Originally Posted by duncan1
Tomb . I take the comment you made about the bias being able to take up to 125 MA as referring to the actual output of the circuit I was referring to the actual total current that can be used through a 2N5486 and that is certainly much less than 125 MA in low figures. The original fet has a higher current load ability and voltage. So you are still putting in an active component that has not been designed for the original circuit--unless you have compensated for it by changing the circuit or you know the current load through the fet is less than the spec. of it.You can get away with putting in a lower voltage active device in a circuit but whenever a fault occurs the safety leeway has been compromised and that component is more likely to blow.
Originally Posted by duncan1
I took your advice and had a look at the circuit. I can see why he wants to use -2N5486 as its used at the output of the AD8610 along with 2N5484. But I take it he is using it as a replacement to Q 4- This is a "different kettle of fish". The voltage difference at the output of the chip is low. But if it is Q 4 then that FET straddles across the combined voltage of the power supply and is used to balance the amp. This is a critical position and personally I would fit the original or failing that there are plenty of equivalents available .It only takes a small fault in either active device at either end to blow the fet and if blown the wrong way would apply the maximum current available through the output. Perceived wisdom due to not just me but may projects on Electronics World Mag . Suggest by experience that the active devices on either side of it should be able to take the combined voltage load[+/-] > I have built plenty of loudspeaker power amps with approx the same biasing [theres a limit to the different types employed] and I always made sure the "amplified diode" type of circuit had an active device that had a working voltage of at least half of the supply rail that VHF FET does not.
Q4 will not "blow." It's in an isolated (so to speak) current loop that sets the bias in the diamond buffer. If its IDSS is too low for the bias desired in the buffer, you will simply not be able to turn the bias current up anymore.
Use of the 2N5486 in a BJT diamond buffer is well documented and has been used for years:
This is the diamond buffer designed for the revMH Millett Hybrid by Steinchen. Literally hundreds were built. Bias can be set up to 60ma in this configuration with the 2N5486 (20-30ma was typical and I believe the setting recommended for the PPAV2).
You can explore all of this in gory detail at:
There's also a helluva lot of detail and real-world applications at the Millett Hybrid support site. Further, the Millett MAX and MiniMAX were originally designed for the 2N5486. We went further than the typical 30ma bias with single output transistors (the above schematic has doubles in parallel) because we were using heat sinks (Steinchen's buffer went without them, as does the PPAV2). It was only then that we went to the PN4392, but that was after we changed the current mirror to 2X - R4 and R5 is only 100 ohm in the MAX/MiniMAX. That gives us the ability to turn the bias up higher than even the PN4392's IDSS limits.
Edited by tomb - 10/3/13 at 8:52am