Page Title

Email: ZEN Amplifiers

Welcome back! I was worried, so I backed up a copy of every page of TCJ on to CD, just in case the journal disappeared from the web. I must admit to being only 30% of a tube head and 70% a transistor kind of guy. Boo, hiss. But before reading your journal, I was only 10% of a tube head, so I am making real progress. 50-50 is probably where I will end up.

I have been wondering why you haven’t commented on Nelson Pass’s Zen amplifiers in the journal. I for one sure like the idea of single ended amplifier that uses a long-lived MOSFET instead of a frail $400 triode. Did I forget to mention that I am very cheap, oops. Wouldn’t one active device make the purest amplifier possible? You did cover the “World’s Simplest SE Amplifier.” but it only put out 3 watts and my speakers need at least 20 watts. Could 10 6BQ5s be but in parallel to get 30 watts of output or would using a single MOSFET with a tube line amplifier be a more reasonable idea?

If you can’t respond to my inquiries, I understand. I am amazed that you have already done so much to advance this hobby of ours (and for free). Thanks.

Fredric

Minnesota

The ZEN amplifier is an intriguing design but with many limitations. First, let’s pull way back and think about what is going on in the amplifier: one MOSFET is doing all the work of amplifying the 1-volt input signal to 16 volts and driving 2-amps of current into the speaker and providing a low output impedance; all at once! What fuels this miracle? It cannot be the wimpy 1-volt input signal. No, it must the MOSFET’s transconductance which gives and voltage and current gain. Thus, the greater the transconductance, the higher the gain and the lower the output impedance. For example, a Gm of 4A per volt would gives us, with an input signal of 1V peak, a gain of 32 into an 8-ohm load.

The output impedance, however, would be next to infinity, as the MOSFET has little if any “rp.” If we apply a feedback loop across the amplifier, the gain might drop to 16 and the output impedance would then be 4 ohms, for example, or a gain of 1 and an output impedance of 0.25 ohms, that is if all the output were returned to the input. Notice the tradeoff: we can have high gain, but high output impedance or a low output impedance, but low gain. We cannot have both high gain and low output impedance, however. Well, what’s the big deal here? Just use a higher transconductance MOSFET or place a few devices in parallel to boost the transconductance.

Two problems: high transconductance MOSFETs are usually meant to be used at high currents, currents much higher than even our Class-A SE amplifier will draw. Unfortunately, the MOSFET’s transconductance is not a perfect constant, as it decreases at low current, which means that an 8A/V MOSFET might only have 4A/V of transconductance at 2A. Second, the higher the transconductance, the higher the input capacitance. Placing many MOSFETs in parallel will also multiply the input capacitance. Have you ever wondered why the ZEN amp uses such a low valued input series resistor (2K) for a voltage driven device? Why not use a 47k input resistor? The high input capacitance limits the value of this resistor, if a wide bandwidth is expected.

Now, in general, triodes have a much lower input capacitance than MOSFETs do, but if we divide the Gm by the input capacitance, I am not sure which device would win. One advantage the triode enjoys over the MOSFET is that it comes with a built-in feedback mechanism in the form of its plate resistance. So, for example, if 10 6BQ5s were placed parallel, the effect rp would be a low 200 ohms, which when divided by the square of the output winding ratio, becomes a small 2 ohms. Not bad. Wait a minute, this value seems to imply that the triode is beating the MOSFET in every regard, but 10 6BQ5s only have a combined Gm of only 0.11A/V. Two tricks were involved here. The first is the world’s simplest SE amplifier would see an input signal closer to 20V peak, effectively increasing the Gm by twentyfold. Second, it takes twenty times more current to drive a capacitance to 20V rather than 1V, which means that a robust line stage would be needed, just as in the MOSFET example.