Back in 1968 Gerald Stanley of Crown patented class AB+B (US3493879A High power high fidelity solid state amplifier) this topology made Crown a huge success used in the DC series, grounded bridge, D series and still in use today...
I see the pre-driver and drivers (high ft devices) run in class AB and the main output devices (slower ft, more rugged) run in a class B when wattage goes over a certain level. Also the output devices are closely matched on VBE and hfe. I see it's a three deep darlington circuit.
The question is does the drivers and pre drivers continue to conduct when the last output devices switch on to fill in the near zero crossing to minimize crossover distortion? I've heard the outputs act like boosters. Surley running in class B would increase harmonic distortion? but I also notice Gerald Stanley was very clever with negative feedback, global and nested also using advance phase techniques to get the THD figures very, very low (as in the D150 and DC300). (I note that because of the high negative feedback the mid-range frequencies sounds very hard, brittle).
Thanks
I see the pre-driver and drivers (high ft devices) run in class AB and the main output devices (slower ft, more rugged) run in a class B when wattage goes over a certain level. Also the output devices are closely matched on VBE and hfe. I see it's a three deep darlington circuit.
The question is does the drivers and pre drivers continue to conduct when the last output devices switch on to fill in the near zero crossing to minimize crossover distortion? I've heard the outputs act like boosters. Surley running in class B would increase harmonic distortion? but I also notice Gerald Stanley was very clever with negative feedback, global and nested also using advance phase techniques to get the THD figures very, very low (as in the D150 and DC300). (I note that because of the high negative feedback the mid-range frequencies sounds very hard, brittle).
Thanks
The class B stage generates quite a bit of crossover distortion. The high NFB lowers the % to a “respectable” figure, but tends to spread that distortion spectrum out in frequency. That’s the reason for the “brittle” sound.
If it weren’t for the output triple with the drivers run hot, the crossover distortion would be intolerable. Even with that much NFB. The driver/predriver can almost do all the work on their own at “light” load.
If it weren’t for the output triple with the drivers run hot, the crossover distortion would be intolerable. Even with that much NFB. The driver/predriver can almost do all the work on their own at “light” load.
Many thanks for your reply. I also note that as the outputs are slow ft devices 2-4mhz, but probably much slower back in the day (RCA,Westinghouse) as in the DC300-D150 the high frequency response at higher levels was not good and with limited negative feedback at those frequencies but these amplifiers measured well.
I measured the THD of my Micro-Tech 1200 the other day and the distortion was very low (0.0021) @ 1Khz.
Is the statement valid, If an amplifier measures very well it doesn't mean it will sound good?
I measured the THD of my Micro-Tech 1200 the other day and the distortion was very low (0.0021) @ 1Khz.
Is the statement valid, If an amplifier measures very well it doesn't mean it will sound good?
Macrotechs use normal class AB output stages. With MJ15024‘s you don’t have to resort to dirty tricks, just bias them normally. They are “underbiased” with respect to the “optimum” 26 mV across the emitter resistor bias, but well within diminishing returns. IMD at 100 mW isn‘t going to be as good as a Honeybadger, but way better than the old days.
The proper measurements on the old DC300 would reveal its limitations. THD at full power would not be as relevant as IMD at 100 mW. The “2 to 4 MHz” devices used at the end of the DC300A’s production wouldn’t have given anybody any trouble if biased into class AB. But the old VERY slow hometaxials did. Barely cracked open, they were often totally unstable. No bias, fine. Hotter then hell, fine. But a few mA? Nah, we’re gonna sing. Then the resulting cross conduction blows them to kingdom come. Solution? No bias. Then you need to run the drivers hot and provide a lot of current gain (triple) just to get it to sound reasonable.
The proper measurements on the old DC300 would reveal its limitations. THD at full power would not be as relevant as IMD at 100 mW. The “2 to 4 MHz” devices used at the end of the DC300A’s production wouldn’t have given anybody any trouble if biased into class AB. But the old VERY slow hometaxials did. Barely cracked open, they were often totally unstable. No bias, fine. Hotter then hell, fine. But a few mA? Nah, we’re gonna sing. Then the resulting cross conduction blows them to kingdom come. Solution? No bias. Then you need to run the drivers hot and provide a lot of current gain (triple) just to get it to sound reasonable.
It's important to understand that the drivers respond to the onset of the class-B outputs because of local feedback in the output stage. And the class-B outputs boost the driver current at the point where the driver gain is falling off. It does not add the distorted class-B signal to an undistorted class-A signal. Perhaps the driver gain has fallen from 99% to 90% when the outputs kick in and prevent it falling much further. I used the idea to protect a marginal output in a couple amps. IE instead of current limiting, large power transistors kick in so that the limiting falls on the power supply/ fuse, but unlike current limiting, there is no clipping. If you could scope the driver current, it would have a flattened but still rounded top where current beyond the class-B threshold is being amplified by the outputs. This transition adds some distortion, but it is past the "first watt" where it would be a significant fraction of the signal.
