74LVC244 Line Driver/Buffer?

[rwgast_lowlevellogicdesin]

I was hoping someone could help me understand a little more about this chip

http://www.datasheetcatalog.org/datasheets/134/42627_DS.pdf

I currently use theses 74lvc244 chips to help do 5 to 3.3v translation. From what I understand a buffer will usually allow you to source/sink more current than a micro can handle on its own. I have been looking over this data sheet and am having a hard time figuring out just exactly how much the pins on this chip can actually source/sync. My best guess would be 24mA but that seems a bit low, most micros can handle that on there own. The other thing I have been wondering if these chips could help me with is, extending signal lines. Lets say I have I micro controller and I want to connect a SPI device to it through 3 feet of wire, would  one of these chips help me do that, is that why they are called line drivers? Also there is the 74xx245 which is a line transceiver what exactly does that chip do over a line driver? I have seen some boards layed out with a a 74xx244 near an CPU/mCU and a 74xx245 near an expansion socket, what exactly is the purpose of that.

[AlphZeta]

Yes, from the recommended operating conditions section the maximum current going through each channel should be capped at +-24mA. Combining this with what you see if the absolute maximum rating for the continuous current through Vcc +-100mA you will get an idea of what you can use this chip for (if you are driving all 8 channels at once, make sure you do not exceed the 100mA maximum rating)

Code: [Select]

My best guess would be 24mA but that seems a bit low, most micros can handle that on there own.
It depends. Some controllers (like the one used in RaspberryPi) have pretty weak driving capabilities and can at most deliver only a few mA per pin.

line drivers are typically used as buffers so that the slew-rates of the output (connected to the buffer) is well defined regardless of the number of connected peripherals.

[rwgast_lowlevellogicdesin]

"line drivers are typically used as buffers so that the slew-rates of the output (connected to the buffer) is well defined regardless of the number of connected peripherals. "

Can you explain that a little more. From what I understand slew rate is how fast a pin can change states? So lets say I have my propeller flipping a pin at it's maximum speed of 12.5ns, this would be the slew rate? If so what do you mean by the buffer makes sure the slew rate is well defined, regardless of the number of peripherals? I am guessing that means that the signal does not lose any integrity as far as rising/falling edge times go? Is this for long distance traces/wires? I don't understand why the middle man i.e the buffer would be needed if not.

[AlphZeta]

By the way for SPI level translating, I haven't used this particular chip before, but in general you would want buffers with three-state output if you have multiple devices on the bus. Other than that you can use this duplexed bi-directional transceivers for SPI communication if the speed requirement is not that critical.

[kxenos]

If you want more current you I think you should look at ULN2003/4 if that suits your needs. They also have diodes built-in, useful when driving inductive loads like relays or (small) motors.

[AlphZeta]

"line drivers are typically used as buffers so that the slew-rates of the output (connected to the buffer) is well defined regardless of the number of connected peripherals. "

Can you explain that a little more. From what I understand slew rate is how fast a pin can change states? So lets say I have my propeller flipping a pin at it's maximum speed of 12.5ns, this would be the slew rate? If so what do you mean by the buffer makes sure the slew rate is well defined, regardless of the number of peripherals? I am guessing that means that the signal does not lose any integrity as far as rising/falling edge times go? Is this for long distance traces/wires? I don't understand why the middle man i.e the buffer would be needed if not.

Sorry, didn't see your question when I hit post last time. Anyway,  in simple terms slew rate determines how fast the signal rises. So if the signal rises too slow, you will get to your desired logical level slower which means you won't be able to clock your signal as fast as you would before. That's where a buffer comes in.

In terms of extending the range, it's more complicated than that and depends on the protocol you use. And if your goal is to minimize the trace capacitance effect on your signal, then yes you can use a buffer for that. But if you are going to go beyond the range for what is specified by the protocol then you will also need to worry about timing, etc. And typically there are dedicated chips for that use.

If you want more current you I think you should look at ULN2003/4 if that suits your needs. They also have diodes built-in, useful when driving inductive loads like relays or (small) motors.

Yes, but only for output these are not bi-directional.

[kxenos]

Yes, but only for output these are not bi-directional.

Neither is 244...

[AlphZeta]

Yes, but only for output these are not bi-directional.

Neither is 244...

I was thinking of 245

[westfw]

My best guess would be 24mA but that seems a bit low, most micros can handle that on there own.

Actually, while most older (PIC, AVR, etc) will do about 20mA on their own, ON THEIR OUTPUT PORTS, this is more than you are likely to get on thing like the memory/address buses of microprocessors (which is where this sort of chip is usually used.)
A lot of the newer microcontrollers (ARM) don't support 20mA even on their IO ports.

[SeanB]

Bus buffers are used to make fast driving of long buses ( more than 10cm on older PC's with a 8080/286 on them) and with a few boards attached. They could drive a full 24mA into the bus, while the processor in many cases could barely do 5mA driving the bus, enough for a single rom in a minimal system and one other RIOT chip. Anything more and you needed bus buffers to get the higher current drive.

Later they were used to handle mixed voltage logic, like 5V processors and 3V3 memory and peripherals.

[rwgast_lowlevellogicdesin]

Im sorry if these questions seem kind of stupid, I am just really starting to grasp signal integrity, impedance matching and all that kind of stuff.

So I know I could just go out and get a 74lvc245 if I wanted a bidirectional buffer, but I have read that there is a trick where you can put an enable low and enable hi buffer back to back to make the buffers work in both directions. I am having a bit of a hard time understanding how that would work, and can not find a schematic. If I understand right would I be able to wire two 244 chips back to back adding an inverter on the OE pin of one to accomplish a bidirectional buffer? This is more of about trying to understand the concept than applying it on a real PCB, it would obviously be a better idea to buy a 30 cent two way buffer chip.

The other thing I was wondering is if someone could enlighten me as to the differences between a chip like the 74xx244 and a unity gain op amp. When is the correct time to use which method. For instance let's say I want to run the output of a DAC in to the reference input of an ADC. Well the VREF of an ADC wants a low impedance load and the best way I know of to do that would be to use an opamp in unity gain mode as a buffer. But what if I had 8 DAC outputs needing to drive 8 ADC reference voltages would a 7vxx244 work just as well as 8 separate op amps?

Lastly lets say I wanted to take a micro and put a bi directional buffer on every pin that would allow the each pin to source/sync 50-100ma each. What would be the best way of doing this, are there fast logic chips capable of this, or would I need to use some time of solution built from discrete components? The buffer would have to be able to switch at about 50mhz.

[SeanB]

Using 2 244 units as a bidirectional buffer is basically the same as a 245, just a little bigger and needing an extra inverter as well. It will not work as a buffer for an ADC though, as it is a digital device and is either high or low on the output. No voltage inbetween, you will need to use a pair of quad opamps to do the buffering as you want.

[rwgast_lowlevellogicdesin]

Well duh, I feel pretty stupid now, not thinking about the 74xx244 only being able to go Hi or Low. But if your Vref was the same voltage as logic hi, would there be an issue?

Do you know where there may be a schematic/explanation of a bi directional buffer using two 244's? I cant really find anything on google searching for Bi Directional buffer using two 74xx244 chips.

[westfw]

Here's a 4-bit bidirectional buffer implemented with one xx244:

[westfw]

PS: note that the bidirectional buffers that we are talking about (245, multiple 244s) are ones that need an external signal to tell them which direction to switch, not something that you can stick on a microcontroller pin that will automagically sense whether the pin is configured as an input or an output.  Doing THAT is much harder.

Also, 50-100mA at 50MHz is also "difficult", and unlikely to happen except in special circumstances.

(yeah, it would be nice if there was a chip you could put between your microcontroller and "the real world" that would boost the micro's output current capability without disturbing any of the other functionality.  As far as I know, such a thing does not exist.)

[Paul Moir]

Paralleling them might be a novel solution, as you would with a buffer.  Could possibly be as simple as stacking them.  But 50MHz is definitely into tricky territory.

The thing about '244s is that their pinout is a bit of a pain for most things.  Sometimes you'd see a '245 used with the DIR pin locked just to simplify routing. 

[rwgast_lowlevellogicdesin]

Thank you for the schematic, that help clear things up a bit.

As far as the pin out of the 244, there are many times I could really make use of the the LVC version for voltage translation but take a different route, just for that reason! Maybe I will just start buying 245 chips and locking there direction. Would paralleling these chips work to drive current? I usually stay away from that unless something specifically states it uses FETs, is it ok to stack 74xx chips?

[westfw]

The thing about '244s is that their pinout is a bit of a pain for most things.

A lot of the old 742xx and 743xx logic with inconvenient pinouts have newer versions with much nicer "flow-through" pinouts, usually with a 74xx5xx part number.  I don't see a 74lvc544, but there is a 74lvc541 octal buffer with all the inputs on one side, all the outputs on the other (one octal driver with two enables, rather than two 4-bit drivers with separate enables.)
The 74373 (8 latch) and 74374 (8 edge-trig latch) have nice 74xx573 and 574 equivalents, as another example.
(The newer chips are somewhat less likely to show up cheap on the hobbyist market.)