[time-nuts] PCB layout question for GPSDO

[Tobias Pluess]

Hoi Attila,

sorry for my late reply as well. And, to make the confusion complete, I
have a different Email address because my old Email provider closed.....

Anyways, thanks for your comments.


> You can use the AC for both 5V and non-5V. The T is for the input voltage
> level compatibility. As you are not using any TTL components, this is of
> no concern here.

No, unfortunately not, the AC has a slightly higher requirement for the
minimum input voltage required for a logic 1, and, taking all tolerances
into account, it won't reliably work with a 3.3V input level. But anyways,
this is not a big deal here.


> Yeah.. That requirement was on the list since before the LEA line
> of ublox modules. I wonder why it dropped out. I would guess the
> modules are less sensitive these days (they integrate more and more
> of the RF electronics into the ASIC which each version), but I would
> not rely on it.

Got it! there is no more traces under the uBlox module.
Besides that, I let a friend who is an EMC/EMI guru have a look on my
schematics as well and he made some similar comments as you, so I added
some more power supply filters. Unfortunately, I have sent my gerber files
to the PCB manufacturer before I got your comments, so therefore some
little things are missing on my board:

* connect the USB pins....
-> this would indeed have been a good idea. However I have a quick
workaround: using a special command on the RS-232 interface, the
microcontroller will be put into a "transparent mode", which means it just
hands over everything it gets from the uBlox module to the RS232 output and
vice versa; this will let me "talk" to the module directly or with u-Center.

* you might want to consider adding a low-capacitance TVS diode on
  the antenna input
-> damn, this would have been a wise idea. Didn't come to my mind.

* connect the PPS pulse to the spare transmitter of the RS-232 transceiver.
  This way you can feed your PC with a PPS pulse
-> this would have been REALLY good! at least one further reason to make a
version 2.1 ;-)

* turn N5/ADG779 around and connect the other side of the switch to GND
-> I did that (my friend said exactly the same, and I wonder why I didn't
see it myself. The schematic I had first is what I reverse-engineered from
the Oscilloquartz, I wonder why they didn't do it like this!)

* I am a little bit concerned that the ADG779 does not list a value for the
  charge injection
-> jup, I also saw that. However there is a couple of alternatives which do
list the charge injection, the best one of them having 6pC. I don't know
whether this is "much" but it sounds quite good, does it!

* R24/R25 are not needed. The feedback resistor of the inverter will do
  that job already. I might use a comparator instead of a schmitt trigger
  here, for lower phase noise/jitter.
-> I changed that part and added a little OpAmp which, I expect, provides a
bit better isolation anyways.

* The whole design looks very nice. Thanks for sharing with us
-> thank you very much :-)
I will see how well it works. At least I got my PCBs this week, and my
Mouser order for some components is on the way. I will start with the
software for the STM32 soon, maybe I will then bother the mailing list
again a bit more ;-)

And I am also working on a bit of a writeup of the whole design.

Attached 2 photos and the new schematic!

Best
Tobias
HB9FSX



[image: pic1.jpg]
[image: pic2.jpg]
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pic2.jpg
Type: image/jpeg
Size: 297338 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20200227/24d2344e/attachment.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: pic1.jpg
Type: image/jpeg
Size: 232463 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20200227/24d2344e/attachment-0001.jpg>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: schematic2.pdf
Type: application/pdf
Size: 95086 bytes
Desc: not available
URL: <http://febo.com/pipermail/time-nuts_lists.febo.com/attachments/20200227/24d2344e/attachment.pdf>