[time-nuts] 35601A as stand-in for 11848A ?
jmiles at pop.net
Sat Jan 5 22:59:14 EST 2008
> John Miles wrote:
> >I have a 35601A, although I've never done anything with it.
> Did you happen to spot any service/calibration manuals
> for the 35601A? I haven't found anything online yet, but one
> of the manual dealers I checked lists a hard copy.
I do have a hardcopy manual for the 35601A but am not sure if I have a .PDF
of it or not. Will check tomorrow and let you know. I can probably scan it
easily enough if not, >11x17 foldouts excepted.
> I've never used the older 3047a, but the impression I have from the
> catalog info I've found so far is that the 3048a updated to the newer
> 3561a for narrowband analysis and added more internal sources
> to the 11848a for convenience; I'm hoping that much of the basic
> signal path for calibration and quadrature lock was left alone.
> >The 11729B/C plus an 8662A (or even an 8640B) has everything most
> > people will ever need for PN measurement.
> I need to measure things that are lower close in than the 11729/8662a
> combo noise floor; i.e. OCXO, direct synthesis from OCXO, amp/divider
> residual noise floors, down to 1-10 Hz offsets.
Keep in mind, though, that without the 11848A's internal sources, you're
going to have to supply some kind of reference source at the same frequency
as the DUT. At that point, all the 35601A, 11848A, *or* 11729B/C is doing
for you is providing a mixer, an LNA, and a quadrature-locking PLL in one
box. They can all be used to compare two external sources directly if you
want, with no 8662A required.
Of those models, only the 11729B/C is really meant for standalone use
(meaning, without a GPIB controller.) The 35601A doesn't have buttons on
the front panel to let you change PLL bandwidths, override the default loop
capture range, or anything else, and I don't believe the 11848A does,
either. You either need a 3047A/3048A system, or some custom software that
you'll have to write yourself.
I can't speak from experience, but I don't know of any magical reason why
the 35601A's or 11848A's noise floor will be much lower than the 11729's.
See http://www.ke5fx.com/regen.htm for a case where the 11729C goes up
against the TSC-5120A in a low-noise measurement scenario and gives a
reasonable accounting of itself. If the 11848A has an advantage, it
probably comes at offsets < 1 kHz, where the 11729C doesn't do so well.
Note that HP is careless with the 11729's noise-floor specs in some places,
in that they report what is really the system noise floor, limited by the
8662A they assume (hope) you are using.
One caveat: to use the 11729 to measure carriers below about 20 MHz, you
*will* have to add an LPF prior to its LNA. Otherwise the upper sideband
from the phase-detector mixer will enter the LNA and saturate it. Not the
end of the world, but the additional LPF wouldn't be necessary with the
11848A, and I assume not with the 35601A either, because they have more
internal filter options. (Which, again, have to be selected via GPIB.)
Analyzer-wise, the fact that you need to look at offsets down to 1 Hz is a
complication. You cannot use a conventional spectrum analyzer for that
unless you have a new/fancy model with sub-1 Hz resolution. An FFT analyzer
would be needed, and like Bruce suggests, a sound card may be the best way
to go there. (I don't currently support sound-card input in my
noise-measurement app, but I probably ought to. If someone needed that
capability and wanted to help test it, it wouldn't be out of the question.)
> >Renting/buying a E5052A or TSC5120A does get to be expensive
> How many significant digits are there in a 5120A price quote?
$2.0*10^4, last I heard. :)
-- john, KE5FX
More information about the time-nuts