EL84/6P14P Output Stage in UL Operation
A measurement Report, by Dipl.- Ing. Wolfgang Trinks.
Introduction:
two years ago, I designed and built a compact amplifier featuring the ECL86 tube. The performance proved to be very satisfactory for such a tiny tube and I decided to create the mature version by implementing the ECC83/EL84/6P14P combo. The amplifier operates single ended in Class-A, Ultra Linear mode and utilizing overall negative feedback. It is an “honest” 2W amplifier (THD <0.6%, 30Hz - 20kHz) with a headroom of up to 3W+ before clipping sets in. A MOSFET stabilized power supply was chosen, providing a 100Hz attenuation of more than 100dB. As a result, both S/N and stereo channel separation are better than 85dB based on 1W@8 ohms.
The relevant publications about Ultra Linear operations, I know, tend to be written with the intend to finally evaluate the application in push pull amplifiers. I was however interested in measuring the Ultra Linear operation of the single ended amplifier I have built. In order to exclude a feedback effect on the driver stage by the input of the power amplifier tube, the signal generator was fed by a low impedance directly into the grid of the power amplifier. The output stage was measured as an independent unit.
By coincidence, the layout of this output stage, as seen from the input of the output tube via the output transformer resembles an impedance converter with a voltage gain of unity. So 4Vrms on the grid of the power tube correspond to 4Vrms on the 8 Ohm load. (This comes in handy, because now the actual voltage amplification takes place in the ECC83 from the system point of view, so the driver tube represents the VAS).
The output transformer is an EDCOR GXSE15-8-5K and alternatively a Lundahl 1663. The efficiencies of both transformers differ, for the EDCOR incorporating an EI-core and set for 100mA DC tolerance was measured for an efficiency of 87%, the Lundahl, incorporating a C-core set for 50mA DC tolerance (and thus has a smaller air gap) for an efficiency of 95%. The different efficiency was not considered for the measurements, as it did not have a relevant influence on the results. The EDCOR provides a 40% UL tap, the Lundahl 33%, 50% and 66%, with only the first two being used.
Measurements:
First, let's have a look at the output power, over the input voltage for the three operating modes. It doesn't come as a surprise, the pentode provides the most amplification. Followed gradually by the UL-variants and finally the triode mode providing the least amplification.
Now let's have a look at the total harmonic distortion (THD) of all operating modes. (K2....K4) As expected, the UL-traces are moving gradually from the pentode-trace towards the triode trace, as the percentage of UL tap increases and the available power, limited by the onset of clipping, is continuously decreasing. Due to it's low gain, the triode operation does not really perform well, as the input signal soon reaches the grid bias voltage. The pentode mode seems superior to all others.
For a closer look, THD and the components k2/3 are measured in the respective operating mode.
And here it gets interesting:
As expected, the triode shows the lowest k3, the pentode shows plenty of k3 above 2W. The UL mode offers a decent gain, while minimizing the K3 content.
Conclusions:
The Ultra Linear operation transforms k3 into k2, I could not observe a general reduction of THD in the course of this evaluation using the UL taps provided by the output transformers. It reduces the proportions of k3, but accordingly more k2 is produced. The THD (Total Harmonic Distortion) remains almost constant. The effect increases the further the tap is located in the output transformers primary winding in the direction of the anode. In other words, the higher the specified percentage of UL tap is used.
At the same time, the available power of the pentode is reduced by more and more local negative feedback.
So the task is to choose the best compromise between output power and reduction of k3. In the diagrams shown above, the Lundahl 1663 represents this optimum at 33% UL, followed by the EDCOR at 40% UL.
In push-pull amplifiers, even-numbered distortion components in the output transformer cancel each other out, whereas odd-numbered ones add up. The reduction of the k3 components through UL operation is extremely advantageous here. They also usually develop more power than a single-ended amplifier, so the reduction in available power can be accepted much easier.
In the case of single-ended amplifiers, UL operation also offers sonic advantages due to the possibility to trade k2 for minimizing k3. However, the power reduction might be of higher relevance.
In the case of an amplifier with over-all negative feedback, k3 of the pentode does not have to be seen as an exclusion criterion from the outset, because the negative feedback also compensates for k3. At the same time the pentode offers a higher open loop gain compared to the UL operating mode.
But if the amount of negative feedback is limited due to available bandwidth or to deliberately allow some harmonic (k2) content (that makes up the sound of a tube amplifier) 3rd order content will also be part in the sound spectrum in the case of the negative feedback pentode operation.
In the literature it is described that the Ultra Linear mode of operation transforms the output impedance of the pentode almost completely to the lower values of the triode, even at taps with low percentages.
I didn't make any research into that subject, however I noticed that there is an analog effect on the Miller capacitance of the power amplifier tube. The wiring of G2 to the UL tap obviously results in canceling out the shielding effect of the screen grid from the anode. This happens even with an UL-tab of small percentage.
The Miller capacitance increases almost to the value of the triode mode and thus poses a much higher frequency dependent load, than the pentode would do on the driver stage.This should be taken into account, especially if a UL operation is to be retrospectively inserted into an amplifier concept that was intended for a pentode or beam power tetrode that is originally operated as such in the circuit.
Summary:
The relevant publications already mentioned in the introduction of this text, describe Ultra Linear operation as a method for the effective overall reduction of distortion (THD). However, this can only be seen as a general statement in connection with the property of the push-pull power amplifier, which almost completely suppresses k2 by extinction due to the operating principle.
- The Ultra Linear operation of the single ended power amplifier measured here does not reduce the total harmonic distortion (THD) in any way.
- 3rd order harmonics are transformed into 2nd order harmonics. The sum of all distortions (THD) remains constant in the first approximation. Even with single ended amplifiers, the use of Ultra Linear operation with little loss of available power, makes it possible to reduce the unpleasant sounding 3rd. order distortion, and get the pleasant 2nd. order distortion instead.
- Ultra linear operation reduces the internal resistance of the power amplifier tube, to lower values, similar to triode mode. This is even so at low percentages of the tap at the output transformer. However, the Miller capacitance of the tube increases as well, which must be considered when designing the driver circuit.
Note: A very detailed data sheet for the EL84 is the edition of Mullard (1961-64). Diagrams of Pentode, Triode and UL Mode, with different taps were published here, but the latter only for push-pull operation. The presentations of the Pentode and Triode Operation, correspond nicely with the measurements carried out here.
About the author
Wolfgang Trinks completed his electronics studies in Germany in 1987, at the University of Applied Sciences, Karlsruhe. After working in R&D at Beckers Autoradio in Ittersbach, Germany, he moved into aviation in 1998, were he served as technical manager.