SN74CB3Q3251: unused analog inputs - must they be pulled up or down.

Hi Louie

Sorry, when I  say not connected I mean

A) The device to which the analog switch is connected to is high impedance, which I assume means the same as not connected

B) The device to which they are connected are unpowered (or the VCCI rail is GND)

In terms of the max value I was thinking that maybe I could find a value for the PD that would be sufficient to meet the analog switch latch up requirement and

Large enough as to not impact the constant current source functionality in the I2C device (I would need to raise a question against this device).

Finbarr

Louie
I generated a block diagram to give you a better overview but alas I do not appear to be able to attach it.

The application comprises two PCB’s. The carrier board contains 5 modem cards which interface with a SIM card interface board which contains up to 16 SIM cards. It should be possible to multiplex any of the 5 modem cards to any one of the 16 SIM cards (if present) through an analog multiplexer array (comprising TI’s SN74CB3Q…. series of devices).
The modem cards act as the SIM card master interface. It provides a clock, reset, a bi-directional I/O interface and it supplies VCC to the SIM cards and associated logic (TCA9800). The VCC supplied by the modem needs to be negotiated with the SIM card WHEN the modem is connected to a SIM card. Before that time i.e. before activation, it is possible that all SIM card interface lines, from the modem, are high impedance.
Because of limited capacitive loading capabilities of the modem cards and SIM cards, and interface rise time requirements, I have had to use the TCA9800 to provide extra drive capability on all the SIM interface lines (and bi-directionally on the I/O line). I need a PU on the I/O bi-directional connection to the SIM card so, based on the TCA9800 data sheet, I need to connect the A side of the device to the SIM card and the B-side of the device to the analogue multiplexer array (as the B side of the device has a constant current PU on the B side and hence no PU is to be applied to the B side).
Both VCCA and VCCB, of the TC9800, will be connected to the VCC output of the modem card. Depending of the “negotiation” between the modem card and the SIM card this voltage may be 1.8V or 3V. The interface will always start with a VCC of 1.8V (all the other interface lines will be at this voltage). If successful negotiation does not occur then the modem will try to connect with the SIM card at 3V.
The analogue multiplexer array is powered independently from the SIM cards, so once the board is out of reset the multiplexers will be powered.
Here’s where it gets a little tricky;
a) SIM card interface side.
If a SIM card is not present, or not connected to a modem card, VCCA and VCCB to the TC9800 will be high impedance. That said VCCA and VCCB can be pulled down to GND via a pull down resistor. The TCA9800 will GND referenced, with VCC inputs driven to GND an OE at logic low. Is this sufficient for the analogue multiplexers to avoid latch-up?
b) Modem interface side.
On the modem interface the SIM card interface lines may be high Impedance until the modem tries to connect to the SIM card. Pull downs can be placed on the clock, reset and VCC inputs, but not on the bi-directional I/O.
So when the modem card is not connected to a SIM card, the input, to the COM port of the associated multiplexer will be high impedance.
I hope this explains things a little better.

Finbarr

Hello,

I will respond to you two latest posts separately as they pertain to two different family of parts.
My team only covers the first mentioned, I will link this to the i2c team for the second post.
As for the large enough value, that will be a combination of our device as well as the i2c device. Our device will protect up to 100mA during latch-up. As for impact to constant current source, you will need to make sure it can source current should this happen.

Thanks,
Louie

Sorry Louie, I am not sure that I follow.

The current source value is 0.54mA on the bi-directional pin. I assume if you use a 10K PD on this pin the CS will switch off ?
So you are effectively turning off the CS and providing a 10K PD to the input of the switch, thus overcoming the potential latch-up issue associated with floating inputs. Would you agree?

Do you want me to raise a question to the I2C guys?

Finbarr