I doubt it's a yield issue. Large transistors and diodes are not hard to manufacture. Also, multi-chip packages tend to be quite expensive - at least in the IC world - so if it was an advantage to integrate the diodes on the transistor die the manufacturers would probably have done so.
I wonder if it's actually a thermal stability issue. Having the two on the same die would give the fastest thermal response, but that also means a higher risk of instability in the thermal system. Designers would have to pay more attention when using these devices.
In the IC world it's the norm that all devices share the same bulk silicon, which would connect the collector of the transistor to one of the terminals of the diode, which is not what you want. There are ways to prevent that, such as silicon-on-insulator, but that's expensive and also not what you'd want in a power device. One could well-isolate the diode, but then you have another PN junction to design around, so designers need to pay attention again.
Obviously these are just guesses, but I think it's plausible that having two separate devices in the package actually makes sense from a manufacturing standpoint. It's better than having the diode mounted on the heat sink and the devices don't require any special precautions in their use.
Anyway. Back to the original topic: The two devices on the right are marked XNJL... the others NJL... Does that have any significance?
Tom
I wonder if it's actually a thermal stability issue. Having the two on the same die would give the fastest thermal response, but that also means a higher risk of instability in the thermal system. Designers would have to pay more attention when using these devices.
In the IC world it's the norm that all devices share the same bulk silicon, which would connect the collector of the transistor to one of the terminals of the diode, which is not what you want. There are ways to prevent that, such as silicon-on-insulator, but that's expensive and also not what you'd want in a power device. One could well-isolate the diode, but then you have another PN junction to design around, so designers need to pay attention again.
Obviously these are just guesses, but I think it's plausible that having two separate devices in the package actually makes sense from a manufacturing standpoint. It's better than having the diode mounted on the heat sink and the devices don't require any special precautions in their use.
Anyway. Back to the original topic: The two devices on the right are marked XNJL... the others NJL... Does that have any significance?
Tom
TT is good technology. But not everyone knows how to cook it correctly. Some did and got phenomenal results.
Attachments
Last edited:
Re the yield issue, I’m just thinking there about the additional process steps in assembly that would not be required with an integrated diode.
Your point about having to isolate the diode within the die is a good one and I think the main reason for the approach ON took. If the diodes were integrated but not isolated from the power part the thermal comp circuit would be more complex for sure.
Last edited:
TT technology allows you to get these results without warming up the amplifier. These characteristics are not affected by cold or heat. They are constant on loud and quiet music and on a sharply changing signal. And this is in class AB
Last edited:
Bias current in a class AB can vary 2 or 3 to 1 with little change in distortion. Probably a subject for s separate thread.
If nothing else the ThermalTrak devices saves you the trouble of having to mount a separate device to the heat sink for temperature sensing. That's attractive from a manufacturing/assembly standpoint.
Tom
Yeah. Much geekery for not much difference. I seem to recall that both Self and Cordell cover the ThermalTrak devices in their respective power amps books.
Tom
The electrical isolation will be the reason for the non-integration. The collector is the die of all mainstream power transistors