Does anyone of you already designed a circuit using the new ThermalTrak ONSemi devices ?
And in particular, how the quiescent current has been set ?
The ON application note seems quite approximate and should produce very high bias when used with more output devices and a simple driver+output stage instead of predriver+driver+output.
Thank You Very Much
And in particular, how the quiescent current has been set ?
The ON application note seems quite approximate and should produce very high bias when used with more output devices and a simple driver+output stage instead of predriver+driver+output.
Thank You Very Much
Those parts are plain transistors but with a diode incorporated. What you have to do is to design some temperature control like a Vbe multiplier or like the application note below.
http://www.onsemi.com/pub/Collateral/AND8196-D.PDF
http://www.onsemi.com/pub/Collateral/AND8196-D.PDF
No, there are two problems with your posted circuit:
1) Assuming that the driver transistors are also on the heatsink with the outputs, you will need to use 4 diodes to compensate for 4 base-emitter junctions in series. If you don't put the drivers on the heatsinks you will need to add 2 diodes that approximate the temperature of the drivers. These other diodes cannot be the ones in the output devices as they will be at the wrong temperature.
2) There is no adjustment for your circuit. I think you will need to replace R20 with a trimpot, at least for the prototype. If you find a resistance value that gives the proper bias, then you can possibly replace it with a fixed resistor.
However, I do not recommend an emitter-follower double. The emitter-follower triple (as shown in the OnSemi app note) provides much less loading on the voltage gain stage, and therefore much less distortion.
1) Assuming that the driver transistors are also on the heatsink with the outputs, you will need to use 4 diodes to compensate for 4 base-emitter junctions in series. If you don't put the drivers on the heatsinks you will need to add 2 diodes that approximate the temperature of the drivers. These other diodes cannot be the ones in the output devices as they will be at the wrong temperature.
2) There is no adjustment for your circuit. I think you will need to replace R20 with a trimpot, at least for the prototype. If you find a resistance value that gives the proper bias, then you can possibly replace it with a fixed resistor.
However, I do not recommend an emitter-follower double. The emitter-follower triple (as shown in the OnSemi app note) provides much less loading on the voltage gain stage, and therefore much less distortion.
Considering that the dissipation in driver transistors is quite constant, especially when mounted on a separate heatsink, You can use 4 out of the six diodes of the output stage to track the rise ...
I ran again the simulation using 4 diodes and from 25° to 85°C I can get a stable enough Iq over the output devices.
I'll let You know
I ran again the simulation using 4 diodes and from 25° to 85°C I can get a stable enough Iq over the output devices.
I'll let You know
- Status
- Not open for further replies.