While experimenting with my clock distribution amplifier, I found that this board creates a lot of interference to my HackRF receivers.
Plus, it’s not very convenient to work with a bare board.
I decided to build metal housing for this board. This is just a little note about its construction and some nuances.
Sure, it’s impossible to find a 100% matching case for this board. I selected aluminum enclosure Gainta G0247.
This case is slightly bigger than the board, and this allows to fit all required connectors and indicators. Also, I decided to build an external 5V regulator using the case as a heatsink.
The board is bolted to one side of the case. All clock outputs are placed on the case back side. Input signals are connected using three short coaxial patch cords. I think it’s most logical.
For the 1PPS, I used KLS15-RCS01-PC10 10 pin circular connector. 8 pins are for 1PPS, and two pins are for ground. I think this is the best choice. An external power supply is connected via KLS15-225-M16 6 pin connector just because I had one.
The initial arrangement inside the case:
The 5V regulator is a simple 7805-based board. This 5V line bypasses the onboard regulator that little bit overheating.
L1 is a common mode choke filter. Just a 9 x 2 wire turns around a 10×25 green ferrite toroid. S1 switch is placed on the front panel.
All LEDs are also moved to a separate board glued to the front panel.
For the 1PPS line, I made a special 8 x 450mm coaxial cable.
3 pin connectors are used for easy connection to my SDR boards.
To ensure that everything works properly, I built a simple setup with my GPSDO, distribution amplifier, and HackRF board to ensure that everything works properly.
And this is what I found at the end of the 1PPS line connected to the HackRF input:
Whoa! A massive 200 ns oscillation with 6V max amplitude. It’s definitely not good. Of course, I forgot about impedance matching!
A 50-ohm transmission line with the HackRF load creates a huge impedance mismatch. The signal can’t be consumed by the destination and reflects back to the LVTH541 device. This creates a massive parasitic oscillation.
The issue could be fixed with an in-series resistor placed Before the 50-ohm line. The resistor value should be selected in a range of tens or hundred ohms. But this should be done very carefully because a bigger resistor value creates additional delay on the line. I found that a 100-ohm resistor fits perfectly. Oscillation is suppressed, and there are no meaningful delays. I precisely selected 8 resistors with my Fluke multimeter.
With in-series resistance the oscilloscope shows a very good result:
Massive oscillations are suppressed and the signal is under the valid limit. Let’s measure delays.
Channel 0 is the 1PPS source. Channel 1 is the point directly after the resistor. No measurable delay.
For example, this is a delay with a 200-ohm resistor, big difference:
All resistors were soldered directly to the connector:
Additionally, I covered all 1PPS wires with a copper foil connected to the chassis. Not sure if it’s really required but I think that 10MHz clock lines could create some trouble. I think now it’s ready for space 🙂
To avoid ground-loops, shielding is insulated from the case’s side.
Completed box with labels. Back side: