Project

General

Profile

Actions

Feature #5626

closed

Bug #5542: Move hub to datacenter colocation

mechanical + electrical setup for GPS receiver at co-location

Added by laforge 4 months ago. Updated 2 months ago.

Status:
Resolved
Priority:
Normal
Assignee:
Category:
hardware
Target version:
-
Start date:
07/22/2022
Due date:
% Done:

100%


Description

We'll have about 100m CAT5 between the GPS02 and the icE1usb + RS485 drivers.

We'll need coaxial + RS4xx overvoltaage protection


Files

nmea_data.png View nmea_data.png 23.9 KB NMEA Rx Data at 100R || 4.7nF termination laforge, 07/23/2022 03:55 PM
1pps.png View 1pps.png 16.2 KB 1pps pulse at 100R || 4.7nF termination laforge, 07/23/2022 03:56 PM
1pps_zoom.png View 1pps_zoom.png 16 KB 1pps pulse (zoomed in, measuring rise time) laforge, 07/23/2022 03:56 PM
cat5em_1.jpg View cat5em_1.jpg 1 MB laforge, 07/23/2022 04:51 PM
cat5em_2.jpg View cat5em_2.jpg 858 KB laforge, 07/23/2022 04:51 PM
1pps_new.png View 1pps_new.png 21.5 KB laforge, 07/23/2022 04:53 PM
1pps_new_edge.png View 1pps_new_edge.png 15 KB laforge, 07/23/2022 04:53 PM
dco_sd2_md_hf5.jpg View dco_sd2_md_hf5.jpg 157 KB laforge, 08/14/2022 07:53 AM
dco_sd2_md_hf5-internal.jpg View dco_sd2_md_hf5-internal.jpg 240 KB laforge, 08/14/2022 07:53 AM
dco_sd2_md_hf5-schematics.png View dco_sd2_md_hf5-schematics.png 6.47 KB laforge, 08/14/2022 07:56 AM
rs422-ovp-board.jpg View rs422-ovp-board.jpg 1.04 MB laforge, 08/31/2022 06:49 PM

Checklist

  • propose OVP parts etc to data centre and ask for review/approval
  • obtain OVP parts
  • end-to-end setup with OVP, GPS, antenna, etc. at laforge
  • investigate if 12V phantom voltage via 100m CAT5 (~20 ohms loop resistance) is viable
  • clarify mechanical mounting (antenna, din-rail) at co-location
  • required coaxial cable length
Actions #1

Updated by laforge 4 months ago

  • Status changed from New to In Progress

The spare Ericsson GPS receivers I have here are GPS03 (successor of GPS02), attached a MA-700G antenna and loked at the current consumption. The absolute peak was 161 mA during start-up. Once the GPS signal is locked (green led continuously on), it's 97..103mA @ 12V.

At the predicted worst-case loop resistance of 100m CAT5 of 20 Ohms, that would be a 2V voltage drop for the average power consumption, and 3.2V on peak.

Will do some voltage range tests next, and also insert a 20Ohm series resistor.

Actions #2

Updated by laforge 4 months ago

Reducing the voltage (after lock was acquired) showed operation down to about 6.5V. Current consuption increased to 140mA, so there's some DC/DC converter inside (good).

Cold-Starting the receiver at 7V worked fine; idle current 140-147mA.

In all of those tests the RS485 drivers were open, i.e. not driving any capacitive or resistive load yet.

Actions #3

Updated by laforge 4 months ago

regarding capacitance, https://en.wikipedia.org/wiki/Category_5_cable states 52pF/m so about 5.2nF for 100m. To that we'd have to add the capacitance of the RS422 receiver input, but for sure it's just going to be some pF, so negligable.

So, I set up as follows:
  • 22 ohms series resistance between PSU and GPS 03
  • 100 ohms parallel to 4.7nF termination of 1PPS or NMEA-Tx pair
  • current consumption increased from 100mA to 120mA
  • scope traces of signal looked fine to me
  • receiver operation down to about 10V PSU output voltage
Furthermore:
  • terminating the second transmitter (NMEA Tx) increased current consumption from 120mA to 148mA
  • operation of receiver still down to 10V from the voltage source
  • cold start also possible from 10V voltage source

So all in all, we can conclude it is safe to operate the GPS03 with the chosen antenna over 100m of CAT5.

some scope traces of the data on the "remote" end of the simulated CAT5:

Actions #4

Updated by laforge 4 months ago

  • Checklist item investigate if 12V phantom voltage via 100m CAT5 (~20 ohms loop resistance) is viable set to Done
Actions #5

Updated by laforge 4 months ago

Building the entire setup (22R on 12V rail, 100R || 4.7nF on each of the three differential pairs) on a small adapter board (mis-appropriating one of my RJ45 jumper box PCB) looks like this:


For some reason the waveform of the 1PPS now looks much more reasonable. Maybe I had a bad probe GND connection or something last time. It's a nice 50us square pulse. Rise time has not changed.

Actions #6

Updated by laforge 4 months ago

  • % Done changed from 0 to 20

the over voltage protection devices for the RS485/422 signals (DCO SD2 MD) have meanwhile arrived. However, the part for 12V was missing from the shipment (despite listed on packing list). Have complained with the supplier.

Actions #7

Updated by laforge 4 months ago

in case anyone is curious, this is the internals of the "DCO SD2 MF HF 5":

dco_sd2_md_hf5.jpg dco_sd2_md_hf5-internal.jpg

Actions #8

Updated by laforge 4 months ago

If I got it right, the schematics look like this:

Actions #9

Updated by laforge 3 months ago

  • Checklist item obtain OVP parts set to Done
Actions #10

Updated by laforge 3 months ago

picture of the APU-sized RS422 OVP board that goes into the 1U enclosure:

rs422-ovp-board.jpg

Actions #11

Updated by laforge 3 months ago

  • % Done changed from 20 to 40

Did some more testing with the following chain:

  • GPS antenna
  • GPS03
  • 100m-rj45-resistance-and-capacitance-simulator
  • rs422 OVP + 12V injection board
  • USB-RS422 converter attached to my laptop / 1PPS to scope

This setup was working fine for hours. It's still missing the DIN-rail OVP next to the GPS03.

I later swapped the 100m-simulator agaist an actual spool of 100m CAT5. Surprisingly, the RS422 signals still look excellent and the NMEA is correctly decoded. However, I think something is wrong with the phantom power, as it looks like the GPS03 is crashing repeatedly. IT does get a fix, but then soon after the first valid standard NMEA sentences, it starts all over again.

Actions #12

Updated by laforge 3 months ago

laforge wrote in #note-11:

I later swapped the 100m-simulator agaist an actual spool of 100m CAT5. Surprisingly, the RS422 signals still look excellent and the NMEA is correctly decoded. However, I think something is wrong with the phantom power, as it looks like the GPS03 is crashing repeatedly. IT does get a fix, but then soon after the first valid standard NMEA sentences, it starts all over again.

false alarm. I introduced an overcurrent situation due to a bad connection, triggering the PTC fuse. The setup is now working fine with 100m of CAT5 on a spool.

Actions #13

Updated by laforge 3 months ago

so with the end-to-end setup with icE1usb and the RS422 daughter board by tnt in place, I'm seeing brown-outs when feeding the 12V phantom supply to the GPS03 over the 100m cable :(

When feeding 12V at the input, I'm seeing about 7V with dips down to 5V on the GPS03, which then makes the 5V output side of its DC/DC converter dip.

Adding a 47uF tantalum capacitor on the 12V input of the GPS03 looks like it helps a bit, but doesn't make the brownout go away.

I don't want to add capacitance to the output side of the DC/DC as that might make it unstable.

Increasing the PSU voltage to 14V resolves the problem: The nominal 12V input to the GPS03 is about 8V with occasional dips to 7.1V. No scope-visible dips in the 5V DC/DC output rail of the GPS03.

So either we increase the 12V to 14V and hope it's ok, or we feed the 12V locally on the roof from a DIN rail PSU. I think the second is actually the safe option without relying on any hacks.

Actions #14

Updated by laforge 3 months ago

I likely didn't see those brownouts in my existing testing as I only attached RS422 transceivers to Rx/Tx pairs but not to 1PPS.

Driving only either 1PPS or GPS-Tx seems to work even in the current end-to-end setup with 12V, so we are apparently just pushing it over the edge.

Actions #15

Updated by laforge 3 months ago

  • Checklist item end-to-end setup with OVP, GPS, antenna, etc. at laforge set to Done
Actions #16

Updated by laforge 3 months ago

  • Checklist item clarify mechanical mounting (antenna, din-rail) at co-location set to Done
Actions #17

Updated by laforge 2 months ago

  • Status changed from In Progress to Resolved
  • % Done changed from 40 to 100

new hub is fully deployed, GPS 1PPS arrives in the rack and icE1usb reports err=0 most of the time, i.e. the 2.048 MHz clock runs at 1PPS-disciplined 2048000 Hz.

Actions

Also available in: Atom PDF

Add picture from clipboard (Maximum size: 48.8 MB)