Push-Pull Triodes, Feedback, and Damping

Development of the EM-7 push pull 10-watt amplifier has led me to some interesting thoughts on the topic of transformer winding resistance and its effect on damping (output impedance). Output impedance and damping factor are the same thing, just stated differently. I like to get a damping factor of 5 to 10 on my amplifiers. Many Single ended amplifiers without feedback are as low as 1 and a pentode amplifier with no feedback can be as low as 0.1. Yes, damping can get to less than one. At that point the amplifier is a current source rather than a voltage source and the output will vary directly with speaker impedance which will certainly result in a different sound, one which was likely very far from what the speaker designer had in mind. A damping factor less than 5 can cause noticeable frequency response differences on many speakers while one above 10 will play most speakers with little difference. Damping factors from 20 on up make no difference at all as the output impedance is now so small as to vanish. 

In reading spec sheets from several amplifier manufacturers, I see no mention of damping factor, regulation, or output impedance. This rather important aspect seems to be ignored, perhaps because the answer is not so good. Achieving good damping without feedback, which seems to be the main selling point of these amplifiers, is no easy task. In making a survey of output transformers that DIYers are using for non-feedback amplifiers I find that rarely is any mention made of insertion loss or winding resistance. These affect the damping of the amplifier greatly. In feedback amplifiers one can reduce the output impedance very easily with feedback. In non-feedback amplifiers one finds that damping is quickly lost in the transformer with no way to recover it. Hence, I am working on some very low resistance transformers.

Just to get a handle on how this affects what you hear let us look at some simple numbers. Damping factor is simply the load impedance divided by the output impedance. So, a damping factor of 10 on an 8-ohm speaker represents an output impedance of 0.8 ohms. There is also another way to look at damping. This is called “load regulation”, which is what I usually measure. To measure load regulation, you simply put a constant sine wave through the amp loaded at the impedance of interest (say 8 ohms) and then disconnect the load and see how many decibels (dB) the output goes up. Since 1 dB is about 10% that implies a damping factor of 10. It also directly tells us what will happen to the frequency response of a varying load much more directly than output impedance in ohms. 

Load, or output, regulation tells us something more interesting than damping. Using it we can more easily predict how un-flat an amplifier's output will be into a widely varying load like an electrostatic speaker where the impedance can vary from 100 ohms to 1 ohm over the audio band. Someone brought me an OTL amplifier to listen to some years ago and we hooked it up to my QUAD 63s which range from about 50 ohms in the bass to 4 ohms in the treble. The amplifier had a damping factor of about 1 or a regulation of 6dB. Which is an output impedance of 8 ohms. Another way to look at that is to imagine a 0-ohm amplifier with an 8-ohm resistor in series with the hot lead. If we know the regulation is 6 dB that tells us that the output can climb no more than 6 dB into an open circuit, which 50 ohms is close to, and the output can also fall 6dB into a load of 4 ohms. This is indeed what I heard. The bass was up 6 dB and the treble down 6 dB thus altering this exceptionally flat speaker into a very un-flat sound. Interestingly, the amplifier's manufacturer sold one to a QUAD 63 owner, who was very pleased with it. There is indeed no accounting for taste. 

In the previous example and in the case of most cone speakers the impedance peak in the bass is quite narrow and the result is “one note bass”. The speaker will literally ring at that frequency and you basically get pitch less bass (ex. thump, thump, thump). Now if an amplifier had 1 dB regulation the output can rise no more than 1 dB no matter how high the impedance peak is. If the speaker is a reasonable load, it likely will not go below 4 ohms and the output will then fall only 1 dB making the total un-flatness 2 dB which is not so bad. A lot of SET, SEP, and non-feedback amplifiers will have low damping/poor regulation, and many manufacturers do not even state these characteristics. No wonder listeners hear great differences with these amplifiers. I would not want to make such an amplifier and therein lies the challenge. How does one get low output impedance without feedback?