Quickswitch / Level Shifter for modern FPGAs

Hi Niels

The CB3T family of devices will translate a high voltage to the supply voltage. The requirement is that the input voltage be 1V above the supply voltage.  Here is an example. SN74CB3T3245 They are switches so they dont need a direction pin.

When going back the other direction. Low to High you will need a pullup to get the signal back up to 5V.

I also attached an appnote with a brief explanation.

They aslo use voltage clamp type devices in this situation. Here is an example of that SN74TVC3306.

You can find appnotes on this type of part in the product folder on the web..  This forum olny allows me to attach one item.

Hi Chris,

well, we tried CB3T, but failed. The Problem is, that the 5V-side is a bus that is sometimes driven with 3.3V Signals. Pullups on the 5V side are not allowed.

An example would be if I want to add a 3.3V device to a PCI-Bus that contains 5V devices.

If the 3.3V device drives a signal (with 3.3V high level), the CB3T limits the voltage on the "bus" side to 2.3V or less, depending on the current drawn by the other devices on the bus. This is not practicable in this case.

What I am looking for is a signal switch that lets pass all voltages between GND and its own supply voltage. All voltages above its power supply and below ground shall be blocked.

Maximum input voltage shall be >5Volt.

This would be a really great device for many kinds of applications with modern FPGAs.

Regards,

Niels

You should be able to use a voltage clamp type part. This would pass voltages up to Vref. The issue is you would need the part powered by a voltage 1V greater than Vref. So if you could power the part with 5V and set Vref to 3.3V then the signals going through the switch would be clamped at 3.3V in either direction.

The SN74TVC3010 is an example.  we also have some new parts about to be released.

I attahced an slide show that has some explination of how they work.

 here is my Email if you need more info. c-cockrill@ti.com

Hi Chris,

I had a look into the TVC documents, and I worry about two points:

(1) the propagation delay is 0-4ns. this is a large spreading and 4ns is too much for e.g. PCI.

(2) what about undershoot clamping? Modern FPGAs accept negative input voltages up to -0.5V and voltage below this may hurt the FPGA.

Thanks in advance,

Niels

Hello Chris and the rest of the comunity,

is there any chance for a solution that allows to connect modern FPGAs to a mixed 3.3V / 5V environment like e.g. a PCI-Bus?

Regards,

Niels