[time-nuts] OT: Power level reference

[Lux, Jim (337C)]

On 12/2/09 12:01 AM, "Poul-Henning Kamp" <phk at phk.freebsd.dk> wrote:

> In message <c793a5fe0912012031k12893065w2682c6e74c920bf8 at mail.gmail.com>,
> Josep
> h Gray writes:
> 
>> I also found lots of Google hits for the AD8307. Within its
>> limitations, this chip seems to be very popular.
> 
> Please notice that most of these chips are intended for AGC purposes
> for transmitters, they very seldom have a frequency-flatness spec
> worth anything

The 8307 and it's ilk are actually quite flat over frequency.  Real
broadband amplifiers in the log chain.  Is it 0.1dB flat? I don't think so,
and in an case, layout might give you more ripple than that.  But certainly
better than 0.5dB practically.

The 8307 datasheet says 0.3dB typical for the middle 80dB range for
frequencies less than 100MHz, and I'd believe that, based on the ones I've
seen hooked up.  That's about 1/10th of the -3dB frequency for the logamps
in the chain, so they're quite flat.


Specs on a logamp are tricky to evaluate against a square law or linear
detector because it's tough to know if it's the log function or the basic
detection that's at issue.

For the thermistor mount, you're lucky to get 30dB dynamic range AND the
accuracy will be worse at the bottom end, because a good portion of the
measurement uncertainty is fixed, not a fraction of the input.

For a diode detector, the newer meters know the cal curve of the diode, so
they're not depending on the squarelaw characteristic, and you get maybe
50dB dynamic range, but again, the accuracy at the low end is worse than the
high end.

I wonder (just haven't looked til now) what they're using inside those USB
power heads (e.g. MiniCircuits)... Is it a AD8307 type widget or a diode?