Go Down

Topic: TTL-compatible 3-state buffer with highish drive current (Read 1 time) previous topic - next topic

ElCaron

Ok, cleanup of my prior question: I am looking for a buffer to supply about 25mA continuous current at 5V or be high impedance, which I can switch with 3.3V.

I have no problems finding buffers that can have 5V Vcc and accept TTL level (HCT series). Those don't have the drive current, or barely have it.
I also don't have problems finding a buffer with e.G. 32mA drive current @5V, but those are strictly CMOS,  meaning they require 0.7*Vcc for high state.

Any ideas? Here is the background:

For a camera triggering project, I need to switch 5V with 3.3V, high-side, to realize a strong, switchable pullup for an open drain network. The current prototype works with a pair of N-channel and P-channel MOSFETs, where the PFET gate is pulled to 5V, if the NFET does not short  it to GND. This works perfectly, but for the final design, it would be handy if this would exist in a single package.

MarkT

A low-side MOSFET driver chip will do several amps from 3.3V logic input...  You'll need one that can be powered at 5V, for instance the MIC4422 which works from 4.5V up.  Its available SMT or DIP
[ I DO NOT respond to personal messages, I WILL delete them unread, use the forum please ]

ElCaron

#2
Sep 01, 2017, 01:47 pm Last Edit: Sep 01, 2017, 03:26 pm by ElCaron
That's great, but I just came back to make an important bit of information more clear :): I need tristate. When The pullup is switched off, it is not supposed to sink several amps, but to be off, that is, gone or high impedance.

I almost thought the NTB0101 would be great, but the auto-directionsensing is undefined in case another (strong) pullup is in the network, so that in the worst case, the network is sensed as an input and the MCU direction as an output, leading essentially to a short-circuit.

Since it is not completely unrealistic (I found almost suitable buffers that only failed V_HI), it would also be great if I could use the same chip at another place, where I have to translate 5V to 3.3V (bi-state, sensing on the trigger line) to keep the BOM down.

ElCaron


MarkT

That doesn't have tristate.

74HCT244 and tie 4 channels together for greater current handling?
[ I DO NOT respond to personal messages, I WILL delete them unread, use the forum please ]

ElCaron

It does have tristate. Look at the top of the datasheet. I admit that it is not absolutely obvious, but it is clearly documented. My design is already set up, for both shifting applications.

Tieing outputs together is a another possibility but I am kind of glad about the small footprint. There is not too much room on the board.

MarkT

Its tristate if you change the direction.  There's no output enable pin.
[ I DO NOT respond to personal messages, I WILL delete them unread, use the forum please ]

ElCaron

No, it is also 3-state if you don't change the direction. Come on, you are better than that.
I discarded some ICs on the same basis (no OE pin), but those where not advertised as 3-state. Sometimes it is worth to actually  read the text.

All pins become high impedance if one of the rails is pulled to zero. That is clearly specified in the second paragraph. Since the input rail does not have to source relevant amounts of current when the direction is fixed, it can be driven from the MCU and switched. So

Code: [Select]

Vcca -> MCU
Vccb -> 5V
A    -> 5V (Pins are 5.5V tolerant regardless of the rail voltage, and 5V is already available)
B    -> Resistor to trigger network
DIR  -> 5V

achieves what I want.

MarkT

Yes I know that, but supply rails need decoupling capacitors on them so that you not be possible at logic speeds, and
its not clear if it will phantom power in that case, I suspect it will.  Its an uncharted area, you'll need to be lucky.  Its is rated for ESD so their must be some sort of over-voltage protection on the pins, but
the datasheet doesn't tell you what sort.
[ I DO NOT respond to personal messages, I WILL delete them unread, use the forum please ]

ElCaron

I don't need logic speeds, I want to switch a pullup on an off (off in case there is another one in the network, we go close to the cameras recommended max ratings for steep rises).
I wonder if you really need decoupling on the input rail ... but anyway, the datasheet clear states "3-state
outputs", so I read this as intended usage, not some hack.

edgemoron

Quote
I have no problems finding buffers that can have 5V Vcc and accept TTL level (HCT series). Those don't have the drive current, or barely have it.
Could you use 2 in parallel with, say, 47 Ohm resistors on the outputs to swamp out any voltage differences?
You need either 5V or Hi-Z, only 2 states, where does the 3rd (low) state come in?
Maybe you need to draw me a picture.   :smiley-confuse:

ElCaron

Could you use 2 in parallel with, say, 47 Ohm resistors on the outputs to swamp out any voltage differences?
You need either 5V or Hi-Z, only 2 states, where does the 3rd (low) state come in?
Maybe you need to draw me a picture.   :smiley-confuse:
Space is not plenty on the PCB. I would prefer not to have multiple buffers there. But actually, I remember finding a 2x4bit (2 OE pins) IC. That could actually be space efficient, since I also do the other way around to listen on the open drain network.
You are right, I only need 2-states. But I never encountered HIGH/high-impedance 2-state in one chip. It can be done with an N- and a P-channel MOSFET, which is what I am doing in the current prototype, but I am trying to get the number of components down with a single buffer IC.
I do not need to look on the single component prices, but simplicity and fast singe unit manual assembly would be nice.

LMI1

There are small so called one gate logic chips, if space is at premium. Every manufacturer has their own name for them but googling one gate logic will help.

A single transistor or fet or two may be enough to do what you want

Go Up