I am looking for the best way to interface a 3.3V to 5V 8 bit data I/O data bus together. Both the 3.3V side and the 5V side can be inputs or outputs on the fly. I currently use a 74VHCT245 with a direction selection pin but the extra overhead in how and who controls this line is not easily attainable. Is there a seamless solution that I have not yet uncovered or is there a creative way to implement this with as little overhead as possible
If you want to use IDT quickswitch as 5V <-> 3.3V translation,
please forget the AN_11 from IDT. It is wrong!!
According the circuits test, the output voltage is not what they said.
By the way, xilinx provides a way to use the IDT quickswitch.
In this application note (**broken link removed**),
it uses quickswitch as 5V <-> 3.3V PCI.
IMO the circuit described in AN_11 is OK. I have an old QualitySemiconductor book, and there is included AN_11A, but with a mistake: there is missing a biasing resistor, without this resistor it would work unpredictably.
There can be another issue. If 3V switch is used then it consist a voltage doubler for steering a gate.
I think the easiest and cheapest way it is use of Philips BSN20 N-channel enhancement mode field-effect transistor. in Logic level translator application. Some time ago I have similar problem and simple transistors ideally solved problem.
regards
I think the easiest and cheapest way it is use of Philips BSN20 N-channel enhancement mode field-effect transistor. in Logic level translator application. Some time ago I have similar problem and simple transistors ideally solved problem.
regards
1. The thread started with data bus (mentioned 245 suggests 8bits), so I do not think that using disrete elements is the cheapest way.
2. The circuit can cause a problem (the one from Xilinx as well). N-MOSFET will cutoff any voltage higher than a gate potential (Vdd) minus VGSth. BSN20 has VGSth(max) 3.5V! It is usless. Lets assume that VDD is 3.135V (3.3V 5%) and VGSthmax is 1.1V. (Then incoming 5V will be cutoff to 2.035V. Is is no longer TTL compatible (smaller than required 2.4V).
3. I expect another issue with Xilinx circuit. It would occur if a 'clamp supply' is bigger than a forvard voltage of an intrinsic diode inside the swith. Problems will start with increased current consumption and can finish with a failure.
Philips also produces a chip that claims to do voltage conversion (I hope it works, I just put it in design I'm working on). The part is GTL2002 and is only 2 bits, but they also make wider versions. The original intent is for I2C bus level conversion, but they have an app note that explains its general use.
For simply circuits yo can use a simple resistor for adapt levels. I use this for adapt a TI MSP430 USART to a MAX232. The high logic level in 3.3 is enough for TTL high level. The clamp diodes of the CI´s that work in 3.3 V protect the inputs. A resitor of 1K is sufficient.