Part search: High speed, low voltage buffer?

[flyingfishfinger]

Hi,
I'm having trouble finding a part that meets my requirements, but I don't know if such a part exists. Maybe someone here knows more?

The requirement is a high-speed 3-state or push-pull buffer that works  from 1.2V to 5.5V and has a transition time of under 50ns (~ 10Mhz frequency). So far, the closest I've found are the Onsemi / Fairchild Tinylogic products which have the speed and output current requirements but only go up to 3.6V, or the NXP 74LV244/5. They have the voltage range but not the speed.

Any other ideas or experience with similar stuff around here?

Thanks,
Rafael

[danadak]

High speed LV comparator ?


http://www.linear.com/parametric/High_Speed_Comparators#!1030_1!1090_<=50!1004_>=5!num_one!pd_50!is_!vsmax_5


Regards, Dana.

[flyingfishfinger]

Hmm thanks, but it doesn't look like those go down to 1.2V though...

R

[T3sl4co1l]

LVC or I think ABT something or other, logic families will do that.  And are available in singles (74xx1G and such, and TinyLogic equivalents).

The problem you will find is that CMOS poops out at low voltages.  Even 74HC still works at 2V (maybe not 1.2V, or, heh, maybe not usefully anyway), and something hot enough to deliver 50ns performance at 1.2V (like 74ABT if I'm remembering it right) will be destroyed at higher voltages (beyond 3.3V usually).  LVC is rated for 5V (though I think it's not a good idea from a lifetime standpoint?) , but maybe not that speed at that low a voltage.

Tim

[flyingfishfinger]

Tangentially, what about an op-amp with a high slew rate as a voltage follower for this application?

R

[David Hess]

I do not think you are going to find any logic which operates over a 1.2 to 5.5 volt supply and input voltage range.  HC, AC, and LVC logic just about cover the logic options.  LVC operates down to 1.5 volts.

Some comparators operate down to 1.2 volts but they are not very fast and an operational amplifier will be slower.

[Cervisia]

Low-voltage families are AUP (optimized for low power, up to 3.3 V) and AUC (optimized for high speed, up to 2.5 V). For 5 V, use the AHC family.

[tggzzz]

In some circumstances the 3-state + low voltage requirements can be achieved by a CMOS analogue switch; part number typically *66*

[flyingfishfinger]

What about discrete transistor solutions in push-pull configurations?

[David Hess]

What about discrete transistor solutions in push-pull configurations?

I was thinking that after I posted.  This is not normally done but a pair of fast saturated switching transistors configured as a complementary common emitter amplifiers with input and output resistors to limit the current while switching might work but will require the input voltage to be within about 0.3 volts of each supply voltage to prevent simultaneous conduction.  Fast saturated switching transistors would be needed to get the speed required.

Of course this circuit inverts which is not normally a problem but if it is, two stages in series would be needed which might have too much delay.  Two stages would also allow hysteresis to be added.

[danadak]

Is the 1.2 to 5.5 v range the power supply variability ?

If so solving, eg. make thru use of a buck/boost regulator, a stable
power supply, then you will have much greater choice in logic families.


Regards, Dana.

[flyingfishfinger]

No, I'm looking for something to increase the drive strength of this part over its full voltage range.

R

[T3sl4co1l]

Perhaps you need a more basic solution.  If one side of that part is always at 3.3-5V, say, then use a "complementary" pair of MOS (e.g., 2N7002 + BSS84) to switch the lower VCC side.  This works all the way to 0V, or a little bit below actually.  (To work below 2 or 3V, you need a slight modification: the PMOS should be driven from a level shifter, to maintain working Vgs(on).  If the signal is switching frequently and the supply voltage changes much more slowly, a coupling capacitor, diode and resistor (DC restorer) does a fine job.)

That's unidirectional of course.  I'd suggest a comparator to level shift in the other direction.

Tim

[David Hess]

Adding a buffer to the PI4ULS5V202 largely defeats the purpose of using it in a bidirectional application.

If you have to add an output stage to improve its output current capability, then just replace it with a better one way level shifter to start with.  If you have a higher supply voltage to work with, then this becomes much easier and tggzzz's suggestion to use an analog switch becomes viable.

[flyingfishfinger]

You have a point.
It doesn't have to be bidirectional, but I didn't have much luck even with unidirectional level shifters that go down to ~1V and up to 5.5V.

I also need 8 channels of this, so discrete solutions aren't exactly ideal from a part count point of view but if that's the best solution I'm open to it.

If you know of a part, I'd appreciate any pointers...

[T3sl4co1l]

This might be an X-Y problem -- could you tell us about the underlying goal?

Tim

[flyingfishfinger]

Not a whole lot, I currently work at a company in stealth mode. This particular project is more or less a high speed USB->UART debugger including handshaking, but the power rails of our product can drop well below 1.8V dynamically under brownout-like conditions hence the 1.2V part.

R

[David Hess]

LVC logic is normally operated between 1.8 and 3.3 volts but some is 5 volt tolerant on inputs and outputs and works down to 1.2 volts.  So what I would consider is using the enable inputs on a 74LVC125 or 74LVC126 as logic inputs and treating the output as an open drain output with a pull-up resistor.  That gives you 4 independent level translating open drain buffers per 16 pin package.

They are a little slow at 1.2 volts but that means 16 nanoseconds from the enable pin which is more than fast enough for what you have in mind.  If their drive capability at 1.2 volts is not sufficient which will need to be tested, then multiple drivers can be used in parallel.

[T3sl4co1l]

Not a whole lot, I currently work at a company in stealth mode. This particular project is more or less a high speed USB->UART debugger including handshaking, but the power rails of our product can drop well below 1.8V dynamically under brownout-like conditions hence the 1.2V part.

R

So, functional operation below 1.8V is not necessary because the product itself doesn't work down there?

Tim

[flyingfishfinger]

Well sadly I can't reveal anything about the product. Let's assume I need to work down to 1.2V; some of the Atmega ATSam products use  a Vcore of 1.2V for example.
The part I cited will do this fine at low drive strengths, but I'm looking for ways to improve this in case of long connection wires to the product (a reasonable possibility given the current state of the system, unfortunately).

Basically, same question as before. Looking for ways of getting a level shifter for UART down to 1.2V with reasonable drive strengths (preferably push pull).
Thanks for the suggestions so far, they've given me a few ideas to investigate.

R

[T3sl4co1l]

Yeah but when does Vcore get exposed on IO pins (IOVCC)?

I'm not aware of many chips that do 1.2V IO.  I've seen 1.8V before.  Not saying they don't exist, but I am saying they aren't going to do all the way from 1.2V to 5.5V in a single range... so why should yours?

If you're just sensing voltage, any logic analyzer circuit will do.  As for driving, you might use an analog switch instead (which also gets you tristate very easily as well).

Tim

[mikeselectricstuff]

Look at MOSFET gate drivers

[Buriedcode]

I just took apart an AVR ISP MKII clone I made 8 years ago.. to see what buffer I used.

http://www.analog.com/media/en/technical-documentation/data-sheets/ADG3304.pdf

Only 4 channel, bidirectional, 1.15- 5.5V.  Not sure why I used this, I think I got a pack of 10 from ebay, randomly added to an order from a surplus store.  Note: doesn't seem to have over voltage tolerance, so the input range is limited to the power supply, unlike the LVC logic or the tinylogic series that can accept up to 5.5V regardless.

[flyingfishfinger]

Ooh, that's a nice part!

It's already better than the one I spec'ed to begin with, although the datasheet is vague as to its output drive capability.

It does specifically say to follow it with a buffer if high drive is required, so for at least that part of the question we're back to square 1.

However, it looks like it can drive a 73pF load at 3.3V and 50MBPS, but I'm having a bit of trouble converting that to drive current.

R