3.3V-5V level shifters

Look at the I2C spec, they do this using mosfets and is bidir.

Don't know that but recently I designed in some 74LVXC3245, yet to power the first board with them, though. But I am pretty experienced with that sort of thing and did not smell anything foul (and I hope saying so now won't come and bite me in 2-3 weeks time when I have the prototype board here...).

Will do - but will take two chips if you need to simultaneously have both directions actively driven. The two sides are separately powered which should make sure no thermal issues like those John mentioned are to be expected.

You must have overlooked something there. OE should be an input only - must be held low for the drivers to be open - the drivers on the 16 data pins, that is. Drive it high and all get tristated.

Didi

------------------------------------------------------ Dimiter Popoff Transgalactic Instruments

------------------------------------------------------

Original message:

It wasn't the LVC spec, that one is not stellar either but ok. Sometimes I wonder if anyone proof-reads these things. But I am not surprised. Right after getting my degree I met an alumni somewhere at an airport or railways station . He had received his masters about the same time but had never done any hobby projects, IOW his knowledge base was all from books and professors. "Hey, what are you doing these days?" ... "Hire on at a big semiconductor manufacturer" ... "What are you doing there?" ... "Writing application notes". I almost froze.

Is that why I2C hangs so often ... ?

If it's any comfort my design is also a white-knuckle ride. Lots of rather unorthodox high voltage stuff on there, all kinds of safety interlocks like HV only to come up when +5V is at least 4.5V but only when there is a full moon, and that sort of thing. Design -> layout -> production. No breadboarding at all.

Not quite. Quote "When the output-enable (OE) input is low, all outputs are placed in the high-impedance state. To ensure the high-impedance state during power up or power down, OE should be tied to GND through a pulldown resistor; the minimum value of the resistor is determined by the current-sourcing capability of the driver."

Low is tri-state. Then further down, quote: "The TXB0104 is designed so that the OE input circuit is supplied by VCCA."

Anyhow, that thing is too hokey for me. I kicked it out and replaced with LVC4245A, two of them. Now I found out that SOT23 LED have all but vanished. Darn ...

Why not 5V>3,3V with resistors only... ?

hth

- Michael Wieser

What's not to like about the 74LCX245?

...Jim Thompson

Come to think of it, I have not used any breadboarding for my designs either last 20 years or so... I had stopped breadboarding for some time back then when SMT came in and made breadboarding impractical anyway :-).

Oh I thought you were referring to the x245, I just don't know that TXB0104 thing, my mistake.

Didi

------------------------------------------------------ Dimiter Popoff Transgalactic Instruments

------------------------------------------------------

Original message:

The customer wants to use series terminators and no loads other than weak pull-ups at the receiving end.

Well, I am using the LVC4245 now. I like the two supplies on those, IOW a "real" level translator. Got to deal with power sequencing though.

driver

T'other month did an 'Xport' network adaptor thing for a customer. I couldn't be arsed ordering or waiting for translator chips, so used a dozen transistors instead. But I wasn't looking for 1nS switching :). I'm happily using more and more of the these transistor things as time goes on, having had a belly full of I.C. stuff being obsoleted and the disruption it causes. Assembly cost of discretes would have been a problem 5 years ago but it's now all pick-n-place and I've yet to hear a peep of complaint from the production company. Ltspice to pick-n-place to customer. Ain't modern technology the bee's knees. Earning a crust is getting much, much easier :)

In sci.electronics.components Joerg wrote: ...

Nowerdays I use the 74LVC8T245 or 74LVC1|2T45 ..

Before reading more that most of the data sheet, the secret of the bidirectional operation is the weak outputs. See near the bottom of page 10 for the drive level for an output that would cause the device to transfer to the other side. In case of a tie both ends can drive safely but no data gets gets sent.

I don't know about you i would be real ticked off about a major late requirements shift. I would be real thoughtful about future dealings that client.

Pre cicely, they are indented to drive one low C CMOS input and loose the fight gracefully to any CMOS output..

To see how it's done properly, see my USB patents (copies on my website).

...Jim Thompson

Thou shalt not have any other outputs besides it on that line ;-)

For us consultants that's a very normal scenario. I recently had an "Oh, we need about 12 times the current" encounter. So I gave them 12 times more current. This stuff is unavoidable when you are working on cutting edge turf where nobody has set foot before.

Well, I just placed LVC4245. Works :-)