[time-nuts] time-nuts Digest, Vol 49, Issue 12
ed_palmer at sasktel.net
Fri Aug 8 12:08:40 EDT 2008
> Date: Thu, 7 Aug 2008 21:50:02 +0100
> From: "David " <t_list_1_only at braw.co.uk>
> Subject: Re: [time-nuts] NavSync CW12 and CW25
> To: <time-nuts at febo.com>
> Message-ID: <000501c8f8cf$2b114550$0200a8c0 at hp6110>
> Content-Type: text/plain; charset="us-ascii"
> I've been playing with a CW12, it basically appears to do what they say but
> I will not confirm any numbers for the moment, it certainly compares to my
> VLF derived 10MHz signals. I was looking to build a GPSDO and 10MHz from the
> GPS Rx seemed a better starting point than 1PPS.
My thoughts exactly. I've been playing with a Jupiter GPS board which
has a 10 KHz output, but it died on me so I'm that's why I'm looking for
> I'm only now near to completing part two of the exercise, phase locking a
> 10MHz double oven XO to the CW12 10MHz, the phase noise on the CW12 output
> is rather significant and needs cleaning for most applications. The ssb
> numbers I see from the CW12 (using an HP8560E) relative to the 10MHz carrier
> are :
> 1kHz -80 dBc/Hz
> 10kHz -102 dBc/Hz
> 100kHz -109 dBc/Hz
> 1MHz -126 dBc/Hz
> The signal is far from a pure line, close in there is a set of sidebands
> based on harmonics of 132 with the fundamental at -22dBc, there are various
> synthesis artefacts. This unit clearly needs filtering.
I thought it was good of Navsync to admit in the app note that their
signal isn't particularly clean. I'm a rank amateur in this area, but
if you changed the output frequency of the CW25 is it possible that you
might find a 'sweet spot' that would give you a cleaner signal - either
by matching or avoiding various harmonics of the internal clock(s)? I
noticed in the datasheet that they spec the processor as 'up to 120
MHz'. If they're saying that the clock speed varies while operating
then forget my question.
> The PLL I've built to clean it up is essentially a simple 2nd/3rd order loop
> with the phase comparison at 78kHz (sigs divided by 128), my model claims
> the 3dB closed bandwidth will be 0.3 Hz and sideband suppression about 127dB
> at 1kHz offset in principle. In reality I'll need to play with it to look at
> the real world limitations of acquisition, loop BW, analogue offsets etc .
Did you divide down the output from the CW12? I see that you can set
this frequency, but I wondered if this setting was non-volatile or if
the frequency would return to 10 MHz after a loss of power. I'm
assuming that the on-board battery would keep it as long as possible.
> The OCXO I'm locking is a Temex part, a 'Doc1478-D" which seems to be from
> their S110 series. It claims to manage -145dBc at 1kHz offset. There were on
> ebay at low cost a while ago, everything else has largely been built from
> scrap which has biased the design ever so slightly, the less it costs the
> more amusing the project if you are not doing it for a customer.
Yeah, it's no fun to just go buy a bunch of parts. Some of my stuff
looks like it came out of Dr. Frankenstein's lab. Of course, sometimes
you have to dig out your wallet. I picked up a couple of those 'naked'
HP Z3801a oscillators (basically HP 10811 oscillators with an second
oven) that showed up on Ebay a month or so ago. I'm planning on using
that for my OCXO. The 'Frankenstein' aspect will be fulfilled because
I'll have to fabricate an outer casing for the oscillator and build a
controller for the outer oven.
> The slight disappointment has been the lack of support information about the
> CW12 on the web, when I spoke to them in March they said their site was
> going to be updated by the end of that month, clearly they have dropped the
> ball somewhere but the email support seems responsive.
I asked for the user manual in June and they responded quickly, but
didn't mention anything about a website update.
> I think I've seen the
> CW25 appear in a recent GPSDO from Quartzlock, have a look at :
Yeah, that little module looks familiar doesn't it? The picture's not
very clear, but you can just make out the Hirose connector in the corner
of the module near the TNC connector.
> This appears to use a DDS/PLL to clean the reference clock rather than a
> large and expensive OCXO.
> Starting afresh I may have gone for the CW25 but I've got the CW12 so not
> thought much more about it.
> David Mackenzie (GM4HJQ)
Thanks very much for the info, David.
More information about the time-nuts