[time-nuts] GPS Antenna Feed Line Decision

[jimlux]

On 9/3/17 3:41 PM, Charles Steinmetz wrote:
> Attila wrote:
>
>> The supplies for LNAs are usually quite benign given two constraints:
>>    *   *   *
>> 1) Low frequency (0Hz to bandwidth of signal) noise is low
>>    *   *   *
>> But 1) is a bit harder as it also includes 1/f noise, temperature,
>> (upstream) supply and load effects.
>
> The LT3042 is better in this regard than anything short of a heroic
> effort with discrete components, IF you use a larger-than-normal bypass
> capacitor on the SET pin.  With Cset = 22uF, the noise density is
> <20nV/sqrtHz down to 10Hz.  220uF is somewhat better, but not by a whole
> decade because of the 1/f noise of the error/output amplifier.
>
> The down side of a large Cset is that the power supply takes longer to
> reach its final voltage.  A 5v supply reaches 90% voltage in about 25
> seconds, and 99% in about 40 seconds.  A 10v supply takes twice as long,
> and a 15v supply, three times.  The LT3042 does have a "fast start"
> mode, which can reduce these times considerably (to ~0.5sec, ~1sec, and
> ~1.5sec, respectively) -- but still longer than usual for a regulated
> supply.  For applications like continuously-on LNA power, there should
> be no problem with any of these times.
>
> The LT3042 is a bit of a pain to use, with its buried ground/heatsink
> tab, but it is very good, and is the lowest noise fully integrated
> regulator available, AFAIK.

I love that part - the PSRR is awesome up to MHz.  So many of those 
linear regulators are great at kHz, but not so hot up higher.
You want quiet, the 3042 is your friend.