[time-nuts] Chinese NTP Time server

[Brian Lloyd]

On 11/26/19 09:28, xaos at darksmile.net wrote:
> Hello everyone,
>
> Does anyone here know about this item:
>
> https://www.ebay.com/itm/Brand-New-NTP-Time-Server-GPSDO-GPS-Disciplined-Oscillator-GPS-Clock/362758051388?hash=item547610963c:g:yFIAAOSwgztdgfM9
>
>
> George, N2FGX

OK, I ordered one. I missed the part about the 10MHz output being
"optional". (Optional? Really?) So I ordered one and it arrived 2 days
ago. Delivery took about 3 weeks. Looks nice, well packed, no damage. It
includes an active GPS antenna with about 20' of RG174 and a 12VDC
wall-wart. Unit, antenna, and PSU, nothing else. No doc and no software.
No problem ... so far.

Plugged it in. Power supply LED comes on and the Sync LED is flashing at
about 2Hz. About 20 seconds later the SV LED comes on. about 2 minutes
later the Sync switches to 1Hz. I am guessing it has achieved some sort
of lock. I connect the 10MHz output to my FA-2. Of course, no 10MHz
output. (More on this later.) I plugged it into the network. Looked at
my DHCP server. No IP address assigned. Huh. How do I find this thing on
my network?

eBay message back to the seller. Seller sends me a link to a zip file
with the software and doc. The first 'uh oh' is that all the file names
are in mandarin. I must admit, I find <mandarin>.pdf and <mandarin>.exe
amusing. I have an old laptop that dual-boots Linux and Windows just for
this sort of thing, i.e. annoying software that only runs on Windows. I
extract everything from the zip file and try things out. One of the
programs shows just an ip address of 192.168.0.100 and has three windows
separated into dotted-quads. Could this be the tool that sets the IP
address, subnet mask, and gateway? I run the PDF that opens with a
picture of the unit through google translate. OK, yes, that is what the
program does. There is a picture of the window and, guessing at the
examples, I was right, IP addr, mask, and gateway. With 5 buttons to
click on in varying orders, it tooks me several tries to finally get it
to change its IP address to one on my network. It is now pingable.

I look at some of the other programs. These are mostly in English, being
open software, and do things like let you look at the status of the GPS
receiver. After successfully setting the IP address of the unit, I was
able to run the utility 'PowerGPS.exe' and have it report GPS status
from the box.

So I pointed ntp on my linux server at the box as a server. It synced
right up and chose it as the primary ntp source. Clearly it works just
peachy as an NTP server and is running in my network that way. Now to
tackle the lack of 10MHz output.

Opening up the box it clearly has a 10MHz OCXO so getting something out
the 10MHz BNC connector should be fairly straight-forward. Examining the
bottom of the board, the path from the OCXO to the 10MHz BNC output is
pretty clear. It goes through a single buffer op-amp and then drives the
BNC jack. The only problem is, the op-amp is missing. The pads are for
an SO-8 package. Following the traces it is pretty clear it is a
standard, single-op-amp pinout, i.e. :

 1. n/c (null)
 2. inverting input
 3. non-inverting input
 4. V-
 5. n/c (null)
 6. output
 7. V+
 8. n/c (null)

Lots of parts could go there. Looks to me like it should probably be
something like an LT1227. The op-amp is operated at a gain of 2 with a
100ohm FB resistor and 100ohm to ground. This should be pretty stable
with 100ohm pretty much swamping any parasitic capacitance in the FB
loop. The input is a 300 ohm/300ohm voltage divider so the overall gain
of the stage is unity.

Input to the buffer from the OCXO is capacitively-coupled as is the
output. There is a 50ohm resistor in series with the output. The
unloaded voltage output of the OCXO is a 3.5V positive square wave.
After passing through the buffer it is probably supposed to approximate
a 3V p-p sine wave. (At least, that is what the marketing verbiage
suggests.)

The board is already populated with linear +5v and -5v regulators (74L08
and 74L09 respectively) and associated bypass/filtering. +5v is present
at pin7. No -5v is present. Tracing the board back to the source of the
-V at the input to the -5v regulator, one arrives at an empty 4-pin
through-hole area marked u14. This has +7.5V on pin 2 which also feeds
the input to the +5V regulator. Pin 3 is the -V output that feeds the
input to the negative analog regulator. So the 64-dollar question is:
does anyone know of a chip/module that is a 4-pin, 0.1" pin spacing,
single-output, inverting regulator module, producing -7V from a +7V
input? I need to find something like that to populate this board and get
the 10MHz output. I haven't been able to find anything on Mouser or
Digi-Key. I am guessing that this is something that already exists as a
module but ...

Yes, I have asked the seller about the parts to populate the board or
the option of returning it and getting one already populated. If the
seller tells me to go pound sand I would like to manage to make this
work anyway.

Hmm ... if I DC couple the input and run the op-amp with a single-ended
+5v supply, it might work without the negative supply at all. My only
worry would be latch-up because the input would be going right to the
negative rail and approaching within 1.5V of the positive rail. OTOH,
most op-amps now are pretty immune to latch-up. Just thinking aloud
here. I am sure someone here has some ideas about getting this to work.

Thanks in advance for humoring me.


-- 


 <https://www.lloyd.aero>

Brian Lloyd
706 Flightline
Spring Branch, TX 78070
brian at lloyd.aero <mailto://brian@lloyd.aero>
+1.210.802.8359