Maker Pro
Maker Pro

3V3 in, 5V out buffer driver

R

Roger

Jan 1, 1970
0
Maybe I am blind but I just cannot seem to find good cheap buffers
capable of taking 3V3 input signals and pumping out 5V in a push pull
line buffer output.

Suggestions anyone?
 
L

LVMarc

Jan 1, 1970
0
Roger said:
Maybe I am blind but I just cannot seem to find good cheap buffers
capable of taking 3V3 input signals and pumping out 5V in a push pull
line buffer output.

Suggestions anyone?
roll your own !

Marc
 
R

Roger

Jan 1, 1970
0
LVMarc said:
roll your own !

Marc

I would but my silicon foundry is down ;-)

What I meant is I would expect to find ready made discrete buffers for
doing this task. In my case I want to get a 5V SPI interface from a 3V3
uP. An oc solution is problematic because the line has to run around
several devices at several MHz......I need a good pp output, but that
requires 2 buffers, an oc to shift and a line driver.
 
LVMarc said:
roll your own !

Marc

Take a look at TI's site:
http://focus.ti.com/logic/docs/logicportal.tsp?templateId=5985
Near the bottom of the page is a section called "Voltage Level
Translation". The parts are similar to a 16-bit version of a '245,
however the dual voltage versions have for example a 3.3V Vcc on left
side of the chip, and a 5v VCC on the right side of the chip. The
SN74ALVC164245 chip, for example has 48 pins with a 0.35mm (0,025")
lead pitch. There are many GND and power pins, which reduce ground
bounce and other nasty things when all 16 signals switch at once.
There are also Bus Switches, which are just FETs that connect side-A to
side-B. Simple logic controls whether the switches are on or not.
These should work OK for 5-TTL to 3.3V CMOS, but AFAIK won't work
correctly with 5V CMOS.

TI has a variety of voltage translation chips as do other
manufacturers, who I can't remember right now ;( .
HTH
-Dave Pollum
 
Y

Yuriy K.

Jan 1, 1970
0
Roger said:
Maybe I am blind but I just cannot seem to find good cheap buffers
capable of taking 3V3 input signals and pumping out 5V in a push pull
line buffer output.

Suggestions anyone?

74HCTxxx, if they are fast enough for your application.
 
N

Noway2

Jan 1, 1970
0
Roger said:
Maybe I am blind but I just cannot seem to find good cheap buffers
capable of taking 3V3 input signals and pumping out 5V in a push pull
line buffer output.

Suggestions anyone?

I might suggest the SN74LVC4245A, bi directional 3.3V<-->5V
transcievers. They cost all of about $0.50 from digikey.
 
J

Jim Thompson

Jan 1, 1970
0
74HCTxxx, if they are fast enough for your application.

Why do you persist in a bad solution when the STATIC overlap current
has been shown to be quite large in previous threads?

The "LVC" parts are the proper way to go.

...Jim Thompson
 
A

Austin Chen

Jan 1, 1970
0
For cheap solution, from 3.3V logic to 5V logic can use the 74HCTxxx series.
These IC use 5V power with near 5V Voh and allow 2.0V as Vih.
For 5V logic to 3.3V logic can use the 74LVC series. These IC have 5V
tolerance in inputs.
For bi-direction ports, only 74LVC4245 series dual power devices can be
used.
 
Y

Yuriy K.

Jan 1, 1970
0
Why do you persist in a bad solution when the STATIC overlap current
has been shown to be quite large in previous threads?
RTFM.

The "LVC" parts are the proper way to go.

You really need to educate yourself; "Art of Electronics" will be a good
primer. After that read the #$%@ datasheet.
LVC family:
Vcc <= 3.6V for most parts,
Vih >= 3.5V for single supply LVC parts if Vcc=5.0V is allowed.
dIcc up to 0.5mA for dual power LVC parts @ Vcc=3.3, Vin=2.7V
 
J

Jim Thompson

Jan 1, 1970
0
You really need to educate yourself; "Art of Electronics" will be a good
primer. After that read the #$%@ datasheet.
LVC family:
Vcc <= 3.6V for most parts,
Vih >= 3.5V for single supply LVC parts if Vcc=5.0V is allowed.
dIcc up to 0.5mA for dual power LVC parts @ Vcc=3.3, Vin=2.7V

Your ignorance is profound. Go back and read all the way through the
original thread on this topic. There you will find that another
poster likewise criticized my opinion, THEN went and MEASURED a
74HCTxxx part at 3.3V input with VDD=5V, and measured the ENORMOUS
overlap I warned about.

You are herewith relegated to s.e.b until you grow up.

PLONK!

BTW, Don't you even dare to tell me what I should read. I could write
the whole fucking book without once looking up a single reference.

And do loop and nodal analysis ;-)

...Jim Thompson
 
D

Didi

Jan 1, 1970
0
74HCTxxx part at 3.3V input with VDD=5V, and measured the ENORMOUS
overlap I warned about.

How much does it get? I would expect low enough so the parts will
be in spec over temperature, but can you give some figure? I am sure I
have
used this configuration and I never had any trouble out of it, but
I have not been looking for the last milliwats. Never noticed a part
getting hot because of that, though.

Dimiter
 
J

Jim Thompson

Jan 1, 1970
0
How much does it get? I would expect low enough so the parts will
be in spec over temperature, but can you give some figure? I am sure I
have
used this configuration and I never had any trouble out of it, but
I have not been looking for the last milliwats. Never noticed a part
getting hot because of that, though.

Dimiter
[snip]

Agent dumps old threads after 10 days so I can't go back and look, but
I believe it was on the order of 4-5mA, which I consider unacceptable
for static consumption.

Maybe the OP will return with his data ??

My main point is: WHY do it when good-engineering-practice solutions
exist?

Of course, if you're a hacker, who cares ;-)

...Jim Thompson
 
W

Winfield Hill

Jan 1, 1970
0
Didi wrote...
Jim wrote...

How much does it get? I would expect low enough so the parts will
be in spec over temperature, but can you give some figure? I am
sure I have used this configuration and I never had any trouble
out of it, but I have not been looking for the last milliwats.
Never noticed a part getting hot because of that, though.

I wish Jim would give us a link or thread title, I can't find the
post he's referring to. TI's 74hct14 datasheet has a plot of the
supply current, and it's only 0.1mA with a 3.3-volt logic input,
for the extreme case of a 5.5-volt supply. If you assume a wimpy
2.8-volts (degraded 3.3V input) it's 0.15mA, not what most people
call an ENORMOUS current. (That is, unless you designed for the
long-life battery-operated field, as I did for 17 years, where
our entire budget to operate 200 ICs was often about 150uA.)

TI specs the maximum current for a HCT14 single 2.4-volt input as
2.9mA and the maximum for a VHCT14 is 1.5mA with 3.4V input. The
real-world currents for both families will be much lower. Neither
is a deal-breaker for most circuits, since even 15mW dissipation
would usually not be a problem. It can't lead to overheating!

BTW, Jim's suggested AVC logic series can't work to convert 3.3
volt logic to nice 5 volt swings, as Yuriy complained, because
the AVC stuff can't be operated on a 5-volt supply. The 3.3V
output of a AVC logic gate would simply pass the problem on to
the next HCTxxxx 5V logic gate.
Dimiter Popoff, Transgalactic Instruments http://www.tgi-sci.com

Interesting company name!
 
D

Didi

Jan 1, 1970
0
I believe it was on the order of 4-5mA, which I consider unacceptable
for static consumption.

4-5 mA per input is indeed a lot, I guess I'll switch over next time I
have
a chance. I remebered now an aplication in which I used HCT367s - 3 of
them - driven by the 3.3V outputs of a DSP, statically, to switch a 3x3
latching relay matrix; at least 6 inputs on all 3 chips are constantly
at 3.3V (just driven low for a few milliseconds when a relay is to be
turned over). This could amount to, let me see, 5*6*5=150 mW! The
entire
device consumes appr. 2W and this would have been the last place
I would look at to save a milliwatt or two, I thought I had a 0
there... :).

Dimiter

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

http://www.tgi-sci.com
------------------------------------------------------

Jim said:
How much does it get? I would expect low enough so the parts will
be in spec over temperature, but can you give some figure? I am sure I
have
used this configuration and I never had any trouble out of it, but
I have not been looking for the last milliwats. Never noticed a part
getting hot because of that, though.

Dimiter
[snip]

Agent dumps old threads after 10 days so I can't go back and look, but
I believe it was on the order of 4-5mA, which I consider unacceptable
for static consumption.

Maybe the OP will return with his data ??

My main point is: WHY do it when good-engineering-practice solutions
exist?

Of course, if you're a hacker, who cares ;-)

...Jim Thompson
 
D

Didi

Jan 1, 1970
0
post he's referring to. TI's 74hct14 datasheet has a plot of the
supply current, and it's only 0.1mA with a 3.3-volt logic input,
for the extreme case of a 5.5-volt supply.

Well the HCT14 is perhaps not typical when it comes to its inputs,
being Schmitt-triggered; I expect special precautions have been
taken with it (and the 132 etc.).
I'll take some measurements next time I have a chance to do so, I
really would not have expected to spend milliwatts of, well, hard to
neglect significance, on _that_.
Interesting company name!

Thanks! To quote from my "who we are" file:

" The name originates from the dark communistic ages (around 1980)
when dreaming of a private high-tech company was regarded more like
a health issue so there had to be some sense of humour in it besides
the hard work (done in an attick back then). "

:) :)

Dimiter
 
W

Winfield Hill

Jan 1, 1970
0
Jim Thompson wrote...
Agent dumps old threads after 10 days so I can't go back and look, but
I believe it was on the order of 4-5mA, which I consider unacceptable
for static consumption. Maybe the OP will return with his data ??

I doubt it was that high. If so, we need to know what manufacturer,
because it's far above the clearly-stated maximum specs.
My main point is: WHY do it when good-engineering-practice solutions
exist? Of course, if you're a hacker, who cares ;-)

Why cast aspersions against standard industry practice, especially
when your suggested LVC alternate simply can't work? What are you
suggesting instead?

I note that ON Semi's mc74vhc1gt32 (were you involved with that
product?) specifically states "to be used as a logic–level
translator from 3.0 V CMOS logic to 5.0 V CMOS Logic", yet it has
exactly the same "problem" and draws 1.5mA max for 3.4-volt inputs
with 5V supply, just like the standard hct and vhct logic families.
 
J

Jim Thompson

Jan 1, 1970
0
Jim Thompson wrote...

I doubt it was that high. If so, we need to know what manufacturer,
because it's far above the clearly-stated maximum specs.


Why cast aspersions against standard industry practice,

Really ?:)
especially
when your suggested LVC alternate simply can't work? What are you
suggesting instead?

I note that ON Semi's mc74vhc1gt32 (were you involved with that
product?) specifically states "to be used as a logic–level
translator from 3.0 V CMOS logic to 5.0 V CMOS Logic", yet it has
exactly the same "problem" and draws 1.5mA max for 3.4-volt inputs
with 5V supply, just like the standard hct and vhct logic families.

I showed a schematic awhile back that had no such current. The *true*
logic translators usually have two VDD's... one at 3.3V and one at 5V.

I don't recognize "VHC"... the stuff I did was "LCX"; and there's a
project file that I know of only as "TinyCkts"... I don't know what
naming convention those ended up under.

...Jim Thompson
 
J

Jim Thompson

Jan 1, 1970
0
Didi wrote...

I wish Jim would give us a link or thread title, I can't find the
post he's referring to. TI's 74hct14 datasheet has a plot of the
supply current, and it's only 0.1mA with a 3.3-volt logic input,
for the extreme case of a 5.5-volt supply. If you assume a wimpy
2.8-volts (degraded 3.3V input) it's 0.15mA, not what most people
call an ENORMOUS current. (That is, unless you designed for the
long-life battery-operated field, as I did for 17 years, where
our entire budget to operate 200 ICs was often about 150uA.)

TI specs the maximum current for a HCT14 single 2.4-volt input as
2.9mA and the maximum for a VHCT14 is 1.5mA with 3.4V input. The
real-world currents for both families will be much lower. Neither
is a deal-breaker for most circuits, since even 15mW dissipation
would usually not be a problem. It can't lead to overheating!

BTW, Jim's suggested AVC logic series

I didn't suggest any such thing. Someone mentioned a part number,
which I can't recall now, that did proper translation... at the time I
checked the data sheet, but can't recall the prefix now.

Win, Are you saying that 74HCTxxx is a good-engineering-practice
solution for 3.3V-to-5V translation?

Maybe at Harvard ?:) But at MIT they periodically show the Galloping
Gerty movie to remind everyone that half-assed design always ends up a
failure ;-)
can't work to convert 3.3
volt logic to nice 5 volt swings, as Yuriy complained, because
the AVC stuff can't be operated on a 5-volt supply. The 3.3V
output of a AVC logic gate would simply pass the problem on to
the next HCTxxxx 5V logic gate.


Interesting company name!



...Jim Thompson
 
W

Winfield Hill

Jan 1, 1970
0
Jim Thompson wrote...
I didn't suggest any such thing. Someone mentioned a part number,
which I can't recall now, that did proper translation... at the time
I checked the data sheet, but can't recall the prefix now.

I'm sorry, I meant to say LVC (correct in my 2nd post). This per
your 18 Dec message, "The "LVC" parts are the proper way to go."

Most of the LVC parts cannot be powered from 5 volts. A few, like
TI's 74lvc1g32 can, but that one shouldn't be used with 3.3V logic.
If there's an ideal LVC part, by your terms, I haven't found it.
Win, Are you saying that 74HCTxxx is a good-engineering-practice
solution for 3.3V-to-5V translation?

One perfectly good practice, yes. If it's specified on the data
sheet, as it is for most if not all HCT families, such as the TI
parts I mentioned. One should check the part in question. As
I said, many of ON Semi's parts have the same specs, including
some of the proprietary ones, like the mc74vhc1gt32, specifically
labeled a logic-level shifter. They all use the same approach
you are ridiculing. What's your problem with < 15mW?

BTW, this has nothing to do with Harvard or MIT (you should see
some of the electronics stuff I see over at the Media Lab).
 
Top