Ways to get a accurate voltage rail?

[TheUnnamedNewbie]

Hi,

For the same project as my other topic.

I want a good accuracy in terms of gain on my board (LNA board with three RF LNA amplifiers, BGA612 from infineon). I can calibrate out the nonlinearity due to the gain changing at high input powers by doing some compensation in my digital processing.
But since the gain of these amplifiers also depends on the Vdd, how do I get a very stable Vdd, in function of temperature? Just get a grossly overspecced part and sink it's heat as much as possible to get as little as possible change in temperature?

[danadak]

What are the goals ?

1) Absolute Vdd accuracy or relative ?
2) Accuracy spec over temp or permissible deltaV/deltaT
3) Permissible noise spec ?
4) Any startup time issue for Vdd ?
5) How much C load on Vdd ?
6) Unregulated max noise/V

Regards, Dana.

[whollender]

It's much easier to get a stable Vdd over temperature than it is to get a stable gain profile over temperature.

If you want a stable gain over temperature, you'll need to compensate for the temperature variation of the amplifier itself, which in this case is really just a pair of transistors.  This would require a temp sensor driving a circuit that changes Vdd or the amplifier current, although you may be able to play some games using the resistor tempco or something like that.

Easiest is to simply characterize the gain variation vs temperature and compensate for it digitally.

You'll still probably want a good, low-noise LDO driving the amps, however.

[TheUnnamedNewbie]

Thanks for the replies all.

Danadak: I don't have such exact specs, and I feel like really doing a thorrough analysis would take too much time in view of the scope of m project. I was just hoping to get some suggstions in terms of circuit techniques or maybe part families to keep in mind to get good performance, without full characterizing this (this is just one part of a large system we have to design, and we have only about 40 hours (on paper) left to work on it).

evb149: I made the assumption that since my signals are very low lever, and te idle consumption of the amplifier will be quite high, it will not change temperature significantly. On op of that, also see the above in terms of how I can't really do a thorough analysis.
Temperature measurement would make it too complex I think, and we are already hitting the limit of the what the arduino we are using is capable of (we are sampleing at 1MS/second with a 10bit ADC, and just getting that much data loaded into the arduino is a pain)...

[awallin]

fwiw heres a low-noise LDO list:
https://hifiduino.wordpress.com/2011/03/29/new-breed-of-ultra-low-noise-regulators/

LT1963 (positive) and LT3015 (negative) seems to work fine in my hands for RF (below 1 GHz) - but YMMV...