Where do voltage level limitations of logic level translators come from

level-shiftinglevel-translationlogic-level

When browsing through available bidirectional level translator ICs, I always see the condition that reference voltage on one side of the translator must be strictly less than voltage on the other side.

For example, ADG3300:

For proper operation, Vcca must always be less than Vccy.

Where does this limitation come from? Aren't bidirectional level translators symmetrical in their internal structure?

Also, are there any level shifters without this limitation? (or is it possible to create such?)

Best Answer

It's likely because usually the designer is specifying the two supply voltages and they are thus known a priori. Since there is likely a cost in performance and/or die size to making a symmetric level converter, the majority of the market can be well served by types with specified Vcca > Vccb or whatever. For example, the common bidirectonal level conversion method using a discrete MOSFET (from here):

http://www.hobbytronics.co.uk/mosfet-voltage-level-converter

..will not work if you swap the supply voltages.

There are available converters with symmetric supply voltage limits, for example, the TI TXB0304, which could come in handy if you needed to program the supply voltage or have it supplied externally in order to deal with customer requirements (the alternative might be to operate all your circuitry at the lowest or highest voltage possible, or to use two converters and similarly set an intermediate supply voltage).