Darlington noise linearity questions..

[Svart]

I'm starting to investigate a discrepancy in some designs here at work. The part's datasheet shows a NPN darlington used in a variable voltage circuit for varactor tuning. Our design uses a standard NPN in it's place.

For some more background, this is part of a varactor tuning circuit, 0-27v varies the frequency of the oscillator tank circuit via the varactor.

I've not done much work with darlingtons and wonder if any of you great analog designers can give me some real world insight in any quirks they might have and any typical noise profiles over voltage/current for both input and outputs. I get a slightly noisy output the closer to 0v I get using this circuit.

 

[bcarso]

Which circuit is giving you the noise---the single NPN or the darlington?

One thing to remember about darlingtons is that unless there are some internal resistors (found on power darlingtons sometimes) the whole of the first device's emitter current is the base current of the second one. So this can be a very small current, and the noise associated with the equivalent emitter resistance of the first device can get fairly large.

I'd have to see more of the circuit to comment further.

 

[CJ]

Another thing to keep in mind is the V sat.

The collector voltage will be quite a bit higher when switched on compared to the regular NPN.

 

[Svart]

It turns out that the circuit in question was a darlington after all. Whomever the designer was didn't even bother to use the correct symbol on the schematic or even call out a part number for that matter. I ended up doing an analysis and everything was in line with a darlington, the base current, the Vsat, everything. Once I decoded the little cryptic marking on the part itself, I found out that this was all futile. Anyway, after all of the red herrings it turned out that the loop filter around the oscillator tank circuit was incorrectly designed. I simulated and replaced a couple of parts and now the phase noise is MUCH better.

The crazy thing is that they had a PhD check this thing out before and he correctly pointed out that there was a phase noise issue with this circuit, the first upconverter IC(essentially LO1 in a dual conversion RF system) but blamed the chip, not the design.

Two of us lowly staffers just saved the company 2.5 million dollars.

And all we got was a (rather good) lunch..

:shock: