This combination provides a complete solution: the vintage paper controls vibration, while the polymer insulation preserves the integrity of the separation between primary and secondary. The result is an output transformer capable of delivering high-quality audio performance while maintaining high safety and insulation standards, even with the relatively low anode voltages mentioned above.
I applied the reverse-winding technique to certain parts of the coil to mitigate potential electrical issues. This approach was key to optimizing the output transformer’s performance. The photo shows the transformer after two impregnation phases: the first with clear varnish, the second with black enamel.
Impregnation in transformers has a dual essential function. First, it creates reliable electrical insulation around the transformer windings, preventing short circuits or discharges between internal parts for safe, efficient operation. Second, the impregnant stabilizes and secures internal parts, reducing vibrations and unwanted noise—an important factor in high-fidelity audio applications. In addition, impregnation forms an effective protective barrier against moisture, preventing damage and deterioration caused by environmental conditions.
The pinnacle of perfection would ultimately be encapsulating the transformer in a metal case sealed with resin. This solution would not only ensure optimal protection but also enhance its appearance. Moreover, sealing in resin plays a key role in preventing mechanical stress on the core, for example when the transformer is screwed onto a surface that might otherwise alter its primary inductance, thereby maintaining stable performance.
At the same time, resin sealing provides an effective barrier against moisture, helping preserve long-term performance and preventing damage and deterioration caused by ambient humidity. All of this ensures that the transformer operates safely and reliably in a variety of conditions and applications. However, for current prototypes we must settle for the present configuration.
Rather than showing expensive measuring instruments with a long sausage of resistors connected to my transformer, I prefer to share the simplicity of my test bench. It consists of a simple piece of wood with a socket screwed on, on which you can see a real UX171 tube. This tube is powered by my laboratory power supply and stabilized high-voltage supply. Using the real tube instead of artificial simulations is a test that is worth far more than any frequency response data obtained with costly instruments. With this setup, the transformer is tested under real-world conditions, ensuring it delivers excellent performance despite the simple test bench.
I now want to share a fundamental yet often underestimated aspect of transformer construction: the complexity of making it perform at its best. A transformer cannot simply be calculated on paper and work perfectly. To illustrate this concept, here are two graphs representing the transformer’s frequency response.
In the first graph you can see the frequency response of the transformer before impregnation with electro-resin. What makes it interesting is the second graph, showing the frequency response after impregnation and about five days of drying.
| Before (note: graph starts at 20 Hz) | After (note: graph starts at 10 Hz) |
 |
 |
Aside from my mistake in setting the first graph to start at 20 Hz—which might mislead the observer—my focus is on the upper part of the spectrum. In the first graph, you can see a frequency response reaching ?1 dB at 30 kHz, with a slight bump (+0.2 dB) starting at 1 kHz and peaking at 10 kHz. In the second graph, however, you see a perfect response from 10 Hz to 25 kHz, measured at ?1 dB instead of the ?3 dB often used by others. The “bump” present in the first graph has disappeared, showing perfect linearity. If the dry transformer had not shown this “feature,” once impregnated it could instead have exhibited a slight attenuation at 10 kHz.
These changes result from impregnating the transformer with resin, which modifies the dielectric characteristics of the insulating materials, especially those that absorb it. This effect is observed not only in vintage papers but also in lateroids and Nomex. Variations may also occur with plastic insulators, since hardened resin fills the space previously occupied by air.
True craftsmanship lies in understanding and managing this behavior. It is not always achieved on the first try; sometimes you need to measure, discard the piece and rewind a new one, refining the approach. This research process takes days of work and repeated tests over time, as the transformer’s behavior changes even as the resin dries inside.
This transformer-design method is extremely time-consuming but is the only way to guarantee a flawless product and to state specifications that truly match reality. That is why transformers made in this way can be considerably more expensive than others. You can find more information about my transformer-design approach and the challenge of comparing it with others’ often cost-driven methods in this article.
Let’s now look at the square waves at 100 Hz, 1 kHz and 10 kHz…
The transformer’s primary impedance is 4800 ohms, with a single 8 ohm secondary. All measurements were taken at 4 Vpp on 8 ohms (0.3 watt). The maximum output power is 0.69 watt (my tube is used and not 100 % efficient, though!). The primary inductance is 26 H, measured at 100 Hz with a 2 volt signal. Leakage inductance is 26 mH, and primary DC resistance (RDC) is 425 ohms.
The graph below shows the harmonic spectrum produced by the UX171 while operating at 0.3 W RMS, obviously with zero feedback. At this low power I did not measure the damping factor, because it is so low that it would hardly cause any issues.
A small note: one of my contacts pointed out that the transformer’s bandwidth is not as extended as in other graphs shown on the site. However, at such low power the tube hardly has enough energy to overcome the absorption of the transformer’s parasitic capacitances, which can be in the range of 1800–2000 pF (as in this case). Still, in zero-feedback conditions a minimal bandwidth is sufficient, because no intermodulation occurs.
In all small-signal transformers I have published or will publish on the site, I have looked for these very characteristics. This is because the tubes involved are of low power, and achieving higher performance would be difficult and, above all, unnecessary, since such circuits do not require feedback loops. As I explained in this article, the need for extended bandwidth is linked to eliminating or limiting the distortions introduced by feedback. The latter is useful to achieve acceptable damping factors, but such issues do not apply to these particular low-power applications.