8 Nov
2011
8 Nov
'11
2:34 p.m.
On 11/06/2011 01:00 AM, David Riley wrote:
Yeah, a 3.3v CMOS output will still be putting out higher than many
5v TTL chips output. A surprising number of modern low-voltage micros
still have 5v-compatible inputs. I worked with a seriously
irritating TI DSP that, among other insults, only took 1.8v inputs
(no higher, no lower). We had 12 voltage rails on that one AMC card
(about 2"x5", highly integrated).
We did, however, like the 74AVC4T245 for voltage translation. That
almost fits your dual-supply bill, only the AVC parts don't go up to
the full 5v. For that, your best bet would be in the LVC family;
they have some dual-rail '245 and similar equivalents (e.g. the
SN74LVC8T245). Nothing in DIP, though, so if you really wanted that
you'd have to mount them to surfboards (I really like the ones by
Bellin at http://www.beldynsys.com/; not cheap, but pretty nice
protoboards the do mount up nicely).
For level conversion, I've just used a really neat chip in two
designs, and it seems to work very well. It's the TI TXS0108E. It's a
bidirectional converter that has separate Vcc inputs for each side.
-Dave
--
Dave McGuire
New Kensington, PA
I don't think so. That's why I use 74LVC
parts for the 5V to
3.3V direction, and 74HCT for the 3.3V to 5V direction, if 5V
CMOS levels are actually needed. Usually they aren't.
If it's a 3.3v microcontroller on the other side, it's a good
reason for selecting one with 5V tolerant inputs, I guess. At
least you can use plain old HCT in both directions.