Surge testing of soil humidity sensors

[nano]

Hey there 

I’m working on a personal project to monitor soil humidity in the garden. For that I plan to design smart sensors connected to a RS485 bus system.
The sensors are powered by a 5V class 2 insulated AC/DC power supply. The secondary is connected to earth (functional earthing). A unshielded cable with two twisted pairs (4 conductors total) is used to connect everything.
A sketch of a sensor is attached.

To have a robust system I try to design the sensors with generic EMC product standards in mind.
Now reading EN 61000-4-5 (Surge Immunity) leaves me with a question:

Section C.1 states: "D.C. ports that are intended to be connected to an a.c/d.c power converter (e.g. laptop power supply d.c. connection) are not regarded as low voltage power supply ports. If d.c. power is fed on conductors included in a signal cable then these connections are not regarded as low voltage power supply ports."
-> 5V/GND can be tested as signal ports. Roger That!

Now because 5V/GND also use twisted pair can they be viewed as symmetrical interconnect and be tested as such? (just as the RS485 connection)

[T3sl4co1l]

Right, you don't need to do surge testing on low voltage stuff like that.

There may be other reasons why you might want it, though.  Lightning induced surge is typical on long runs through facilities; your data pair might be better classed as a TNV type, which could potentially have hazardous transients on it.  The surge conditions are relaxed here, mainly because of induced rather than direct influence, and because of assumed wiring resistance, that helps limit current a lot.

Don't forget about possible ground loop, and common mode range for the soil sensors and the RS485.

Tim

[nano]

Thank you Tim!

This is the first time I hear of TNV type circuit. I will have to investigate further.

In the meantime I had another idea:
The sensor PCB will be coated completely. If the dielectric strength of the coating is big enough there is no electrical path to earth.
So basically I would not even have to worry about line-to-earth surges as the current has no where to go. Or am I missing something?

[AndersJ]

You might want to consider voltages induced in your cable from
ground currents caused by lightning strikes.
Your insulation will not help there.

[T3sl4co1l]

The capacitance of the cable will still carry return current at high frequency, which is true with respect to ground, as well as between wires in the cable.

Most importantly, between the two twisted pairs, which will not necessarily be at the same AC potential at the far end -- the RS-485 common mode range will come in handy, make sure not to compromise it!

This is roughly where surge reduces to ordinary operational immunity concerns.  You might still provide ESD diodes to account for ESD hazards, and possibly charged cable transients.

Tim

[calexanian]

When you said Surge in irrigation my mind went to an entirely different place. That used to be a thing. In row and furrow irrigation there was a fad where they would would wet the rows in an alternating pattern and that somehow did something. They would wet the soil, let it seep in, hen wet it again sort of thing. With modern drip and overhead irrigation i think surge irrigation has largley gone out of fashion, but at a time it was a big thing.

Here is some info on that old stuff, but nothing about electronics.

http://www.cropinfo.net/BestPractices/bmp-SurgeIrrigation.php

[nano]

I attached the schematic of my current input section.
If I understood correctly there may be voltages/curents on the input wires induced by ground currents from lighting strikes.

For the ports of the RS485 transceiver I’m less concerned: They are rated for 12kV ESD and 4kV EFT.
More problematic is the LDO input:
Input voltage shall be in the range of 4.5V-13.2V
I plan to add a small TVS diode here aswell as a standard diode for reverse polarity protection.

Do you think the transceiver internal protection and a TVS diode (e.g. MMBZ15V) is sufficient for unshielded outdoor cable runs up to 100m?
Can you recommend me other parts for the protection of the LDO?

When you said Surge in irrigation my mind went to an entirely different place

Funny how the same term can mean totally different thing

[T3sl4co1l]

Max input 13.2V means you should pick a TVS with that as the peak maximum voltage.  Which is going to be more like a 9V rated diode.  Which is perfectly fine for a 5V system, and can still be powered from, say, unregulated 5-7V DC adapters.

A unidirectional TVS will also cover reverse voltage protection.  For both reverse and overvoltage conditions, you might want to add a polyfuse*, a regular fuse, or a fusible resistor.

*Polyfuses take forever to "open" (~100ms), so that your circuit won't really be protected from damage.  In that time, the TVS will happily cook off and fail shorted (and maybe burn out and open up again, who knows).  There are combo TVS+polyfuse devices out there, which use the TVS to heat the fuse element: it opens faster, protecting itself.

Polyfuses also need the fault current to be limited to a safe value, otherwise they'll blow open (a monofuse ).  With some shopping, you can usually find a device with the right current rating, and enough minimum resistance, to handle that without adding external resistors.

If you wish to give the RS-485 additional protection, I'd recommend a 6 to 12V bidirectional TVS for each line.

Tim

[nano]

Thank you again!

My last post regarding the input voltage was a bit misleading.
I want to support an input voltage range of 4.5V to 13.2V. The LDO maximum voltage is actually 16V.

Attached is a picure of two ways to protect the input against overvoltage impulses and reverse polarity.
For the first circuit I chose a polyfuse (300mA Trip Current) that trips when reverse polarity is applied. It never trips during normal operation because the overcurrent protection of the LDO (typ. 250mA) will most likely be faster.
The second circuit simple uses a diode for reverse polarity protection. To prevent a high reverse current a biderectional TVS was chosen.

Which of those two circuits is better suited for the application?

[floobydust]

My $0.02 I prefer reverse-protection to "do nothing" instead of shunt. Like automotive designs- If a unit is connected backwards, or GND opens to it, better to have no current flow.
Polyfuse, series-diode (silicon not Schottky) then unidirectional TVS is what I like. Your RS-485 port will also be vulnerable.

Polyfuses are horrible at high temps so make sure you have enough headroom for highest ambient temp and the RS-485 line driver, which can use a lot of current depending on baud rate. 0ZCK0010FF2G is also lossy 0.7-6 ohms, and add the series-diode drop and Vreg headroom, and the field wiring voltage drop and you might need more than 5V PSU. It's a big calculation.

[T3sl4co1l]

My $0.02 I prefer reverse-protection to "do nothing" instead of shunt. Like automotive designs- If a unit is connected backwards, or GND opens to it, better to have no current flow.
Polyfuse, series-diode (silicon not Schottky) then unidirectional TVS is what I like. Your RS-485 port will also be vulnerable.

Polyfuses are horrible at high temps so make sure you have enough headroom for highest ambient temp and the RS-485 line driver, which can use a lot of current depending on baud rate. 0ZCK0010FF2G is also lossy 0.7-6 ohms, and add the series-diode drop and Vreg headroom, and the field wiring voltage drop and you might need more than 5V PSU. It's a big calculation.

^^^

The second one, and transpose the TVS and series diode, and use unidirectional.

Tim