Adaptive class A

[jane]

In short time I am going to start building a PX25 class A push pull amplifier. I will like to incorporate a (solid state) circuit that controls the bias. A sort of class A on demand or adaptive bias circuit. The idea is to run the amplifier at class AB at idle and class A when there are input signal. This will reduce heat and power consumption.
As a start I will do it very simple with two fixed bias levels, for the PX25 let's say 4x20mA at idle and 4x60mA for class A under normal conditions. I could have made a circuit that is sensing the input signal all the time and controlling the bias as a function of signal level, but that will alter the working point of the output tubes continuously, and I'm not sure what kind of sonic impact that will bring.



Looking at the graph you will see some time constants. There is attack time: from a signal is detected to the bias is full class A (10 sec). A silence delay time constant: after 60 sec without signal the timer goes from logical 1 to 0. Decay time: time constant to decrease the bias from class A to class AB hibernating (20 sek). I probably have to experiment with these time constants.

The circuit will be made of an op-amp, a 555 timer and a power transistor. The power transistor will control the C- voltage that supplies the bias networks. That means that I still can adjust the bias for each tube in the old fashion way. The controller will just increase the C- supply from one pre defined level to another.

The power supply is quite stiff so a change from class A to AB (from 4x60mA to 4x20mA) will not raise the B+ too much.

Any comments, warnings or bright ideas?

Jan E Veiset

[rdf]

Originally posted by jane
Any comments, warnings or bright ideas?

How about a stolen idea? The jukebox amp I'm currently modifying employs what Rowe's schematics label a 'mute circuit'. The rectifier output feeds a 20 kohm, 10 watt wirewound with shorting relay contacts. When open the B+ falls far enough to bring the output tubes near shut-off while still keeping the front end lit, though at reduced standing current.

If this jukebox is any indication they generally ran near Class B to maximize tube life, and minimize service calls, so it won't be as effective with pure A. Perhaps a combination of this along with your adaptive bias will provide double bang for the buck at the cost of an extra pair of contacts and a wirewound.

Edit: Forgot, this method also allowed Rowe to get long life from a 5U4GB with 80 uF directly on its output. Apparently it doubles as a soft-start.

[ErikdeBest]

I think a MOSFET standby swith could also work to accomplish this. If it is better? Do not know, but, you have another example. Look at

http://www.geofex.com/Article_Folder...osfetfolly.htm

Scroll down a bit for the MOSFET standby switch.

[jane]

Thanks for sharing information.
I could use a longer time constant (600+ seconds) to semi shutdown the amplifier. The front end has its own power supply but I suppose I could reduce both B+ and B2+ for long term silence periods.

Jan E Veiset

[Geek]

Hi,

So, we're talking a "dynamic class-A" for tubes, like SS uses.

John Broskie at tubecad.com has an autobias circuit that could quite probably be adapted to this purpose.

http://www.tubecad.com/2005/May/blog0045.htm
http://www.tubecad.com/2005/May/blog0046.htm

I played with it a bit too here:
http://geek.scorpiorising.ca/GeeK_Zo...p?topic=1728.0

Hope this helps.

Cheers!