G400 IO protection

I am finishing our new custom board based on G400D. Previous board was EMX based and EMX ios are 5V tolerant however in three years we had some boards that fried. Most of problem occurs when customers are plugging power supply in wrong polarity, when they shortcuts relays output (24VDC) or when they plug the power supply on IOs…

For polarity inversion, a diode have been placed.

What is used to protect analog IOs?
I have read https://www.ghielectronics.com/community/forum/topic?id=10653&page=4

and it seems that this device provide a good ESD protection http://www.onsemi.com/pub_link/Collateral/CM1213-D.PDF.

However It seems to me this won’t protect the IO permanently if 24V is connected to the IO.

Do you have any idea?

May be a PTC fuse between the outside world and each channel of the TVS array could do the job. The major problem is to determine the PTC… I do not see what is teh max current that can support a channel…

Here’s what I found on the Atmel datasheet when doing my own custom board:

Max Digital Output Current = 8 mA (for pins PAx, PBx, PDx, and PEx)
Max Digital Output Current = 4 mA (for pins PCx)

Hello Iggmoe, at this stage I am more focused on inputs protection, but you’re right, outputs need also to be protected against over current. I will check that.

For now I have 8 analog input on the G400 that need to be protected against surge and permanent over voltage.

I am planning to use the WE-TVS 824013 TVS array:
http://katalog.we-online.de/pbs/datasheet/824013.pdf

This won’t protect against permanent overvoltage so the idea was to use a poly fuse but I don’t succeed to choose one.

The system need to keep accuracy of ADC, while efficiently protecting the IO…

Here is the plan

Analog/digital inputs are high impedance by design, so I don’t think you need to worry so much about the current that will flow through the analog input pin, since it will be negligible.

The worst case current will flow from Voltage_To_Measure through the TVS and ultimately to Gnd for the situations when the input voltage exceeds the TVS breakdown voltage, in which case the TVS’ job is to short the input voltage to Gnd.

The main consideration then is simply to ensure that the TVS itself can handle the anticipated current by looking at the datasheet and looking at the device’s rated maximum power or rated peak pulse current.

If you are protecting against ESD events, which tend to be very brief, I think the TVS by itself will suffice and you don’t need the PTC. If you think an overvoltage event can occur for an extended period of time, where the TVS may be shorting the current to Gnd for a long time, then an extra protection element like a PTC may be needed to protect the TVS.

I’ve used TE’s PolyZen devices with an integrated PTC, but they tend to be bulky and take up a lot of board space.

For input protection I have used a zener diode and 100R resistor. This has protected inputs to +24V before. :slight_smile:

What about some of the ideas here.

http://www.digikey.com/en/articles/techzone/2012/apr/protecting-inputs-in-digital-electronics

1 Like

Hi Dave, Hi Iggmoe

I based the solution after reading the digikey article. It uses 2 clamp diodes for ESD protection and a zener to not propagate voltage on power rail. For permanent overvoltage, the idea is to use the PTC as current limiter. I don’t talk here about IO current, but the current that can handle the TVS diode.

The PTC has been chosen to have a trip point at 30mA (PRG18BB221MS1RB from Murata). The doubt is about the permanent current that can support the TVS Diode… This is not described in the datasheet (http://docs-europe.electrocomponents.com/webdocs/107e/0900766b8107e9d4.pdf)

Here is a reference circuit from Analog Devices for an industrial application that I’m using now. It is designed for +/-10VDC inputs so you may have to scale for your particular input range.

Just a little story about IO protection:

My collegue told me he burned two G400-D by applying overvoltage on the board.

First time was 11.2V which results in 3.58V on the analog input. This is bellow max rating described in http://www.atmel.com/Images/Atmel-11055-32-bit-ARM926EJ-S-Microcontroller-SAM9X35_Datasheet.pdf on page 1297. Unless I misunderstand the document. Can anyone confirm my assumption? I will receive the SOM in next few days.

The second time he missconnects the power supply and therefore apply 28V on 0-10V input of the board. This should results in 9V on the input but the TVS diode start clamping and only 4V have been applied on the analog input of the SOM. I already receive this SOM and start investigating. First check, the SOM is still alive :slight_smile: . Let’s go to check the analog input… it still works also… therefore where is the problem?.. still don’t know but is there a problem…

My feeling is that the protection based on TVS diode and polyfuse works but due to several doubts and customer pressure he didn’t investigate enough.

After investigating during 2 hours this morning, I finally discover on a camera picture of the all system, that two wires was inverted on the analog input…

Similar here. Client connected in reverse and put 24V input on the ADC input. Blew the chip up nicely. New chip and all was working again. Has been for other 8 months now.

Found a nice protection IC from Maxim for the 4-20mA inputs. All new designs use this :slight_smile:

The weird thing is:

If the cable is inverted therefore the voltage is negative on ADC input but Atmel datasheet mentions that:
Voltage on Input Pins with Respect to Ground…-0.3V to VDDIO+0.3V(+ 4V max)

My protection shouldn’t work for negative voltage…

What is VDDIO (should I consider 3.3V?)…knowing exactly what’s is happening can be really tricky… May be the board is faulty…

As the comedian says, “You can’t fix stupid!”

That being said, I have a plaque a friend made that was a little 8051 based MCU glued to it with the words “You broke it good!”. Yes, I hooked 24V up to the analog input :-[ , and it was just before we were doing a demo. Luckily we had other boards and it was easy to rectify and the demo went well.

Can everybody on this forum like that comment; please, for the sake of us all.

1 Like