DAQ input protection (overvoltage & overcurrent) question

[ricardosilva]

Hello Everyone

I´m designing a DAQ board as my thesis work.

I have several questions about input protection, I have searched and searched and I have found only 2 circuits, a crowbar and another one using 2 zenner and a PTC resetable fuse or a ordinary fuse. 

My question is, if there is any  other circuits that can protect inputs.

The measurement limits are +-20V with a maximum frequency of 100 kHz. At the input there will be a INA163 IA, and then the rest of the circuit. I have to protect these inputs to 230 Vac. 

If someone knows any other method of doing this, please reply. I'm pending to the PTC+diodes.

Thank you

[johnsold]

This sounds like an interesting thesis and a very important and useful topic.

Does the protection need to work only at power line frequencies or can the 230 V occur over the entire input frequency range?  Does the protection need to work when no power is available to the DAQ board or can the inputs have 230 V when the DAQ board is unpowered?  How far above the measurement limits can the input voltages go before damaging something?  What is the accuracy specification and how much error can the protection circuit introduce without exceeding the specification?  If you use a crowbar, how will it be reset?

Both passive and active protection circuits exist.  Which ones might be suitable depend on things you have not specified.  PTC devices and Zener diodes are a good choice for many applications.  The diodes are fast and clip at a consistent level when conducting and have low leakage current when off.  An active circuit might turn on an FET when the input voltage gets too high.  Whether this could work would depend on the series impedance available at the inputs.

Lynn

[ricardosilva]

Thank you for the reply

The specified maximum input range will be +-20V at a maximum frequency of 100 kHz. This project will be used for educacional purposes, so as you can imagine it needs to be protected from the students. I´m more worried at voltages above 20V, because they can damage the rest of the circuit. The 230V protection are for power lines protection (some student may think that it is a good ideia to connect that directly). 

The +-20V input will be attenuated to a +-5V signal that will be supplied to an AI (INA163), they can take +-18V, but I need some signal conditioning, and those +-5V will be used in the PGAs and with the ADCs, these will all be controlled by a PIC32

I´ve been searching for crowbar examples, and I think that I prefer using a circuit something like this

http://www.zen22142.zen.co.uk/Design/overvoltage.htm

I understand that the diode activates the tyristor gate, but apart of that I have no idea how to choose the components.

On the other side, the diodes+PTC will be much simpler.An active circuit might be a good idea too, but right now i'm using alot of the PIC pins, there is also a USB communication using a FT2232H using a 16 bit data bus, and the space is running out too.

[johnsold]

One approach would be to use the input resistors of the 20 V to 5 V divider as part of the protection circuit.  Assuming that they have fairly high resistance values, select the power rating so that they can withstand exposure to the line voltage.  Then connect the zener diodes at the 5 V points.  See the image below.  With the values in this diagram 1/4 W resistors and low power zener diodes will work.  Make sure that the resistors are rated for 230 V (> 325 V peak). If you cannot get resistors with suitable voltage ratings (yes, resistors do have voltage ratings), you may need to connect several resistors in series.  Also consider the spacing required on your printed circuit board for safety at line voltages.

Lynn

[ricardosilva]

That seems a good approach, to keep it simple. About the resistors I don´t know yet if I´m going to use SMD type or the normal type, I have to see if there is enough space on the PCB. Probably if I choose the SMD type I will have to connect several in series as you said (230 V SMD resistors with those values are a bit hard to find, Ive checked farnell), but I will test it.

About the zenner, when choosing it, It should whitstands also 230V of direct applied voltage right?

I´ve attached a file with the circuit, as I pretend to measure diferential signals with the INA163, I think that this circuit is correct using your sugestion.

Now about the PTC, for the overcurrent protection, if I´m not wrong there are 3 parameters that must be taken into account:

- Holding current (I think this is the normal working current)

- Tripping current (At this value the PTC cuts the current going into the circuit)

- Operating voltage (I think I must choose the 230V version)

The holding current I´ve pointed it to 50mA with a tripping current of 100mA (I have no ideia if this is ok), the maximum current input of the INA163 is 10mA, the resistive divider will limit also the input current, but I would like to have some more protection. 

Shoud I put one PTC in each signal branch (Vin+ and Vin -)?

[johnsold]

The zener diodes will never see more than 6.8 V because they start conducting at that voltage (and 0.7 V for the opposite polarity of voltage).  You only need to be concerned about the current and power ratings.  The current is limited by the 1 megohm resistors to less than 1 mA, so power dissipated in the diodes is quite small.

In your circuit, what happens if 230 V is applied to one input while the other input is grounded (earthed)?  There is no current path to ground so the current will pass through the amplifier.  That is why I grounded the diodes.

PTC: If the tripping current is 100 mA and the voltage is 230 V, something in your load circuit will be dissipating 23 W before the device trips.  The line voltage is specified as the rms value for a sinusoidal waveform.  This means that the peak voltage is sqrt(2) times higher or about 325 V. The PTC rating is probably the instantaneous voltage it can withstand.  You need to allow for normal variations in the line voltage and the tolerance of the voltage rating of the device.  So probably 350 or 375 V rating.

However, your circuit will not draw more than a fraction of a milliampere even without protection, so you will never trip the PTC device.  The PTC device and a crowbar ahead of the megohm resistors could be used, but it is not clear that it would add any protection.

Lynn

[ricardosilva]

OK, you are right, I forgot the GND connection.

About the zenner diodes again, the lower DAQ voltage range will be +-0.1V, the diodes won´t interfere with these small signals right? They only limit the voltage when the input surpasses their breakdown voltage.

This circuit is already current limited by the 1Mohm resistor, there will be ~1mA with 230V on the input, but about the PTC, I have to talk to my tutor and discuss with him this option. He said that I should use a PTC or a fuse, but I will discuss this with him.

I can test the circuit only on Monday, thanks again for the reply.

[johnsold]

Yes.  The zener diodes draw very little current when the applied voltage is less than the zener breakdown voltage.

Lynn

[MarkMoss]

If you decide to use the circuit Lynn suggested, there are a couple of caveats you need to look out for:

1. Make sure that the diode's leakage current is low enough not to affect the accuracy of your voltage reading. (100 nA of leakage will cause a 100 mV drop across the 1 MOhm resistor).

2. You may have problems if you use the circuit with a DAQ device which only has once A/D converter.  Basically, the A/D converter will have some amount of input capacitance, and the high impedance of the voltage divider may not allow the capacitance to charge/discharge fully between channels.

Mark Moss

Electrical Validation Engineer

GHSP 

[johnsold]

Good points.  Thanks Mark.

Lynn