Dan, close but no ceegar! On this one I just needed fair accuracy, at 10 MHz only, and didn't want to be handling little teeny things and/or laying out an ECB. Instead I bought packaged components from Mini Circuits. and cable them up with 0.086" "hand-flex" cable I bought from Amazon. I did all this on Cu-clad board, and soldered down some BNC connectors from the junk box.
The MCL components are: > 1 ea MSC-2-1W+ 0 deg power splitter, used as the LO splitter. > 1 ea PSCQ-2-51W+ 2-way 90 deg splitter, used as the RF splitter > 2 ea SBL-1 Mixer This was all wired up in the most naive possible configuration, with no precautions taken to avoid cross-coupling of harmonic components between mixers, etc. The only thing a bit unusual is that I did not terminate the IF ports in 50 ohms. Instead each is terminated to GND with a 510 ohm resistor, then a series resistor of 10 k, then a shunt capacitor of 0.1 uF, to make an RC lowpass filter of about 160 Hz BW. Then there is a 51 ohm resistor to the center pin of the I or Q BNC connector (as the case may be), which is expected to connect to a Hi-Z load such as an o'scope input. This arrangement yields significantly higher signal voltage output below the compression point. I usually drive the RF input at about -18 dBm ((IRC) and get I & Q outputs of about 60 mV p-p. I should be driving the LO input at about +10 to +11 dBm, but it seems to work OK at levels of +8 to +9 dBm, which is all I can usually manage to give it. If I plot I & Q on an X-Y 'scope display through at least one full cycle of phase difference, the expected circle is visibly slightly distorted into an ellipse slanted at approx 45 deg, but it's not real bad. I would not use this as-is for precision phase measurement, but for my present purposes of tuning an Rb onto GPS it's more than good enough. For my precision work at "all" frequencies, I use a Signal Hound SA44B (USB Spectrum Analyzer). It does a nice job in terms of IQ accuracy, but finding out *exactly* where it's tuned can be a bitch. Hope this helps... Dana On Mon, Aug 23, 2021 at 10:33 AM Dan Kemppainen <[email protected]> wrote: > Dana, > > Just curious, can you share some details on what you are using for the > IQ demodulator? Is this a 'roll your own' solution Gilbert cell mixer, etc? > > Thanks, > Dan > > > On 8/23/2021 3:30 AM, [email protected] wrote: > > Message: 4 > > Date: Sun, 22 Aug 2021 10:16:56 -0500 > > From: Dana Whitlow<[email protected]> > > Subject: [time-nuts] Re: Efratom LPRO-101 Rubidium question.... > > To: Discussion of precise time and frequency measurement > > <[email protected]> > > Message-ID: > > < > cadhrwpeyiqa1pwrvtb_htmzc8oxe-wr+fzyz4oz4vucnc4h...@mail.gmail.com> > > Content-Type: text/plain; charset="UTF-8" > > > > C'mon guys- when you speak of a frequency error in Hz, you should also > > specify "at > > what frequency"*in the same breath*. Or better yet, always specify > > frequency errors in > > relative terms. > > > > Hundreds of Hz at 10 Mhz is unthinkable for a Rb. Even at 8.2 GHz, 100 > Hz > > error is about > > 12E-9, which is likely to be outside the EFC tuning range of some (if not > > most) Rb standards. > > Both of my Rbs have a tuning range of only roughly 2E-9 via the EFC > input. > > Outside that > > range, much sterner measures must be taken, which I frankly dread. > > > > The two Rbs that I own (an L-Pro and a PRS-10) both tend to drift upwards > > in frequency > > to the tune of about 1E-11 or 2E-11 per month. Superimposed on that are > > random > > variations of around 1E-11 on a time scale of a few hours. > > > > I make phase comparisons between the 10 MHz outputs of a GPSDO and the Rb > > under > > test using a simple quadrature demodulator, with the I & Q outputs > > displayed on a 2-chan > > DSO. I use the "roll mode" display feature on the DSO at its slowest > speed > > (1000 sec/div) > > and just leave things running continuously for up to several days. On my > > DSO a full screen > > width is 14000 sec (slightly under 4 hours), and I just take a glance > from > > time to time as I > > happen to pass by. > > > > Someone suggested a 100 sec measurement with a counter, but that is right > > in the realm > > where GPSDOs are typically the most noisy, so a single measurement is > > likely to have > > rather large errors. One would have to record a fairly large number of > > such measurements > > (several hundreds of them) and plot them out to get a good assessment of > > what the Rb is > > actually doing. WIth the IQ phase difference display, one can get a > pretty > > decent estimate of > > the needed tuning correction, without doing any real work at all, in a > day > > or so. > > > > When I'm doing something requiring the best frequency accuracy, I keep > the > > 'scope display > > running while I'm doing the serious work, and note the frequency error of > > the Rb at the time > > for use in correcting the final result. BTW, I don't see much "settling" > > effect after making > > tuning changes- the correction made seems to take effect essentially > > immediately (as best > > as one can tell in the presence of GPS noise). By comparing two Rbs, I > can > > investigate > > settling effects quite well without the noise having anything to do with > it. > > > > Dana > > > > > > On Sun, Aug 22, 2021 at 8:41 AM Matthias Welwarsky< > [email protected]> > > wrote: > _______________________________________________ > time-nuts mailing list -- [email protected] -- To unsubscribe send > an email to [email protected] > To unsubscribe, go to and follow the instructions there. > _______________________________________________ time-nuts mailing list -- [email protected] -- To unsubscribe send an email to [email protected] To unsubscribe, go to and follow the instructions there.
