Power Amplifiers

Earthquake Sound Cinenova Grande Multi-Channel Power Amplifier with Class A Bias


On the Bench

I regret to inform you that you are about to be presented with a S*** load of graphs, mainly because I had to test the Class A modules and the normal bias modules, so there are twice as many graphs for you to wade through as you might otherwise. Distortion measurements were made within an 80 kHz bandwidth. Since the amplifier is modular, I only drove one channel for each test.

OK, here we go:

Here is 1 kHz, 5 volts output into an 8 ohm load (3.12 watts), in a normal bias module. Distortion was 0.008% THD+N.


Now, the same test using the Class A module. Notice that although the total amount of distortion was the same as with the normal bias module, the odd-order harmonics were higher with normal bias.


Below are the same tests, but with a 4 ohm load (6.25 watts)

With normal bias, the emphasis on odd-order harmonics is even greater than it was at 8 ohms. This is to be expected, as 4 ohms is a more stressful load. You can see also that there is an absence of the high-order harmonics in Class A. This makes for very pleasant listening, because high-order harmonics make the sound harsh.



Are you already sold on getting the amp with two of the modules biased into Class A for your two-channel listening? I thought so.

So, now let's move into 20 volts output, which for an 8 ohm load, is 50 watts, and for 4 ohms, 100 watts, well outside our 10 watt Class A bias in those two modules.

The distortion was about the same for both amplifier modules, both at 8 and 4 ohm loads, because the power output was beyond the 10 watts of Class A, and the modules were both now operating in Class AB.





For IMD, the differences were dramatic. Notice the relative size of the harmonic peaks all the way out to 42 kHz. The actual measurement number was about the same, but it only included peaks within 250 Hz on either side of the 7 kHz fundamental. This is why it is very important to show the actual graphs.



At 20 volts, again, we are beyond the 10 watts of Class A, so the performance between the normal bias module and the Class A module was the same for 8 and 4 ohm loads.





THD+N vs. Frequency, using a normal bias module, showed low distortion all the way out to 50 kHz. I used one of my Carver Mark IV speakers as the load.


Here is an impedance plot of the Carver speaker.


The measured frequency response at 20 volts, and a normal bias module, was down 0.2 dB at 20 kHz into an 8 ohm load, and down 0.4 dB using a 4 ohm load.


Power output is shown below. Using a normal bias amplifier module, at 8 ohms, the soft knee was at 20 watts, the hard knee (the practical output of a power amplifier) at 320 watts, and clipping (1% THD+N) occurred at 360 watts. At 4 ohms, the soft knee was at 30 watts, the hard knee at 500 watts, and clipping was at 605 watts. This is about right on specification.



One more graph to show. This one illustrates what you give up in power output into 8 ohms when you bias a module 10 watts into Class A. You can see that instead of the hard knee being at 320 watts, as it was in normal bias, it was at 280 watts with 10 watts of Class A bias. This is a 40 watt reduction, which is what we would expect (Class A operation is very inefficient). So, if you purchase a Cinénova Grande, and bias two modules 20 watts into Class A for your two-channel music listening, you can expect the hard knee to be 80 watts less than it would be in normal bias, or 240 watts. Is it worth the loss in total power? I think that after you hear what you get with Class A bias, you would agree that the answer is yes.