Level Translation with possible floating inputs

Eric Blanchard

So I've got a battery powered device with the majority of the system running on a single 1.8v rail. I've got a 3.3v regulator but am looking to only turn it on occasionally based on some infrequent events as it is mainly for high current functions. The device will have an external connector to interface to customer sensors, switches, and uCs. We'll have 4 dry contact inputs for their sensors as well as a 4 pins for a UART with CTS/RTS.

I need to provide voltage dropdown on the dry contacts and level translation for the serial communication as most of our customers are accustomed to working with our legacy products that run at 3.3v. Of course we recommend that they only connect the dry contact inputs to a switch, but in practice there will be users who try to drive the lines from a micro.

I'm pretty comfortable with the UART level translation as we can simply use a dual supply chip with 1.8v acting as Vcca which will always be there and 3.3v being enabled and therefore provided on the Vccb side whenever serial communication is required.

However, as not all installations of the device will use either the dry contacts or the UART, I wasn't sure what the best practice would for providing the overvoltage protection/dropdown function for the dry contact inputs which could be toggled at any time (i.e. not necessarily when 3.3v is turned on). While using something in the 74LVC family would do the trick, if there is nothing attached to the external connector, the inputs would float and that's generally not a good situation to have. If I pull the input up to Vcc (1.8v), it would seem as if that would defeat the purpose of having the protection. Is there a particular logic family or level translation method that is tolerant of input voltages higher than Vcc but can also handle floating inputs or should I just pull the inputs up to 1.8v through a diode?

We're equally size, cost and power consumption sensitive so I was trying to reduce part count wherever possible.

Thanks,

Eric

over 9 years ago

0 Jennifer Joseph over 9 years ago

TI__Intellectual 2530 points

Hi Eric

The most effective method to provide defined levels for a floating bus is to use built-in bus-hold feature. But as you mentioned leaving the inputs floating is not recommended.

Please take a look at the TXS- type translators. TXS0xxx has internal pull-up resistors. The OE pin cannot be left floating. The pass-transistor voltage translators and their voltage clamping feature makes TXS type translators an ideal choice for applications requiring over-voltage protection.

Thanks

0 Eric Blanchard over 9 years ago in reply to Jennifer Joseph

Prodigy 70 points

Hi Jennifer,

So that family looks like it would be good for my UART bus as the 3.3v rail will always be enabled any time the UART is used. When 3.3v is turned off, Vccb will go to zero and it looks like the chip handles that scenario well.

However, for the dry contact inputs, they could be toggled at any point in time...likely not when my 3.3v rail is enabled. In that case, I think I would need some type of single supply (1.8v based) over voltage protection. And since the possible input voltage I'm trying to protect from is higher than my system voltage, I can't use a pullup. Also, since the pins are generally intended to be used with switches or at the very least open collector devices, I can't just pull the input down. That was why I was thinking maybe I can just use a simple buffer and supply the input pullup resistor with my 1.8v rail through a diode so it would always be pulled up but not expose my system to 3.3v in the event that a user drove that line high from a 3.3v source.

0 Conor Orourke over 9 years ago in reply to Eric Blanchard

Prodigy 190 points

I have a similar situation with my system so I thought I could tag on and maybe help. I'm using 74LVC level shifters but I'm unclear as to whether floating inputs are allowed when Vcc is off. I have local 3.3V circuits with external 1.8V inputs to the LVCs - these will float if the 1.8V power line is also disconnected.

The LVC datasheets do say "Vcc isolation - if either VCC is at gnd, both ports are high impedance". Which looks good. But then they also say "input circuitry is always active and must have logic high or low applied". Can I assume that that latter statement only applies when both VCC inputs are active? Isn't "high impedance" incompatible with "inputs active" !!

With Eric's issue I've done two things

Two MOSFET's face to face (as per page 12 of this old Philip's app note) which is rather a lot of parts perhaps but it does work
Just using a diode to simulate an "open-drain" operation. Your 1.8V input connects to the anode of a Schottky, the Cathode goes to the pin. That way an external driver can only pull down your input line. You'll need a pull up to 1.8V to bring it back up. Like this

0 Jennifer Joseph over 9 years ago in reply to Eric Blanchard

TI__Intellectual 2530 points

This application note has some part suggestions that are simpler and can address overvoltage protection. Bus-Hold AND clamping diodes might do the trick.

http://www.ti.com/lit/an/scba004c/scba004c.pdf

Thanks

0 Jennifer Joseph over 9 years ago in reply to Conor Orourke

TI__Intellectual 2530 points

"input circuitry is always active and must have logic high or low applied". Can I assume that that latter statement only applies when both VCC inputs are active? Isn't "high impedance" incompatible with "inputs active" !!

The input circuits are active if the power supply of the respective side is powered. For example, A side input circuits are active when VCCA=1.8 and VCCB=0.

Thanks

0 Conor Orourke over 9 years ago in reply to Jennifer Joseph

Prodigy 190 points

Great! That makes complete sense. Thank you, Conor.

Logic

Logic forum

Level Translation with possible floating inputs