Cable interface vs. Radio interface

Hi All,

I’m sure You guys considered in the past option to connect sensor head to PC/laptop using cable instead of some sort of radio interface. I do understand that Radio link guarantees galvanic isolation and makes sensor head way safer for end user. With that in mind I just want to mention another option: use of silicon isolators (thru induction) to provide galvanic separation. Here is a quick summary of such approach:

Pros:

• throughput no longer limited by RF interface – assuming cable connection provides good signal integrity,
• less components - cheaper design,
• simpler design – less components that required to be separately  pre-programmed before they are operational in desire mode,
• lighter design - provides power so no need for battery,

Cons:

• still potential risk of electrical shock due to bad PCB soldering or unforeseen hardware malfunction. Risk of electric shock is somehow minimized by using USB interface which in itself generates voltage up to 5V only,
• there might be some power supply noise problem introduced by power source being generated by DC-DC converter. This will have to be evaluated before deciding that this is real problem here. DC-DC will operate in MHz range where sensor head is looking for 10s to 100s Hz.
• if differential connection is necessary minimal number of cable connection will increase,
• cable might limit free movement. Sensor head might have tendency to apply more pressure around the cable end (assuming cable is heavy).

To backup this discussion I put together few diagrams.


 

Diagram 1 – existing solution  - outlines current approach to whole interface between sensor head and host PC. Left hand side outlines USB dongle which contains:

• USB to UART interface (FTDI) and
• Bluetooth module (RFduino).

Right hand side outlines sensor head with:

• Bluetooth module (RFduino),
• battery (and whole circuitry around power generation),
• UART to SPI converter (ATmega or PIC processor),
• two ADS1299 (one on main board and one on daisy module) and
• accelerometer

Based on Diagram 1 I put together Diagram 2 – UART interface. Here are main differences:

• RFduino modules are gone. Part which replaces them is one that is capable to provide interface isolation as well as power isolation. Some examples of them are ADUM5211 http://www.analog.com/en/products/interface-isolation/isolation/isopower/adum5211.html#product-overview or ADUM5201 http://www.analog.com/en/products/interface-isolation/isolation/isopower/adum5201.html#product-overview depending how much power sensor head will consume in the end.
• There is obviously a cable now. There are two scenarios described in Diagram 2 – UART interface. One on the top outlines option with single ended connection. This will have to be evaluated before deciding that drive strengths on both sides of the link are high enough to drive ~2m long cable at few Mb speed. If it turns out that single mode is not workable (signal integrity is not preserved) or it is too noisy back-up option might be to use LVDS transceivers. For example SN65MLVD205A http://www.ti.com/product/sn65mlvd205a or similar from LVDS TRx family. That of course will add a bit of cost to the total solution and increase requirement for number of cables in the interconnection.

The Diagram 3 – SPI interface1 takes slightly different approach. It removes UART to LVDS translator at the sensor head side (ATmega or PIC serves this function at the moment). Since FTDI as a standard provides UART type interface (virtual COM port) there is still need for UART to SPI translator. This time this translator seats before isolation circuitry. This makes dongle a bit more complicated but sensor head is simpler. This gives user easy access to the trigger signals and there is no need to provide galvanic isolation for them.
Since now whole SPI interface is transmitted thru the cable more wires will be needed. This is a drawback for this option. Bottom part of Diagram 3 – SPI interface1 diagram shows option with differential interface. As can be noted the connection in such scenario will require 14 separate wires. This might be a bit of overkill.



Finally Diagram 4 – SPI interface2 shows different take on SPI only interface. It assumes that FTDI chip will directly provide conversion for USB to SPI interface (skipping UART). The FTDI crowd seems to have such IC on their shelves: http://www.ftdichip.com/Products/ICs/FT220X.html Also Silicon Labs has something: http://www.silabs.com/products/interface/usbtouart/Pages/usb-to-spi-bridge.aspx But I don’t know what is the story with drivers here. That will require bit more digging. Disadvantage with this approach is increase number of cables required for connectivity. Advantage is the fact that sensor head in now simple ADS1299 plus power supply and that is it.

So guys as I have stated at the beginning this is more just food for thoughts. Maybe You already went thru that exercise and decided that any chances of electric shock thru sensor head are not worthy hence radio is the only viable option. If that is the case I can understand it.

Cheers,
Henryk