The Power Output Stage Pt.3: Configuration Performance
The previous information on internal resistance and output stage configuration is important to have a basis of understanding as to how and why various popular modifications to the output stage affect the sound as they do. The common understanding is that it is tubes themselves that produce a unique sound quality when in reality, it is the interaction of a particular tube's characteristic with the circuit it is installed in that produces a given sonic quality. Therefore, changing the circuit can have just as much impact on sound quality as changing tubes can, if not more. Changing the output stage configuration in the ST-70 then is no different.
FEEDBACK -- Claims are often made how feedback hurts performance, so some modifications result in reduced, or potentially no feedback at all. Feedback primarily affects frequency response, distortion, and internal resistance. It would take having a feedback modified unit on the test bench to determine the impact produced on these areas, but suffice to say, that a stock ST-70, when operating with the NFB loop disabled, will display an internal resistance of 19 ohms on the 8 ohm tap, and a distortion of 1.1% at 1 kHz, at 1 db below 35 watts output.
Since amplifier internal resistance can have a significant effect on the sound produced, it needs to be minimized as much as possible, so that the speaker and amplifier can each be judged on their own merits. After all, when the characteristics of one component affects the operation of the another, the ability to make an accurate assessment of either then is destroyed.
Besides amplifier internal resistance, the other electrical element that influences speaker damping is the DC resistance of the voice coil itself, with a typical 8 ohm speaker often having a DC resistance of about 7 ohms or so. With a quality amplifier then, it is the DC resistance of the speaker itself that primarily determines how effective the electrical damping action is. So what level of internal resistance should a quality amplifier represent?
For the internal resistance of an amplifier to effectively be a non-factor, it should represent no more than 1/10th the DC resistance of the speaker. Applying this back to the UL output stage of the ST-70 then, it can be seen that with an internal resistance of 19 ohms on the 8 ohm tap, some help is needed from a feedback network to reduce this value low enough so as to remove its influences on the effect of speaker operation. Even if the output stage is configured for the lowest possible internal resistance (triode), that value (7.8 ohms on the 8 ohm tap) is still plenty high to influence the bass response of most speakers.
As a result, any modification that does not include SOME reasonable level of feedback is almost certainly going to color the sound produced, either by way of its internal resistance, or distortion produced. Therefore, designs that claim that feedback is not needed due to the use of a UL or triode output stage should be held in question, with regards to accuracy produced. The stock ST-70 displays an internal resistance of .72 ohms on the 8 ohm tap, effectively removing any coloration it might contribute due to the effects of internal resistance.
Internal resistance is only one amplifier characteristic that can influence reproduced sound. Like testing for frequency response, it is typically measured at low power levels. The power response produced over the audio bandwidth is another. But just as important is the quality of power response produced in the face of a varying load impedance, like that of a speaker. The configuration of the output stage can have a significant impact on this issue as well.
ALTERNATE OUTPUT STAGE CONFIGURATIONS
TRIODE OPERATION -- As mentioned, the internal resistance of a triode connected ST-70 output stage is about 7.8 ohms on the 8 ohm tap. Since triodes have the least gain of all output stage configurations, the feedback loop is not as effective in reducing this down to an acceptable level as when operating in UL. But that's OK, because triodes start with a lower internal resistance to begin with. With a triode connected output stage, the nominal 20 db of global NFB employed in the ST-70 is reduced to about 16 db, which is still enough to reduce the overall internal resistance of the amplifier down to about .74 ohm -- low enough to effectively remove it from introducing any coloration of sound.
Against this however, is the poorer HF response that triodes display, causing the stock frequency response at 20 and 30 kHz to go from -.25/-.5 db, to -1.0/-2.25 db respectively. A 10 kHz square wave presentation clearly shows the effects on frequency response, when the output stage is converted to triode operation.
One major benefit of triode mode however, is the power response curve produced with respect to variations in loading impedance. Triode mode will produce a flat power response of about 17 watts RMS within the audio bandwidth. But importantly, as the load impedance swings from 1/2 rated load, to 2X rated load, the available power becomes about 16 watts and 14 watts respectively across the same frequency bandwidth, while still retaining quite reasonable distortion figures in the process (<.4% at 1 kHz).
So while the triode operation presents challenges in the frequency response department, its ability to handle varying loads while still maintaining low distortion performance is excellent.
PENTODE OPERATION -- A pentode output stage has the highest internal resistance. So connected, the ST-70 output stage has a whopping 105.7 ohms of internal resistance on the 8 ohm tap! But pentodes also have the most gain of all output stage configurations, so the feedback loop becomes more effective in reducing this down to an acceptable level, than with any other configuration. With a pentode connected output stage, feedback rises to 23+ db, which is still effective enough to reduce the overall internal resistance of the amplifier down to .82 ohms on the 8 ohm tap, and enough to prevent it from introducing any coloration of sound.
Pentode operation also causes a rise in HF response, showing +.10, +.25, and +.5 db at 10, 20, and 30 kHz respectively. A 10 kHz sqare wave presentation clearly shows the effects on frequency response, when the output stage is converted to pentode operation.
At a constant impedance and normal loading, the power response in pentode mode is flat across the audio bandwidth at 35 watts RMS, although distortion is at least double that of either triode or UL operation. And while power and distortion are acceptably maintained when the load drops to 1/2 of rated value, distortion deteriorates notably at 2X rated load, effectively becoming a 20 watt amplifier under this condition, with distortion climbing yet another three times over normal loading of pentode operation (to ~ 2.4% at 1 kHz). This is as much a product of pentode operation itself, as it is the A-470 output transformer providing marginal loading conditions for pentode operation of EL34s.
So pentode operation produces elevated frequency response, and an acceptable internal resistance level when feedback is applied, but has increased distortion with a rising load impedance.
ULTRA-LINEAR OPERATION -- UL operation displays a low internal resistance as mentioned earlier (.72 ohms on the 8 ohm tap), preventing it from introducing any coloration of sound from this characteristic.
Since the amplifier was designed with this output stage configuration from the ground up, frequency response is as is stated in the Measured Stock Performance Post, failing to meet published specifications, but still being down only .25 db at 20 kHz. A 10 kHz square wave presentation of the stock unit was provided in that post.
With a constant impedance and normal loading, a flat power response at 35 watts RMS across the full audio bandwidth is produced, at the lowest distortion levels of all configurations, as posted in the Stock Performance Post.
When the load impedance varies by 1/2 or 2X rated load impedance with UL, the power response is very similar to that of pentode operation (with maximum power at 2X rated load being ~23 watts). HOWEVER, because of the UL connection, distortion does not climb above .4% at 1 kHz under either condition.
SUMMARY
While UL and pentode operation produce basically the same power response across the audio bandwidth and with respect to loading conditions, UL operation ALSO shows the same INsensitivity to distortion with varying load impedance levels that triode operation does -- while watt for watt, producing the lowest overall distortion of all three output stage configurations. This is the beauty of UL operation, combining the best that the other two modes have to offer in one package.
In the ST-70 then, UL operation of the output stage with feedback would appear to be the best compromise, considering the power, distortion, frequency response, and damping factor targeted for the unit. Triode operation can certainly be a viable alternate configuration, with the advantage of that mode typically producing lower order distortion products, but at a cost to HF response and power. Pentode operation of the ST-70 would not be recommended with EL34 tubes, although may prove acceptable with 6L6 class tubes. This was not investigated for this project.
Pics include:
1. A 10 kHz square wave in triode mode.
2. A 10 kHz square wave in pentode mode.
The last stop will be the driver board.
Dave