Using standard subrack components to build a DAQ platform

Over the time it became clear that the original NatWatch platform is not well suited for many intended applications and there are reasons to make some improvements. The NatWatch-R name was selected because the new specification is based on standard components of 19“ rack and subrack cabinets.

Moving away from the original NW

After designing a couple of boards I was able to fulfill the needs of some of my projects. There are designs available for low-cost GPRS enabled main board, solar MPPT charger, 2Msps 4 channel digitizer, CAN interface, 48V isolated power supply, WiFi interface board, internal environmental monitor board (not yet finished).

However, when actually using them, I found it very hard to perform some basic tasks I was expecting from the platform and some of the components turned not to be easily available:

  • It is not possible/hard to insert or add a board to the stack in the field. Using M2 screws to mount the boards together is apparently not the best choice. Even if this is not considered problematic, try to replace a single board in the middle of the stack. If you still don’t believe, try to update firmware on the main board located on the bottom.

  • Debug connector position is not defined. Usually a 10pin 1.27mm pitch Cortex debug connector is fitted somewhere on the board. This means you are not able to attach a debug cable to the board if another board is sitting above it.

  • The stack of boards is not mechanically protected in any way

  • It is not possible to properly mount the stack itself to most of the cases. An adapter board/plate fastened to the case must be made to mount the stack on top of it.

  • Stacking connectors used on the boards are very hard to source. There are only ~2 companies making compatible connectors and usually they are not in stock of common distributors. Apart from that, they cost nearly 3e/1pc.

  • The stack as a whole or individual boards cannot be properly shielded.

  • The stacking connector does not have enough signal lines for additional buses/interconnects, such as power lines, high speed buses, low power low speed buses, etc.

The proposed changes

  • Cards of the system must be independent of each other. It must be possible to remove a single card without touching the rest of the system. The card may be additionally fastened using screws. The stack of boards is no longer a stack.

  • The system must be enclosed in a rugged frame. The frame size must be customizable either during manufacturing or by the system user using common tools.

  • The rugged frame must comprise only standardized stock components available from multiple manufacturers and easily manufacturable parts.

  • The system shall use a backplane instead of stacking connectors. The backplane should be fully passive. The backplane must use standardized connectors available from multiple manufacturers. For bigger systems, it should be possible to interconnect multiple smaller backplanes to form a single bigger backplane.

  • Access to the system console must be possible using standardized interface with a commonly available hardware/cable/adapter.

  • It must be possible to connect a debug probe to every board in the system during operation.

  • It must be possible to insert/remove a card from the system during runtime (hot plug, hot unplug).

  • The system frame must be designed in a way which allow it to be mounted: - in a waterproof and shockproof case for mobile installations - in a electronic cabinet for fixed installations - in an extruded aluminium case for desktop use - in nothing for debug use (it must be fully operational and self-contained in this case)

  • All components replaceable during operation must be accessible from the front panel.

Trying to formally specify the new platform

Draft of the specification is kept on GitHub. Keep in mind, it is a draft.