Single-Ended Transformer Design Considerations

There is much misinformation about Single Ended (SE) transformers resulting mostly from the expressed views made by "experts" which come from unfounded theories of this complex device. Due to silver tongued advertising it has become necessary for Electra-Print Audio to call attention to incorrect information in order to speed up the process of ordering a correct and practical transformer.

Bandwidth

A SE transformer cannot have a frequency response from 2 Hz to 80 kHz. A more realistic and practical average bandwidth is -1db 25Hz to 25 kHz. This bandwidth is measured with the tube plate resistance included from the signal source. Measured this way the transformer will now operate exactly as stated with the tube type indicated.

How Not To Measure A Wider Bandwidth

If you remove the series plate resistance from the signal source and connect the signal generator (600 or 50 ohms) directly to primary the measured bandwidth of the same transformer will be very wide, about -1db10hz to 70 kHz. This is because the low source impedance shunts out most of the inductive reactance and the transformer becomes a capacitor. Now you are measuring the output of the generator through a capacitor. The reason low plate resistance triodes work well here is the same process. These incorrectly applied measurements have simply been published to embellish advertising and price.

Core Materials

Nickel laminations are used for low-level audio, which have little or no winding current. But nickel saturates quickly and distortion will result at the bass end due to the primary inductance that decreases with higher signal level. Nickel cannot be of any practical use in SE type transformers because an extremely large amount of nickel laminations would need to be used with a large gap to slow its saturation. In the end, this would act just like standard lamination material. It's important to note that large nickel laminations are not manufactured and that standard laminations are a fraction of the cost of nickel.

Amorphous cores have their use with higher frequency service. They were originally developed for aircraft equipment operating at 400Hz. These amorphous transformers weighed less per Volt-Ampere (VA). For full bandwidth audio, amorphous transformers can be difficult to design. This material shows less permeability at low frequencies and peak at 1 kHz. Low frequency power response versus core mass could be compromised. A flat frequency response across the usable audio spectrum would then be a problem. These cores are also very expensive. These materials when used with a SE type transformer offer an output that varies with the change in permeability with signal level and frequency unlike the standard materials.

M19 & M6 Laminations

Core material made of M19 and M6 has been used for many years. The metal alloy is made of metals offering uniform increase in inductance as flat as possible for power transfer over a given bandwidth. These materials were developed to sell product to audio transformer companies for the high fidelity amplifier industry back in the 1950's. Designers had the other materials (except amorphous core) and did not use them for these products. Nickel had its place in low-level audio for the recording industry.

These standard materials, when used for SE types, would offer the flattest, most efficient power transfer, and overall acceptable measurement. M19 is used for SE types due to no zero crossing. M6 is more costly and its attributes not used in SE service. M6, as well as M19 can be used for push-pull.

What Are The Attributes Of M19 & M6?

When someone states that the sound created with exotic core materials is far superior to standard lamination materials you must ask yourself what happened to this reviewer's amplifiers bandwidth and power? This sound is just the new setting of the bass and treble controls. Our reviewer will probably just say, "it sounds better" having no idea what happened. We know that if there was an improvement then it should be measurable.

To date there are many materials available but with only two overall differences in the peak permeability minimum (Gauss/Oersted) with an average of 2000 for M15 to M56 lamination materials and 6800 for M6 (75% Grain). This indicates that realistically only two types are any good for a full bandwidth flat response and these are M19 and M6 laminations.

Wire

Copper wire is used to wind transformers due to it being one of the lowest Direct Current Resistance (DCR) of all the metals used to create a magnetic field. Copper is also used to receive the flux variations in the secondary windings. If silver wire is used for the primary, it will create the same field due to the same amount of turns needed as the copper. The only difference is a lower DCR than the equivalent copper primary, a costly decision for no performance increase. However, silver wire used for the secondary is more sensitive than copper. If the primary has very low-level flux motion from higher harmonic content sounds, silver wire will reproduce it. This is the only increase in performance offered by silver wire when used as a secondary.

Size

The size of a SE type transformer will be much larger than the equivalent push-pull transformer. The size is mainly determined by the amount of current that passes through the primary winding, but lower frequency response and power level also have an effect.

Primary Current

When a specific primary current is needed, this is what determines the primary winding wire size and core mass. As an example, if one desires a transformer with 100ma primary but with a 200ma maximum capability then it can only be built as a 200ma transformer with a core mass for 200ma. SE transformers should be specifically built for the tube and its single operating point.

Secondaries

In an ideal world, the SE type transformer would be a device that makes 100 turns on the secondary laying right next to every turn of a 3000 turns primary. In the real world, the most efficient arrangement of primary/secondary that can come close to the ideal is to use single secondary impedance interleaved with the primary. For example: a] 4 ohms or 6 ohms rather than 4 plus 8 ohms or b] 12 ohms for 8 and 16. Also, any other singular impedance (ex. 5, 8, or 10 ohms) are equally efficient. The reason for this is that multiple taps leave unused wire within the winding. These will cause dips or peaks across the bandwidth. Note that the early four secondary arrangements as the ST3KB style are this type with no unused windings.

Determining The Proper Plate Load

The plate load for a 300B is 3000 ohms. Not necessarily. Our Western Electric reference book indicates the 300B will operate from 1500 to 6500 ohms load depending on the operation point and job needed. This goes for all tubes – triode, tetrode, or pentode. When the SE output primary impedance is lower than what the tube is setup for, it will distort. If the primary impedance is higher than what the tube is setup for, it will work well. Best operation design for any power triode is to operate it very close to full plate dissipation for highest power output with 1% Total Harmonic Distortion (mid-power) or less.

To find the plate load that a tube likes, use a SE transformer that you know the ratio of. For example, 3000 ohms to 8 ohms use the math 3000/8=375 square root yielding 19.36 turns ratio. This is the test procedure: a] set up a breadboard with the tube selected, b] set up power supplies and such for full operation of this tube, c] connect a 1 kHz drive signal or viewing signal with a variable secondary load, d] then measure power output by taking AC Root Mean Squared (RMS) voltage, at the secondary with its load, and finally e] take voltage squared divided by load impedance used. This will give you the output power of this tube at its operating point you selected. To see what plate load the tube likes at this operating point let's use as an example that the highest power at good waveform is 10 ohms, and knowing the transformer is 19.36 turns ratio and it is a higher primary load because its now 10 instead of 8 ohms, lets say 4000 ohms to 10 ohm, this gives 20.0 turns ratio, so we are not there yet. Take 3800 and 10 ohms resulting in a 19.49 turns ratio and we are now close enough to 19.36. So plate load this tube likes is 3800 ohms to an 8 ohm speaker.